278 58 4MB
English Pages 1033 [1031] Year 2017
‘And’ Conjunction Reduction Redux
Barry Schein
The MIT Press Cambridge, Massachusetts London, England
© 2017 Massachusetts Institute of Technology All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher. This book was set in 10/13 pt Times Roman by Toppan Best-set Premedia Limited. Printed and bound in the United States of America. Library of Congress Cataloging-in-Publication Data Names: Schein, Barry, author.
Title: ‘And’ : conjunction reduction redux / Barry Schein. Description: Cambridge, MA : The MIT Press, [2017] | Includes bibliographical references and index. Identifiers: LCCN 2016033674 | ISBN 9780262035637 (hardcover : alk. paper) Subjects: LCSH: And (The English word) | English language--Conjunctions. | Grammar, Comparative and general--Conjunctions. | Grammar, Comparative and general--Coordinate constructions. | Grammar, Comparative and general--Sentences. | Generative grammar. Classification: LCC PE1345 .S34 2017 | DDC
https://Iccn.loc.gov/2016033674
10
98
7
6
5
4
3
2
1
425/.7--dc23 LC record available at
Contents
Acknowledgments
xiii
1
1
Introduction
1.0 1.1 1.2 1.3 1.4 1.5
1.6
1.7 1.8 1.9
Univocal and 1 The slippery slope to Conjunction Reduction 3 Conjunction Reduction restrained 5 A new clausal architecture 11 Supermonadicity 14 Descriptive event pronouns 22 1.5.0 Descriptive event pronouns and the Davidsonian 24 33 1.5.0.0 Selective perspectives for nonmaximal reference 1.5.1 Reference to events and the syntax of descriptive anaphora 37 1.5.2 Event pronouns and modal insubordination 40 Adverbialization 47 1.6.0 Adverbialization in logical form 52 1.6.1 Naive reference and substitution under identity 54 1.6.2 Adverbs denoting events of counting 56 1.6.3 Scenes to count by 57 1.6.4 Singular and plural frames of reference 60 Spatiotemporal orientation 61 Identity statements simpliciter and conditioned 64 Contextualism logicized 68
DP and DP 75 2.0 Coordinating generalized quantifiers, and the syntax and semantics of collectivized Right-Node Raising 75 2.0.0 Translating away DP and DP 77 2.0.1 Conspectus 89 93 2.1 Syntax and semantics for DP and DP 2.1.0 Collectivized Right-Node Raising 93 2.1.1 Description without commitment 94 2.1.2 “Telescoping” event pronouns 97 2.1.3 Minding the gap 100
viii
Contents
2.2
2.3
24 2.5
2.6
2.1.4 Translation myths and rules 105 2.1.5 Phrasing 106 Number agreement 107 2.2.0 (In)definite descriptions vs. distributive quantifiers 111 2.2.1 Number agreement under a frame of reference 114 2.2.2 Thematic relations and number agreement 119 Number agreement in Lebanese Arabic 120 2.3.0 The distribution and interpretation of event quantifiers and descriptions 126 2.3.0.0 Summary 134 2.3.1 Number agreement and subatomic “telescoping” 135 2.3.2 Vagaries of number agreement amid the layers of event quantification 149 2.3.2.0 Interaction with the linear order of subjects and adverbs 150 2.3.2.1 Molecular event quantifiers licensing distributivity 152 2.3.2.2 Subatomic event quantifiers: Plural reference to events without phrasal distributivity 157 2.3.2.3 Pre- and postverbal subjects, the distribution of tacit each in Lebanese and Slavic, and further evidence of Conjunction Reduction 159 Coordination and subordination in Davidsonian logical form 164 First-conjunct agreement (with conjoined (in)definite descriptions) 170 2.5.0 The logical form of first-conjunct agreement 172 2.5.1 The meaning of first-conjunct agreement 175 2.5.2 Summary remarks on the logical form and meaning of first-conjunct agreement 190 Comitative phrases and number agreement in other languages 192 2.6.0 Counting participants and their accomplices at the scene 198 2.6.0.0 The chains that bind accomplices 199 2.6.0.1 A locality condition on the comitative phrases that count for number agreement
2.7
202
2.6.1 Number agreement and comitativity Summary 207
205
PredP and PredP: Of Subjects and Ancient Grievance 209 3.0 Of subjects and ancient grievance 218 3.1 Scope and reconstruction into subject position 219 3.1.0 Predicative coordination and the distribution of event quantifiers 3.2 Case’s place 223 3.2.0 Tailoring coordination to size 227 3.2.1 Small-clause sizes 235 3.2.1.0 The Case position of small-clause subjects 237 3.2.1.1 Right-Node Raising 241 3.2.1.2 A free event variable in small clauses? 244 3.2.1.3 Quantifier Lowering and scope inversion 249 3.2.14 Number agreement on the lam 255 3.2.1.5 Focus and reconstruction into subject position 260 3.2.2 Opacity in coordinate structures 262
221
Contents
3.3 3.4
3.5 3.6 3.7
Null coordinative pronouns 271 Reference under the eaves 278 3.4.0 Disjunctive interpretation excluded 279 3.4.1 Reconstruction interrupted 280 3.4.1.0 Reconstruction or “telescoping”? 281 3.4.1.1 One or two null coordinative pronouns 283 3.4.1.2 Scope inversion and the position of null definite descriptions 284 Quantifier Lowering into collectivized Right-Node Raised constituents Appendix: Economy and reconstruction 291 Summary 302
PredP and PredP: Coordination vs. Subordination
4.0 4.1 4.2
286
307
PredP and PredP: Coordinating supermonadic PredPs.
Phrasing—complementation — quantification
309
(Tense+) Aux sharing 310 Bound morphemes and affixation under coordination 311 4.2.0 Why not the semidistributive of the disjunctive? 314 4.2.1 Appendix: Bound morphology vs. phrasal complementation
5
PredP and PredP: (Tense+) Aux Sharing 321 5.0 Progressive be 323 5.1 Perfect have 327 5.1.0 Consequent states of the perfect 327 5.1.1 have [[be+en...] and [V+en...]] 329 5.1.2 have [[V+en...] and [...]] 331 5.1.3 The present perfect puzzle 332
6
PredP and PredP: Complementation as a Condition on Subatomic Event Anaphora 335 6.0 The distribution of the disjunctive interpretation 340 6.1 The logical syntax of the disjunctive interpretation 341 6.2 Deriving the distribution of the disjunctive interpretation 342 6.3 Coordinating simple tensed verbs 349 6.3.0 Mixing tenses in coordinated PredPs 354
7
PredP and PredP: Conclusion
8
Introducing Adverbialization and Cinerama 369 8.0 Previously ... 369 8.1 The several faces of NP, and NP,: Puzzles of extensional substitutivity 8.2 Spatiotemporal orientation and adverbialization 377 8.3 Coming attractions 384
318
359
372
X
Contents
9
Cinerama Semantics 389 9.0 Reference mise-en-scene 404 9.1 Referring to frames of reference 409 9.1.0 By definite description in the object language 409 9.1.1 Frames of reference frame events 413 9.2 Scenes for spatial orientation and navigation 417 9.3 The narration of visual experience, and narrative as artifact 426 94 Cinema verité 436 9.4.0 Scenes, projections, and frames of reference for spatial orientation and navigation 437 9.4.1 Path integration and anticonvergence 447 9.4.2 Resolution and reticulation for visual counting 456 9.42.0 Summation of visual counts 462 9.4.2.1 Visual counting and spatial orientation 464 9.4.22 Anomalous counts 466
10
Adverbialization in Logical Form 469 10.0 The dependence of Tense and temporal reference on the descriptive content of nominal quantifiers 471 10.0.0 Symmetric relations under Figure and Ground 473 10.0.1 The selective lifetime effect 476 10.0.2 Adverbial restriction and shift in the lifetime effect 481 10.0.3 Semantic innocence regained 483 10.0.4 Adverbialized logical form 484 10.0.5 The lifetime effect reconfigured 486 10.0.6 Multiple lifetime effects and multiple adverbs, asynchronous but coordinated 490 10.0.7 That how things were with the subject is no longer 497 10.0.8 Lifetime effects and intentional events 500 10.0.9 When that which was and is no longer is perspectival 502 10.1 Mode-of-presentation effects: Substitutivity failures in simple sentences 504 10.2 Analytic, nonlogical substitutivity 509 10.3 Neighborhood watch: The view from symmetric predicates 515 10.3.0 Trans-scene identity 524 10.3.1 Identity without entity 530 10.4 Further consequences of adverbialization and supermonadicity: NPs as event descriptions 538 10.5 Summary 545
11
Naive Reference for the Cinéaste 551 11.0 A certain sequence of events 561 11.0.0 Witness to a consecution 563 11.0.1 Local cardinality preservation 570 11.0.2 Now Playing at Cinerama Wherever
580
Contents
11.1
Xi
Recounts
587
11.1.0 The anticonvergence condition and scenes of counting 11.1.1 Scene changes for counting and nominal syntax 597
595
12
Measuring Events 601 12.0 Counting with reference to events 603 12.1 Numerals 607 12.1.0 Syntactic evidence that number words denote events of counting 616 12.1.1 Semantic evidence that number words denote a relation between an event of counting and the events counted 618 12.2 Singular plurals: [A(n) AP k NP.pL], [A(n) (AP) NP.sG and NPsG] 620 12.2.0 Generalized singular plurals: [QX : ... sG ...] 627 12.3 Singular plurals, distributive plurals, and distributive singulars 632 12.3.0 Distributive plurals and distributive singulars 634 12.3.1 Counting many 635 12.3.2 Singular plurals and distributive plurals as antecedents for anaphora 637
13
Antisemidistributivity vs. Conjunction Reduction Redux 641 13.0 Semidistributivity 652 13.0.0 Semidistributivity without plural reference to persistent objects 653 13.0.1 Semidistributivity via singular reference to collective events 656 13.0.1.0 Reference to collaborative vs. individual histories 662 13.0.1.1 Singular reference to events 664 13.1
A null determiner
671
13.1.0 Disoriented counting 679 13.1.1 Defining the space referred to via the nominal denotation of its landmarks
687
13.2 Antisemidistributivity and its antidote
= 698
14
[or D AdrP and AdrP] 707 14.0 Within the same DP: The context for conjoined AdrPs 712 14.1 AdrPs coordinated 716 14.1.0 Landmark [AdrP and AdrP] 716 14.1.1 In the neighborhood of coordinated AdrPs 734 14.1.2 Nondenoting AdrPs and existential commitments within coordinated AdrPs 740 14.2 Kinematic and object-tracking scenes and frames of reference 750
15
The Ordered-Pair lllusion 763 15.0 Perspectival relations within nominal conjuncts 15.0.0 Scene correspondence 766 15.0.1 “Similarly orienting” 771 15.0.2 Silent ADR 776 15.1 Scenes in the neighborhood 783
16
QED
791
764
xii
Contents
Appendix 1 Descriptive Anaphora and Nonmaximal Reference under Selective Perspectives 793 Appendix 2 Eventish Notes
819
Bibliography Name Index Subject Index
973 991 997
811
Acknowledgments
In 1989, Jim Higginbotham’s counterexamples, now in section 14.1.1, launched a pair of job talks with equivocal results. After nearly three decades of critical commentary, attempted rescue, and comic relief, my benefactors will find the results no less equivocal and only these words of thanks for their pains. Those among the living who suffered the longest the details herein are Elena Herburger, Paul Pietroski, Roger Schwarzschild and Anna Szabolcsi. Joining them in the metatheoretical excursions about meaning or vinification have been Richard Larson, Peter Ludlow, and Robert May. Through most of it, Donca Steriade has endured progress reports, returning pitch-perfect advice. But, our young son, Aaron Joseph Steriade Schein, escaped quickly and painlessly, it seems, from bewilderment to a higher science. Other lifers putting up with my game include Toni Borowsky, Neil Briskman, David Eisenberg, Victor Eisenberg, Joey Freed, Alessandra Giorgi, Nina Hyams, Pino Longobardi, Richard Mendelson, Melissa Monroe, Joel Rotenberg, Ken Safir, Sandra Schein, Jerry Silverman, Dominique Sportiche, Richard Usatine, and Karina Wilkinson. Cheering on the last three years and twenty-six miles of this marathon were Ladan Shams and Leon and Ella Shams-Schaal, current and future cognitive scientists. For merely a decade, Elena Guerzoni sustained the semantics program and in our seminars posed questions so deftly that I would not grasp the full force of it until she had safely left the room. We caught up in postmortems to review the week in semantics and semantics gossip. More recently, joining the mash-up of workday research and after-hours speculation have been Yael Sharvit and Alexander Williams. To Alexander Williams, I am especially grateful for a close reading and comments on this and an earlier manuscript. The Poet says that “there are in our existence spots of time, / That with distinct pre-eminence retain / A renovating virtue, whence, depressed/ By false opinion and contentious thought, / Or aught of heavier or more deadly weight, / In trivial occupations, and the round / Of ordinary intercourse, our minds / Are nourished and invisibly repaired.” Lots of spots. Nourished and invisibly repaired I have been by
xiv
Acknowledgments
many occasional poets, some who may better remember their submission as a protracted inquisition: Pranav Anand, Chris Barker, Lucas Champollion, Cleo Condoravdi, Danny Fox, Martin Hackl, John Hawthorne, Norbert Hornstein, Kyle Johnson, Kathrin Koslicki, Angelika Kratzer, Peter Lasersohn, Chris Laterza, Terje Lohndal, Philippe Schlenker, Alexis Wellwood, and Eytan Zweig. In corridor and classroom, touching on this work, I have learned much from colleagues and fellow seminarians: Joseph Aoun, Hagit Borer, Thomas Borer, José Camacho, Lina Choueiri, Bridget Copley, Elena Guerzoni, Jim Higginbotham, Roland Hinterhölzl, Hajime Hoji, Utpal Lahiri, Katy McKinney-Bock, Toby Mintz, Sarah Ouwayda, Roumyana Pancheva, Philippe Schlenker, Robert Shanklin, Saurov Syed, Barbara Tomaszewicz, Antonella Vecchiato, Jean-Roger Vergnaud, Karina Wilkinson, and Maria Luisa Zubizarreta. José Camacho went as far as a 1997 dissertation and 2003 book on the syntax of coordination just as I was alleged to be finishing up its semantics. At critical junctures for this book, three, Ernie Lepore invited presentations at the Semantics Workshop of the Rutgers Center for Cognitive Science, his home for the care and feeding of interdisciplinary linguistics and philosophy. On four occasions of equal import, Massimo Piattelli-Palmarini and colleagues at the University of Arizona welcomed presentations of this work at their forum for linguistics and cognitive science. I am grateful to Jay Keyser for his years of support and support of the completed manuscript and also for his example of personal courage with schtick—I meant, grace. To my editors, Marc Lowenthal and Marcy Ross, at MIT Press, my copy editor Elizabeth Judd and indexer David Hill, many thanks for shepherding a very fat sheep to publication.
1
1.0
Introduction
Univocal and
It is an old prayer that the and conjoining the Determiner Phrases (DPs) in (1) be the same and that connects sentences, if only the logical form of (1) were something like (2): (1) At Acropolis Pizza, the sauce and the cheese are fresh. (2) At Acropolis Pizza, the sauce is fresh, and the cheese is fresh. Aristotle laments (Lasersohn 1995) that this could not in general be so because the sensible (3) would be rendered nonsense if its logical form were like (4): (3) At Acropolis Pizza, the sauce and the cheese are a perfect marriage of two rivals. (4) *At Acropolis Pizza, the sauce is a perfect marriage of two rivals, and the cheese is a perfect marriage of two rivals. The and of (3) must instead be one that composes DPs rather than sentences, and it applies to them so that the result refers to a plurality consisting of their referents, to the toppings of a plain cheese pizza that the subject of (5) also refers to: (5) The pizza toppings are a perfect marriage of two rivals. Aristotle’s brief for a nominal and is as good as the impoverished logical forms he assumes. If there is nothing more to be a perfect marriage than a one-placed predicate, there will indeed be no sentential source for (3). There is nothing for a sentential connective to connect except the conjuncts in the nonsensical (4), “Ms & Mc.” But if there are unspoken relations that logical form makes explicit (Davidson 1967; Castañeda 1967; Parsons 1990), it becomes possible to render (3) with a logical form in which the ordinary sentential connective conjoins clauses about the sauce’s and the cheese’s participation in the same event: (6) For some event e, ((Participates(e, s) and Participates(e, c)) & be a perfect marriage(e)) ‘The sauce participates and the cheese participates & it’s a perfect marriage.’1
2
Chapter 1
Here is the only and, the sentential one. Analysis of (3) as (6) explains why natural languages pronounce the sentential connective and the nominal conjunction alleged for (3) as if they were the same—they are. The collective reference of nominal conjunction finds a counterpart in the divided reference of predicative conjunction: (7) The pizza toppings are ladled and sprinkled over an outstretched pizza. Ladling and sprinkling divide the pizza toppings. That is, there is an event of ladling and there is an event of sprinkling and the pizza toppings participate in them, referring to these events with a plural pronoun: (8) For some events e1,e2, (Participates(them1,2, t) & ladled(e1) and sprinkled(e2)) As the ladling is not the sprinkling, there are the two events. As it is left open which toppings go on which way, all that can be said is that the toppings participate in them. The reference to distinct events in (7) is as obvious as the distinct verbs describing them. Yet plural reference to events is also latent in nominal conjunction, although (6) did not reflect it: (9) The sauce simmering and the cheese burbling are a perfect marriage of two rivals. For some events e1, e2, ((Participates(e1,s) & simmering(e1)) and (Participates(e2,c) & burbling(e2)) & perfect marriage(they1,2)) (10) The sauce slowly and the cheese quickly are ladled and sprinkled over the face of the outstretched pizza. For some events e1, e2, e3, e4, ((Participates(e1,s) & slow(e1)) and (Participates(e2,c) and quick(e2)) & Be(they1,2, them3,4) & (ladled(e3) and sprinkled(e4)) As the simmering is not the burbling and what is slow is not quick, the sauce and the cheese each participate in their own state or event in (9) and (10). Their states while simmering and bubbling, these constitute a perfect marriage. The slow and quick events in which they participate, these are those, the events of their being ladled and being sprinkled. To Aristotle’s question of why this and of collective or divided reference isn’t different from sentential and, the answer is that in a language with unspoken relations to events, and’s collective or divided reference is just an illusion of plural event pronouns cloaking their plural reference to events in silence. If and is always a sentential connective, then much of what it connects often goes unspoken—Conjunction Reduction (Chomsky 1957; Ross 1967)—and unspoken must be grasped in context from what is not unspoken.
Introduction
3
So plural event pronouns rescue and. But even if a word’s integrity did not mean much, there is further argument concluding this chapter that any syntax and semantics deriving DPs or Predicate Phrases from their conjunctions—[DP[DPthe sauce] and [DPthe cheese]], [PredP[PredPladled] and [PredPsprinkled]]—are mistaken both in syntax, facing a dilemma with no exit (section 1.5.1), and in semantics. What is expressed with an and of collective or divided reference proves not to be synonymous with the use of plural event pronouns (section 1.5.2). Natural language again conforms in form and meaning to a language, Eventish, with a univocal sentential and, unspoken relations to events, and plural event pronouns. This argument launches a grammar in Eventish for and and its surroundings in natural language developed in the following chapters. An overview of the major results fills out this one (sections 1.2–1.8), but I begin with an apology (section 1.1). Prior to any thought of events and what is to be said in their name, a few facts of natural language show that answering Aristotle in a logical language in which and is univocal and sentential is the only game in town, whether or not it will be played well here. 1.1
The slippery slope to Conjunction Reduction
The semantic fact of (3) that inclines Aristotle toward another meaning for and recurs in (11)–(15) despite a sentential connective that cannot be anything else: DP-Tense-V Δi and DP-Tense-V Φi (11) No well-behaved child will grab at Δi and no well-behaved sibling will tear into [a piece of pizza that they have been told to share]i. (12) Not many a student proposed and not many a professor (of his) accepted that they should collaborate even more than they already have. (13) Most philosophers by noon had overrun and most linguists no more than 10 minutes later had mobbed some twenty pizzerias between them in the neighborhood of their joint session. (14) Most of the philosophers (while) hollering crushed and most of the linguists (while) hooting squashed forty pizza delivery vans between them. (15) Marvin made a grand entrance and Bernice swept in at the gala with each other’s spouses on their arms. In (11)–(15), and conjoins fully tensed sentences containing spoken subjects and tensed verbs. The coordinate structures chosen exemplify Right-Node Raising, so called in that the first of the conjoined sentences contains a gap Δi, an unspoken phrase, related to the Right-Node Raised phrase, spoken at the rightmost edge of the second sentence. The Right-Node Raised phrases of (11)–(15) are all collective
4
Chapter 1
(Perlmutter and Ross 1970; Jackendoff 1977; McCawley 1981a) like be a perfect marriage in (3). Their spoken repetition in the first sentence fails to preserve the meaning of the original, just as (4) fails (3). The sentential connective, it appears, is no impediment to collective meaning from a Right-Node Raised phrase. Its collective meaning continues into (16)–(25), where the sentences conjoined have now been reduced to a spoken subject and adverb, as already seen in (9) and (10) too: DP AdverbP Δi and DP AdverbP [i Tense-V Φ] (16) No well-behaved child selfishly and no well-behaved sibling angrily will tear from each other pizza they have sworn to share. (17) Not many a student before an oral exam and not many a professor (of his) afterward agree to collaborate more than they already have. (18) Few philosophers (while) boasting and few linguists (while) bragging converse about each other’s children. (19) Few Russian chess masters in one move and few Armenian chess masters in the very next one stalemate. (20) Most philosophers (while) hollering and most linguists (while) hooting drowned out the lecture heckling each other. (21) Saul hollering and David hooting drowned out the lecture heckling each other. (22) Saul hollering and David hooting heckled each other. (23) Most philosophers by this morning and most linguists by nightfall yesterday had plundered more bordeaux from each other’s wine cellars than the region produces in a week. (24) Saul from early morning in New Jersey and David later in the day in California have been swapping wine futures in online trading. (25) Marvin this afternoon from Great Neck and Bernice this evening from Syosset are arriving at Leonard’s with each other’s spouses in rented Mercedes. Whether of fully tensed sentences or of sentences reduced, their conjunction in (11)–(25) does not refer to a plurality or to an object of any kind, a point made plainer by the truth of (11) and (16) when there are no well-behaved children or siblings to refer to (and similarly (12), (17)–(19)). Adverbs being adverbs, there must be something for them to modify in the logical forms for (22) and (24), although it is unspoken: (26) Saul hollering participates and David hooting participates, & it all was a heckling of each other.
Introduction
5
(27) Saul from early morning in New Jersey participates and David later in the day in California participates, & it all has been a swapping of wine futures in online trading. Adverbs being adverbs, they are optional. Omitting them in (28) and (29) but leaving in place what was modified but unspoken sketches logical forms for (30) and (31): (28) Saul participates and David participates, & it all was a heckling of each other. (29) Saul participates and David participates, & it all has been a swapping of wine futures in online trading. (30) Saul and David heckled each other. (31) Saul and David have been swapping wine futures in online trading. Pace my pleading in the chapters to follow, whatever semantics next solves collectivized Right-Node Raising and is then supplied to the conjunction of the more fullblooded clauses in (11)–(25) applies without amendment when the clauses happen to be greatly reduced, dropping as in (30) and (31) (and (3)) adverbs and any other spoken verbal elements. It thus promises a general account of the semantic facts on display in (3), (9), and (11)–(31). In contrast, an and that fashions a plurality only from phrases that themselves refer is good only for conjoined DPs and may be dismissed on these grounds. What was Aristotle thinking, imagining that collective predication in (3) propels an argument that and has a sense other than that of the sentential connective when only the latter occurs with Right-Node Raising and its collective meaning? In concluding that the sauce and the cheese refers to a plurality, Aristotle succumbs to an illusion, misparsing as conjoined DPs a conjunction of sentences in which much is unspoken. The semantics of natural language it leads him to thus rests on an error of syntax.2,3 1.2
Conjunction Reduction restrained
Almost as old as the prayer for univocal and and a perfect logical language is foreboding about translation into it. Conjunction Reduction, in insisting that much of a sentence’s logical form goes unspoken, threatens runaway translations. What will spare (32) and (33) from becoming synonyms if an unspoken subject copies its antecedent, and will it not confound (34) and (35) if unspoken subjects can be null pronouns? (32) A rocker shimmied and shook. (33) A rocker shimmied and a rocker shook. (34) No rockeri screeched and (then) smashed hisi Stratocaster. (35) *No rockeri screeched and (then) hei (then) smashed hisi Stratocaster.
6
Chapter 1
There is no respite from this foreboding without grammar, which will consist here largely of variations on two themes. The first is special to the Davidsonian setting to which Conjunction Reduction is restored. No matter that the subject is copied in (36); there cannot be two different rockers unless there are also two independent existential event quantifiers as in (37) approximating (33): (36) [A x: rocker(x)] Agent(e, x) & shimmied(e) and [A x: rocker(x)] Agent(e, x) & shook(e) (37) [A x: rocker(x)] ∃e(Agent(e, x) & shimmied(e)) and [A x: rocker(x)] ∃e(Agent(e, x) & shook(e)) Should however the event variable or pronoun in the second clause refer to the event described in the first, the second clause comes to repeat harmlessly that this same event has for its unique Agent a rocker.4 Thus (32) and (33) are spared synonymy by the grammar of event quantification, withholding from (32) the two existential event quantifiers necessary for (33). Pronunciation in the second clause of (33) cues articulation of a second existential event quantifier, and silence in (32) signals resumption in the events referred to (chapter 3). The distribution of existential event quantifiers fixes the meaning not only of unspoken subjects but of unspoken phrases in general, including the gaps that RightNode Raising (section 1.1) leaves behind. The sentences of Lebanese Arabic (38)– (41) (chapter 2) all report that the pines were sparse as were the oaks, two sparsities, while the forest itself may be dense:5 (38) keenou l-snoubraat w keenou l-sendyeeneet xfeef bi mantʕa Hadd were.3m.pl the-pines and were.3m.pl the-oaks sparse in region near mantaʕa bašariyyeh region human ‘The pines were and the oaks were sparse in a region near human settlement.’ (39) keenou l-snoubraat w l-sendyeeneet xfeef bi mantʕa Hadd mantaʕa were.3m.pl the-pines and the-oaks sparse in region near region bašariyyeh human ‘The pines and the oaks were sparse in a region near human settlement.’ (40) keenou l-snoubraat men 5000 seneh w l-sendyeeneet men 1000 seneh were.3m.pl the-pines from 5000 year and the-oaks from 1000 year xfeef bi mantʕa Hadd mantaʕa bašariyyeh sparse in region near region human ‘The pines 5000 years ago and the oaks 1000 years ago were sparse in a region near human settlement.’
Introduction
7
(41) l-snoubraat w l-sendyeeneet keenou xfeef bi mantʕa Hadd mantaʕa the-pines and the-oaks were.3m.pl sparse in region near region bašariyyeh human ‘The pines and the oaks were sparse in a region near human settlement.’ Despite the reported two sparsities, the sentences differ in allowing them different regions. In a coordination of Tensed Phrases (38), with Tense introducing existential event quantification, the unspoken and spoken tokens of a region near human settlement fall within the scope of distinct existential event quantifiers and may thus be understood to describe distinct events. In (41), the DPs include a Tensed Phrase within their scope. With an unspoken Tensed Phrase in the first conjunct, (41) comes to resemble (33) in structure and interpretation—two tokens of Tense, two existential event quantifiers, the possibility of two regions. In contrast, in (39), a solitary existential event quantifier corresponds to the single token of Tense. It is a plural quantifier, denoting the two sparsities, about which it is said that they are sparse and they are in a region near human settlement, the same one. The introduction in (40) of existential event quantifiers in the guise of Adverb Phrases within each conjunct restores the interpretation that allows for different regions. The presence of this interpretation in (40) is evidence of an unspoken token of a region near human settlement within the first conjunct.6 That the unspoken token is harmless in (39) again reflects the distribution of existential event quantifiers. The second theme in the grammar of Conjunction Reduction (chapter 3) turns to the modern syntax of clauses that discovers among them different sizes according to how much morphology is present beyond the minimal subject and predicate. There is certainly some difference of syntax between (34) and (35). Yet, argument that the conjuncts in (34) are less than sentences and therefore without subjects is as sound as an argument from the contrast between (42) and (43) that the conjuncts in (42) are likewise without subjects: (42) No rockeri is shimmying and any of hisi mamas shaking to that funky disco beat. (After McCawley 1993) No rockeri has shimmied and any of hisi mamas shaken to that funky disco beat. (43) *No rockeri is shimmying and any of hisi mamas is shaking to that funky disco beat. *No rockeri has shimmied and any of hisi mamas has shaken to that funky disco beat. The contrasts are rather evidence of a constraint confining the scope of no rocker within fully tensed clauses, so that when these large clauses are conjoined, as in (35)
8
Chapter 1
and (43), no rocker will either fail to include within its scope the bound variable in the second conjunct, (45a), or violate this constraint on its scope, (45b): (44) [TPNo rockeri Tense+Aux [[ti shimmying] and [any of hisi mamas shaking]]] (45) a. *[TPNo rockeri Tns+Aux [ti shimmying]] and [TPany of hisi mamasj Tns+Aux [tj shaking]] b. *No rockeri [[TPti Tns+Aux [ti shimmying]] and [TPany of hisi mamasj Tns+Aux [tj shaking]]] A coordination of smaller clauses, (42) and (44), leaves room enough for no rocker to remain within the fully tensed clause while including within its scope both conjuncts. What is to be said about the contrast between (34) and (35), where the coordinations in both are of fully tensed clauses? It must be that a clause is larger with its subject pronounced than unpronounced, and only the largest clauses, fully tensed with subjects pronounced, confine the scope of no rocker. The suggestion that pronunciation induces a larger clause goes hand in hand with the earlier one that it cues articulation of an existential event quantifier. This sketch of the contrast between (34) and (35) illustrates the approach throughout to the evidence alleged for nonsentential, phrasal conjunction. The facts are real enough and indicative of healthy sentences in S, M, and L, which should not be tortured into an XXS impossible-to-fit sentential and between. A grammar developed with attention to both the distribution of event quantifiers and clauses tailored to size promises to rein in Conjunction Reduction against runaway translation. Foreboding about runaway translation allayed, this grammar of event quantification and clausal size is also an argument in favor of the answer to Aristotle at the center of this work. Clausal size combines with morpheme weight to restrict divided reference in conjoined PredPs: (46) Kunstler is sitting and standing (to annoy the judge and entertain the gallery). According to (46), the flamboyant attorney divides his antics between events of sitting and events of standing, a division not effected by a special meaning for and in the composition of a coordinate participial phrase. From outside the coordination—so goes the answer to Aristotle’s lament—a plural event pronoun refers to the events that it describes and finds a relation, again outside the coordination, to relate these events to those of the subject’s participation (cf. (10)): (47) For some events e1, e2, e3, Participates(e1,k) & Be(that1, them2,3) & (sitting(e2) and standing(e3))
Introduction
9
For all of that, an event pronoun and a morpheme expressing the relation—here represented by the copula be—there must be room enough outside, of which there is not in a coordination that already takes up Tensed Phrases: (48) #Kunstler is sitting and is standing. Cf. Kunstler is glaring at the judge and is formulating his rebuttal. Sentence (48) only means that the attorney is also a contortionist sitting and standing in a single posture. Without room enough for the apparatus of divided reference, the silence of an unpronounced subject just signals resumption to the events antecedently described, as already noted for (32). If it were left to and to effect divided reference, it is unclear why it should succeed in (46) but fail in (48). Observe that Kunstler can be spared his contortions if again pronunciation of the subject cues a second existential event quantifier (cf. (33)): (49) Kunstler is sitting, and Kunstler is standing. (50) Kunstler is sitting, and he is standing. Sentences (49)–(50) can still put Kunstler in a compromising position, electing the second event quantifier to be anaphoric to the first. But the clauses can also now be understood to advance the narrative so that (49)–(50), in contrast to (48), can describe successive postures.7 With what has been said to relate unpronounced subjects to the resumption of events antecedently described, crucially in the explanation of why (32) has no interpretation in common with (33), it might appear that (51), itself a large coordination of Tensed Phrases, is a striking counterexample in finding equivalent expression in (52): (51) An easy model theory textbook is badly needed and will surely be written within this decade. (Partee and Rooth 1983, 369) (52) An easy model theory textbook is badly needed, and an easy model theory textbook will surely be written within this decade. But, for a logical syntax—silence signaling an unspoken pronoun—it is salutary that here a spoken one also just happens to result in an expression equivalent to (52): (53) An easy model theory textbook is badly needed, and it will surely be written within this decade. In fact, constructing sentences where spoken pronoun and indefinite description diverge from the equivalence of (52) and (53) shows the unspoken subject and spoken pronoun behaving alike. Sentence (54) about many tons of rice needed and many tons paid for, asserting that the poor are hungry and the rich will be well fed,
10
Chapter 1
leaves room for the hope that the poor will eventually be delivered from their hunger: (54) Many new tons of rice are needed and many new tons will surely be farmed and distributed to those who pay. (And maybe the rice that is needed will be delivered to those who need it.) (55) #Many new tons of rice are needed and they will surely be farmed and distributed to those who pay. (And maybe the rice that is needed will be delivered to those who need it.) The spoken pronoun in (55) destroys this hope in promising that the many tons of rice needed will go elsewhere. Omitting the pronoun in (56), the sentence with an unspoken subject is just as pessimistic, as if, indeed, the subject were a null pronoun: (56) # Many new tons of rice are needed and will surely be farmed and distributed to those who pay. (And maybe the rice that is needed will be delivered to those who need it.) A sentential connective connecting fully tensed clauses cannot escape conjoining clauses with subjects. With a null pronoun in the second clause of (56), the conjuncts in fact have different subjects, an indefinite quantified phrase in the first and definite pronoun in the second. This asymmetry preempts alternative analysis that would derive a subsentential, predicative coordination, [PredP[PredPare needed] and [PredPwill surely be farmed and distributed to those who pay]]. For, if this alternative means anything at all, it means that when this coordinate Predicate Phrase meets its solitary subject, many new tons of rice, the subjects understood within it, however they are to be understood, are understood exactly the same way, symmetrically.8 Although it did not have to turn out this way, the fact of a different subject in the second clause is precious argument that and is sentential. Since these introductory remarks are meant as much to entice as instruct, I have skated by detail and argument to flag a few punchlines … results with broad theoretical implications for the logical syntax of constructions in and. Chapters 2–7 provide the more measured development. Chapter 2 dissolving Aristotle’s illusion of collective reference is about the logical syntax of alleged DP-coordination, where what is unspoken is predicative. Chapters 3–7, turning to divided reference, are about constructions where nominals go unspoken to create an illusion of predicative coordination. The grammar for constructions in and regiments the distribution of event quantifiers, the sizes of clauses, the morphology expressing relations to events, the distribution and interpretation of unspoken pronouns, which prove to be as wideranging in their interpretation as their overt counterparts, and the scope relations among all of the above.
Introduction
1.3
11
A new clausal architecture
With and always the sentential connective, a corrected syntax entails four emendations in the logical language Eventish. Three are purely formal, supermonadicity (section 1.4), descriptive event anaphora (section 1.5), and adverbialization (section 1.6). The fourth emendation, spatiotemporal orientation (section 1.7), augments the nominal and clausal morphology so that temporal reference becomes spatiotemporal. In the classic citation form for (57), the sentence with several arguments decomposes into several conjuncts all about the same event e in (58) (Davidson 1967; Castañeda 1967; Parsons 1990): (57) The waiter buttered the toast in the pantry with a knife at midnight. (58) Agent(e, w) & butter(e) & Patient(e, t) & in(e, p) & with(e, k) & at(e, m) Supermonadicity (section 1.4) is a resolution into smaller events, emending (58) so that every argument is isolated in its own event. Various relations between the several events then hold the sentence together: (59) Participate(e1, w) & Cause(e1, e2) & butter(e2) & Participate(e2, t) & in(e2, e3) & Participate(e3, p) & with(e1, e4) & Participate(e4, k) & at(e2, e5) & Participate(e5, m) Isolated here in his own event is the butterer from the buttered. Yet, even when a coordination describes participants participating the same way, every DP is occasion for a new event: (60) Jones gracefully and Jeeves clumsily buttered 613 pastries for brunch. Although Jones and Jeeves are both butterers, if what is done with grace is not what is done clumsily, the isolation extends into the coordination in (60) so that Jones and Jeeves are on their own (see (9)–(10)): (61) Participate(e1, j) & graceful(e1) and Participate(e2, k) & clumsy(e2) & butter(that1,2) (62) Jones participated gracefully and Jeeves participated clumsily, and it all was a buttering of 613 pastries for brunch. The buttering subsequently referred to must therefore derive from plural reference to their efforts e1 and e2. It is thus a quick step from contrary adverbs to at least some plural event pronouns. Prompted by what logic dictates here, the basic device of cross-reference to events is emended everywhere to be an unspoken plural, descriptive event pronoun (section 1.5) instead of a naked event variable. Plural event pronouns, as remarked in section 1.0, shoulder the burden of dissipating the illusion of an and of collective or divided reference. Empirical justification follows
12
Chapter 1
when it is shown that the unspoken pronouns display the same vagaries of reference as their spoken counterparts. Emending how DPs prefix to their scope, adverbialization (section 1.6) interposes an adverb derived from the DP’s descriptive content so that what is spoken as (63) is always parsed along the lines of (64): (63) The waiter poured. (64) The waiter while a waiter poured. Adverbialization is to apply to every DP as a constituent of its basic mode of combination. (65) The man surpassed the boy of his childhood in regrets and not much else. Since the events of a man while a man are not those of his while a boy, adverbialization like the adverbs in (60) can only be entertained in the company of supermonadicity. The surpassing itself, if there were only the one event to speak of, could not have been both in middle age and childhood. Taking up supermonadicity and adverbialization at once (without clutter from the event pronouns) delivers for a simple sentence (66) a logical form along lines suggested in (67) and the paraphrase in (68): (66) The waiter buttered the toast in the pantry with a knife at midnight. (67) [The x: ∃e waiter(e,x)] [while e: waiter(e,x)][∃e1: within(e, e1)] Participate(e1,x) & Cause(e1,e2) & [The y: ∃e toast(e,y)][while e: toast(e,y)] [∃e2: within(e, e2)] Participate(e2, y) & butter(e2) & [The z: ∃e pantry(e,z)][while e: pantry(e,z)][∃e3: within(e,e3)] Participate(e3, z) & in(e2,e3) & [The k:∃e kitchen(e,k)][while e:knife(e,k)] [∃e4:within(e,e4)]Participate(e4,k) & with(e1,e4) & [∃m:∃e 0h00(e,m)] [while e: 0h00(e,m)][∃e5:within(e,e5)]Participate(e5,m) & at(e2,e5) (68) The waiter while a waiter participates & it caused a buttering & the toast while toast participates in that, which is a being-in in which the pantry while a pantry participates & it has assistance which a knife while a knife participates in, & it is a being-at that the midnight hour while midnight participates in. With just three emendations, supermonadicity, descriptive event pronouns, and adverbialization, logical form looks to have fallen down a rabbit hole in (68) into a world of events an unaltered speaker is hardly aware of. It gets worse with the fourth emendation, spatiotemporal orientation (sections 1.7 and 1.8). Annotations to logical form will indicate perspectives, frames of reference, and scenes under a certain resolution and so on—that is, events or conditions that do not purport to deconstruct the goings-on of the speaker’s conscious report but instead choreograph her epis-
Introduction
13
temic stance toward them.9 For NPs, it has been a commonplace in the semantics of modal and temporal languages that their interpretation varies with worlds and times, which may be represented more or less explicitly in the object language. A finer-grained resolution is to be imagined for NPs relating to frames of reference. As it has been a point of grammar that VPs do not occur in certain sentences unless accompanied by other morphology such as Tense, it is to be stipulated that NPs do not occur in DPs without a morphology that describes location in a spatiotemporal frame of reference as in (69) and is applied to every NP when they are coordinated as in (70): (69) [DPthe X : ∃E (here[E,f] &NP[E,X])] [DP D [AdrP Adr NP]] (Cf. [CP C [TP Tense VP]] ) ‘The fielders’ (70) [DP D AdrP and AdrP] ‘The infielders and outfielders’ The addressing “here[E,f]” assigns the address here in the frame of reference f to all and only the events E. Correlative with the finer-grained resolution, the Tensed clause becomes a canonical structure for the expression of spatiotemporal relations, including a fix on a perspectival center (now) and navigation from there (ℋ) to the events Tense locates: now-ℋ-Tense-VP (section 10.3.0). All four emendations— supermonadicity, descriptive event anaphora, adverbialization, and spatiotemporal orientation—formal and substantive, are necessary to confer on the logical language the expressive power to translate natural language while and remains throughout the univocal sentential connective. The following sections sketch some of the details and arguments for the new clausal architecture, which are culled from chapters organized to address empirical problems confronting the thesis that and is univocal and sentential. Consider one problem already solved and exemplified in the contrast between (71) and (72). If the implied temporal sequencing that distinguishes them is semantic (rather than pragmatic), then and is certain to be ambiguous between its atemporal meaning elsewhere and a temporal one here if what is on display here is of the form ⌜p and q⌝ and ⌜q and p⌝: (71) They got married, and she got pregnant. (72) She got pregnant, and they got married. A difference of meaning between (71) and (72) and the sameness of and can be had only with a difference elsewhere, parsing them ⌜p and q⌝ and ⌜q′ and p′⌝. What is spoken is an illusion masking that no two of the four clauses are identical. To that end, it is quickly observed that none of the clauses are identical if Tense as temporal anaphora (Partee 1973, 1984) explicitly represents that the time of the event described
14
Chapter 1
by the second clause follows that of the event described by the first. The dissociation of temporal meaning from and is corroborated in that the difference between (71) and (72) survives and’s absence: (73) They got married. She got pregnant. (74) She got pregnant. They got married. Today it is banal to say that what varies in (71)–(74), if semantic, has nothing to do with and and everything to do with Tense. But if there were not already in hand well-understood and broadly accepted accounts of Tense, temporal logic, and anaphora, one can imagine—especially, say, for a language without Tense morphology—what would be needed to defend that and means no more than it does. It would be the invention of Tense and the rest of it to explain away the illusion that and has a temporal meaning, and it would have to be all about sentences that are not themselves conjunctions, (73)–(74), if the claim that and has no special sense is not to ring hollow. And so this book about nothing—that and is never other than the trite sentential connective—is several hundred pages about what and is not. What and is not depends on what it can be mistaken for, which varies with its context. The chapters are thus each about a context and the empirical problems encountered there for the thesis that and itself never means much: chapter 2 about DP and DP, chapters 3–7 about PredP and PredP, and chapters 8–15 about NP and NP.10 The last chapters especially, introducing adverbialization and spatiotemporal reference to basic clause structure, are first of all about what these innovations do for simple sentences and only then does the solution to the problems of conjunctions emerge, appended as a corollary to the more general theory. 1.4
Supermonadicity
The relations unspoken under Conjunction Reduction have been glossed in (6) and (26)–(29) as participation with a deliberate vagueness that proves itself in sentences with multiple conjunctions: (75) The Columbia students and the Harvard students surrounded the Pentagon and were crowded into the Mall. (76) The Columbia students participate and the Harvard students participate; & it all was a surrounding the Pentagon and a crowding into the Mall. Sentence (75) commits a student to neither venue, leaving vague whether she acted at the Pentagon or was acted on, a victim of crowd control on the Mall (see (8) and (10)). All that is implied is that the Columbia students participate in some way as do the Harvard students, and what they do constitutes the events subsequently described, so that some of the Columbia and Harvard students must have been at the Pentagon while the rest were on the Mall. The hidden sentences conjoined in
Introduction
15
(75) are necessarily vague as in (76) about what the students are up to. That vagueness however threatens the translation of (77) in the event that Harvard students mercilessly tease Columbia students who decline to tease themselves: (77) The Harvard students and the Columbia students teased the Columbia students. (78) The Harvard students participate and the Columbia students participate; and it was a teasing. Sentence (77) in that event is false, but its considered translation (78) is mistakenly true in that both the Harvard students and the Columbia students have participated, the former as aggressors and the latter as victims in the same event of teasing. To divorce the vagueness necessary for (75) from its ill-effect on (77), the logical form for the latter must resolve two events, cause and effect: (79) The Harvard students participate and the Columbia students participate; & that caused a being teased11 & the Columbia students participate in that. The vagueness of their participation notwithstanding, what they do is said to cause the Columbia students’ humiliation. The Columbia students had no hand in the causal event, and (79) is false, as a translation of (77) should be. This argument repeats itself mutatis mutandis wherever DPs can occur allegedly conjoined,12 and thus a resolution into smaller events occurs across the board at every argument position. In a (neo-)Davidsonian logical form (81) for (80), relations to events relate to the same event e: (80) The waiter buttered the toast in the pantry with a knife at midnight. (81) (Davidson 1967, 1985; Castañeda 1967; Parsons 1990) Agent(e,x) & butter(e) & Patient(e,y) & in(e,z) & with(e, v) & at(e,w) (82) Supermonadicity Participate(e1,x) & Cause(e1,e2) & butter(e2) & Participate(e2,y) & in(e2, e3) & Participate(e3,z) & with(e1,e4) & Participate(e4,v) & at(e2,e5) & Participate(e5,w) It is now urged that even a simple sentence speaks of several events as in (82), which are (at least) as many as the relations in the original (neo-)Davidsonian logical form. Holding the sentence and disparate events together are relations between them such as “Cause(e1,e2)” for butter or kill, where cause and effect are especially vivid. In homage to generative semantics, Conjunction Reduction (Gleitman 1965; Lakoff and Peters 1969; McCawley 1968; Ross 1967) restored restores the causative analysis of transitives (Lakoff 1965) in the guise of supermonadicity. The decomposition into cause and effect finds support in classic arguments that the V-ing is not the being V-ed. Vlach (1983) cites the contrast between (83) and
16
Chapter 1
(84) to conclude that the being shot is not the shooting. Example (84) reports truly that Jackie saw the effect, and (83) reports falsely that she saw the cause: (83) Jackie saw Oswald shoot Kennedy. (84) Jackie saw Kennedy shot by Oswald.13 Beardsley (1975) notes that what can be admitted in (85) as an alternative description of Lincoln’s signature action, his alarming the Northern slaveholders, cannot in turn describe its effect, (86).14 And its effect, the Northern slaveholder’s being alarmed (87), is not to be mistaken for the cause Lincoln’s alarming the Northern slaveholders, (88)–(89): (85) Lincoln’s signing the Emancipation Proclamation was his alarming the Northern slaveholders. (86) # Lincoln’s signing the Emancipation Proclamation caused his alarming the Northern slaveholders. (87) Lincoln’s signing the Emancipation Proclamation caused the Northern slaveholders’ being alarmed. (88) #Lincoln’s signing the Emancipation Proclamation was the Northern slaveholders’ being alarmed. (89) #Lincoln’s alarming the Northern slaveholders was the Northern slaveholders’ being alarmed. The precedent welcome, supermonadicity goes on to hold that the causative analysis of these transitive verbs is merely a special case of a more general, formal point about polyadicity driven by the semantics of sentences with multiple conjunctions, which applies across the board: (90) Brooklyn borders Queens. Participate(e1,x) & O(e1,e2) & border(e2) & Participate(e2,y) (91) Kings County contains Brooklyn. Participate(e1,x) & O(e1,e2) & contain(e2) & Participate(e2,y) Despite absent cause and effect, Brooklyn, Queens, and King County in (90) and (91) still all inhabit their own events or states, to be related perhaps by a topological relation O, the final inventory of relations enlisted in the decomposition of transitive verbs being contingent on a complete survey. Supermonadicity is still not done multiplying events. It has so far enforced that Saul and David’s action in (92) not be confounded with the wine futures being swapped or Jones and Jeeves’s action with the toast piling high. The former cause the latter, perhaps according to (93) and (95):
Introduction
17
(92) Saul and David have been swapping wine futures in online trading. (93) Participate(e1,x) and Participate(e1,y) & Cause(e1,e2) & swap(e2) & Participate(e2,z) (94) Jones and Jeeves piled high the pastries for brunch. (95) Participate(e1,x) and Participate(e1,y) & Cause(e1,e2) & pile(e2) & high(e2) & Participate(e2,z) Yet, reason to reject (93) and (95) is that what is early morning in New Jersey is not what is afternoon in California and what is graceful is not clumsy: (96) Saul from early morning in New Jersey and David later in the day in California have been swapping wine futures in online trading. (97) Jones gracefully and Jeeves clumsily piled high the pastries for brunch. Saul and David operate their own events, e1 and e3, which together cause the reported effect, as must Jones and Jeeves: (98) Participate(e1,x) & a.m.(e1) & in(e1, NJ) and Participate(e3,y) & p.m.(e3) & in(e3, CA) & Cause(that1,3, e2) & swap(e2) & Participate(e2,z) (99) Participate(e1,x) & graceful(e1) and Participate(e3,y) & clumsy(e3) & Cause(that1,3, e2) & pile(e2) & high(e2) & Participate(e2,z) Contrary adverbs recur in (100) with similar mission: (100) Jones gracefully and Jeeves clumsily separated the brioches to the left and the croissants to the right. (101) At the loom, Arachne gracefully and her apprentice clumsily crisscrossed silk threads horizontally and linen threads vertically. Neither the leftward nor the rightward parade of pastry is a separating on its own, and the separating is itself neither to the left nor to the right, having components of both. So neither the leftward event nor the rightward event is described by the verb, and the adverbs modify a relation separate from it, describing the brioches’ and the croissants’ participation (pace Kratzer 1996, 2003): (102) Participate(e1, x) & graceful(e1) and Participate(e2, y) & clumsy(e2) & Cause(that1,3, e2) & separate(e2) & Coincide(e2, that3,4) & Participate(e3,z) & to the left(e3) and Participate(e4,w) & to the right(e4) ‘Jones participated gracefully and Jeeves participated clumsily; that all caused a separating, which coincided with all this in which the brioches participated to the left and the croissants participated to the right.’
18
Chapter 1
It is small events all the way down, with every DP launching its own event and its own relation to it. Life under supermonadicity couldn’t be lonelier spinning around in a subevent solo, unrelated to anything else except through the mediation of further relations between events. So lonely is it that the fact of participation in one event being also participation in another is strictly an inference, without direct representation in logical form. Recall that it cannot be said who was at the Pentagon and who at the Mall and thus it cannot be said who was an agent of the surrounding or a theme of the being crowded into. Thus existential quantifiers close off the subject positions of the conjoined clauses: (103) The Columbia students and the Harvard students surrounded the Pentagon and were crowded into the Mall. (104) [the X: C(X)] Participate(e1,X) and [the Y: H(Y)] Participate(e2,Y) & O(that1,2, that3.4) & (∃Z Agent(e3,Z) & surround …) and (∃Z Theme(e4,Z) & … crowded …) ‘The Columbia students participated and the Harvard students participated, & it was someone’s surrounding the Pentagon and someone’s being crowded into the Mall.’ There can be no variable-binding into these positions from the Columbia students and the Harvard students. It is only to be inferred from the coincidence of their participation in some events with events at the Pentagon and on the Mall that they were indeed agents and themes in the latter. If all this is not an eccentricity of conjunction, the same holds of simple sentences: (105) The students demonstrated. (106) [the X: S(X)] Participate(e1,X) & O(that1, that2) & (∃Z Agent(e2,Z) & demonstrate …) (107) [TPThe studentsi [Tense [VP ti [demonstrate]]]] The subjects of embedded phrases despite some syntactic relation to the subjects of higher phrases as in (107) are existentially closed off from them, (106). That is, the traces of movement are not translated as bound variables. From an assertion that the students’ participation was a demonstration by some agents, we infer that the students were its agents, without direct representation of the fact. This conclusion descending from the semantics of sentences with multiple conjunctions such as (103) finds corroboration in a distant corner of syntax. To begin, note that the participial phrases conjoined in (108) and (110) assign contrary thematic relations to their subjects like in (103). Logical forms that conform to what has been concluded on the evidence of (103) are shown in (109) and (111):
Introduction
19
(108) Pollock was attacked by his critics in print and attacking them on canvas. (109) [℩x: Pollock(x)] Participate(e1,x) & O(that1, that2,3) & ((∃Z Patient(e2,Z) & attacked(e2) …) and (∃Z Agent(e3,Z) & attacking(e2) …) (110) The Abstract Expressionistsi were [[ti attacked by theiri critics in print] and [ti attacking them on canvas]]. (111) [The X: Abstract Expressionists(X)] Participate(e1,X) & O(that1, that2.3) & ((∃Z Patient(e2,Z) & attacked(e2) …) and (∃Z Agent(e3,Z) & attacking(e2) …) As the Abstract Expressionists may be divided between those attacked and those on the attack without it being said who are aggressors and who are victims, it can at best be said as in (111) that their participation, vague about the manner, coincides with some Patients being attacked and some Agents attacking. As in the above discussion of (77), so that it is entailed that Abstract Expressionists are both attacked and on the attack, it must be that the event of being attacked is not the same event as the attack, and that to participate at all in the former is to participate only as a victim, a Patient, and to participate in the latter is only to be an aggressor, an Agent. In turn, it can hardly matter for these aspects of logical form that the plural the Abstract Expressionists is replaced in (108) with the singular Pollock. Thus the logical form (109) carries over the vague participation relation and the existential closure of the thematic relations within the conjoined VPs. Pollock as both victim and aggressor is merely to be inferred from his vague participation in two events, one a being attacked in which its only participant is a Patient, some victim, and another, an attack in which its only participant is an Agent, some aggressor. In earlier literature focused on the singular case, to represent that Pollock is both victim and aggressor, the VP-Internal Subject Hypothesis15 also supposes representation of the subject positions internal to the coordinated participial phrases from which Pollock has been moved across the board as indicated in (112) and then interpreted, without the intervention of existential quantification, as direct variable-binding: (112) Pollocki was [[ti attacked by hisi critics in print] and [ti attacking them on canvas]]. (113) [℩x: Pollock(x)] was [[x attacked by hisi critics in print] and [x attacking them on canvas]]. Heycock (1995) points out that the hypothesis that such a participial phrase contains a subject position that moves with it in (114) is inconsistent with the facts about pronoun coreference in (114) and (115):
20
Chapter 1
(114) How pleased with the pictures Pollocki painted in hisi youth do you think hei really was? (115) *Hei really was pleased with the pictures Pollocki painted in hisi youth. For if the participial phrase fronted in (114) contained a subject, it should be disjoint from the name Pollock, just as it is in (115).16 Elena Guerzoni, Katy McKinney-Bock, and Saurov Syed (p.c., March 2011) observe that the problem is more general and independent of any hypothesis about subject position in the largest phrase fronted in (114) or (116). It arises for any subject position that (116) might contain: (116) How eager to appear to be indifferent to the pictures Pollocki painted in hisi youth do you think hei was? (117) *Hei was eager to appear to be indifferent to the pictures Pollocki painted in hisi youth. (118) *Hei appeared to be indifferent to the pictures Pollocki painted in hisi youth. (119) *Hei was indifferent to the pictures Pollocki painted in hisi youth. Surely, (116) contains at least a position for the subject of the smallest infinitival clause: (120) *[ti to be indifferent to the pictures Pollocki painted in hisi youth] But such a subject should also be disjoint from Pollock, just as it is in (119). This inconsistency arises only if it is assumed, as Heycock implicitly does, that any subjects in the phrases fronted in (114) or (116) are to be interpreted as coreferent to the subject pronoun he. If, however, the subjects within the fronted phrases are closed off by existential quantifiers from any direct relation to a higher subject, such as the pronoun he, then we can respect a disjoint reference condition between the name Pollock and the fronted phrases’ subjects, as in (121) and (122): (121) How ∃yj yj pleased with the pictures Pollocki painted in hisi youth do you think hei was? (122) How ∃yj yj eager ∃zk zk to appear ∃wl wl to be indifferent to the pictures Pollocki painted in hisi youth do you think hei was? As in (106), we only infer that Pollock and the person pleased with the pictures are the same; and it is only inferred that Pollock and the persons eager, appearing, and indifferent are all the same. The disjoint reference effect observed in (115) and (117)–(119) is due to the relation between the name Pollock and the highest subject, the pronoun he. It seems that only existential quantification in (121) and (122) can reconcile the presence of subjects internal to the fronted phrase with the facts of disjoint reference. This result then converges with what the logic of conjunction independently requires for (103).
Introduction
21
Supermonadicity is a revision to clausal architecture defended throughout the following chapters, warranted here, after the slippery slope into Conjunction Reduction, by three elementary observations: (i) that sentences with multiple conjunctions (75) demand a certain vagueness about the manner of participation in the events described, (ii) that this vagueness should not then confuse aggressors and victims (77), and (iii) that any participant’s participation may be described by an adverb that is contrary to how anyone else participated (96). Thus a few facts about natural language draw a direct line from Conjunction Reduction to supermonadicity, to a revision of clausal architecture generalizing the causative analysis of transitivity to polyadicity. For Eventish,17 the necessary vagueness about the manner of participation demands a thematic relation “participate(e,x)” (abbreviated “W(e,x)” in the following chapters), and then the precision not to confuse aggressors and victims demands resolution into smaller events so that every argument is isolated in its own event and various intereventive relations such as “Cause(e1, e2)” or “O(e1, e2)” hold the sentence together. Supermonadicity demands both that the thematic relation “participate(e,x)” apply indifferently to Agents, Themes, Patients, Goals, Locations, and so on—that is, to all arguments—and that it apply separately (thematic separation; see Schein 1993, Lohndal 2014, and Williams 2015) to the isolated event, since any token of “participate(e,x)” may itself be the occasion for an adverb that describes e and no other event referred to in the sentence. Thus, “participate(e,x)” is a primitive lexical relation of Eventish both separated in logical syntax from any other morphology and divorced from the meaning of any particular thematic relation or verb. Supermonadicity demands vagueness and generality from the meaning of “participate(e,x).” It demands little from the content of the other thematic relations and intereventive relations that populate the logical syntax: (123) Booth killed Lincoln. a. Agent(e1,x) … Cause(e1,e2) … kill(e2) … Patient(e2,y) b. Killer(e1,x) … kill(e1,e2) … Killed(e2,y) (124) Brooklyn borders Queens. a. Theme(e1,x) … O(e1,e2) … border(e2) … Location(e2,y) b. Borderer(e1,x) … border(e1,e2) … Bordered(e2,y) (125) Lucretius enjoys life. a. Experiencer(e1,x) … Entrain(e1,e2) … enjoy(e2) … Stimulus(e2,y) b. Enjoyer(e1,x) … enjoy(e1,e2) … Enjoyed(e2,y) As far as supermonadicity (or thematic separation) is concerned, it could be that each verb is an idiosyncratic collection, as in (123b)–(125b), and the thematic and intereventive relations as numerous as the verbs themselves n-fold (Schein 1993,
22
Chapter 1
85ff., 331ff. n. 2; 2002). Were the language to contain only the three verbs, kill, border, and enjoy, the argument for supermonadicity and its formal syntax would be unaffected, although such a language would no longer contain a lexicon that warrants subclassification into causative verbs, experiencer verbs, and so on, as well as the thematic relations intended to be common across the verbs of the same subclass. Supermonadicity is robust even in despair that Agent, Theme, Patient, Experiencer, Cause, etc., will ever make sense. Once this formal point about supermonadic decomposition is recognized, one should of course go on to inquire about the content of its terms. Thematic and intereventive relations that aspire to generality—Agent, Theme, Patient, Experiencer, Cause, etc.—join and enlarge vocabulary with similar aspirations—prepositions such as to and light verbs such as make, have, or let, or causative bound morphemes—which must all contend with whether or not a univocal meaning can overcome or accommodate their apparent polysemy. This question about the content of verbal morphemes as it pertains to Davidsonian semantics is addressed in Schein 2002. Here I help myself to “Agent(e,x),” “Theme(e,x),” “Patient(e,x),” “Experiencer(e,x),” “Cause(e,e′),” “O(e,e′),” etc., as much for readability opposite the likes of (123b)–(125b) as out of conviction, but not without a tendentious remark. If univocal meaning eludes these terms, it is as good a reason to reject them as it is to reject the preposition to, light verbs, or causative morphology, as these are equally elusive in their meaning. Rather, in a Davidsonian semantics, there is no daylight between, say, “Theme(e,x)” and “to(e,x).” I should first begin to worry if the former enjoyed an ease of definition or meaning denied the latter. They happen to differ in their pronunciation and are otherwise subject to the same vagaries of meaning as everything else in the vocabulary.18 1.5
Descriptive event pronouns
What it means for and not to mean much is a logical syntax with supermonadicity, adverbialization, a nominal and clausal structure augmented for spatiotemporal orientation, and plural descriptive event pronouns. This chapter surveys selectively the empirical problems encountered when and is the univocal, sentential connective and anticipates the analyses and arguments mustered in the chapters to follow for Eventish. The story, however, began simply with Aristotle’s lament that there seems to be an and of collective or divided reference different from sentential and. This section defends the answer to Aristotle: in a language with unspoken relations to events, and’s collective or divided reference is just an illusion from plural event pronouns cloaking their plural reference to events in silence. Section 1.5.0 begins with the argument that descriptive event pronouns are here to stay, whether or not they deploy to rescue and from ambiguity. Essential for any Davidsonian logical form, they are the basic device for cross-reference to events,
Introduction
23
holding the sentence together. They remain a constant feature of Davidsonian logical form throughout its evolution, including its revisions under supermonadicity and adverbialization. Any Davidsonian honest enough to reconcile event talk with natural language quantification resorts to them extensively. Supermonadicity extends their use even further: the junctural, intereventive relations introduced, such as Cause, come with a certain logical structure interpreting their complementation, ⌜[vP Cause VP]⌝, and occasion for another token of an event pronoun. These unspoken pronouns prove to be as glib as their spoken counterparts. In particular, the unspoken event pronouns holding the sentence together across its subatomic clauses shadow the expressive power and vagaries of reference that overt pronouns show across clauses in what is called descriptive anaphora: (126) Not many a professor endorses operant conditioning, and they must fill no more than a single classroom. (After Evans 1977, 1980) (127) No more than 18 barrels filled bottles with two Rhône varietals, it filling no more than 60 bottles each (bottle) with 750 mL of two Rhône varietals. The pronoun in (126) appears to stand in for the description fixed by its antecedent clause, what professors if any that endorse operant conditioning, and the pronoun in (127), for the event description what events there were if any of barrels filling bottles with two Rhône varietals. In this section it will be shown that Eventish for (128) is akin to (127), with an unspoken counterpart for the boldfaced event pronoun: (128) No more than 18 barrels filled no more than 60 bottles each (bottle) with 750 mL of two Rhône varietals. At the end of the day, whatever else is chanced, any account of descriptive anaphora will do fine, even if none is equal to capturing the brevity and grace with which simple pronouns, it, they, etc., pull it off. Rather, it is this last observation that figures large in what is an empirical rather than a logical thesis. Speakers are glib in their usage of pronouns of the slightest weight to mean all that they mean. It is on these grounds a modest emendation to logical form to replace event variables with pronouns as the basic event anaphor and thus to import into subatomic clauses equivalent expressive power. A corollary to the empirical thesis is that all of the vagaries of that usage are reflected in the subatomic usage too, as surveyed below. Finding it so then corroborates that the unspoken pronouns are where they are, even if any particular account of descriptive anaphora is rejected. Nature has no compact with us to locate problems only where they can be seen. If the nature of pronouns is elusive, the logical syntax proposed at least offers to reduce the problem of an and of collective and divided reference to the problem of descriptive anaphora.
24
Chapter 1
If, despite the event pronouns ready for the asking, Conjunction Reduction is still refused, section 1.5.1 steps forward summarizing an argument that a syntax foolish enough to deny that and conjoins sentences when conjoining DPs or Predicate Phrases—[DP[DPthe sauce] and [DPthe cheese]], [PredP[PredPladled] and [PredPsprinkled]]— will always founder on a dilemma of syntax from which there is no exit. Worse, the meaning that would be derived from an and of collective or divided reference is not, as section 1.5.2 shows, synonymous with the use of plural event pronouns. The latter prove truer to natural language. 1.5.0
Descriptive event pronouns and the Davidsonian
Prior to any thoughts about conjunction, the descriptive event pronoun proves essential for Eventish (Schein 1993, 11–15, 139–153, 179–193). On independent grounds, it must already be the basic device for cross-reference to events: (129) In perfect four-part harmony, every songstress sang solo a single melodic line. (Davies 1991; Taylor 1985) According to (129), each songstress is the solitary agent, the only voice, in her own event; and none of these, by definition, is the four-part harmony. Rather, they each sang a part: (130) [∃e: In perfect four-part harmony(e)] [every x: songstress(x)] [∃e′: ≤ e] sang(e′) solo a single melodic line. (Davies 1991; Taylor 1985) Yet, sentence (129) is false if the two songstresses sang soprano and alto in harmony with two songsters singing tenor and bass. Showing its weakness as the logical form for (129), (130) would be true. Distributive quantification is concurrent as shown with quantification over solo events, and often these events are subevents of a larger one described elsewhere in the sentence, as in (129) and (130). But, if so, the many solo events constitute in its entirety the larger one. Sentence (129) is false because the events of songstresses singing solo single melodic lines are not the perfect fourpart harmony, and yet nothing in (130) says so. The spoken descriptive event pronoun in (131) says it, managing somehow to denote exactly what singing the songstresses did, and is correctly false: (131) Every songstress sang solo a single melodic line, and it was perfect fourpart harmony. It being perfect four-part harmony, every songstress sang solo a single melodic line.
Introduction
25
An unspoken counterpart is then to be introduced into the logical form for (129): (132) a. [∃e: In perfect four-part harmony(e)] (e = thati & (i[every x: songstress(x)] ∃e′ sang solo a single melodic line[e′])); or, b. [∃e: In perfect four-part harmony(e)] (O(e, thati) & (i[every x: songstress(x)] ∃e′ sang solo a single melodic line[e′])) The reasoning that enlists a descriptive pronoun to mediate between adverbial description and host sentence extends to thematic relations (essential separation; see Schein 1993, 8–11, 57–84, 146–150; Schein 2006, 751–752) and generally to any juncture between event descriptions, whether the decomposition is supermonadic ((b) examples) or not ((a) examples): (133) Eighteen barrels filled 5400 bottles each with 750 mL of at least two Rhône varietals. a. ∃e ([∃X: 18 barrels[X]] Source(e,X) & e = thati & (i[∃X: 5400 bottles[X]][each x: Xx] ∃e′ filled with at least two Rhône varietals[e′])) b. ∃e ([∃X: 18 barrels[X]] Source(e,X) & Cause(e, thati) & (i[∃X: 5400 bottles[X]] [each x: Xx] ∃e′ filled with at least two Rhône varietals[e′])) (134) F One hundred eighty barrels filled 5400 bottles each with 750 mL of at least two Rhône varietals. a. F ∃e ([∃X: 18 barrels[X]] Source(e,X) & e = thati & (i[∃X: 5400 bottles[X]] [each x: Xx] ∃e′ filled with at least two Rhône varietals[e′])) b. F ∃e ([∃X: 18 barrels[X]] Source(e,X) & Cause(e, thati) & (i[∃X: 5400 bottles[X]] [each x: Xx] ∃e′ filled with at least two Rhône varietals[e′])) (135) T ∃e ([∃X: 18 French barrels[X]] Source(e,X) & [∃X: 5400 bottles[X]] [each x: Xx][∃e′: e′≤ e] filled with at least two Rhône varietals[e′]) Sentence (133) is read as saying that each of 5400 bottles was filled as described, the supply coming from a total of 18 barrels. Eighteen 225 L barrels is 5400 750 mL bottles. In (133)–(134) and all the following examples, each is distributive for the direct object, 5400 bottles.19 In a context verifying (133), the producer has blended 5400 bottles of various Rhône blends from 18 assorted barrels of Grenache, Syrah, Mourvèdre, Cinsault, and Carignan. But (134) is false by an order of magnitude in relating the contents of 180 barrels to the 5400 bottles, despite a vintage in which the 5400 bottles were each filled as part (135) of a larger production in which 180
26
Chapter 1
barrels filled 54,000 bottles. Absent the pronoun thati with the same reference as it in (136), (135) mistakes (134) for true. (136) 5400 bottles were each filled with 750 mL of at least two Rhône varietals. It took 18 barrels. (It) taking 18 barrels, 5400 bottles were each filled with 750 mL of at least two Rhône varietals. Descriptive event pronouns must be introduced just to reconcile Davidsonian event talk with the distributive, first-order quantification that leads immediately to talk about subevents of individual participation. Thus event pronouns litter the Davidsonian landscape.20,21 An argument in two steps next shows that logical forms like (132) and (133), with only the one event pronoun, do not have enough of them to represent the intended meaning when nonincreasing, distributive quantification substitutes for the universal: (137) In perfect four-part harmony, not many a singer sang even a single note off-key. (138) [∃e: In perfect four-part harmony(e)] (O(e, thati) & (i[not many a x: singer(x)] ∃e′ sang solo a single note off-key[e′])) (139) Eighteen barrels filled no more than 5400 bottles each with 750 mL of at least two Rhône varietals. (140) a. ∃e ([∃X: 18 barrels[X]] Source(e,X) & e = thati & (i[No X: more than 5400[X] & bottles[X]] [each x: Xx] ∃e′ filled with at least two Rhône varietals[e′])) b. ∃e ([∃X: 18 barrels[X]] Source(e,X) & Cause(e, thati) & (i[No X: more than 5400[X] & bottles[X]] [each x: Xx] ∃e′ filled with at least two Rhône varietals[e′])) Note that in the context described, with 18 barrels filling exactly 5400 bottles, (139) is also true. In contrast, (141), in this context, is doubly false:22 (141) No more than 5400 bottles were each filled with 750 mL of at least two Rhône varietals, and it took 18 barrels. Tenfold as many bottles were produced, and it took 180 barrels to fill them. Since (140) says the same thing as (141), it cannot be the correct logical form for the true (139). What is missing from (140), which is the first step in the argument, is the relation of the individual bottle’s content to its source in the 18 barrels. Ignoring super-
Introduction
27
monadicity for the moment (see (150) and (151)), what is missing relates the individual bottle to a part of the event in which the barrels are sources: (142) ∃e ([∃X: 18 barrels[X]] Source(e,X) & e = thati & (i [No X: more than 5400[X] & bottles[X]] [each x: Xx] [∃e′: e′ ≤ e] filled. with at least two Rhône varietals[e′])) There was an event of 18 barrels filling bottles, and no more than 5400 bottles were each filled in that, exempting bottles filled from other barrels. A sentence in which the distributive quantification is both nonincreasing and nondecreasing, exactly 5400 bottles, illustrates in a single sentence truth conditions requiring both event pronoun thati and the quantification over subevents “[∃e′: e′ ≤ e],” combining the considerations sketched above: (143) Eighteen barrels filled exactly 5400 bottles each with 750 mL of at least two Rhône varietals. Having of necessity reintroduced “[∃e′: e′ ≤ e],” the first step, the second step of the argument is to show that the resulting logical form is again too weak, unless yet another event pronoun is introduced. Suppose that the 18 barrels empty at an equal flow rate, 50 mL/sec per barrel. At any point, the process may be interrupted to count the bottles filled so far, and given the regulated flow rate, it can be asserted with confidence that the bottles filled to this point have been filled from 18 and no fewer than the 18 barrels. With no such interruption in mind and speaking after production has ended, (144) is unequivocally false: (144) F Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. The 18 barrels filled 5400 bottles, and it took 75 minutes, during which 60 bottles are filled in 50 seconds. Yet the purported logical form for (144) is true of any interval of the 18 barrels’ pour lasting 50 seconds or less: (145) ∃e ([∃X: 18 barrels[X]] Source(e,X) & e = thati & (i[No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′ ≤ e] filled with at least two Rhône varietals[e′])) The 18 barrels are all sources during that interval and no other is, and no more than 60 bottles are filled in any part of it. Events verifying (145) must exist, since these same 18 barrels, it was remarked, do fill 60 bottles in 50 seconds, so that (146) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals.
28
Chapter 1
An event (146) is true of, (145) is also true of. As a logical form for (144), the initial existential event quantifier in (145), with scope over the entire sentence, and the subsequent quantification over its subevents would undermine the force of the nonincreasing quantifier no more than 60 bottles. The only recourse to represent (144)’s meaning is to deny the existential event quantifier “∃e” wide-scope, falling back on an event pronoun, singular or plural (or mass) to relate the bottles to their source: (147) (j[∃X: 18 barrels[X]] ∃e Source(e,X)) & thatj = thati & (i[No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′ ≤ thatj] filled with at least two Rhône varietals[e’])) (148) Eighteen barrels filled bottles with Rhône varietals, (it) filling no more than 60 bottles each with at least two Rhône varietals. Eighteen barrels filled bottles, and it was no more than 60 bottles each being filled with at least two Rhône varietals. Eighteen barrels filled bottles, and there/in that no more than 60 bottles were each filled with at least two Rhône varietals. Eighteen barrels filled bottles, and no more than 60 bottles were each filled (then and there) with at least two Rhône varietals. For any 18 barrels, the pronoun, spoken or unspoken in (147) and (148), manages to refer to their bottle filling and not to that by any other barrels. There is no surprise here, since the subject “[∃X: 18 barrels[X]]” includes within its scope the entire sentence and thus the pronoun thati denotes what they X did, “[℩E: Source[E,X]].” But, crucially, it refers to their entire pour, all 75 minutes for the 18 barrels under discussion, and never to any fraction thereof. After all, how could definite reference to their bottle filling, their event(s) of bottle filling, that or it, be reference to 50 seconds or less? Which 50 seconds or less (Evans 1977, 130)? The only definite reference here with any purchase is to the maximal event or events of filling bottles for 75 minutes. Of course, it is false that that was of no more than 60 bottles, 5400 having been filled, exactly as desired. It had better be then that the logical form for (144) does not place it entirely within the scope of an existential event quantifier, so that any reference to the barrels’ bottle filling is by descriptive event pronoun, inducing conveniently a failure of definite reference if anything less than all their bottle filling is intended. This argument for two event pronouns, thatj and thati in (147), precedes consideration of coordinate structures and, as evident from its absence in (147), it also precedes supermonadicity,23 which itself was an argument for multiple event pronouns. Here, even a more genteel Davidsonian starting from a more modest decomposition, when faced with the facts of distributive quantification in natural language,
Introduction
29
must concede that event pronouns come with the territory. It can then be no surprise that these pronouns recur with coordinate structures for their antecedents. The argument for two event pronouns repeats itself mutatis mutandis with supermonadicity on board, leading to a supermonadic counterpart to (147) with a junctural, intereventive relation (see note 19) joining the distinct events of source-ing, the 18 barrels’ pour, and the filling up within the bottles: (149) (j[∃X: 18 barrels[X]] ∃e Source(e,X)) & Cause(thatj, thati) & (i[No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′ ≤ thatj] filled with at least two Rhône varietals[e′])) But, now something is amiss. The individual bottle’s filling up is not, as “e′ ≤ thatj” says, part of the barrels’ pour. It should rather read that as a result of the 18 barrels’ filling bottles no more than 5400 bottles were each filled in a subevent of its effect. Under supermonadicity, the sentence no longer describes some superordinate event that the subevents supermonadic clauses describe are part of.24 It is rather that the subevents demanded for distributivity, “… [each x: Xx][∃e′: e′≤ e″] …,” are subevents of the events determined by the matrix verb Cause to be described by its complement. It is the logical structure of complementation,25 I will assume, to say as much: The logical structure of complementation (150) a. [vP Cause VP] ⇒ [vP Cause(ei, ej) & [℩ej: Cause(ei, ej)]VP] b. [vP Cause VP] ⇒ [vP Cause[Ei, Ej] [℩Ej: Cause[Ei, Ej]]VP]26 To Cause-VP is to cause some events and these events are the effects that VP. The supermonadic counterparts to (147) are in (151), with (151b) closer to official Eventish: (151) a. (j[∃X: 18 barrels[X]] ∃e Source(e,X)) & Cause(thatj, thati) & (i[℩e″: Cause(thatj, e”)][No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′ ≤ e″] filled with at least two Rhône varietals[e′])) b. (j[∃X: 18 barrels[X]] ∃E Source(E,X)) [℩Ej: proj][℩Ei: proi]Cause[Ej, Ei] [℩Ej: proj](i[℩Ei: Cause[Ej, Ei]][No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: Eie′] filled with at least two Rhône varietals[e′])) With or without supermonadicity, it has been argued that the logical form for (144) conceals two event pronouns: (1) a forward anaphor, thatj or [℩Ej: proj] in (147) and (151), describing the barrels’ filling bottles, their pour; and (2) a backward anaphor thati or [℩Ei: proi] describing the bottles’ filling up. In (144), it happens that the antecedent for the forward anaphor is a subatomic clause in the scope of an (in)definite description, 18 barrels, (j[∃X: 18 barrels[X]] …), and the backward anaphor is anaphoric to a clause in the scope of a decreasing quantifier, no more than 60 bottles, (i… [No X: more than 60[X] & bottles[X]] …). The logical form of sentences (152)
30
Chapter 1
is expected to be the same as in (147) or (151) except for an exchange of (in)definite description and decreasing quantifier: (152) a. Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. b. Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. c. No more than 18 barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. d. No more than 18 barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. But it is across the full paradigm that the two event pronouns come to show the vagaries of reference shared with overt descriptive anaphora, both with their spoken counterparts occurring in paraphrases as in (153) and with anaphora across clauses in general, (154): (153) No more than 18 barrels filled bottles with two Rhône varietals, it filling no more than 60 bottles each (bottle) with 750 mL of two Rhône varietals. (154) Not many a professor endorses operant conditioning, and they must fill no more than a single classroom. (After Evans 1977, 1980) Essential to this observation, the problem of nonmaximal reference distinguishes anaphora to (in)definite descriptions from anaphora to other quantificational antecedents:27 (155) Two men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. (After Donnellan 1978) (156) No more than two men came to the office today. They tried to sell encyclopedias. (*Perhaps there were even others who did the same.) If they in both (155) and (156) is the description “the men who came to the office today,” why and how—so the problem goes—does they in (155) differ from (156) in referring to not all the men who came to the office today? (157) Two men came to the office today. These (two) men who came to the office today tried to sell encyclopedias. Perhaps there were even others who did the same. (158) *No more than two men came to the office today. These (no more than two) men who came to the office today tried to sell encyclopedias. Perhaps there were even others who did the same.
Introduction
31
The definite descriptions in (157) and (158) differ the same way, and thus it would gain no purchase on the problem to suppose that the pronoun in (155) is a bare variable. It could be that the pronouns in (157) and (158) have as much descriptive content as the definite descriptions in (157) and (158), are as much logical definite descriptions (i.e., with the iota-operator) as they are, and are as similar in structure to each other as they are. It is plain that the definite description in (158) is a logical, plural one with an iota-operator,28 and therefore so is the definite description in (157) in any solution to the problem of nonmaximal reference that purports to solve it rather than rename it. Any such solution will apply without amendment to the pronouns, and pending such, the existence of the problem is no argument against the uniform treatment of pronouns as descriptive, definite descriptions.29 Note that in (159)–(160) the problem of nonmaximal reference iterates: (159) Two men came to the office today. They sold two encyclopedias. These were discounted. Perhaps they sold more. (160) Two men came to the office today. They sold no more than two encyclopedias. These were discounted. (*Perhaps they sold more.) In both (159) and (160), the first pronoun they is nonmaximal in its reference to two of the men coming to the office today. In (159), given an (in)definite description as its antecedent, the demonstrative these is also nonmaximal in its reference to two of the encyclopedias that these two men coming to the office today sold. But the decreasing quantifier that is the antecedent in (160) forces the demonstrative to refer maximally to all the encyclopedias the two sold. The problem of nonmaximal reference recurs among the unspoken event pronouns of Eventish and their spoken counterparts, a good thing if they are all the same. Consider first the backward pronoun thati in the logical form (162) for (161) (the forward thatj is taken up shortly): (161) F Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals (in no more than 50 seconds). (162) [∃X: 18 barrels[X]] ((j [∃e Source(e,X)) & Cause(thatj, thati) & (i[No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′≤ thatj] filled with at least two Rhône varietals[e′]))) It denotes maximally all the bottles’ filling from the 18 barrels, which was of 5400 bottles in 75 minutes. Yet these same 18 barrels, it was remarked, do fill 60 bottles in 50 seconds, so that (163) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals, in no more than 50 seconds.
32
Chapter 1
Assuming the same structure for (163), thati in (164) must denote nonmaximally just some 60 bottles filling and not the filling of all 5400 that the 18 barrels filled: (164) [∃X: 18 barrels[X]] ((j [∃e Source(e,X)) & Cause(thatj, thati) & (i[∃ X: 60 bottles[X]] [each x: Xx] [∃e′: e′≤ thatj] filled with at least two Rhône varietals[e′]))) Spoken event pronouns show the same variation, backward too:30 (165) a. i. F With 18 barrels as the source, it lasting no more than 50 seconds, no more than 60 bottles filled each with 750 mL of at least two Rhône varietals. ii. F For some 18 barrels, it lasting no more than 50 seconds, no more than 60 bottles were filled each with 750 mL of at least two Rhône varietals. b. F No more than 60 bottles filled from them each with 750 mL of at least two Rhône varietals, 18 barrels did it and it was in no more than 50 seconds. c. F With no more than 60 bottles filling from them each with 750 mL of at least two Rhône varietals, 18 barrels did it and it was in no more than 50 seconds. (166) a. i. With 18 barrels as the source, it lasting no more than 50 seconds, 60 bottles filled each with 750 mL of at least two Rhône varietals. ii. For some 18 barrels, it lasting no more than 50 seconds, 60 bottles were filled each with 750 mL of at least two Rhône varietals. b. Sixty bottles filled from them each with 750 mL of at least two Rhône varietals, 18 barrels did it and it was in no more than 50 seconds. c. With 60 bottles filling from them each with 750 mL of at least two Rhône varietals, 18 barrels did it and it was in no more than 50 seconds. So it is for the unspoken event pronouns and those spoken in (165)–(166) that their reference is nonmaximal only when the antecedent description is about 60 bottles, an (in)definite description, rather than no more than 60 bottles. Like all pronouns and definite descriptions caught up in the problem of nonmaximal reference, they betray no outward sign of varying their reference, and yet they do. If the empirical thesis is sound, that descriptive event anaphora are the basic device in logical form for cross-reference to events, then the unspoken anaphora should be just as mercurial as their spoken counterparts. Any solution to the problem of nonmaximal reference should do, applying equally to descriptive anaphora spoken or unspoken, referring to objects or events. To underscore its detachment from the present argument, what I say at some length about nonmaximal reference
Introduction
33
is exiled to appendix 1, “Descriptive Anaphora and Nonmaximal Reference under Selective Perspectives.” 1.5.0.0 Selective perspectives for nonmaximal reference
But what has already been observed above in arguing for the two event pronouns does constrain approaches to nonmaximal reference as applied to the forward thatj in (164), referring to 18 barrels’ pour—constraint congenial to the use of selective perspective to derive nonmaximal reference (Schein 1993, 219–237). Recall that (134) is false, with only a tenth of the 180 barrels as sources for 5400 bottles, and a fortiori (167) is false, with even less room for wine from 180 barrels: (134) F One hundred eighty barrels filled 5400 bottles each with 750 mL of at least two Rhône varietals. (167) F One hundred eighty barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. (144) F Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. (146) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. The contrast between (146) and (167) now becomes a puzzle. With only a very small, proper fraction, .11% of the pour filling 60 bottles, (167) is false; and also with a rather small proper fraction, 1.11% of the pour filling 60 bottles, (146) is nevertheless true. The contrast between (146) and (167), for example, would be unremarkable if their logical forms were respectively (168) and (169), in that the fraction of the 18 barrels’ pour filling 60 bottles is an event to which 18 bottles all contribute in some measure: (168) ∃e([∃X: 18 barrels[X]] Source(e,X) & filled 60 bottles each with 750 mL of at least two Rhône varietals[e]) (169) F ∃e([∃X: 180 barrels[X]] Source(e,X) & filled 60 bottles each with 750 mL of at least two Rhône varietals[e]) Of the 180 bottles, 162 contribute nothing at all, and thus there is no filling 60 bottles for which they are all sources. But such logical forms have already been rejected in denying (144) the logical form (145): (145) ∃e ([∃X: 18 barrels[X]] Source(e,X) & e = thati & (i[No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′ ≤ e] filled with at least two Rhône varietals[e′]))
34
Chapter 1
Assuming instead that (167) and (146) have the same structure as (144) (=(147)) delivers (170) and (171): (170) F [∃X: 180 barrels[X]] ((j [∃e Source(e,X)) & thatj = thati & (i[∃ X: 60 bottles[X]] [each x: Xx] [∃e′: e′≤ thatj] filled with at least two Rhône varietals[e′]))) (147) F (j[∃X: 18 barrels[X]] ∃e Source(e,X)) & thatj = thati & (i[No X: more than 60[X] & bottles[X]] [each x: Xx] [∃e′: e′≤ thatj] filled with at least two Rhône varietals[e′])) (171) [∃X: 18 barrels[X]] ((j [∃e Source(e,X)) & thatj = thati & (i[∃ X: 60 bottles[X]] [each x: Xx] [∃e′: e′≤ thatj] filled with at least two Rhône varietals[e′]))) The logical forms (170), (147), and (171) are identical except for the substitutions between 180 barrels and 18 barrels and between no more than 60 bottles and 60 bottles. As remarked, thatj in (147) denotes for some 18 barrels, their entire pour, which fills 5400 bottles and falsifies (147) and (144), as desired. In (171), thatj thus also denotes the entire pour from some 18 barrels, and in (170), the entire pour from some 180 barrels. The ensuing and desired falsehood of (170) and (167) is expected, since plainly the pour from 180 barrels ( = 54,000 bottles) does not amount to filling 60 bottles (“thatj = thati”) nor is that the pour that causes 60 bottles to fill, anticipating the supermonadic relation, “Cause(thatj, thati).” It is rather the supposed truth of (171) that offends stark and robust intuitions about event coincidence and event causation. We know exactly how much flow results in how much filling up, measure by measure, and that fraction of the 18 barrels filling 5340 bottles elsewhere had no part in filling 60 bottles. The grounds for rejecting the truth of (170) and (167) apply equally to (171), and yet (146) is judged true. Intuitions about events and the contrast between (146) and (167) would all be reconciled if (168) and (169) were their logical forms, and so in excluding them, it is fair to wonder if the apparent puzzle is more than an injury self-inflicted by the regimentation in (170), (147), and (171). The puzzle is however real enough to be found among spoken pronouns too, as in (172)–(174): (172) F One hundred eighty barrels filled bottles with Rhône varietals, (it) filling 60 bottles each with 750 mL of at least two Rhône varietals. (173) F Eighteen barrels filled bottles with Rhône varietals, (it) filling no more than 60 bottles each with 750 mL of at least two Rhône varietals. (174) Eighteen barrels filled bottles with Rhône varietals, (it) filling 60 bottles each with 750 mL of at least two Rhône varietals, (in no more than 50 seconds).
Introduction
35
The gerundive clauses in (172)–(174) are closest to wearing on their sleeve the logical forms proposed for (167), (144), and (146)—serial, nonfinite, subatomic clauses introduced with pronominal reference to antecedent events. The event pronouns, spoken and unspoken, appear to have as their antecedent the same clause, 18(0) barrels filled bottles with Rhône varietals, and they refer unambiguously for some 18(0) barrels to their filling bottles with Rhône varietals, which in the context under discussion amounts to 5400(0) bottles. As expected, (173) is false in claiming there were no more than 60, and (172) and (174) contrast the same way as (167) and (146). Whatever puzzle resides in the interpretation of event pronoun and gerundive clause, there is no denying the pronoun or its definite reference for some 18(0) barrels to the filling of theirs of bottles with Rhône varietals—that is, their entire pour on this occasion. Existential event quantification is precluded from including within its scope the entire sentence in (144) and (173), lest a proper fraction of their pour, enduring no more than 50 seconds, make these sentences mistakenly true. Instead, a descriptive event pronoun refers maximally for some 18 barrels to whatever they did as sources. With the parity of structure assumed for (167), (144), and (146), and in plain sight for (172)–(174), it carries over that wide-scope existential event quantification is precluded and replaced with a descriptive event pronoun throughout. With the loss of (168) and (169) to represent them, the contrast between (146) and (167) indeed becomes a puzzle, but one that in any case must be confronted in the contrast between (172) and (174). Again, that the unspoken event pronouns behave like their overt counterparts subject to the same puzzle supports their conjectured presence, prior to the puzzle’s resolution.31 The suggestion in appendix 1 and in Schein 1993, 219–237, is that nonmaximal reference derives from a narrowed perspective Π on the spatiotemporal context, so that (173) and (174) are effectively glossed: (175) F Eighteen barrels filled bottles with Rhône varietals, (it) filling no more than 60 bottles each with 750 mL of at least two Rhône varietals. (176) Eighteen barrels filled bottles with Rhône varietals, zooming in to Π (it = their filling bottles under Π) filling 60 bottles each with 750 mL of at least two Rhône varietals, (in no more than 50 seconds). (177) F 180 barrels filled bottles with Rhône varietals, zooming in to Π (it = their filling bottles under Π) filling 60 bottles each with 750 mL of at least two Rhône varietals, (in no more than 50 seconds). The event pronoun in (176) is a nonrigid definite description localized to Π, which allows it to refer to a fraction of the 18 barrels’ pour, said subsequently to fill 60
36
Chapter 1
bottles. That fraction of 18 barrels’ pour that is a filling of 60 bottles is an event they all participate in as sources, as the nonrigid definite description describes it. In welcome contrast, note that despite perspective, it remains false that 180 barrels filled 60 bottles although 18 of them did just that. There are no events under any perspective of them the 180 filling bottles that fill only 60 bottles, since there are not 60 bottles to which all 180 contributed. Next, in contrast to (176), to derive the absence of nonmaximal reference in (175), some justification is needed and provided that precludes a selective perspective in (175) from confining the interpretation of the decreasing quantifier no more than 60 bottles. It begins with the observation that selective perspective threatens to trivialize decreasing quantification, since there is always a perspective narrow enough so that nothing in it happens to more than 60 bottles. In Eventish, the logical forms for (144) and (150), the counterparts for (173) and (174), are (144) Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. (178) (j[∃X: 18 barrels[X]] ∃E Source(E,X)) [℩Ej: proj][℩Ei: proi]Cause[Ej, Ei] [℩Ej: proj](i[℩Ei: Cause[Ej, Ei]][No X: 60 bottles[X]] [each x: Xx] [∃e′: Eie′] filled with at least two Rhône varietals[e′])) (146) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. (179) (j[∃X: 18 barrels[X]] ∃E Source(E,X)) ∃Π[℩Ej: proj Π][℩Ei: proi]Cause[Ej, Ei] [℩Ej: proj](i[℩Ei: Cause[Ej, Ei]][∃X: 60 bottles[X]] [each x: Xx] [∃e’: Eie′] filled with at least two Rhône varietals[e′])) ‘Eighteen barrels sourced; in some of what was happening, their source-ing there had this effect: (no more than) 60 bottles fill from their source-ing each with 750 mL of at least two Rhône varietals.’ If a selective perspective is the correct approach to nonmaximal reference and it is correct to find here a token of an unspoken event pronoun interpreted nonrigidly, therein is further argument on behalf of descriptive anaphora rather than bare event variables, which of course afford no such possibility. The logical forms (178) and (179) are analysis—provisional and arbitrary in its details—of the empirical claim that (144) and (146) deploy unspoken event pronouns in a way that closely tracks (173) and (174). What (178) and (179) accomplish is that they render (144) and (146) with the same logical forms except for the substitution of 60 bottles for no more than 60 bottles and the contrast in selective perspective that distinguishes (in)definite description from decreasing quantification.
Introduction
1.5.1
37
Reference to events and the syntax of descriptive anaphora
Rejecting (181), translation from the natural language deploys more than existential quantifiers and variables to close off the argument positions that supermonadicity opens. (180) Saul and David have been swapping wine futures in online trading. (181) * ∃e1∃e2(Participate(e1,x) & Participate(e1,y) & Cause(e1,e2) & swap(e2) & Participate(e2,z)) To start, (96) (repeated in (182)) already requires a second-order definite description, signaled by “that1,3” in (98), to divide its reference between Saul’s and David’s events in referring to what causes the wine futures swap. (182) Saul from early morning in New Jersey and David later in the day in California have been swapping wine futures in online trading. (183) ∃e1∃e2∃e3(Participate(e1,x) & a.m.(e1) & in(e1, NJ) Participate(e3,y) & p.m.(e3) & in(e3, CA) & [℩E: Ee1 & Ee3]Cause(E, e2) & swap(e2) & Participate(e2,z))32 Yet (183) in its details should not be accepted in haste. Phrases embedded within the conjuncts—the adverbs and tokens of participate—occasion mention of e1 and e3; (183) makes reference to these events dependent on quantifiers external to the coordination, detached and remote from the linguistic content about them, and including within their scope all of (183). Perfecting the analogy between (182) and (184) or (185), the subatomic pronoun in (183) referring to the cause should be taken to fall within the scope of its antecedent quantifiers just in case the same is to be said about it in (184) or (185) and its antecedents: (184) Saul traded early morning in New Jersey, and David traded later in the day in California, and it was them swapping wine futures online. (185) Saul traded early morning in New Jersey. David traded later in the day in California. It was them swapping wine futures online. As remarked in section 1.5.0, the unspoken pronouns populating logical form are as glib as their spoken counterparts, including they in (186), which manages its reference without falling within the scope of either antecedent quantifier: (186) Not many a professor endorses operant conditioning, and not many a student does either, but they must all be behaviorist rats. This is a familiar constraint (Evans 1980) imposed on pronouns with decreasing quantifiers for antecedents, and it carries over to subatomic event pronouns. The analysis that promises in the logical form for (187) Right-Node Raising and
38
Chapter 1
Conjunction Reduction also must in referring to events antecedently described manage it without it falling within the scope of any antecedent event quantifiers— certainly not an existential event quantifier when neither a Columbia nor a Harvard student is guilty of participating in anything at all: (187) No Columbia student and no Harvard student collaborated to overthrow the government. ‘No Columbia student participated in it and no Harvard student collaborated in that, whatever they did, in collaborating to overthrow the government.’ Without an analysis of collectivizing Right-Node Raising (see chapter 2) yet explicit about its treatment of decreasing quantifiers, the argument can still be made that subatomic event pronouns track their spoken counterparts, comparing the synonymous (188) and (189): (188) Biff did something and possibly Tiff did something, and it started the riot. (189) Biff and possibly Tiff started the riot. (190) ∃e1Participate(e1,x) and Possibly ∃e3Participate(e3,y) & [℩E: pro1,3] Cause(E, e2) & start(e2) & Participate(e2,z)) What does it refer to in (188)? If Tiff didn’t do anything, then it refers to just what Biff did. If Tiff did do something, then it refers to just what Biff and Tiff did and not to what anyone else did. The unspoken pronoun in (190) does the same. The conditional nature of these pronouns puts their translation beyond reach of a purely logical vocabulary, ruling out anything like ‘[℩E: Ee1 & Ee3]’ in (183) and implicating some further descriptive content—the events if any that Biff participated in and Tiff participated in—either in direct translation or off in a pragmatics for the resolution of pronouns if that approach is favored. Again, as remarked in section 1.5.0, any account of the pronouns’ reference in this and similar contexts will do fine, even if none emulates the simplicity of the pronouns it, they, etc., themselves. In the logical form (193) for (75) (repeated in (192)), the unspoken, subatomic event anaphora ((193ii)) run backward and forward as their spoken counterparts do in (191): (191) An ethnic slur triggering it, Biff shouted obscenities and possibly Tiff shoved an officer of the peace, and it caused the DC riots. (192) The Columbia students and the Harvard students surrounded the Pentagon and were crowded into the Mall.
Introduction
39
(193) [the X: C(X)] ∃e1 Participate(e1,X) and [the Y: H(Y)] ∃e2 Participate(e2,Y) & (i) [℩E: pro1,2][℩E′: pro3,4] O(E, E′) & (ii) ∃e3(∃Z Agent(e3,Z) & surround …) and ∃e4(∃Z Theme(e4,Z) & … crowded …) (iii) ‘The Columbia students did something and the Harvard students did something; this was that; there was some surrounding the Pentagon and some being crowded into the Mall.’ The forward event pronoun refers plurally to what the Columbia students and Harvard students did, and the backward event pronoun refers to the events at the Pentagon and on the Mall. The only relation conveyed between student and venue is a relation between what these pronouns denote, vague as to the location of any one student at either venue. In (193), the sentential connectives and conjoin sentences merely asserting the existence of some events as described. The first and does not itself compose a phrase referring to students or to their events; the second and does not compose from predicates a predicate that divides anything it denotes between a surrounding the Pentagon and a being crowded into the Mall. Within the conjunctions of sentences and means “&”; the illusions to the contrary betray the presence of event pronouns, as (193) illustrates, referring plurally from positions outside the conjunctions that deliver them their antecedents. This error discovered in section 1.1 in a comparison with (11)–(25), collectivized Right-Node Raising that cannot be mistaken for DP-coordination, resurfaces in a syntactic dilemma for the misparsed structures themselves: (194) No more than a single senator and no more than a single congressman from his home state have cosponsored an antiwar resolution. (195) Not many a professor and not many a student of hers are conspiring to revive operant conditioning. If the subject of (194) or (195) is a conjunction of DPs, how is it to be composed so that the first DP excludes from within its scope the second, as the meaning requires, even as the second contains a pronoun covariant with quantification in the first? If, instead, (194) and (195) conceal Conjunction Reduction, it can be anticipated that this pattern of quantifier scope and pronominal covariance will share an analysis with its more full-blooded counterparts in (196) and (197) (see chapter 2): (196) No more than a single senator sponsored an antiwar resolution, and no more than a single congressman from his home state helped him; and they were junior.
40
Chapter 1
(197) Not many a professor endorses operant conditioning, and not many a student of hers does either, but they must all be behaviorist rats. The dilemma for a conjunction of DPs is yet more perplexing. Whatever is to be said for it, it must support combination with both singular, distributive predicates (198) and plural predicates (199): (198) No rocker and no roller has ever grooved to a funky disco beat. (199) a. No rocker and no roller have ever each grooved to a funky disco beat. b. No rocker and no roller have ever grooved together to a funky disco beat. But if, deep within that conjunction of DPs, the second contains a pronoun covariant with the first ((200), (201)) or is otherwise dependent on it in falling within its scope ((202), (203)), the derived Determiner Phrase no longer combines with singular predicates: (200) *No rocker and none of his mamas has ever grooved to a funky disco beat. (201) No rocker and none of his mamas have ever grooved to a funky disco beat. (202) *No rocker and any roller has ever grooved to a funky disco beat. (203) No rocker and any roller have ever grooved to a funky disco beat. Shouldn’t it perplex Aristotle that number agreement between predicate and what he takes to be its subject in (198)–(203) turns out to be dependent on the grammar internal to his subject, as that would seem to violate, especially if number agreement reflects semantic reference, cherished ideas about compositionality and meaning? It will be seen in sections 2.1.2 and 2.3.1 for (200) and (201) as well as section 3.2.2 for (202) and (203) how Conjunction Reduction does better with these curious facts while conforming to the ordinary ways scope, anaphora, and number agreement interact. 1.5.2
Event pronouns and modal insubordination
Introducing possibly into the conjunctions in (204) and (205) launches an argument that collective reference in (204) and divided reference in (205) is all a trick of unspoken event pronouns on the outside: (204) The Columbia students and possibly the Harvard students formed the unbroken chain around the Pentagon. (205) Three Harvard students cooked up the Harvard beets and possibly garnished them with pearl onions. Were it otherwise, (204) might be said to contain a phrase the Columbia students and possibly the Harvard students referring to those who are the Columbia students
Introduction
41
and possibly the Harvard students, of whom it is then said as in (206) that they formed the unbroken chain around the Pentagon: (206) [℩X : ∃X1∃X2 (∀x(Xx ↔ (X1x ∨ X2x)) & [the X: C(X)]∀x(X1x ↔ Xx) & possibly[the X: H(X)] ∀x(X2x ↔ Xx))] ∃E formed(E,X) the unbroken chain around the Pentagon ‘Those who are the Columbia students and possibly the Harvard students formed the unbroken chain around the Pentagon.’ (In (206), and disappears under translation of its complex meaning into familiar logical language.) Alternatively, the Columbia students and possibly the Harvard students might be taken as an indefinite description of some who are the Columbia students and possibly the Harvard students as in (207). In either case, no event pronoun or any other mediates between (in)definite description and predication. (207) [∃X : ∃X1∃X2 (∀x(Xx ↔ (X1x ∨ X2x)) & [the X: C(X)]∀x(X1x ↔ Xx) & possibly[the X: H(X)] ∀x(X2x ↔ Xx))] ∃E formed(E,X) the unbroken chain around the Pentagon ‘Some who are the Columbia students and possibly the Harvard students formed the unbroken chain around the Pentagon.’ The argument for an event pronoun will apply a fortiori to these now distant analyses. After the slippery slope to Conjunction Reduction (section 1.1), let it be granted what reflection on the meaning of (204) confirms, that the possibility entertained is whether the Harvard students participated in the unbroken chain: (208) ∃E∃e1∃e2 (∀e(Ee ↔ (Ee1∨ Ee2)) & [the X: C(X)]Participate(e1,X) & possibly[the Y: H(Y)] Participate(e2,Y)) & formed(E) the unbroken chain around the Pentagon ‘There are some events in some of which the Columbia students participated and in some of which it is possible the Harvard students participated, and those events are the forming of the unbroken chain around the Pentagon.’ Given the event talk and plural reference to events that the Columbia students participated in with possibly the Harvard students, the question is whether an event pronoun intervenes in the assertion that they were a forming of the unbroken chain as in (209) (cf. (193))—or not, as in (208).33 (209) ∃Eloc([the X: C(X)][∃e1: Eloce1]Participate(e1,X) and Possibly[the Y: H(Y)][∃e2: Eloce2] Participate(e2,Y) & [℩E: pro1,2] formed(E) the unbroken chain around the Pentagon) ‘Somewhere the Columbia students participated in something and possibly the Harvard students participated in something there too, & it was the forming of the unbroken chain around the Pentagon.’
42
Chapter 1
Within the setting of supermonadicity, is it event pronouns that effect plural reference from outside the conjunctions describing the events referred to as in (209), or is a meaning of and to be contrived so that the Columbia students and possibly the Harvard students emerges as an expression of plural reference, resulting in something equivalent to (208)? The same question attaches to (205) (repeated in (210)): (210) Three Harvard students cooked up the Harvard beets and possibly garnished them with pearl onions. Is cook up the Harvard beets and possibly garnish them with pearl onions a Predicate Phrase that denotes some events just in case they divide between those of some cooking up the Harvard beets and those of possibly garnishing them with pearl onions, or a phrase that denotes some students just in case they divide between agents of the one kind of event and agents of the other, or a phrase perhaps that denotes pairs of students and events just in case they can be divided accordingly? Or is it rather that this phrase is just the conjunction of sentences asserting that there are events of cooking up the Harvard beets and possibly some events of garnishing them with pearl onions, with event pronouns doing the rest? The argument for event pronouns in the translations of (204) and (205) concerns a restriction on the epistemic possibilities that possibly entertains. Consider two sources of doubt that might lead a speaker to hedge her assertion in (204) (repeated in (211)) that the Harvard students participated to say only that it was possibly so, as in (212): (211) The Columbia students and possibly the Harvard students formed the unbroken chain around the Pentagon. (212) It is possible that the Harvard students participated in it. For the first scenario, suppose that the SDS organizers have planned that the Columbia students will surround the Pentagon and only if they are shorthanded will the Harvard students join them. One of the Columbia students is to signal when the chain is finally completed. The SDS lookout from a vantage point on one side of the Pentagon sees the signal and only a thin blue line of Columbia students. Unable to see the other sides of the Pentagon but knowing from the signal that the mission is completed, the lookout believes (212) and so utters (211), which is true. The lookout’s uncertainty concerns whether or not the event referred to involves only the Columbia students—an uncertainty about the body count: (213) Scenario I (“Call in the reinforcements”). The SDS organizers have planned that the Columbia students will surround the Pentagon and only if they are shorthanded will the Harvard students join them. One of the Columbia students is to signal when the chain is finally completed. The SDS lookout, from a vantage point on one side of the Pentagon, sees the signal and only a thin blue line of Columbia students.
Introduction
43
(214) Scenario II (“Too few Columbia students and impostors among us”). Observing the demonstration, the lookout either sights crimson jerseys or calculates that the Columbia students are too few to encircle the Pentagon on their own, but Harvard students at demonstrations are sometimes impostors. In the second scenario, the Harvard students at demonstrations are sometimes impostors. They have been known to hire substitutes to wear the crimson for them. The SDS lookout knows this and therefore believes nothing stronger than (212). Now, observing the demonstration, the lookout either sights crimson jerseys or calculates that the Columbia students are too few to encircle the Pentagon on their own. It is anyway certain that there are extra bodies whatever doubt there may be about their identity, and the lookout knows therefore that the demonstrators are not only the Columbia students. In this scenario it would be entirely infelicitous for the lookout to utter (211). It’s not hard to see why. Sentence (211) is equivalent to (215)—that there was a demonstration and the demonstrators were the Columbia students or maybe the Columbia students and the Harvard students: (215) The Columbia students formed the unbroken chain around the Pentagon, or the Columbia students and the Harvard students formed the unbroken chain around the Pentagon. So, why would a speaker assert such a disjunction already knowing one of its disjuncts to be false? Compare the simpler examples in (216) and (217). Example (217a) is unacceptable, it already being known that Biff could not have met: (216) a. b. (217) a. b.
Biff, Miff, and possibly Tiff met. Biff and Miff met, or Biff, Miff, and Tiff met. *Biff and possibly Tiff met. *Biff met, or Biff and Tiff met.
The logical form for (211) should explain why it is felicitous in the first scenario but not the second. A logical form equivalent to (208) (repeated in (218)) (or others equivalent to (206) or (207)), without event pronouns, would however be true and appropriate in either scenario: (218) ∃E∃e1∃e2 (∀e(Ee ↔ (Ee1∨ Ee2)) & [the X: C(X)]Participate(e1,X) & possibly[the Y: H(Y)] Participate(e2,Y)) & formed(E) the unbroken chain around the Pentagon ‘There are some events in some of which the Columbia students participated and in some of which it is possible the Harvard students participated, and those events are the forming of the unbroken chain around the Pentagon.’
44
Chapter 1
After all, (212), true in either scenario, is just the middle conjunct of (218) and the other two conjuncts are true too. With a meaning like (218), it doesn’t matter which doubts prompt the speaker to hedge the assertion, either that the Columbia students were insufficient or that the Harvard students were impostors. In contrast, the event pronoun in (209) (repeated in (219)) serves to distinguish the two scenarios: (219) ∃Eloc([the X: C(X)][∃e1: Eloce1]Participate(e1,X) and Possibly[the Y: H(Y)][∃e2: Eloce2] Participate(e2,Y) & [℩E: pro1,2] formed(E) the unbroken chain around the Pentagon) ‘Somewhere the Columbia students participated in something and possibly the Harvard students participated in something there too, & it was the forming of the unbroken chain around the Pentagon.’ (220) [=(188)] Biff did something and possibly Tiff did something, and it started the riot. (221) [=(189)] Biff and possibly Tiff started the riot. In the second scenario, the lookout sees crimson jerseys or calculates that the Columbia students are too few, and therefore knows that the Columbia students are not alone. The event pronoun in (219) refers to whatever the Columbia students or the Harvard students (and no one else) participated in there, just as the spoken and unspoken event pronouns in (220) and (221) refer to exactly what Biff and Tiff did. What Columbia or Harvard students did and no others, (219) goes on to say, was a forming of the unbroken chain around the Pentagon. If so, then any in the chain other than Columbia students must have been Harvard students. Of course the lookout in the second scenario does not intend this because she has reservations about the identity underneath those crimson jerseys. Example (219) is inconsistent with these reservations. Knowing for certain that not only Columbia students were involved, the use of possibly in (211) is infelicitous and cannot be grounded in suspicions about impostors. This then is the objection to (218) as the meaning for (211). It deprives us of an account of (211)’s infelicity in the second scenario where the speaker knows that not only the Columbia students participated in the demonstration. An account is forthcoming when an event pronoun outside the conjunction and therefore outside the scope of the modal possibly has as its content description of whatever if anything the Columbia students participated in and the Harvard students participated in. With an event pronoun in place, and need be nothing more than the truth-functional connective. Next is conjunction in predicative position, (222) and (223). Here too plural reference by “[℩E′: pro3,4]” in (224) to events of cooking up the Harvard beets and garnishing them with pearl onions allows the three Harvard students to divide the tasks,
Introduction
45
without supposing that cook up the Harvard beets and (possibly) garnish them with pearl onions is a predicate denoting them just in case they can be so divided, as in (225). (222) Three Harvard students cooked up the Harvard beets and garnished them with pearl onions. (223) [=(205)] Three Harvard students cooked up the Harvard beets and possibly garnished them with pearl onions. (224) [∃X: 3(X) & H(X)] ∃e1 Participate(e1,X) & [℩E: pro1][℩E′: pro3,4] O(E, E′) & ∃e3 (∃Z Agent(e3,Z) & cooked …) and ∃e4 (∃Z Agent(e4,Z) & … garnished …) ‘Three Harvard students did something; this was that; there was some cooking up of the Harvard beets and possibly some garnishing them with pearl onions.’ (225) [∃X: 3(X) & H(X)] λX0∃X1∃X2(∀x(X0x ↔ (X1x ∨ X2x)) & ∃e3(Agent(e3, X1) & cooked …) & possibly∃e4(Agent(e4, X2) & garnished …))(X) ‘There are three Harvard students some of whom cooked up the Harvard beets and the others of whom are among those of them who possibly garnished the beets with pearl onions.’ A curious asymmetry in the meaning of (223), where possibly intrudes, goes without explanation unless the event pronoun is present. If (223) is understood with the so-called full Conjunction Reduction reading equivalent to (226), then, like (226), the grounds for using possibly remain fairly open. (226) Three Harvard students cooked up the Harvard beets, and possibly they garnished them with pearl onions. It may be that the speaker is uncertain whether the beets were ever garnished, although if they were, she is certain it was the three Harvard students who did it. Or it may be plain that the beets were garnished, and certain that three Harvard students cooked them; she nevertheless suspects that the three she later saw with the garnish were impostors. In contrast, now consider the interpretation of (223), where we explicitly mean to allow that the Harvard students be divided between the cooks and the garnishers. As we have seen earlier, the possibility conveyed seems to render the sentence equivalent to a disjunction: (227) Three Harvard students cooked up the Harvard beets, or three Harvard students cooked up the Harvard beets and garnished them with pearl onions.
46
Chapter 1
Something went on. It was at least a cooking up of the Harvard beets and maybe also a garnishing, and three Harvard students did it. No impostors here. If there was no garnish, then three Harvard students did what was left to do, the cooking. Example (223) would be infelicitous if it were known that the one Harvard student Biff cooked up the beets by himself and suspected that Yalies Eli and Nathan posing as Harvard students Miff and Tiff garnished the beets with pearl onions and who knows what else. On the other hand, the so-called full Conjunction Reduction interpretation equivalent to (226) is appropriate on analogous conditions: it is known that Biff, Miff, and Tiff cook up the Harvard beets and suspected that Eli, Nathan, and Handsome Dan posing as them did the garnish. All this is quite puzzling if and composes in (222) a Predicate Phrase where the sentence is true when the Harvard students are divided between those that satisfy the first conjunct, cooking up the Harvard beets, and those that satisfy the second conjunct, garnishing the beets with pearl onions. Modifying the second conjunct with possibly in (223) undermines the interpretation, and yet it should be acceptable to divide the Harvard students between those that cooked and those of whom it is possible that they (or their impostors as the case may be) garnished the beets. Instead, and never deviates from its ordinary sense conjoining sentences, describing events in (223) that a backward pronoun refers to in (228), analogous to the spoken pronoun in (229): (228) [∃X: 3(X) & H(X)] ∃e1 Participate(e1,X) & [℩E: pro1][℩E′: pro3,4] O(E, E′) & ∃e3 (∃Z Agent(e3,Z) & cooked …) and ∃e4 (∃Z Agent(e4,Z) & … garnished …) ‘Three Harvard students did something; this was that; there was some cooking up of the Harvard beets and possibly some garnishing them with pearl onions.’ (229) Three Harvard students doing it, the Harvard beets were cooked up and possibly garnished with pearl onions. If the beets were cooked and garnished, then the pronouns in (228) and (229) refer to exactly that; and if there was no garnish, then the pronouns refer to just the cooking. Since nothing referring to the students falls within the scope of possibly, questions about their true identity are irrelevant. Indeed whatever the pronoun does refer to, whether it is cooking and garnishing or just cooking, the three Harvard students, themselves, not impostors, must all be its agents. The introduction of possibly has probed the meanings of the conjunctions containing it, exposing that the Columbia students and possibly the Harvard students is not a term of plural reference and cook up the Harvard beets and possibly garnish them with pearl onions is not a Predicate Phrase dividing anything it denotes between
Introduction
47
what the conjoined predicates denote. If not, these conclusions carry over to the conjunctions that omit possibly. These conjunctions conjoin sentences, and illusions to the contrary are an effect of pronouns positioned outside the conjunctions that refer plurally to the events described by the sentences conjoined. 1.6
Adverbialization
Granted a univocal sentential and, supermonadicity and descriptive event pronouns follow almost as a matter of logic with just a few empirical observations to egg them on. Adverbialization and an apparatus for spatiotemporal orientation (chapters 8– 15), in contrast, relate to and in the solution to puzzles of substitution under identity that coordinate NPs are prey to. In one class, opposing simple NP0 and coordinate NP1 and NP2 that are coextensive, musicians and instrumentalists and vocalists, DPs such as many musicians and many instrumentalists and vocalists do not substitute salva veritate. Another class opposes an (in)definite description containing a coordinate NP, some/the NP1 and NP2, and the coordinate description of its coordinated NPs, some/the NP1 and some/ the NP2. In such puzzles, neither a sunrise and sunset and a sunrise and a sunset, for example, nor the apostles and saints and the apostles and the saints substitute salva veritate. (230) ⌜Ft⌝ (231) ⌜[D ξ : NP[ξ]] Φ[ξ]⌝ Under the spell that (230)–(231) is the structure of predication and quantification in natural language, logic will insist after a failed substitution of many musicians and many instrumentalists and vocalists that the simple NP and the coordinate NP do not denote the same things after all, implicating and in some sleight of hand to distinguish their reference. If, in turn, the sunrise and the sunset is taken to be a referring expression, then also whatever it refers to is not what the sunrise and sunset refers to, failing to substitute salva veritate, and it looks again like and is up to no good coordinating NPs. Illustrating the first, to be a musician is to be an instrumentalist or to be a vocalist (232), and so the musicians are the instrumentalists and vocalists (233): (232) musician(e,x) ↔ instrumentalist(e,x) ∨ vocalist(e,x))34 (233) ∃E musicians[E,X] ↔ ∃E instrumentalists and vocalists[E,X] Yet simple NP and coordinate NP do not substitute salva veritate, alternating between the banal truth of (235) and the false implication of (234) that there are many one-person bands: (234) ??? F Many musicians are an ensemble. (235) Many instrumentalists and vocalists are an ensemble.
48
Chapter 1
For a syntax conforming to (236) or (237), all that should have mattered to preserve truth is that the things X that simple NP and coordinate NP denote are the same, as in (233) (whatever passes in the events E): (236) [Many X : ∃E NP[E,X]] are-an-ensemble[X]. (237) [Many x : ∃X(Xx & ∃E NP[E,X])] are-an-ensemble[x]. The puzzle becomes more acute in the following, where the baseball players are the fielders, who are also the batters;35 but they are not the fielders and batters, so it appears, under substitution: (238) ??? F Many baseball players are a powerhouse squad. (239) Many fielders and batters are a powerhouse squad. If the baseball players, the fielders, and the batters are all the same, then except for what and might manufacture from them, how else could simple NP and coordinate NP come to denote a difference?36 Whatever and is up to in (235) and (239), it looks to be up to something else again in certain examples where identity statements host the substitutions: (240) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned. (241) The lover and belovèd in either love note are the lovers exchanging them. The substitution in (240) under the identity (241) is false: (242) F The lover and belovèd in the first love note are not the lovers exchanging them.37 When a simple NP lovers and a coordinate NP lover and belovèd refer to the same things and yet their substitution does not occur salva veritate, there is an illusion that and is to blame and temptation for and under new guise to fashion from lovers their ordered pairs and for the coordinate NP to denote. Simple lovers and coordinate lover and belovèd end up referring to different things, sparing substitutivity on impulse but ending all the same in contradiction and confusion in (241) and (242). Alongside substitutions between simple and coordinate NPs, the other class of puzzles relate a DP containing a coordination of NPs to a coordination of DPs containing the coordinated NPs, comparing a/the sunrise and sunset and a/the sunrise and a/the sunset: (243) A sunrise and a sunset have preceded two twilights’ bloody battle. The sunrise and the sunset have preceded two twilights’ bloody battle. (244) A sunrise and sunset have preceded two twilights’ bloody battle. The sunrise and sunset have preceded two twilights’ bloody battle.
Introduction
49
Despite plural reference to two sun transits, as reflected in the plural number agreement in (243) and (244), only the coordination of DPs (243) may report events that interleave sun transits and battles, sunrise—battle—sunset—battle. With coordinate NPs, (244) demands the sun transits to both precede both battles, sunrise—sunset— battle—battle. The single temporal interval demanded in (244) finds an analog in a single scene or presentation that (246) demands. Both the hapless cub reporter Jimmy Olsen, always missing the action, and senior reporter Clark Kent, Superman’s alter ego, never manage to catch Superman at a crime scene: (245) A reporter and a superhero were at the crime scene interviewing witnesses. The reporter and the superhero were at the crime scene interviewing witnesses. (246) A reporter and superhero were at the crime scene interviewing witnesses. The reporter and superhero were at the crime scene interviewing witnesses. Both (245) and (246) may indifferently narrate the adventures of Jimmy Olsen and Superman. But only (245) may do the same for Clark Kent and Superman, as if the single DP in (246) containing coordinate NPs demanded reporter and superhero to sit for the same photograph, which of course Clark Kent and Superman cannot do. If it is held that the reporter and the superhero and the reporter and superhero are both referring expressions, and one has not already gagged on the metaphysics of a twosome of Clark Kent and Superman, it is just more poison to swallow that these two, unlike Jimmy Olsen and Superman, are not two enough for the reporter and superhero to refer to. On the other hand, it appears that the sunrise and sunset refers to less a twosome than the sunrise and the sunset as far as what there is that can be said to be separated in time. A contrast between coordinate NPs and their counterpart coordinate DPs extends to the interpretation of relative clauses restricting them: (247) Some Apostles and some Saints who were (both) twelve in number were martyred in the All-Star game last Friday with many injuries. (248) Some Apostles and Saints who were (*both) twelve in number were martyred in the All-Star game last Friday with many injuries. (249) The elms and the beeches that are (both) dense in the middle of the forest darken the forest floor most of the day. (250) *The elms and beeches that are (*both) dense in the middle of the forest darken the forest floor most of the day. In (248), twelve is unambiguously the number of all those Apostles or Saints martyred on the playing field. In contrast to (247), it cannot be understood that such Apostles were twelve and such Saints were too.38 Similarly, (250), in contrast to (249), cannot be taken to describe the elms as dense and the beeches too. It describes as dense only the whole grove of elms and beeches.
50
Chapter 1
Given the received syntax for predication and quantification in natural language, the substitution puzzles are argument that the musicians are not the instrumentalists and vocalists that they are; the baseball players, not the fielders and batters; and the two lovers, not the lover and belovèd. The sunset and sunrise are not the same as the sunset and the sunrise, nor are the reporter and superhero the same as the reporter and the superhero; and they are two-ish to different standards. These substitution puzzles look to implicate and in metaphysical mischief. In the first failed substitutions under identity, NP0 and NP1and2 are coextensive in the objects X they describe (251), and thus the DPs containing them either refer to or quantify over the same things. Yet the DPs fail to substitute salva veritate in natural language sentences that, as spoken, seem to differ only in the substitution itself, (252): (251) ∃E NP0[E,X] ↔ ∃E NP1and2[E,X] (252) a. … [the NP0] … ⊬ … [the NP1and2] … b. … [many NP0] … ⊬ … [many NP1and2] … The small difference is however an illusion under adverbialization. To substitute the such-and-such for the so-and-so in a sentence of the natural language is always a double substitution in logical form—the such-and-such while such-and-such-ing for the so-and-so while so-and-so-ing. To be valid, it is not enough that the such-andsuch are the so-and-so. It must also be that while such-and-such-ing is while so-andso-ing—that is, that the derived adverbs describe the same events too: (253) NP0[E0,X] ↔ NP1and2[E1and2,X] (254) … [the X : ∃E0 NP0[E0,X]] [℩E0: NP0[E0,X]][∃E:while[E0, E]] … ⊢ … [the X : ∃E1and2 NP1and2[E1and2,X]] [℩E1and2: NP1and2[E1and2,X]][∃E:while[E1and2, E]] … A puzzle of substitutivity may yet be resolved without ontological dereliction if under substitution different events are described. Even where the events of NP0-ing and of NP1and2-ing coincide,39 music being instrumental or vocal, if NPs always occur in DPs with spatiotemporal morphology, as in (69)–(70) above, scenes and the spatiotemporal orientation of the events within may restrict the events described and thus how the matrix events are framed: (255) ??? F Many musicians in scenes while musicians in those scenes are an ensemble. (256) Many instrumentalists at addresses in scenes and vocalists at other addresses in those scenes while instrumentalists at their addresses in those scenes and vocalists at their addresses in those scenes are an ensemble. As described in (256), the scenes of ensembles are scenes presupposed to be wide enough for both instrumentalists and vocalists to find addresses within. In contrast,
Introduction
51
the scenes of musicianship in (255) are not so structured, affording then some difference of syntax and meaning to be exploited for substitution non salva veritate without the conclusion that the musicians are not the instrumentalists and vocalists. If, as a further point of syntax, DPs—but not NPs—may introduce new scenes or frames of reference, the substitution in natural language of ⌜[DPa/the NP1 and NP2]⌝ for ⌜[DPa/the NP1] and [DPa/the NP2]⌝ is also not a substitution of DPs into an identical context in logical form: (257) A sunrise in a scene while a sunrise in that scene and a sunset in a scene while a sunset in that scene have preceded two twilights’ bloody battle. ‘A sunrise and a sunset have preceded two twilights’ bloody battle.’ (258) A sunrise in a scene and sunset in that scene while a sunrise and sunset in that scene have preceded two twilights’ bloody battle. ‘A sunrise and sunset have preceded two twilights’ bloody battle.’ The logical forms now invite surmise that the substitution failure has something to do with it being easier to interleave two scenes among the battles than just one. Likewise, it may just be harder on Clark Kent and Superman to appear in the same scene than in different ones: (259) A reporter in a scene while a reporter in that scene and a superhero in a scene while a superhero in that scene were at the crime scene interviewing witnesses. ‘A reporter and a superhero were at the crime scene interviewing witnesses.’ (260) A reporter in a scene and superhero in that scene while a reporter and superhero in that scene were at the crime scene interviewing witnesses. ‘A reporter and superhero were at the crime scene interviewing witnesses.’ In another turn, if DPs introduce new scenes or frames of reference, perhaps speakers are mindful enough of continuity from one scene to the next to comment on their orientation: (261) In a mutual declaration of requited love, the lover here in a scene1 and belovèd there in that scene1 in the first love note while lover here and belovèd there in that scene1 are not the lover here in a scene2 and belovèd there in that scene2 in the note returned while lover here and belovèd there in that scene2, that scene1 oriented the same way as that scene2. In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned oriented the same way.
52
Chapter 1
According to (261), the speaker relies on a tacit adverb to deny that they are them arranged a certain way. In using a coordination of addressed NPs, the speaker intends scenes that segregate lover and belovèd, and she denies in (261) that the scenes in which they appear share the same orientation. The speaker never entertains that the two tokens of the lover and belovèd might differ in their reference or differ from that of the lovers exchanging love notes. Any conclusion otherwise rests on an error of syntax, a parse of the puzzle sentences that omits adverbialization and the morphology of spatiotemporal reference. In the corrected syntax, all DPs retain their customary and ordinary reference, and when all is said and done, and remains throughout the univocal sentential connective. The impetus for these preliminary remarks has been an empirical problem, puzzles of substitution under identity, that look to corrupt and with adventitious meanings at least when conjoining NPs. The resolution sketched supposes that there is so much else going on in the grammar, it is hardly necessary to look at and at all. Argument for the new grammar begins with the recognition that what has been at issue with coordinate NPs is just a special case of classic substitution puzzles that include coextensive NPs both simple and both coordinate, where a new meaning for and is no help. Adverbialization proves necessary for the resolution of all the puzzles of substitution under identity. In turn, some of these puzzles—still with simple NPs in mind—are about number and counting, for which the substitution puzzle is not resolved without reference to the epistemic conditions for counting, including a frame of reference and protocol for the measurement events. With adverbialization and frames of reference in hand, it is then a small step to suppose that every NP is addressed under a frame of reference. If the speaker has in mind to segregate the instrumentalists and the vocalists, instrumentalists here and vocalists there, at different addresses, it finds no equivalent expression with a simple NP, not even musicians here and there. Rather than a special and, the only thing special about the coordination is that there are (at least) two NPs and therefore (at least) twice the addresses of a simple NP. Chapters 8–15 are a study detached from and of the empirical phenomena, the semantic puzzles, and other aspects of meaning, that reflect the effects of adverbialization and spatiotemporal orientation. The good it all does and, emphasized in these introductory remarks, is a corollary. A summary follows of the argument for these fundamental revisions to clausal architecture. 1.6.0
Adverbialization in logical form
Prior to any substitution puzzles, there is independent evidence of adverbialization from two unrelated quarters. First, descriptive anaphora, both implicit and explicit as in (262), covarying with the events described, must fall within the scope of the event quantifier adverbialization derives as in (264):
Introduction
53
(262) Every fugitive from a federal prison is soon apprehended near it. (After Enç 1986) (263) Every fugitive from a federal prisonj as of somei time is soon after thati time apprehended near itj. (264) [Every x: ∃e fugitive from a federal prison[e,x]] [while40 e : fugitive from a federal prison [e,x]] ∃e′ (soon(e,e′) & apprehended[e′,x] & [℩y: federal prison(y) from which x is a fugitive in e] near(e′,y))41 Second, certain adjectives such as respective can only be interpreted as in some way modifying the matrix scope of their host DP, as if an adverb, respectively (Gawron and Kehler 2002, 2004). Suppose, for example, that twins rent the same make and model vehicle and pose for photographs with the two cars in which it is variously true that (265) The twins stood beside their respective rentals. The twins stood behind their respective rentals. (266) The twins stood between their rentals. In contrast to (266), there can be no photograph for which (267) *The twins stood between their respective rentals. For the twins to be in respective possession of their rentals, as the nominal describes them, puts them in a pair of states in each of which a twin is opposite her own. Such a state frames a standing beside (or a standing behind) if its twin is beside (or behind) her rental, (265). But, neither of the states that respective possession denotes frames a standing of a twin between the rental in it, (267). This nonsense that (267) cannot escape is adverbialization demanding that the twins’ being with their respective rentals frames their standings between. The scope of adverbialization tracks the scope of the nominal phrase adverbialized: (268) At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer pulled their respective champions apart. (269) *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer pulled apart their respective champions. Given the scope of their respective champions in (268), it is immediate that the states of respective possession separately frame only the champions’ being pulled: Ali’s trainer grabbing Ali and Frazier’s trainer grabbing Frazier pulled hard until the champions are parted. Later noticed is the macabre interpretation in which the states of respective possession each frame a champion’s being pulled apart by his trainer. The macabre becomes more salient in (269) to the extent that stylistic movements favor the scope worn on the sleeve, here the one now including apart within
54
Chapter 1
the scope of their respective champions.42 Note that for the sensible interpretation of (268) and (269), the scope that adverbialization affords to respective is the minimal one that the host DP itself requires, a relation expressing participation in some events—a scope that is too small for the felicitous use of the lexical adverb respectively: (270) *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer respectively pulled their champions apart. (271) *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer pulled apart their champions respectively. (272) ??At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer pulled their champions respectively—apart. And why not? Nominal descriptions just are the adverbs dedicated to the smallest thematic relations, reflecting an economy of expression and compactness in the natural language. Lexical adverbs are reserved for larger phrases such as might contain verbs. Any DP may contain respective or an antecedent for descriptive anaphora, and thus adverbialization is mandated at any position a DP occurs. As reflected in these two phenomena, adverbialization is a fact of grammar. 1.6.1
Naive reference and substitution under identity
If (273) is the structure of predication in natural language and adverbialization is absent, testimony in a world of exactly forty nations that (274) and (276) are true and (275) false (and even unnatural) is proof that what their forty allies refers to is not what their forty enemies refers to, and the things that neither of these definite descriptions refers to are the same as what the forty nations refers to: (273) ⌜Ft⌝ (274) The forty nations rallied their forty allies, and the forty nations thwarted their forty enemies. (275) The forty nations rallied their forty enemies, and the forty nations thwarted their forty allies. (276) The forty nations rallied the forty nations, and the forty nations thwarted the forty nations. The alliances and counteralliances that matter for the truth of (274)–(276) should better manage it without fragmenting nations. Departing then from (273), adverbialization interposes an adverbial phrase so that what is spoken as (277) is always heard and parsed along the lines of (278): (277) The forty nations rallied their forty allies. (278) The forty nations being nations rallied their forty allies being in alliances.43
Introduction
55
The definite descriptions retain their customary and naive reference as merely alternative descriptions of the same things; yet to rally the nations in their alliances is not to rally them in counteralliances, leading to the divergence in truth among (273)–(276). The descriptions serve not only to fix reference to the objects referred to but also to stipulate conditions under which they participate in the events described—adverbialization. The effects of adverbialization are pervasive, as they should be if it defines how any DP combines with its scope: (279) The ponds the temperatures of which were recorded early this morning were colder than the ponds the temperatures of which were recorded early this evening. (280) Every pond the temperature of which was recorded early this morning was colder than every pond the temperature of which was recorded early this evening. (281) Venus Pond early this morning was colder than Venus Pond early this evening. (282) #Venus Pond was colder than Venus Pond (was). From neither (279) nor (280) is there implication that the morning ponds are other than the evening ponds. If, in fact, Venus Pond is among them both, (279) and (280) imply (281). Insofar as adverbial qualification is necessary to escape contradiction (cf. (282)), its only source in (279) and (280) are the nominal phrases themselves that, according to adverbialization, do indeed supply adverbial phrases to qualify participation in the events described. Of course, if adverbialization is pervasive and applies to all DPs, it applies equally in (282) but to not much avail, deriving the same adverb:44 (283) #Venus Pond while Venus Pond was colder than Venus Pond while Venus Pond (was). As morning is not evening and manhood, not boyhood, the adverbs derived by adverbialization in (279) and (280) and in (284) cannot apply to the same events without contradiction: (284) The man surpassed the boy of his childhood in regrets and not much else. Thus, only in a logical language that conforms to supermonadicity will adverbialization resolve the puzzles of substitutivity under identity. Naive reference, always a virtue, and the solution to substitutivity under identity thus affirm the logical syntax, supermonadicity, essential to the language in which and is univocal and sentential.
56
1.6.2
Chapter 1
Adverbs denoting events of counting
The logical forms for (274)–(280) resolve their substitution puzzles without the ontological dereliction of nations not the allies they are, men not the same beings as the boys they were and endless chaos under heaven.45 These are avoided by supposing that the same things are always participating but do so under different arrangements or under different temporal conditions—conditions on the participants’ behavior. In other substitution puzzles, the metaphysics is spared only when conditions step in on the speaker’s epistemic stance toward the events under report, granting that the nominal description contains expressions of such: (285) Three million passengers crowded National Airlines’ routes last year. (286) ⊭ Three million persons crowded National Airlines’ routes last year. (After Gupta 1980, 23; Moore 1994) Frequent fliers recounted under (285)’s protocol keep it from implying (286). If, still under the spell of (273), the arithmetic predicate three million is a simple property “3M(_)” determined by the numeric identity of what is referred to, three million passengers does not refer to the one million whole persons who flew frequently. Away we go, denying the identity of passengers and persons to thwart the inference from (285) to (286) and asserting it in (287), despite the mystery of how what is three is also one if number is such an intrinsic property: (287) The three million passengers who crowded National Airlines’ routes last year were the one million frequent fliers loyal to it. Worse, the many who are the few in (285) do not always exist as the many, subject to a condition that Doetjes and Honcoop (1997) aptly call sequencing of events: (288) Three million passengers had three million opinions about the food on National Airlines. (289) *Three million passengers have three million opinions about the food on National Airlines. Sentence (285) can be followed by (288), recounting the passengers’ experience on board and continuing to count as many the fewer. In the present tense, however, (289) cannot count more passengers than persons who fly, despite the fact that these persons still have three million opinions about flights they remember all too well. The relation between the count and present tense wants explanation. Instead of the bare “3M(_),” imagine that three million translates as something like “now counted this way to three million,” a description of events of measurement (the counter clicks, as it were) under an explicit protocol. What is counted and what the indefinite description three million passengers refers to is naively what there is, persons, who
Introduction
57
are counted and under some protocols recounted. With the counting now entered into the description, adverbialization derives (290) and (291): (290) Three million passengers while counted to three million had three million opinions about the food on National Airlines. (291) *Three million passengers while counted to three million have three million opinions about the food on National Airlines. Clicking the counter as passengers go by frames or locates opinion recorded contemporaneously as in (288), but there is no counting to three million that frames current conditions, which, according to (291), is what defeats (289). The speaker is secure in her constant reference to the naive, familiar objects she swears to but only by a sleight of hand that manipulates her epistemic conditions, in this case, conditions of measurement. 1.6.3
Scenes to count by
There is a fundamental relationship here between the conditions of measurement and the scenes of what is counted, which under adverbialization frame the events reported. In the narrow scene or frame of reference in (292), there are exactly two green regions if finitely many. (Green is shown as gray in (292).) There is only one in the wide scene or frame of reference. This is not a metaphysical claim, for surely the regions counted under the narrow frame of reference still exist as parts of the ring under the wider frame of reference. They are just not to be counted there, under conditions inappropriate for their measurement. (292)
Think of the design of an optical counter, blind to what is counted—passengers or planets if you say so. To utter two is to take on a perspective from which what is reported projects a scene the optical counter measures as 2. Counting two commits to the narrow scene or frame of reference in (292) and to a report of what the events
58
Chapter 1
counted as two may frame within it, whether counting passengers or planets, as reflected in number agreement in the following: (293) The morning star and the evening are playing hide-and-seek in the twilight, season after season. (294) *From the vantage point of Jupiter, (you can see that) the morning star and the evening star are circling the sun. (295) From the vantage point of Jupiter, (you can see that) the morning star and the evening star is circling the sun. Narrow scenes while the morning star and while the evening star exist only from earth. From Jupiter, if anything, what is while the morning star and while the evening star converges on the same large scene of Venus in orbit, where there is exactly one Venusian region, undermining the plural number agreement.46 Scenes and the discrimination between scenes of one and scenes of more than one that (292) illustrates reach down into the primitive vocabulary, to the relation that has been glossed participate, to confer on it perspectival content. In dividing its reference between civilians and superheroes, some civilians and superheroes denotes some participants in events that coincide with some events of being civilians and some events of being superheroes: (296) [Some X : ∃E0 participate(E0,X) & O(that0, that1and2) & ∃E1∃X civilians[E1,X] and ∃E2∃X superheroes[E2,X]] ‘Some civilians and superheroes’ Recall the contrast between (297) and (298). Whereas both can be about Jimmy Olsen and Superman, the coordination of DPs in (297) can be about Superman and his alter ego Clark Kent, but the coordination of NPs in (298) cannot. (297) A reporter and a superhero were at the crime scene at two different times. The reporter and the superhero were at the crime scene at two different times. (298) A reporter and superhero were at the crime scene at two different times. The reporter and superhero were at the crime scene at two different times. The two DPs introduce their own scenes or frames of reference, but the one introduced by the solitary DP in (298) appears to crowd Clark Kent and Superman too much into the same scene. Two DPs will always escape the restriction. The problem presenting Clark Kent and Superman in the same scene or frame of reference seems, however, to attach only to a coordination with a singular article as in (298). In the plural, (299) seems to impose no sanction against the participation of Clark Kent
Introduction
59
and Superman, Bruce Wayne and Batman, and Peter Parker and Spiderman—all superheroes and their alter egos. (299) Some civilians and superheroes were at the crime scene at six different times. The civilians and superheroes were at the crime scene at six different times. Six civilians and superheroes were at the crimes scene at six different times. What is there one of in (298) to rule out Clark Kent and Superman, and more than one or none of in (299) with plural morphology throughout? In a nearby two motorists in (301), it is a single being nearby: (300) Last night at opposite ends of the city, two lightning strikes in heavy traffic injured two nearby motorists. (301) #Last night at opposite ends of the city, two lightning strikes in heavy traffic injured a nearby two motorists. Absent a singular article, (300) allows two nearbys, each locating a motorist near a lightning strike at an opposite end of the city. In (301), on the other hand, the two motorists are in a nearby, the same one, which cannot be at opposite ends. The only interpretation for the sentence is distributive, a nearby for each lightning strike, reporting four victims. In (298), what there is one of is a participation, as glossed in (296) and left to be plural participations in (299). Of course, there is not much to the meaning of participate to banish Clark Kent and Superman from the same event of participation. If Clark Kent’s visit to the crime scene is an event in which he participates and Superman’s is another event in which Superman participates, there are the two events in which they participate. Who is to say that the mereological fusion of these events is not an event in which both participate? A scene, instead, is kinematic in representing all the motions of bodies within the spatiotemporal region it is a scene of, and it is object-tracking only if any object within it is tracked within it for its duration. An object-tracking scene of Clark Kent and Superman that is long enough to frame their visits to the crime scene, beginning when the first one enters and ending when the last one leaves, is not one that either may exit from for those moments when he is not at the crime site. Within such a scene—an aerial in effect—their trajectories coincide, the wide frame of reference in (292), and therefore they are one and not to be counted two. Narrowing the frame of reference to the crime site to record only the traffic in and out of civilian and superhero is a cinema of many object-tracking scenes, each no longer than an entrance and exit and perhaps a close-up of just one character. A single frame of reference for the several scenes that (299) allows is thus room enough for superheroes and their alter egos. The bland ‘participate(E0,X)’ in (296) is to be replaced with
60
Chapter 1
‘object-tracking-scene(E0,X)’ or ‘participate(E0,X) & object-tracking-scene(E0’)’, which is further qualified in a reporter and superhero to be a single scene ‘an(E0)’: (302) [∃ X : ∃E0 an(E0) & object-tracking-scene(E0,X) & O(that0, that1and2) & ∃E1∃X reporter[E1,X] and ∃E2∃X superhero[E2,X]] ‘A reporter and superhero’ The discovery of latent perspectival content in every NP-coordination, a reporter and superhero and some civilians and superheroes, serves up two tendentious points. First, the specific, topic-laden content counters what must have been Aristotle’s thought about an and of collective or divided reference, namely, that it is a (more or less) topic-neutral operator, logical, set-theoretic, or mereological, without cinema. Second, central to the answer to Aristotle is an apparatus for divided reference that sits outside an ordinary conjunction of sentences. That apparatus carries morphological weight in providing the singular article in (302) something to modify, answering what there is one of, when two participate. So there it is in the logical syntax for all to feel. An object-tracking scene, after all, is as good a thing for a(n) to denote as a nearby is in (301). 1.6.4
Singular and plural frames of reference
For the three million passengers to be the same X as the one million frequent fliers, it has been remarked that it cannot be that three million(X) and one million(X) are primitive arithmetic concepts. Rather, the same X may be subject to measurements with different results under different frames of reference and protocols: count[Ep,X,fp,3000000] & count[Eff,X,fff,1000000]. Under fp, counter clicking Ep, as it were, to three million, the protocol allows for an X to be at what is clicked for more than once—to be in more than one of the Ep. Since 3000000 ≠1000000, the same X are counted only if the measurements and the conditions for measurement have been different—that is, (count[Ep,X,fp,3000000] & count[Eff,X,fff,1000000]) → (Ep ≠ Eff & fp ≠ fff). Now that different events are described, different adverbials are derived, with adverbialization making the difference between (304) and (305). (303) Three million passengers had three million opinions about the food on National Airlines. (304) *Three million passengers have three million opinions about the food on National Airlines. (305) One million frequent fliers have three million opinions about the food on National Airlines. Counting comes with spatiotemporal conditions. The events so counted under one protocol may differ in their temporal resolution from those counted under another, as in (304)–(305), or in their spatial resolution, (293)–(295). To utter in the natural language three million or one million invokes a protocol and frame of reference for
Introduction
61
events within which are represented the temporal and spatial properties of theirs that have satisfied the conditions for counting. Expanding the natural language three million, from ‘3000000(X)‘ to ‘count[E,X,f,3000000]’, to include the process of measurement E and its relation to frame of reference and protocol f solves a class of substitution puzzles and spares the metaphysics. The extra parameters find independent welcome addressing other semantic problems too. As it turns out, no two of many fielders, many a fielder and many a one or more fielders are synonymous. In appreciation of the problem, note that to begin with the thought that fielders denotes pluralities of one or more fielders, ‘fielders(X)’, and many quantifies over them, ‘[many X : NP(X)]’ is to forgo daylight between many fielders and many a one or more fielders. The latter manages however to break away from the former if it quantifies over frame of reference and protocol, exploiting the variable f made available, ‘[many f : ∃E∃X … a … count[E,X,f, ≥1] …]’ or as eventually adopted, ‘[∃F: many(F) & ∃E∃X … a … count[E,X,F, ≥1] …]’. A contrast between singular and plural frames of reference is behind a contrast between perfective measurement (307) when the article is present and an imperfective, scanning measurement (306) when the article is null: (306) Baseball legends by the hundreds were Jews. (307) #Some baseball legends by the hundreds were Jews. This contrast between articles in referring to singular or plural frames of reference is also reflected in the failure to apply a genuinely collective predicate (see note 2) across multiple frames of reference in (308) and (309): (308) #Unsolved murders were a cluster near the Green River in Washington. (309) Unsolved murders were in a cluster near the Green River in Washington. (310) Some unsolved murders were a cluster near the Green River in Washington. (311) #Unsolved murders that were a cluster had a common modus operandi. (312) Unsolved murders that were in a cluster had a common modus operandi. (313) Some unsolved murders that were a cluster had a common modus operandi. Distributive quantification and singular and plural reference to frames of reference will thus warrant their explicit representation in the object language as supposed in the solution to the substitution puzzles. 1.7
Spatiotemporal orientation
Cardinality predicates in natural language are relativized to the contingent, fragile circumstances of measurement and in expressing a relation to protocol and frame
62
Chapter 1
of reference presuppose some fix or perspective on the goings-on being reported. It is not to be imagined that all this springs to life only with a thought about measurement and number agreement. Other vocabulary and thoughts presuppose it too. If, for example, the frame of reference for (314) is taken to be that of the ballerina’s proximate visual experience, the sentence puts her partner in synchronous orbit around her (possibly leading from an outstretched arm): (314) The ice ballerina pirouetted with her partner on the left across the rink. Alternatively, the frame of reference could be fixed at a moment by the line of sight pointing across the rink and tangent to the ballerina’s curved trajectory, tracking the partner in a parallel course, on one side or the other depending on whether the speaker faces the ballerina and the frame of reference is centered on the speaker’s point of view or on the ballerina’s. The sentence is thus three-ways ambiguous, and it communicates no determinate thought at all unless the speaker and hearer agree on the frames of reference intended. A lexical ambiguity—left1, left2, and left3—to sort them out would insult the concept’s integrity, which must rather be left in f, for different choices for frame of reference f. In contexts that are not demonstrative for f, as in third-party transmission, to have understood (314) is to have grasped some descriptive intention like in (315), absent which, again, no determinate thought is expressed: (315) In the orbital frames of references f centered on the agent’s proximate visual experience, the ice ballerina pirouetted with her partner on the left in f across the rink. Sentence (316) is also three-ways ambiguous—mercifully, not the nine ways it would be if the frame of reference for the second token of left were independent and possibly discontinuous from the first’s: (316) The ballerina will pirouette with her partner on the left to the middle of the ice, do a triple somersault, and pirouette with her partner on the left the rest of the way. The instruction and knowledge not to switch frames of reference within the sentence is itself a point of grammar. There is no thought here without description of a frame of reference, and its representation is subject to grammatical constraint.47 For purposes of spatial orientation and navigation, analysis of the visual and auditory scene includes parameters justified by the agent’s accurate navigation even as she is unaware of them in her experience of coarse-grained objects in location and in motion. Among the communicative intentions the design of language supports, I include the intention to convey to a companion witness accurate enough to guide her navigation,48 for which she has to acquire a scene from words of mine encoding
Introduction
63
parameters that I do not consciously attend to or assert, such as a frame of reference. How the world is at any given moment, as fixed in true propositions, is not accurate witness for where you are in it, for which the logical form of utterances should be tricked out to choreograph and calibrate the speaker’s and hearer’s thoughts and perceptions. If narrative is rich enough to support the communicative intentions of its narrator, real or imagined (and always imagined if not real), it is shortsighted to suppose that the only parameter for narrative continuity is temporal—a timeline for world history and a succession of points or intervals along it as the narrative advances its “now.” Even a “here and now,” the indices for a context of utterance, will not fix the thoughts expressed in a spatial vocabulary including left. Here we now are at the landing site on schedule, and yet we are disoriented and disabled for visual navigation if the faulty periscope from our windowless rover swivels about randomly, no matter how detailed the image transmitted nor acute my report of it. A sequence of scenes or their narration is orienting and navigational for observer, speaker, or hearer only if their lines of sight are known and they constitute a survey conforming to some natural conditions on their continuity—conditions that regulate over the course of survey and narrative such parameters as needed to constitute a natural (stereoscopic, i.e., 3-D) cinematic experience.49 As with any contextual or discourse parameter, thoughts are more or less dependent on them, and some narratives may wash out that dependency in adopting an omniscient narrator or an arbitrary, unknown, or neutral perspective. Nevertheless those parameters are always there for the asking. It has been a commonplace in semantics that interpretation varies with the world and time at which what is described is located, representing two parameters of discourse more or less explicitly in the object language. This is fine as far as it goes, which is not far enough to model the cinematic structure of narrative intentions. Addresses within and reference to a frame of reference f are a small step toward the representation of the thoughts embedded in a narrative, some of which might very well pause to count as in section 1.6.2 under the present conditions of observation. For Tensed clauses, the cinematic scene means a canonical structure expanded to include relations to perspectives and frames of reference. For NPs, it means that they do not occur in DPs without a morphology that describes location in a frame of reference as in (317) and applied to every NP when they are coordinated as in (318): (317) [DPsome X : ∃E (here[E,f] &NP[E,X])] [DP D [AdrP Adr NP]] (Cf. [CP C [TP Tense VP]] ) ‘Some chessmen’ (318) [DP D AdrP, AdrP and AdrP] a bishop, rook, and pawn
64
Chapter 1
The addressing “here[E,f]” assigns the address here in the frame of reference f to all and only the events E. Eavesdropping on the chess tutor teaching my son the endgame, I overhear: (319) A bishop, rook and pawn will always checkmate in ten moves, but a rook, pawn, and bishop only stalemate. The well-paid tutor does not intend contradictory falsehoods, namely, that the same three chessmen will always checkmate and only stalemate. He rather quantifies over chessboards, frames of reference within which the chess pieces are assigned fixed addresses. Frames of reference and addresses do not rescue the tutor from contradiction unless they are joined with adverbialization in a logical form for (319) along the lines of (320): (320) A bishop here1, rook here2, and pawn here3 while a bishop here1, rook here2, and pawn here3 will always checkmate in ten moves, but a rook here1, pawn here2, and bishop here3 while a rook here1, pawn here2, and bishop here3 only stalemate. The frames of reference quantified over frame the events reported—games—none of which is in progress during the lesson. As with (314), the frames of reference like the events themselves may very well be elsewhere at some remove from the actual context of the utterance. Now even if a speaker is in the dark and unprepared to identify frames of reference and addresses, grammar (see (317)–(318)) and narrative structure compel (321) to be parsed as (322), and (323) as (324): (321) Many instrumentalists and vocalists are an ensemble. (322) Many instrumentalists at an address in a frame of reference and vocalists at another address at that frame of reference while instrumentalists at an address in a frame of reference and vocalists at another address in that frame of reference are an ensemble. (323) ??? F Many musicians are an ensemble. (324) Many musicians at an address in a frame of reference while musicians at an address in a frame of reference are an ensemble. The instrumentalists and vocalists are of course the musicians. But, in choosing (321), the speaker commits herself to frames of reference that segregate instrumentalists and vocalists, which under adverbialization will result in the divergence in truth and felicity between (321) and (323), an effect of the general structure of DPs and narrative discourse. 1.8
Identity statements simpliciter and conditioned
If supermonadicity and adverbialization are everywhere, they must also touch the most elementary assertions of identity. In any case, parsing (325)–(327) [=(240)–
Introduction
65
(242)] without them as closures of ⌜x = y⌝ is a garden path into the substitution puzzles without exit, since the sentences can only chase after different values for x and y: (325) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned. (326) The lover and belovèd in either love note are the lovers exchanging them. (327) ⊬ The lover and belovèd in the first love note are not the lovers exchanging them. The logical forms and lexical items contained in (325)–(327) must admit modification by Tense at the very least and by the adverbs that, by hypothesis, adverbialization introduces with every DP, including those of (325)–(327). Besides hypothesis, modification in identity statements finds overt expression in (330), which differs from (329) in truth and differs only in the presence of a secondary predicate: (328)
(a)
(b) (a) Still life with fruit; (b) Still life with ginger jar, sugar bowl, and oranges
(329) T The oranges in the first still life are the oranges in the second. (330) F They are them arranged the same way. To be them is not to be them that way. It wants explanation how a classical identity statement simpliciter comes to be expressed in the same words as conditioned identity except for the condition itself. There is no room for modification in ⌜x = y⌝ and no place for ⌜x = y⌝ in Eventish absent relation to events. Instead, an identity relation ‘ℐ[E1, E2]’ is said to hold between events or states only if the participants in
66
Chapter 1
E1 are identical to the participants in E2. Eventish ℐdentity holds between the oranges’ participation in the event or state depicted in the first canvas and their participation in that of the second canvas, verifying (329) (see (331)), but that is not to say that their state in the second canvas is also the same arrangement. The further comment describing it as such results in the falsehood (330) (see (332)): (331) [the X : ∃E NP[E,X]][℩E: NP[E,X]][∃E1:while[E, E1]] participate[E1, X] & [℩E1: that1][℩E2: that2] ℐ[E1, E2] & [the X : ∃E NP[E,X]][℩E: NP[E,X]][∃E2:while[E, E2]] participate[E2, X] (332) [the X : ∃E NP[E,X]][℩E: NP[E,X]][∃E1:while[E, E1]] participate[E1, X] & [℩E1: that1][℩E2: that2] ℐ[E1, E2] & [the X : ∃E NP[E,X]][℩E: NP[E,X]][∃E2:while[E, E2]] participate[E2, X] & arranged the same way[E2] Given the common syntax that (329) and (330) warrant, adverbialization just joins in, as in (331) and (332), to further condition the events related by Eventish ℐdentity. As weak as Eventish ℐdentity is, if (325) is just a negated identity simpliciter, it falsely denies that there are events with the same participants in one of which the lovers participate while lover and belovèd and in the other of which they participate with roles reversed. The sentence as intended is however to be understood as a true denial of a conditioned identity, where the condition is to be conveyed by an appropriately crafted, tacit secondary predicate: (333) The lover and belovèd in a love note are not the lover and belovèd in the note returned oriented the same way. (334) The lover at the 1st address in a frame of reference1 and belovèd at the 2nd address there1 are not the lover at the 1st address in a frame of reference2 and belovèd at the 2nd address there1, that frame of reference1 oriented the same way as that frame of reference2. Now if all the NPs are addressed, as supposed, to a scene or frame of reference, the secondary predicate is definable as a relation between the two frames of reference described as the paraphrase in (334) hints at. Again without encroaching on the meaning of and or the naive reference of the coordinated DPs, the last class of substitution problems is resolved, once the Eventish internal structure of identity statements is realized. With all the event arguments in a supermonadic clause that adverbials, tense, and secondary predicates might modify, it becomes an empirical question to discover its grammar, leading to a canonical structure for the expression of spatiotemporal relations in a clause (section 11.1).
Introduction
67
(335) For some evenings in 1892, Venus was Hesperus (aligned with a crescent moon). For some evenings in 1892, Venus was the evening star (aligned with a crescent moon). (336) For some evenings in 1892, Hesperus was Venus *(aligned with a crescent moon). For some evenings in 1892, the evening star was Venus *(aligned with a crescent moon).50 From the contrast (335)–(336), for example, it appears that the temporal-frame adverbial, Tense, the secondary predicate, and the adverbial derived from the predicate nominal—in (335), while Hesperus or while the evening star—coincide in the events they describe, which are to be related by Eventish ℐdentity to events described by the subject’s derived adverbial, while Venus in (335). Both modification and the asymmetry it induces in (335)–(336) put the clause structure beyond reach of an analysis based on ⌜x = y⌝.51 Supermonadicity affords a dissociation despite Eventish ℐdentity between the events E1 and E2 separated by temporal location and by description that applies to just the one, such as that supplied by a secondary predicate. This coincides with a perspectival asymmetry of Figure and Ground (Talmy 1978, 1983; Gleitman et al. 1996) that attaches to all clauses in virtue of their canonical structure despite the symmetry of the relation expressed by lexical verb or adjective: (337) The humblest citizen is equal to the President. (338) The President is equal to the humblest citizen. (339) The President and the humblest citizen are equal(s). The humblest citizen and the President are equal(s). Sentence (337) is an elevation of the humblest citizen; (338), a demotion of the President; and (339) is of no apparent motion either way. Similarly, (340) implies a transposition in the prelude, and (341) is uninformative about the circumstances for the lapse in their identity: (340) The prelude was identical to the étude until a few notes were transposed. (341) The prelude and the étude were identical until a few notes were transposed. Both (342) and (343) imply that Soul and Kaddish ceased to be equally dead—but (342) says that Ray Charles revived Soul, and (343), that he revived the Jewish mourners’ prayer: (342) Soul was as dead as Kaddish until Ray Charles recorded “What’d I say?” in 1959. (343) #Kaddish was as dead as Soul until Ray Charles recorded “What’d I say?” in 1959.
68
Chapter 1
Since it is the same morpheme equal in both (337)–(338) and (339), it cannot be that it expresses an asymmetric relation or denotes states with an intrinsic asymmetry in (337)–(338) but describes symmetry in (339). If any concepts of symmetry are within reach, equal and the like would have to be its vocabulary. I assume that equal conveys the symmetric, metaphysical bedrock common to all these sentences. Factoring it out exposes that the asymmetry latent in the meaning of (337)–(338) (or of (340) or (342)–(343)) must derive from morphology elsewhere in the clause and independent of the lexical adjective. Whatever it means it must not obscure that the humblest citizen and the President are seen to arrive at the very same symmetric state in all of (337)–(339). That is, the asymmetry engages relations to other events or states related in some way to the state of equality, a fine-grained resolution into events that supermonadicity is designed for. These other states or events are presumably perspectival with no great metaphysical import, such as being the figure at the center against being background. In this flat-footed approach to meaning, if sentences differ, there is an Eventish morpheme expressing a relation to events that tells them apart; where the sentences agree, there is an Eventish morpheme for that too: (344) The Red Sox with the Yankees brawled on the pitcher’s mound. The Yankees with the Red Sox brawled on the pitcher’s mound. (345) The Red Sox and the Yankees brawled on the pitcher’s mound. (346) The Red Sox with the Yankees are in a tie for first place in the AL East. The Yankees with the Red Sox are in a tie for first place in the AL East. (347) The Red Sox and the Yankees are in a tie for first place in the AL East.. Thus, to the extent that how things are in the world is exactly the same in (344)–(345) and the same in (346)–(347), a fragment of (344)–(345) says that the Red Sox and the Yankees were the agents of a brawl, and a fragment of (344)–(345), that they are the parties to a tie. The rest of any of these sentences elaborates some other confection of the speaker’s interests and perspective on these events or on events leading up to them that discriminates being an accomplice from being the focus of attention.52 1.9
Contextualism logicized
Apparently, logical form is a shameless riot of inaudible phrases in a tawdry abstract syntax, unsung since Generative Semantics.53 A pursed-lipped chorus calling itself Compositionality holds nowadays that one cannot think without moving one’s lips, as linguistic semantics has fallen captive to slow readers: it isn’t part of the thought the utterance expresses if it isn’t a part that is itself uttered. Philosophy is anxious too, voicing it in work that frets about contextualism, which this section is meant to address. No one who understands the sentence Sam kicked Robin can fail to think
Introduction
69
of feet on hearing it and to understand that the sentence is not true unless a foot is involved. If unanchored to what is audible, is the analyticity of this understanding of the sentence sufficient for the analysis Sam foot-struck Robin? If not, then truth conditions conveyed aren’t always flagged in the analysis of the sentence uttered. In this book’s practice of descriptive semantics, I worry about getting the truth conditions right, when maybe they shouldn’t be gotten at all, according to the skeptic. How to tell when a twitch in truth conditions (or in anything else in understanding a sentence) deserves inaudible mention or no mention at all? The short answer is that grammar tells and grammar corroborates with linguistic evidence that is not limited to a language’s sound patterns. But something more deliberate ought to be said about the practice that lets truth conditions shape logical form, a practice I think robust enough to get by with little need for doctrine. To begin, logical form is grammar, the “real structure” of the sentence in a scientific description of natural language (Harman 1972; Stanley 2000) that reflects truth conditions in a way Sperber and Wilson ([1986] 1995, 72) explain: A logical form is a well-formed formula, …, which undergoes formal logical operations determined by its structure. … What distinguishes logical operations from other formal operations is that they are truth-preserving [my italics]: a deduction from a true representation P yields a true representation Q. … Given this relationship between truth and logic, it might seem that only a conceptual representation which is capable of being true or false can have a logical form. We see things rather differently. In essence, we will argue that for a representation to be amenable to logical processing, all that is necessary is for it to be well formed, whereas to be capable of being true or false, it must also be semantically complete: that is, it must represent a state of affairs, in a possible or actual world, whose existence would make it true. We take it that an incomplete conceptual structure can nevertheless be well formed, and can undergo logical processing.
The deductive calculus defines inference, well-formed proofs, in the syntax of the object language (i.e., the natural language’s logical form). As Sperber and Wilson ([1986] 1995) remark, it is a logic just in case its well-formed proofs are sound: (348) (Soundness) If Γ ⊢ φ, then Γ ⊨ φ. ‘If there is a well-formed proof of sentence φ from sentences Γ, then whenever Γ are all true, so is φ.’ For a natural language, a deductive calculus defined over its syntax is a theory of inferential competence for that language. Given that deduction is at the mercy of syntax, that is, logical form, linguistic behavior with truth-conditional import must often be explicitly reflected in logical form if inference in the deductive calculus is itself to remain sound: (349) Aristotle is a philosopher. Aristotle is not a philosopher.
70
Chapter 1
(350) Pa ¬Pa (351) I am a philosopher. I am not a philosopher. (352) She is a philosopher. She is not a philosopher. (353) That is a philosopher. That is not a philosopher. If for example (349) instantiates the schema for contradiction in (350), and speakers sometimes deduce contradiction from utterances of (351)–(353) and sometimes do not, then when they do not, it can only be that the logical forms of the sentences then uttered do not instantiate (350). Otherwise, speakers and hearers would deduce unsound contradiction from Aristotle’s utterance of the first sentence in (351) and Alexander’s utterance of the second. It will not eschew contradiction and the revulsion it provokes to suppose that all utterances of (351) have logical forms that instantiate (350) and then to plead that Aristotle and Alexander are spared disagreement because the alleged semantic interpretation of their otherwise identical logical forms provides them with distinct semantic objects so called: to Aristotle, the set of possible worlds in which he is a philosopher and to Alexander, the set of possible worlds in which he isn’t. What comfort is this to them or to their audience if the syntax of their deductive calculus, the grammar of well-formed proof that circumscribes their inferential competence, is not defined over the alleged semantic objects, sets of possible worlds? (Modus ponens, anyone?) So long as (351) has the syntax of (350), they will discover contradiction where there isn’t any and resent each other for it. That the logic defined for the logical form of natural language must be sound affords enough of a commitment to truth and truth conditions to warrant an articulation of logical form when truth-conditional content is encountered, just so that deduction in the logical language does not go astray. In this remark, I concur with Stanley’s (2000) rejection of “unarticulated constituents” (discussed further below) and with his arguments against allegedly found examples of such. But I do not see in this, a recommendation for a truth-conditional semantics that consorts with semantic objects, representations of an utterance’s truth conditions that semantic interpretation—some process or something—assigns to them. One may join Sperber and Wilson ([1986] 1995) against Stanley (2000) in denying that an utterance has— as determined only by its language and public facts about it (spatiotemporal coordinates, speaker, audience, etc.)—a literal meaning that is the utterance’s complete truth conditions.54 The articulation of logical form is not such pie in the sky. Due diligence in pursuit of a theory of inferential competence is enough to drive logical
Introduction
71
form toward more explicit representations, as the following improvisation on muchdiscussed examples illustrates. Suppose that schoolchildren are competent and confident in their usage of “x is at noon,” using it in conformity with the truth conditions in (354) or (355): (354) For any utterance of u = ⌜vi is at noon⌝, σ satisfies ⌜vi is at noon⌝ iff σ(⌜vi⌝) is at (℩t)(t = time(u) & the sun is at its zenith for location(u) at t). (355) For any utterance of u = ⌜vi is at noon⌝, σ satisfies ⌜vi is at noon⌝ iff σ(⌜vi⌝) is at (℩t)(t = time(u) & speaker(u)’s watch reads 12:00 at t & t is in daylight). They schedule lunch and ballgames and avoid scheduling conflicts accordingly, all New Yorkers from the same sandlot. Half then leave on vacation in California, and all of a sudden their interstate scheduling—Skype sessions, online baseball—is off 3 hours, while their intrastate scheduling is unaffected. At some point they catch on without quite understanding how it works, as reflected in their updated scheduling and inferential behavior. All continue to accept the inference in (356) in their intraand interstate communications: (356) I will sign on at Red Sox game time (when live broadcast begins). I will also sign on at Red Sox game time (when live broadcast begins). We will be signed on together. (357) I will sign on at noon. I will also sign on at noon. We will be signed on together. Before vacation, they also all accepted (357) in all their communications. Now (357) is sound only intrastate, to which they have adapted. What was learned? What now rescues (357) from the deductive calculus that validates (356)? It is not an adjustment to the truth conditions of their utterances, which continue as before to conform to (354)–(355). It is rather some revision “articulated” in logical form that will distinguish (357) from the logical form of (356): (358) I will sign on at noon for me. I will also sign on at noon for me. We will be signed on together. It could be an inaudible, implicit argument, as in (358), with the proviso that the two tokens of noon for me, containing first-person pronouns, not be mistaken to instantiate the same logical form ⌜noon for a⌝, just as the logical form of (351) is not to be mistaken for (350) if a contradiction there is not to be inferred. Or it could be that the kids learn that interstate communication is interlinguistic
72
Chapter 1
communication (Ludlow 2008, 2014), from which no inference follows when homonyms from different languages are in play: (359) I will sign on at noonNY. I will also sign on at noonCA. We will be signed on together. As Stanley (2000) argues, from arguments met in Partee 1989, the ambiguity of (360) between (361) and (362) is dispositive for the presence of an implicit argument in the logical form of (360):55 (360) Every kid will sign on at noon. (361) Every kid will sign on at noon his time. (362) Every kid will sign on at noon my time. Yet what has been said about the kids’ usage of noon can be said mutatis mutandis about their usage of tomorrow, as they come to discover that (364) in contrast to (363) is also invalid in interstate communication: (363) I will be signed on throughout the Red Sox game. I will also be signed on throughout the Red Sox game. We will be signed on the same time throughout. (364) I will be signed on throughout tomorrow. I will also be signed on throughout tomorrow. We will be signed on the same time throughout. And yet there is no ambiguity in (365) to corroborate the implicit argument, since tomorrow unlike noon is strictly indexical to the speaker’s clock: (365) Every kid will be signed on throughout tomorrow. If one is reluctant to conclude that invalid instances of (364) rest on an interstate, interlinguistic equivocation between homonyms tomorrowNY and tomorrowCA, take comfort that implicit arguments show similar, arbitrary restrictions elsewhere. Before and ago express nearly synonymous temporal relations, and yet only the former’s implicit argument behaves as a bound variable while the latter’s is obligatorily indexical to the speaker (Mitchell 1986; Partee 1989; Schlenker 1999): (366) (367) (368) (369)
Every Bar Mitzvah at Leonard’s was booked 18 months before. Every Bar Mitzvah at Leonard’s is booked 18 months before. Every Bar Mitzvah at Leonard’s was booked 18 months ago. *Every Bar Mitzvah at Leonard’s is booked 18 months ago.
The implicit argument for noon is either a bound variable or indexical, the one for tomorrow and ago, only indexical, and the one for before, only a bound variable. This variability is attested in the overt pronominal vocabulary of other languages.
Introduction
73
In Amharic (Schlenker 1999), literally Every soldier thinks that I am a hero can mean either what it means in English or that every soldier thinks that he is a hero. The Amharic first-person pronoun is either a bound variable or indexical, like the implicit argument for noon and unlike English I. Whether to engage the comparative grammar of implicit and overt pronouns or pursue an account of the code switching that might occur between interstate speakers is an empirical question (Ludlow 2008, 2014).56 But one way, the other, or a third, some articulation of logical form must come out to explain the inferential behaviors that distinguish (357) from (356) and (364) from (363). Except for here, the book does not mention inferential competence or behavior, and in its practice of descriptive semantics, it worries much about getting the truth conditions right. Reflecting now on that practice, it is a worry about just those aspects of a sentence’s truth conditions that threaten to undermine the soundness of inference in the natural language and a theory of inferential competence unless they are explicitly represented in logical form. Nothing in my preoccupation with truth conditions or in the representational details of any other aspect of meaning tells against the skepticism that has been aimed at truth-conditional semantics, presaged in the passage quoted from Sperber and Wilson and voiced in due deference to the problem for literal meaning it faces from language’s open texture (Bezuidenhout 2002; Ludlow 2008, 2014; Searle 1978, 1980) and from the ineffability of a sentence’s truth conditions—not to mention that I do not traffic in propositions, structured propositions, properties, or any other semantic objects. Logical forms are purely linguistic, syntactic objects—abstract descriptions classifying cognitive events—with the truth-conditional import that derives from the soundness of the proofs in which they are premises in a proof theory that is a theory of inferential competence. As linguistic objects, they are subject to the same scrutiny and justification in their details as any other fragment of grammar. Stanley’s (2000) evidence in (360)–(362) for an inaudible, articulated constituent in (360) is good. But I don’t know whether every effect of context with truth-conditional import is articulated in logical form or should be. Perhaps almost all are, as for any one of them it may be easy enough to show that its omission from logical form results in unsound inference somewhere. Anyway, for the articulation of logical form that proceeds piecemeal in this book, there will also be good, independent evidence for the linguistic object proposed. Even so, it could in truth be less meager. I have not yet found that poetic tradition in which the inaudible phrases of logical form count for metrical scansion—where I will sign on at noon is a line of iambic tetrameter. Now that would be evidence! Appreciate that it would be, to appreciate the sense in which logical form is here claimed to be a linguistic object. As it is, I may have to settle for the dull forecast of some neuroimaging correlate.
2
DP and DP
2.0 Coordinating generalized quantifiers, and the syntax and semantics of collectivized Right-Node Raising
The slippery slope to Conjunction Reduction (section 1.1) launches from the observation that (1) contains a sentential connective and that cannot be anything else, conjoining tensed sentences as it does, and yet the meaning of the Right-Node Raised constituent is collective (Perlmutter and Ross 1970; Jackendoff 1977; McCawley 1981a): (1) No more than one Columbia student openly convened and no more than thirty-five Harvard students privately formed a strike committee of thirty-six that the Columbia student chaired and the Harvard students staffed. The sentential connective, it appears, is no impediment to collective meaning. It persists in (2), where the sentences conjoined have now been reduced to a spoken subject and adverb: (2) No more than one Columbia student openly and no more than thirty-five Harvard students privately formed a strike committee of thirty-six that the Columbia student chaired and the Harvard students staffed. Adverbs being adverbs, there must be something for them to modify in the logical form for (2), although it is unspoken: (3) No more than one Columbia student openly participated and no more than thirty-five Harvard students privately participated, & it formed a strike committee of thirty-six that the Columbia student chaired and the Harvard students staffed.
76
Chapter 2
Adverbs being adverbs, they are optional. Omitting them in (4) but leaving in place what was modified but unspoken then points at a logical form for (5): (4) No more than one Columbia student participated and no more than thirty-five Harvard students participated, & it formed a strike committee of thirty-six that the Columbia student chaired and the Harvard students staffed. (5) No more than one Columbia student and no more than thirty-five Harvard students formed a strike committee of thirty-six that the Columbia student chaired and the Harvard students staffed. Whatever semantics solves collectivized Right-Node Raising and is supplied to the conjunction of the more full-blooded clauses in (1) applies without amendment when the clauses are greatly reduced, dropping as in (5) adverbs and any other spoken verbal elements. It thus promises a general account of the semantic facts on display in (1), (2), and (5). This cannot be said for a nonsentential and fashioned ad hoc for a coordination of DPs, so misparsing (5), which can be dismissed for the reason that its generalization to (1) and (2) is hopeless. The only problem in sight is the problem of collectivized Right-Node Raising—the only game in town—to which I now turn. The argument that and is always a sentential connective dead ends if all that has been wrung out of (6) in the previous chapter does not generalize to (7) and any other alleged conjunction of quantifiers including (5): (6) The Columbia students and the Harvard students formed the unbroken chain around the Pentagon. (7) No Columbia students and no Harvard students conspired to overthrow the government. If and is a sentential connective and quantifiers are sentential operators, the problem is to solve for Φ in (8): (8) No Columbia students Φ and no Harvard students conspired to overthrow the government. Putting the question this way, I depart from a line of research (e.g., Barwise 1979; Sher 1990; Westerståhl 1987; Szabolcsi 1997) that points to sentences such as those in (9) as evidence for branching or binary quantification. The failure of (9) to mean the same thing as (10) suggests to these researchers that the conjoined quantifiers must be compounded into a single quantificational prefix (which also robs and of its meaning).1 (9) Most stars and most dots are all connected. Few stars and few dots are all connected. (Barwise 1979)
DP and DP
77
(10) Most stars are all connected, and most dots are all connected. Few stars are all connected, and few dots are all connected. If however the ellipsis in (8) can be solved, the quantifiers and the conjunction and will appear in their normal guise. As argued in section 1.1 and just rehearsed, collectivized Right-Node Raising as in (11)–(13) already demands as much. Whatever is its semantics in (11)–(13) will apply to (7), in which the remnant Φ of Right-Node Raising happens to go unspoken: (11) No Columbia students secretly and no Harvard students openly conspired to overthrow the government. (12) No Columbia students drafted and no Harvard students redacted their joint manifesto for the revolution. (13) Not many a Columbia student proposed to the strike committee to adopt a resolution to encourage violent resistance to—and not many a Harvard student accepted to endorse to the strike committee a call for illegal action against—attempts by police to prevent them from chaining each other to the White House fence. In (7), if, as in the preceding chapter, ‘participate(e,x)’ or ‘W(e,x)’, as it is here and in appendix 2, is the thematic relation related to the subject, then the expansion of Φ at the very least contains this thematic relation, in which respect the logical forms for (11)–(13) could hardly differ: (14) No Columbia students … W(e,x) …, and no Harvard students … W(e,x). … Of course Φ should not be expanded to the point where it represents the logical form of (15), which, even if judged grammatical, does not mean the same thing as (7) (cf. (9) vs. (10)). (15) No Columbia students conspired to overthrow the government, and no Harvard students conspired to overthrow the government. Translating away DP and DP
2.0.0
In fixing Φ, the logical form for (7) (and (11)) must render three facets of (7)’s meaning, the last of which is explained below: (16) i.
The conspiracies to overthrow the government concerned involve only Columbia students and Harvard students, and ii. the quantifiers apply to such conspiracies and not to other actions that the students may have engaged in, and iii. the quantifiers emulate branching or binary quantification.
78
Chapter 2
In one way or another, the paraphrases (17)–(18) neglect (16). Only (19) is equivalent to (7), where each conjunct in (19) seems to import the italicized information from the other. (17) No Columbia students participated (‘W(e,x)’), and no Harvard students conspired (with Columbia students) to overthrow the government. (18) No Columbia students participated in conspiracies to overthrow the government, and no Harvard students participated in conspiracies to overthrow the government. (19) No Columbia students participated with Harvard students in conspiracies to overthrow the government, and no Harvard students participated with Columbia students in conspiracies to overthrow the government. The logical form for (7) will derive the equivalent of (19) from occurrences of the unspoken plural, event pronouns, which as argued in chapter 1, are the bearers of collective reference that and, the sentential connective, is not. For the definite descriptions conjoined in (6), it would suffice as in chapter 1 for a single token of an event pronoun referring to the Columbia students’ participation and the Harvard students’ participation to refer to events that are then said to be a formation of the unbroken chain around the Pentagon—The Columbia students participated, and the Harvard students participated, and they, these events, were formation of an unbroken chain around the Pentagon. The step up to generalized quantifiers requires a second token of a plural event pronoun at the gap in the first clause left by Right-Node Raising. The two tokens will, in effect, supply as part of their own content the italicized information in (19) that the conjuncts seem to import from each other. How it is imported, what exactly the content of the event pronouns is, will render the third facet of meaning, (16iii), that branching or binary quantification is emulated, as (20) will illustrate conjoining a pair of singular quantifiers: (20) No Columbia student and no Harvard student were a strike committee all by themselves. Suppose that exactly two Columbia students are on strike committees, each leads a committee consisting of herself and thirty-five Harvard students, and there are no other strike committees. Since neither strike committee is a pair, (20) is unequivocally true. To which end, the logical form for (20) should reach for (21) (cf. (19)): (21) No Columbia student participated with a Harvard student in being a strike committee all by themselves, and no Harvard student participated with a Columbia student in being a strike committee all by themselves.
DP and DP
79
In contrast, the imagined context falsifies (22) and (24) and their target paraphrases, while leaving (26) and its paraphrase true: (22) No Columbia student and no Harvard students were a strike committee all by themselves. (23) No Columbia student participated with Harvard students in being a strike committee all by themselves, and no Harvard students participated with a Columbia student in being a strike committee all by themselves. (24) No Columbia students and no Harvard students were a strike committee all by themselves. (25) No Columbia students participated with Harvard students in being a strike committee all by themselves, and no Harvard students participated with Columbia students in being a strike committee all by themselves. (26) No Columbia students and no Harvard student were a strike committee all by themselves. (27) No Columbia students participated with a Harvard student in being a strike committee all by themselves, and no Harvard student participated with Columbia students in being a strike committee all by themselves. Each of (20), (22), (24), and (26) appears to quantify over pairs, Columbia students individual or plural paired with Harvard students individual or plural as number on the NPs dictates, and to assert that no such pairs are of students constituting a strike committee all by themselves. Forgoing direct quantification over pairs, if one is to emulate it, it seems at first blush essential that the italicized phrases in the paraphrases be indefinite descriptions (singular or plural as indicated). After all, in interpreting the first sentential conjunct in (20), there is no pair of a Columbia student and a Harvard student allowing a definite description the Harvard student to refer to h, and it would seem that in this context with seventy Harvard students on strike committees, a singular definite description would fail to refer. Similarly, (22) is unequivocally false in this context, but a paraphrase in which the indefinite description Harvard students in (23) were replaced with a definite description referring to the seventy Harvard students would be mistakenly true, since the individual Columbia student is on a strike committee all alone with only half the Harvard students. Granted the intended paraphrases, with all this talk of definite event pronouns in logical form, nothing has yet been said about reference to the events’ participants, definite or indefinite. However it goes, the meanings at issue find no resolution in branching or binary quantification itself despite the emulation. The slippery slope to Conjunction Reduction again slickens the way to a more general problem and solution. In (28)–(29), with sentential and beyond doubt and quantifiers at too great a remove for a single
80
Chapter 2
compound quantifier, the truth conditions in this context are nevertheless exactly the same mutatis mutandis: (28) No Columbia student(s) secretly and no Harvard student(s) openly were on a strike committee all by themselves. (29) No Columbia student(s) launched and no Harvard student(s) joined a strike committee with only themselves in it. On the same grounds and as above, the target interpretations are equivalent to paraphrases that help themselves to indefinite descriptions, and in no paraphrase for (28)–(29) will any term make definite reference to the Columbia students, the two of them in this context, or to the seventy Harvard students, and no term will attempt singular reference to the Columbia student where there are two or to the Harvard student where there are seventy. Yet, within each conjunct of (28)–(29), it remains as if one were still quantifying over Columbia and Harvard pairs so that any definite description actually spoken in situ refers to the Columbia or Harvard student or students of the pair:2 (30) No Columbia student(s) secretly and no Harvard student(s) openly were on a strike committee in which the Columbia student(s) led and the Harvard student(s) followed. (31) No Columbia student(s) launched and no Harvard student(s) joined a strike committee in which the Columbia student(s) led and the Harvard student(s) followed. Apparent quantification over such pairs within the scope of the quantifier in each conjunct is evident in those interpretations of (32)–(34) that the imagined context falsifies with more than thirty-five—in fact, seventy—Harvard students among the strike committees described:3 (32) Not many a Columbia student—exactly two, in fact—and no more than thirty-five Harvard students (in total) were a strike committee in which the Columbia student led and the Harvard students followed. (33) Not many a Columbia student secretly—exactly two, in fact—and no more than thirty-five Harvard students openly were on a strike committee in which the Columbia student led and the Harvard students followed. (34) Not many a Columbia student—exactly two, in fact—launched and no more than thirty-five Harvard students joined a strike committee in which the Columbia student led and the Harvard students followed. If overt definite description, the Columbia student in (32)–(34), manages to refer to a Columbia student as described in the antecedent clause, there ought to be no
DP and DP
81
further impediment to referring to the events in which that Columbia student participates as antecedently described. The apparent arrangement of Columbia and Harvard students into pairs and quantification over them is a pale shadow of spatiotemporal orientation and the grammar of perspectives, frames of reference, and sequence of spatiotemporal reference (sections 1.7–1.8, 9.0; chapter 14). Recall that the chess tutor overheard in section 1.7 to have said (35) does not intend the contradiction that the same three chessmen will always checkmate and only stalemate: (35) A bishop, a rook, and a pawn will always checkmate in ten moves, but a rook, a pawn, and a bishop only stalemate. He rather quantifies over chessboards, frames of reference within which the chessmen are assigned fixed addresses, which rescue (35) after adverbialization along the lines of (36): (36) A bishop here1, a rook here2, and a pawn here3 while a bishop here1, a rook here2, and a pawn here3 will always checkmate in ten moves, but a rook here1, a pawn here2, and a bishop here3 while a rook here1, a pawn here2, and a bishop here3 only stalemate. Singular reference is supported relative to the frame of reference, referring as it were to the bishop on the chessboard, the rook there, and the pawn there: (37) A bishop, a rook, and a pawn will always checkmate in ten moves but not without sacrificing the bishop, the rook, or the pawn. The nonmaximal reference of the singular definite descriptions in (37) indicates its tacit restriction to a selective perspective or frame of reference, such as is provided if all NPs are addressed under one (sections 1.7–1.8, 9.0; chapter 14; see also appendix 1). Reference to frames of reference is equally effective in uncontroversial instances of Right-Node Raising at safe remove from the illusion of branching or binary quantification: (38) A bishop (here) retreats, a rook (here) advances, and a pawn (here) remains in place to checkmate in ten moves with the sacrifice of the bishop, the rook, or the pawn. And it generalizes to generalized quantifiers and to nondemonstrative frames of reference or chessboards: (39) Not many a bishop from any position on a chessboard ever retreated, not many a rook from a second position ever advanced, and not many a pawn at a third position ever remained in place to checkmate in ten moves without the sacrifice of the bishop, the rook, or the pawn.
82
Chapter 2
Any chessboard under consideration—(39) is likely about one side’s endgames—has on it just one of the side’s bishops, one of its rooks, and one of its pawns if singular reference is to work as intended. Given that one of each is never many, the plain meaning of (39) is not the triviality in (40): (40) [∀c: chessboard(c)] (Not many a bishop from any position on c ever retreated, not many a rook from a second position on c ever advanced, and not many a pawn at a third position on c ever remained in place to checkmate in ten moves without the sacrifice of the bishop on c, the rook on c, or the pawn on c.) It is rather to say that not many a bishop, not many a rook, and not many a pawn have been in the games in which the three of them checkmate without the sacrifice of one of them, implying that not many such games have ever been. For this quantificational force, an existential quantification over chessboards or frames of reference occurs within each DP—that is, within the scope of not many—parallel to the quantification over pairs apparent in (32)–(34). Yet, concurrently, none of this quantification makes claims about any chessboards other than those with one bishop, one rook, and one pawn, as if as in (41): (41) Not many a bishop from any position on a chessboard with a rook in a second position and a pawn in a third position ever retreated, not many a rook from a second position on a chessboard with a bishop in first position and a pawn in a third position ever advanced, and not many a pawn at a third position on a chessboard with a bishop in first position and a rook in second position ever remained in place to checkmate in ten moves without the sacrifice of the bishop, the rook, or the pawn. If NPs occur in DPs with a morphology that describes location in a spatiotemporal frame of reference as in (42), and a sequencing of spatiotemporal address, like sequence of tense, relates successive addresses to antecedent addresses, the logical form for (39) is along the lines of (43) with a chessboard for frame of reference: (42) [DP Q X : ∃E ∃f (Address[E,f] NP[E,X])] [DP Det [AdrP Adr NP]] (43) Not many a bishop from any position on a chessboard ever retreated, not many a rook from a(nother/second) position on such a chessboard ever advanced, and not many a pawn at a(nother/third) position on such a chessboard ever remained in place to checkmate in ten moves without the sacrifice of the bishop, the rook, or the pawn. Whether the addresses are an ordinal series, first, second, and third, or each is described simply as other than those previous, relative position among them makes
DP and DP
83
little sense except under the same frame of reference. These particulars and the general pragmatics that accommodates antecedent clauses to the presuppositions of later definite descriptions imply that what takes the form of (43), reflecting formally just the structure in (42) and the sequence of spatiotemporal addresses, is understood as in (41). A certain kind of chessboard is at issue—a chessboard with a bishop, a rook, and a pawn—but its description unfolds only as subsequent clauses elaborate with further details of what is on location. Only that same kind of chessboard does each clause promise to make any claim about right from the start, even though the content of those claims is filled in only after the three clauses are uttered and the description of the chessboard at issue is completed.4 If the DPs have their content relativized to addresses under the same frame of reference, then so will the definite descriptions, the bishop, the rook, and the pawn, for which they are antecedents. Revisiting the Columbia and Harvard students, a paraphrase more explicit than (34) is thus (44), which in turn becomes (45) paraphrasing an event pronoun too: (44) Not many a Columbia student anywhere—exactly two, in fact—launched and no more than thirty-five Harvard students anywhere (in total) joined a strike committee in which the Columbia student there led and the Harvard students there followed. (45) Not many a Columbia student anywhere—exactly two, in fact—launched and no more than thirty-five Harvard students anywhere (in total) joined so that the events of the Columbia student there launching and the Harvard students there joining amounted to a strike committee in which the Columbia student there led and the Harvard students there followed. It is thus a few steps from (7) or any other alleged conjunction of quantifiers to a conjunction of sentences. For fixing Φ in (8), the three facets of meaning in (16) that a logical form for (7) must render are all in plainer sight in (32)–(34) and (39). Any logical forms for (32)–(34) and (39) will carry instructions for a logical form for (7) that likewise coordinates sentences given the subatomic sentences that Davidsonian decomposition and supermonadicity introduce. The argument that and is always a sentential connective, even where under duress in the alleged coordination of generalized quantifiers, could end here with exhibit (32)–(34) and (39) in its defense. Just do unto (7) as you would do unto (32)–(34) and (39). What Eventish does unto them (section 2.1) is first to derive the collectivization of collectivized Right-Node Raising from plural descriptive event pronouns, with content the likes of which has been paraphrased in (45). Since (44) deploys both quantifiers to measure just the events of a Columbia student launching and Harvard students joining (see (16i–ii)), one token of the plural descriptive event pronoun must refer to them in the first conjunct within the scope of not many a Columbia
84
Chapter 2
and another token refers to the same within the scope of no more than thirty-five Harvard students in the second conjunct. Collective reference to what Columbia student and Harvard students do is by plural event description. But then its restriction to just what a single Columbia student does in a collective with some Harvard students, when there are many a Columbia student (see (16i–ii)), is by further restriction of that event description to what goes on there, at an address at which there be only one Columbia student, itself referred to by descriptive anaphor covariant with students quantified over. Thus, the fundamental semantics of collectivized RightNode Raising (and as a special case, the illusory coordination of DPs), namely, its collectivization across the conjuncts from which it is raised and yet its continued dependence on the quantification within those conjuncts, derives from the distribution of plural event descriptions that themselves contain dependent parameters. In Eventish (recall from chapter 1), there is little daylight between the overt sentences and descriptive pronoun in (46) (see (188) in chapter 1) and the subatomic sentences and descriptive pronouns that Eventish unveils in (47): (46) Biff did something and possibly Tiff did something. It started the riot. (47) Biff and possibly Tiff started the riot. How it is in whole paragraphs for pronouns, sequence of tense, and any other referential device from one sentence to the next, so it is for them in the subatomic realm of Eventish.5 The vagaries of reference that engage pronouns at large also engage its unspoken event pronouns, descriptive anaphora that in various contexts are singular or plural, nonmaximal in reference despite definite description, nonrigid and “sloppy” when identical copies are copied into the scope of different operators, subject to “telescoping”, and so on. Exploiting some of these features, it has just been outlined how the event pronoun is a focus in the Eventish solution to the problem of collectivized Right-Node Raising. It proves itself in a broader syntax and semantics for Right-Node Raising, collectivized or not, and in the explanation it affords for how the construction patterns in natural language, as one might expect from what purports to be bedrock grammar. In section 2.3 especially and in subsequent sections, whether or not a Right-Node Raised phrase, its number agreement and the (in)definite descriptions, quantifiers and anaphora within it are interpreted distributively—that is, dependent on the quantification within the clauses from which it is Right-Node Raised—or collectively is a question about the content and interpretation of the event pronoun that prefixes the phrase (and of the token of this pronoun that is copied into the gap in the first conjunct). This variation in the meaning of the Right-Node Raised phrase will probe the locations of event pronouns and quantifiers in sentence structure. Word order (pre- vs. postverbal subjects), adverb placement, and the choice of DP
DP and DP
85
subject (distributive quantifiers vs. (in)definite descriptions) interact with the syntax of event quantification. Some features of Eventish and translation into it ought to be highlighted as they recur throughout. First, unpacking Right-Node Raising populates sentences with unspoken copies of what is pronounced only once. In copying an indefinite description, a Columbia student, or any other scope-dependent phrase, its multiple tokens do not therein imply that the sentence is to be construed as a report of more than one Columbia student. That rather depends on whether the subatomic clauses in which these copies appear, such as ‘[a(n) x: Columbia student(x)] Agent[e,x]’, describe the same event e or not. Only if these clauses fall within the scope of different event quantifiers can it report the actions of different Columbia students. Correlatively, definite descriptions that are exact copies with the same content denote differently only if they fall within the scope of different operators that quantify into them, inevitably different event quantifiers, or quantifiers over frames of reference. An empirical claim about translation into Eventish is that number agreement in natural language pronounces an Eventish event pronoun. Singular number agreement means something like “the eventery therei all with the same solo participant that proj” and plural number agreement, “the eventery therei with one or more participants that proj,” where agreement supplies a copied event description to proi, typically a thematic relation. In (48), the singular number agreement pronounced ‘-s’ within the Right-Node-Raised phrase in the second conjunct and copied unpronounced with the exact same content into the first conjunct means in both “the eventery therei all with the same solo participant who is its Agent”: (48) Every Columbia student and every Harvard student has sung a song of peace and love. As above, this definite description of events manages to denote different events, a Columbia student’s action in the first conjunct and a Harvard student’s action in the second, only if different operators bind the two tokens of the parameter therei. It is in the nature of Eventish (section 1.5.0; Schein 1993) that the distributive quantification in (49)–(50) is concurrent with event quantification within its scope if a fourpart harmony is ever to comprise four solo subevents: (49) In perfect four-part harmony, every songstress sang solo a single melodic line. (Davies 1991; Taylor 1985) (50) Every songstress sang solo a single melodic line, and it was perfect four-part harmony. It being perfect four-part harmony, every songstress sang solo a single melodic line.
86
Chapter 2
That much can be counted on to license therei and singular number agreement in (48). In contrast (Hoeksema 1983, 1988; Lasersohn 1995, 109ff.), (in)definite descriptions (names included) do not license singular number agreement: (51) *A Columbia student and a Harvard student has sung a song of peace and love. (52) A Columbia student and a Harvard student have sung a song of peace and love. So it becomes a further empirical claim in this chapter that distributive quantifiers and (in)definite descriptions are distinct grammatical classes that differ in the event quantification licensed. As a result, in (51), the event description that translates singular number agreement will fail to refer in the absence of independent existential event quantifiers within the conjuncts to bind therei. If number agreement pronounces a definite description of events and the singular and plural morphemes mean what they mean, then it is foretold (section 2.2) what to conclude about languages like Lebanese Arabic that have first-conjunct agreement, an option to agree with the first conjunct as in (53) as an alternative to the expected agreement in (54): (53) DeHk-et alia w marwaan Laughed.3fs Alia and Marwan ‘Alia and Marwan laughed.’ (54) DeHk-u alia w marwaan Laughed.3mp Alia and Marwan ‘Alia and Marwan laughed.’ If Alia and Marwan participate in exactly the same manner, say, as Agents, then singular number agreement must fail in its attempt to refer to the eventery there all with the same solo participant who is its Agent, as it does in (51). There must rather be a difference between them, for example, that Alia is the principal agent and Marwan, her accomplice, so that singular number agreement in (53) refers to the eventery there all with the same solo participant who is its principal agent—indeed to what only one, Alia, has done. If number agreement is semantic as suggested, then any difference of number agreement implies a difference of meaning. In all the languages known to me to have it, first-conjunct agreement is semantically conditioned, for which section 2.5 elaborates an Eventish semantics. Now if event pronouns are essential in the Eventish solution to all collectivized Right-Node Raising, then event pronouns must be free to collect up rather heterogeneous events, Columbia students’ launchings but Harvard students’ joinings in the logical form for (55) as paraphrased in (56):
DP and DP
87
(55) Not many a Columbia student launched and no more than thirty-five Harvard students joined a strike committee in which the Columbia student led and the Harvard students followed. (56) Not many a Columbia student anywhere launched and no more than thirtyfive Harvard students anywhere joined so that the events of the Columbia student there launching and the Harvard students there joining amounted to a strike committee in which the Columbia student there led and the Harvard students there followed. If number agreement is yet just another event pronoun, the occasion might be expected when it too divides its reference between heterogeneous events, as it does in those languages where a comitative phrase contributes to plural number agreement: (57) Marta con María levantaron el piano Martha with Maria lifted.pl the piano (Camacho 1997, 206)
[Spanish]
The meaning of plural number agreement in (57) is approximately “the eventery therei with one or more participants that are Agents and With-ers” (see section 2.6). Like first-conjunct agreement, comitative plural number agreement is semantically (and syntactically) conditioned. Imagine for the love of mischief a language with both constructions. Join it to a theory of semantic number agreement that held as common coin that number agreement refers to what a nominal subject refers to. Then Martha with Maria refers to that thing of Martha and Maria that the comitative plural refers to, Martha and Maria refers to that other thing of Martha and Maria that first-conjunct singular agreement refers to when it occurs so, and Martha and Maria refers to that third thing of Martha and Maria that plural number agreement refers to when agreement is plural. An error of grammar paves the way to metaphysical hell. That number agreement is an Eventish event pronoun and that distributive quantifiers and (in)definite descriptions license different schema for event quantification are two empirical claims about the translation between natural language and the logical language Eventish. Three more, in anticipation of the work they do in later sections, should be mentioned here because of their general import for translation. First, as far as Eventish is concerned and the thesis that and is always a sentential connective, (59) is a compliant translation for (58), as appropriate as (61) is for (60): (58) The Columbia students and the Harvard students surrounded the Pentagon. (59) [The X: C(X)] ∃ei Agent(ei,X) and [The X: (X)] ∃ej Agent(ej,X) & [℩Ei,j: proi,j] Cause[Ei,j, …] …
88
Chapter 2
(60) The Columbia students surrounded the Pentagon. (61) [The X: C(X)] ∃ei Agent(ei,X) & [℩Ei: proi] Cause[Ei, …] … There is, however, in (59) no trace of Right-Node Raising as such—there is no obvious gap in the first clause. To force the translation of (58) to include a gap as in (62), (62) [The X: C(X)] ∃ei Agent(ei,X) & Δk and [The X: (X)] ∃ej Agent(ej,X) & [k [℩Ei,j: proi,j] Cause[Ei,j, …] …]k , it will be said (section 2.4) that conjunction in natural language is not merely the sentential, logical connective ‘&’. It means only that, but natural language conjunction comes with the further grammatical property that it renders its conjuncts opaque to complementation. Thus, any bound morpheme, such as the thematic relation ‘Agent’ that affixes to the verb surround, must find its host within the conjunct in which it occurs—hence, the gap ‘Δk’ in (62), into which the verb must be copied. The second and third empirical claims to mention are also programmatic. Logical form and translation are grammar and syntax. Thus, all its elements carry grammatical weight—the next empirical claim (section 2.3). A clause in logical form that is longer than another by virtue of the presence of some unspoken event quantifier is strictly a larger clause of syntax, which may therefore differ in its grammar, perhaps becoming opaque to movement, much as the coordination of larger clauses in (64) renders it opaque to the movement seen in (63): (63) What did Nixon know about and admit to? (64) *What did Nixon know about and did he admit to? *What did Nixon know about and he admitted to? Finally, with the same literal-minded regard for the syntax of logical form, copies in logical form are identical even to the alphabetic identity of their variables, a motive for which is illustrated in the following contrast (section 2.3.1). In (65)–(67), Sam and Joe can be taken to work on different houses unless a temporal relation between their work intrudes, in which case they are understood to work on the same house: (65) a. Sam repainted and Joe re-sided a new house. b. Sam repainted and later Joe re-sided a new house. (66) a. What Sam repainted and Joe re-sided was a new house. b. What Sam repainted and later Joe re-sided was a new house. (67) a. What Sam repainted and what Joe re-sided was a new house. b. What Sam repainted and what Joe later re-sided was a new house. As above, for each to work on a house of her own, is for two syntactically identical tokens of ‘[a(n) x: new house(x)] Patient[e,x]’, in the same variable e, to fall in
DP and DP
89
the scope of different event quantifiers within their respective conjuncts. But, ‘later(e,e′)’ is an asymmetric relation in alphabetically distinct variables, relating the repainting e′ to the re-siding e. With later present, the same event variable cannot be used in both clauses, which apparently disrupts the narrow-scope reading of a new house. 2.0.1 Conspectus
The slippery slope to Conjunction Reduction dismisses any semantics that neglects to see the same handiwork in (68)–(70): (68) Not many a Columbia student and no more than thirty-five Harvard students (in total) were on a strike committee in which the Columbia student led and the Harvard students followed. (69) Not many a Columbia student secretly and no more than thirty-five Harvard students openly were on a strike committee in which the Columbia student led and the Harvard students followed. (70) Not many a Columbia student launched and no more than thirty-five Harvard students joined a strike committee in which the Columbia student led and the Harvard students followed. Slickening the slippery slope, section 1.5.1 also pressed dilemmas for the syntax of DP-coordination that confound it even if there were no collectivized Right-Node Raising for it to assimilate to in a uniform semantics for (68)–(70). Never to let slip from sight the argument, this section revisits the dilemmas in syntax (section 1.5.1). The disparate features of Eventish and empirical claims mentioned in the previous section that find in this and the next chapters independent justification and fuller development rally to resolve DP-coordination’s dilemmas the moment sentential coordination and Right-Node Raising are acknowledged. Singular number agreement (section 1.5.1) is precluded with a coordination of singular quantificational DPs in case there is a dependency internal to it of the second DP on the first. In the scopal dependency illustrated in (71)–(76), the first DP must include within its scope both the second DP and the entire sentence in order to license the negative polarity items any and ever: (71) *No rocker and any roller has ever grooved to a funky disco beat. (72) No rocker and any roller have ever grooved to a funky disco beat. (73) No rocker and no roller has ever grooved to a funky disco beat. (74) No rocker and no roller have ever grooved to a funky disco beat. (75) *Not any student of mine yesterday and any one of yours today has conjectured and proven p and not p.
90
Chapter 2
(76) Not any student of mine yesterday and any one of yours today have conjectured and proven (between them) p and not p. (77) No student of mine yesterday and none of yours today has conjectured and proven p and not p. (78) No student of mine yesterday and none of yours today have conjectured and proven (between them) p and not p. Absent a dependency, as in (73) and (77), there is no constraint against singular agreement. If coordination composes a derived DP, it is unexplained why the result should preclude singular agreement just in case the first conjoined DP includes the second in its scope. In chapter 3, it is argued that the extra layer of event quantification needed to license the dependent interpretation of singular agreement amounts to a larger clause so that the clauses coordinated with singular agreement in (73) and (77) and in (71) and (75) are somewhat larger than those with plural agreement in (74) and (78) and in (72) and (76). Larger enough, it appears (section 3.2.2), to entrap within the first clause its subject, no rocker or not any student of mine, preventing the quantifier from reaching the scope needed to license the negativepolarity items. In contrast, the smaller clauses coordinated with plural number agreement afford the quantifier scope outside the coordination. Singular agreement is also precluded when the second DP is dependent in containing a pronoun covariant with the first. But prior to this worry, it is already worrisome how the second conjoined DP in (79)–(80) can contain a pronoun covariant with the first without falling within its scope: (79) No more than a single senator and no more than a single congressman from his home state have cosponsored an antiwar resolution. (80) Not many a professor and not many a student of hers are conspiring to revive operant conditioning. Recognizing sentential coordination, it reduces to the more familiar problem of “telescoping” attested in (81)–(84). It will benefit from the same account whatever that may be, which here is one relying on the intervention of an unspoken adverb to license the singular pronoun, as in (85) and (86): (81) No more than a single senator introduced and no more than a single congressman from his home state cosponsored an antiwar resolution. (82) Not many a professor secretly and not many a student of hers openly are conspiring to revive operant conditioning. (83) No more than a single senator sponsored an antiwar resolution, and no more than a single congressman from his home state helped him; and they were junior.
DP and DP
91
(84) Not many a professor endorses operant conditioning, and not many a student of hers does either, but they must all be behaviorist rats. (85) No more than a single senator sponsored an antiwar resolution, and in any event of a senator sponsoring an antiwar resolution, no more than a single congressman from his [ = the senator in that event] home state helped him; and they were junior. (86) No more than a single senator participated, and in any event of a senator participating, no more than a single congressman from his [ = the senator in that event] have cosponsored an antiwar resolution. Singular agreement, as remarked in section 1.5.1, is yet more worry for alleged DPcoordination. Supported in (87), it is ruled out if the DP-coordination embeds a covariant pronoun, as in (89) and (91): (87) No rocker and no roller has ever grooved to a funky disco beat. (88) a. No rocker and no roller have ever each grooved to a funky disco beat. b. No rocker and no roller have ever grooved together to a funky disco beat. (89) *No rocker and none of his mamas has ever grooved to a funky disco beat. (90) No rocker and none of his mamas have ever grooved to a funky disco beat. (91) *No math genius and no protegée of hers has conjectured and proven p and not p. (92) No math genius and no protegée of hers have conjectured and proven (between them) p and not p. A covariant pronoun, however, disrupts neither plural number agreement nor collective interpretation in (90) and (92). Once again number agreement is implicated in the syntax and semantics of apparent DP-coordination, and something is to be said about its strange interaction with coordination and covariant pronouns within. It remains to be discovered what is defective about twinning “telescoping” and singular agreement, which requires both the extra layer of quantification in support of singular agreement and the interpolated adverbial phrase for the benefit of the pronouns’ singular reference: (93) No rocker somewherei participates … & in those events therei all with the same one participant, … and in those events therei, none of his [ = the rocker in those events] mamas somewherei participates … & in those events therei all with the same one participant+have ever grooved to a funky disco beat. Perhaps the clause is now so large that it rejects the adverbial modification derived from a tenseless subatomic clause, as in (93) or (86), although it may yet accept
92
Chapter 2
those adverbs derived from fully tensed antecedent clauses as in (87). The extra layer of quantification, ‘somewherei’ essentially makes the clause opaque to further subatomic modification, all of which should have taken place properly within—the subatomic being subatomic after all. The alternative explored here (section 2.3.1) and mentioned earlier does not exploit the difference in clause size and locates the defect underlying (93) elsewhere. Note that the “telescoping” adverb and the event pronouns translating singular number agreement all take for their antecedent the same subatomic clause, the first in (93). Note also that the two tokens of singular agreement affix to two tokens of the same description, ‘have ever grooved to a funky disco beat’, the first deleted under identity with the second, the Right-Node Raised phrase in the second conjunct. In tokening identical descriptions—that is, in copying the content of an antecedent into its anaphor—suppose that the identity imposed extends to the use of variables, so that a deleted phrase uses ‘Ei’ just in case the spoken phrase does, ‘have ever grooved to a funky disco beat[Ei]’. The same holds among the variables used in the “telescoping” adverb and the two event pronouns translating singular agreement, all of which share the same antecedent. Alongside these constraints on the variables used, note that the events described by the adverb are to be linked by some relation, paraphrased ‘in’ in (93), to the events described by the clause the adverb modifies. Of course this relation must use distinct variables, ‘in[Ei,Ej]’ for i≠j and not ‘in[Ei, Ei]’; thus, if the adverb describes events Ei the clause it modifies must describe events Ej. This sets up competing constraints on the variables’ identity that threaten to be inconsistent, and it remains to be seen (in section 2.3.1) that the logical syntax of anaphora, “telescoping,” and number agreement makes impossible a parse like (93) for (89) and (91) in the singular and yet parses (90) and (92) in the plural. Once “telescoping” joins in in (87)–(92), there is an embarrassment of riches to look at in the syntax. It could be that the extra layer of quantification simply pushes out the “telescoping” adverb, as first entertained, or perhaps it is an illicit use of variables subject to a condition that anaphora copy the same variables as their antecedents. Anyway, recognizing a concealed sentential coordination has prospects; it is hardly the blind alley that denying the sentential coordination has run into cornered in (71)–(76) and (87)–(92). There is thus abundant reason, in semantics and syntax independently, to reject DP-coordination as such and to turn elsewhere. There is the slippery slope to collectivized Right-Node Raising, and there is this mysterious interaction that holds number agreement with the alleged DP-coordination hostage to what goes on inside it between its conjuncts.
DP and DP
2.1 2.1.0
93
Syntax and semantics for DP and DP Collectivized Right-Node Raising
In an example such as (12) (or (13) or (34)), it is uncontroversial that the first clause contains a gap Δ where the remnant of Right-Node Raising is plainly unspoken: (12) No Columbia students drafted and no Harvard students redacted their joint manifesto for the revolution. (94) No Columbia students drafted Δδ6 and no Harvard students redacted [δ their joint manifesto for the revolution]. It however does not follow—not from the thesis that and is a sentential connective—that (7) or (11) need contain such a gap, as and’s demand for a sentence may be satisfied by a token of the participation relation W alone without indication of anything else missing: (7) No Columbia students and no Harvard students conspired to overthrow the government. (11) No Columbia students secretly and no Harvard students openly conspired to overthrow the government. (95) No Columbia students W-ed (secretly) and no Harvard students W-ed (openly) conspired to overthrow the government. So it is to be stipulated as a point of grammar that (7) and (11) are indeed instances of Right-Node Raising, gap and all, as in (96), as its meaning equivalent to (19) seems to warrant: (96) No Columbia students W-ed (secretly) Δδ and no Harvard students W-ed (openly) [δ conspired to overthrow the government]. (19) No Columbia students participated with Harvard students in conspiracies to overthrow the government, and no Harvard students participated with Columbia students in conspiracies to overthrow the government. Section 2.4 offers a wonky rationalization for this fact of grammar. Presumably the same structure occurs in (6), here not for meaning’s sake but to maintain a syntactic uniformity conjoining DPs: (6) The Columbia students and the Harvard students formed the unbroken chain around the Pentagon. At issue then is how to interpret the deletion sites Δ of Right-Node Raising, the leading problem of this chapter. A second event pronoun at the deletion site denoting the events described by the Right-Node Raised phrase joins the one already
94
Chapter 2
introduced for the sake of collective predication (section 2.1.3). A spoken analog to this backward event pronoun is something like such in (97): (97) a. No Columbia students participated in (any) such, and no Harvard students participated in—a conspiracy of theirs to overthrow the government. b. No Columbia students participated in it/that, and no Harvard students participated in—a conspiracy of theirs to overthrow the government. Aristotle’s and of collective or divided reference proves to be an illusion of unspoken, plural event pronouns—two of them in the general case of an apparent coordination of generalized quantifiers (i.e., two in every Right-Node Raising, to which the coordination of generalized quantifiers is assimilated). In stepping up to generalized quantifiers, no Columbia students, for example, these event pronouns that purport to refer to what events the Columbia students participated in must do so without commitment to their existence (section 2.1.1), like the pronoun in (98): (98) Not many a professor endorses operant conditioning, and they must all be behaviorist rats. That is, the event pronouns like the one in (98) are descriptive. They are sometimes also “lazy” or “sloppy” like the pronoun in (99) and “telescoping” as in (100)— various contexts in which anaphoric and antecedent descriptions find themselves within the scope of and dependent on different quantifiers, obstructing coreference between antecedent and pronoun: (99) Every wise investor invests her nest egg in the financial markets, and every foolish investor invests it also in the financial markets. (After Geach 1962) (100) No man can be friends with a woman he finds attractive. He always wants to have sex with her. (Billy Crystal in When Harry Met Sally (1989), cited in Poesio and Zucchi 1992) In replacing bare event variables, event pronouns thus import into subatomic clauses the greater expressive power of their overt cousins, which sections 2.1.2, 2.2, and 2.3 document throughout. 2.1.1
Description without commitment
If there is to be an appeal to plural event pronouns and they are the same in (6), (7), (9), and (11)–(13) and in the coordination of DPs of any type, they are descriptive. At the very least this is true of the pronoun prefixed to the Right-Node Raised
DP and DP
95
constituent referring to the plural events amounting to a conspiracy or pertaining to a joint manifesto: (101) [No Columbia students Wj-ed (secretly) Δδ] and no Harvard students Wk-ed (openly) [δ [℩Ej,k : proj,k] conspired to overthrow the government]. (102) [No Columbia students draftedj Δδ] and no Harvard students redactedk [δ [℩Ej,k : proj,k] their joint manifesto for the revolution]. Any reference from this position to what Columbia students do does not fall within the scope of no Columbia student in the first clause, since this would imply that no Columbia student also included within its scope no Harvard student, which from the meaning, it plainly does not (see also section 1.5.1). Outside the first clause, any reference to what Columbia students do is therefore descriptive.7 Sentences (7) and (11)–(13) are true if there are no Columbia students and no Harvard students and therefore no events in which they participate. It is safe to assume that the plural event pronouns are definite descriptions without existential commitment, referring to whatever, if anything at all, Columbia student and Harvard student did, and there is argument that it has to be this way. If many Harvard students redact revolutionary tractates, then (103) and (104) are unambiguously false even if Columbia students are nowhere to be found on this occasion: (103) [[DPNot a single Columbia student] [Wj-ed (quickly) Δδ]] and [[DPnot many a Harvard student] [Wk-ed (carefully) [δ [℩Ej,k : proj,k] have redacted revolutionary tractates]]] ‘Not a single Columbia student (quickly) and not many a Harvard student (carefully) have redacted revolutionary tractates.’ (104) [[DP Not a single Columbia student] [quickly draftedj Δδ]] and [[DP not many a Harvard student] [carefully redactedk [δ[℩Ej,k : proj,k] revolutionary tractates]]] ‘Not a single Columbia student quickly drafted and not many a Harvard student carefully redacted revolutionary tractates.’ The pronoun [℩Ej,k : proj,k], dividing its reference between what Columbia students do and what Harvard students do, as in the events such that the Columbia students’ participation is among them and the Harvard students’ participation is among them, must not imply the existence of the Columbia students’ events. For, if it did, the pronoun itself would denote no events, affording to (103) and (104) an interpretation where they are vacuously true. This conclusion follows from the observation that the pronoun is outside the scope of not a single Columbia student but within the scope of not many a Harvard student. Sentences (103) and (104) show that the
96
Chapter 2
pronoun must paraphrase at least as the events among which are whatever events the Columbia students’ participated in and the events the Harvard students’ participated in, leaving uncommitted the existential commitment of the in this paraphrase. It is however safe to further assume the absence of existential commitment throughout—whatever events there are of whatever events the Columbia students participated in and whatever events the Harvard students participated in—which I assume it is, with nothing to recommend a mix. Descriptive pronouns without existential commitment find precedent among the spoken in examples such as (105) after Evans 1977, 1980, and subsequent literature (e.g., Davies 1981; Elbourne 2005; Heim 1990; Ludlow 1994; Schein 1993, 215ff.; De Swart 1991): (105) Not many a professor endorses operant conditioning, and they must all be behaviorist rats. The pronoun does not fall within the scope of its antecedent, and it denotes whatever professors if any endorse operant conditioning, where the description appears to take up the antecedent professors and its scope professors endorsing operant conditioning. With greater affinity to the unspoken event pronoun, ‘[℩Ej,k : proj,k]’ above, the pronoun in (106) has a split antecedent, again without commitment to the existence of what either describes: (106) Not many a professor endorses operant conditioning, and not many a student does either, and they must all be behaviorist rats. This pronoun is equivalent to the definite description in (107), interpreting the operator in (108) as Sharvy (1980) does (see Cartwright 1996), but again without existential commitment: (107) [℩Xij : [℩Xi: professors who endorse operant conditioning[Xi]][∀x: Xix]Xijx & [℩Xj: students who endorse operant conditioning[Xj]][∀x: Xjx]Xijx] ‘the Xij such that the professors endorsing operant conditioning are among them & the students endorsing operant conditioning are among them’ (108) [℩X: Φ[X]] Ψ[X] ↔df [∃X: Φ[X] & ∀Y(Φ[X/Y] → ∀y(Yy → Xy)) & ∀Z((∀Y(Φ[X/Y] → ∀y(Yy → Zy))) → ∀x(Xx → Zx))] Ψ[X] (Sharvy 1980, with amendment) Taking up the antecedents and their scope, the description says that whatever professors there are endorsing operant conditioning are among those described and whatever students there are endorsing operant conditioning are among them, too. It is left to the meaning of the iota-operator itself to enforce reference to only such professors and students, saying that those described are the fewest with all such
DP and DP
97
professors and students among them. The pronoun in (106) is equivalent to the definite description (107). Ham-handed and flat-footed as I am, I will be content to speak as if the pronoun in the logical form for (106) just is (107) (cf. skeptical remarks in section 2.1.4). But, anything more nimble rendering the equivalence will do as well. The point is that there are unspoken plural pronouns interpreted in much the same way as their spoken counterparts, and they do not part company no matter how it comes about that the pronoun in (106) refers to what it does. For the sake of concreteness and translation into an explicit logical language, descriptive anaphora here are indeed descriptions. Descriptive event anaphora proper, spoken and unspoken, forward and backward, surface in the following: (109) Few Columbia students drafted (press releases) and fewer Harvard students redacted joint manifestos, and it lasted no longer than a study break. (110) a. Lasting no longer than a study break, few Columbia students drafted (press releases) and fewer Harvard students redacted joint manifestos. b. Few Columbia students drafted (press releases) and fewer Harvard students redacted joint manifestos, lasting no longer than a study break. The mass term pronoun it8 divides its reference between whatever drafting Columbia students did and whatever the Harvard students’ redaction was, again without implying that either exists. The same is to be said about the unspoken subjects of the adverbial phrases in (110). 2.1.2
“Telescoping” event pronouns
In solving for Φ in (8), subatomic pronouns referring to events have been uncovered that are descriptive, backward, crossing, and in section 2.2, sometimes “sloppy”—like many of their overt counterparts. (7) No Columbia student and no Harvard student conspired to overthrow the government. (8) No Columbia student Φ and no Harvard student conspired to overthrow the government. It would be a fair response to recoil from the overwrought logical forms that translate this usage, and yet one should not be provoked into denying speakers their wayward way with pronouns. As in (111) from chapter 1, a monomorphemic pronoun in (112) glibly accomplishes what I have only been able to render as a complex description. (111) Biff did something and possibly Tiff did something, and it caused the riot. (112) No Columbia student cheered the other school’s football team, and no Harvard student did that either.
98
Chapter 2
The argument remains that subatomic pronouns occur with the usages suggested, while alternative analyses of such pronouns so construed are invited to do better. If the conjoined quantifiers occur in separate clauses as they do in (8), we have no choice but to resort to devices very much like ‘[℩Ei,i : proi,i]’ from the previous section. The first clause in the logical form for (7) must not assert that no Columbia students participated in arbitrary events, since we know them to be quite active sometimes. Rather, there were none who participated in events of Columbia students and Harvard students collaborating, exactly as the first clause would assert if its event argument were a backward pronoun. Thus some treatment of the descriptive content of ‘[℩Ei,i : proi,i]’ is warranted if the logical syntax in (8) is correct. In this connection, consider (113) and (114) from section 1.5.1 and the further examples in (115)–(118): (113) Not many a professor and not many a student of hers are conspiring to revive operant conditioning. (114) No more than a single senator and no more than a single congressman from his home state have cosponsored an antiwar resolution. (115) No student and no professor she knew well collaborated. (116) No well-behaved child and none of her well-behaved siblings fought with each other over the bigger piece of cake. (117) Exactly one student and exactly one professor that shared the prize with her met at the award ceremony. (118) Most any student and most any professor of his disagree at some point in graduate school. The quantifiers are independent, as the coordination of nonincreasing quantifiers in (113)–(118) makes plain. Thus the first quantifier does not include within its scope the second, and yet a pronoun within the second quantifier covaries with the first. The disposition of the pronoun recalls the “telescoping” (Roberts 1989; Ludlow 1994; Poesio and Zucchi 1992) that has been observed across sentences: (119) No man can be friends with a woman he finds attractive. He always wants to have sex with her. (Billy Crystal in When Harry Met Sally, cited in Poesio and Zucchi 1992) In (119), always is understood as ‘whenever a man finds a woman attractive’ and the pronoun as a description of the woman he finds attractive. Similarly, in (120), a tacit adverbial preceding the second clause, whenever there was a linguistics professor, is taken to support the pronoun as a description of the linguistics professor then there.9 (120) Not many a linguistics professor endorsed operant conditioning, and not many a student of hers did either. (121) No more than a single senator sponsored an antiwar resolution, and no more than a single congressman from his home state helped him.
DP and DP
99
Sentences (113)–(118), if taken the same way as (119) and (120)–(121), suggest subatomic “telescoping,” paraphrasing (113), for example, as (122): (122) Not many a professor … W …, and whenever a professor W, not many a student of that professor are conspiring to revive operant conditioning. Like its full-clausal counterpart in (124)–(126), subatomic “telescoping” in (123) does not iterate comfortably, even where meaning does not preclude one conjoined quantifier from including the next in its scope: (123) ??Every boyk, every onej of hisk aunts, and every onei of herj pets were proud (of itsi showing) at the dog-and-pony show. (124) a. Every boy assists every one of his aunts. He escorts them to their starting positions at the dog-and-pony show. b. ???Every boy assists every one of his aunts. He escorts her to her starting position at the dog-and-pony show. (125) a. Every boy assists every one of his aunts with every one of her pets. He escorts them and their prize pets to their starting positions. b. ???Every boy assists every one of his aunts with every one of her pets. He escorts her and her prize pet to their starting position. (126) ?Every boy attends the dog-and-pony show. He escorts every one of his aunts to her starting position. She then preps and preens her prize pet.10 Perhaps it becomes awkward to stack tacit adverbials, ‘whenever a boy assists (such that) whenever an aunt is assisted …’ or to enter a restriction to the adverbial that strips the antecedent clause of all of its (universal) quantifiers, ‘whenever a boy assists an aunt with a pet’. Whatever accounts for the degraded acceptability, its noniterativity speaks in favor of treating subatomic “telescoping” as something that intrudes on the coordination. The covariance of pronoun and quantifier in no student and no professor of hers and in every boy and every one of his aunts cannot be a reflex of the structure that the iteration of coordination itself iterates. Thus, in no student and no professor of hers and in the other examples of (113)–(118), the pronoun covaries with a quantifier in whose scope it cannot occur (the coordinated quantifiers being scopally independent), and it cannot be a fact about coordination itself that such covariance is licensed. Rather, such covariance occurs, as it may sometimes occur, across clauses, with the intervention of an adverbial. Like the modal argument of section 1.5.2 based on the Columbia students and possibly the Harvard students, the intervention of an adverbial argues for the sentential coordination in (8) (for a similar argument, see Camacho 1997, 19ff., 56ff.).11,12 In summary, an alleged coordination of generalized quantifiers is rather a coordination of subatomic sentences held together by the cross-reference of pronouns
100
Chapter 2
that describe events. Like spoken descriptive anaphora, the unspoken, subatomic anaphora sometimes depends on “telescoping” for its interpretation. Later (in section 2.3), with number agreement taken as the spoken trace of an event pronoun, semantic conditions on number agreement in Lebanese Arabic will be found to track its scopal context, including whether or not it is “telescoped.” 2.1.3
Minding the gap
The meaning of Right-Node Raising, it was announced, engages two event pronouns, one to be introduced at the deletion site, and the other already in place. Necessary for any collective predication, it refers to the Columbia students’ participation and the Harvard students’ participation as shown in (127)–(129): (127) [[DPNo Columbia students] [Wj-ed (secretly) Δδ]] and [[DPno Harvard students] [Wk-ed (openly) [δ[℩Ej,k : proj,k] conspired to overthrow the government]]] ‘No Columbia students (secretly) and no Harvard students (openly) conspired to overthrow the government.’ (128) [[DP No Columbia students] [draftedj Δδ]] and [[DP no Harvard students] [redactedk [δ [℩Ej,k : proj,k] their joint manifesto]]] ‘No Columbia students drafted and no Harvard students redacted their joint manifesto for the revolution.’ (129) [[DP Not many a Columbia student] [proposed to the strike committee to adopt a resolution to encourage violent resistance toi Δδ]] and [[DP not many a Harvard student] [accepted to endorse to the strike committee a call for illegal action againstj [δ [℩Ei,j : proi,j] attempts by police to prevent them from chaining each other to the White House fence]]] ‘Not many a Columbia student proposed to the strike committee to adopt a resolution to encourage violent resistance to—and not many a Harvard student accepted to endorse to the strike committee a call for illegal action against—attempts by police to prevent them from chaining each other to the White House fence.’ In (127), in which the remnant of Right-Node Raising is minimal, the plural event pronoun refers to whatever events there are of Columbia students participating (W-ing) (secretly) and of Harvard students participating (openly), and the sentence is about whether these are a conspiracy to overthrow the government. These are in fact claimed to involve no Harvard student, and it is the token of this pronoun within the scope of no Harvard student that confines the claim to conspiracies with Columbia students to overthrow the government, exempting conspiracies with Yale
DP and DP
101
students and conspiracies with other nefarious ends in mind, as the paraphrase in (19) reflects. When Right-Node Raising leaves behind a larger remnant, as in (128), the event pronoun finds its antecedents in the same place, and the equivalent description is only that much longer—and even longer in (129). The pronoun refers in (128) to whatever events there are of Columbia students drafting and Harvard students redacting. Recall that sentence (130) with multiple conjunctions appears in chapter 1 in an argument for supermonadicity concluding that its logical form tokens at least two pronouns, one ‘[℩Ej,k : proj,k]’ in (131), to denote the participation of the Columbia and the Harvard students, and another, ‘[℩El,m : prol,m]’, to denote events of surrounding the Pentagon and being crowded into the Mall: (130) The Columbia students (quickly) and the Harvard students (slowly) surrounded the Pentagon and were crowded into the Mall. (131) [[DPThe Columbia students] [Wj-ed Δδ]] and [[DPThe Harvard students] [Wk-ed [δ[℩Ej,k : proj,k][℩El,m : prol,m]O[Ej,k, El,m] ∃El(surroundedl the Pentagon) and ∃Em(were crowdedm into the Mall)]]] Sentences with multiple conjunction also engage Right-Node Raising with larger remnants. Sentence (133) differs from (132) (identical in structure to (130)) in leaving behind larger remnants and thus longer antecedent clauses for ‘[℩Ej,k : proj,k]’ in (134). The antecedents for the backwards pronoun ‘[℩El,m : prol,m]’ are now smaller clauses in a coordination that, absent verbs, is traditionally and mistakenly called Nonconstituent Coordination, reflecting (133)’s surface appearance: (132) The Columbia students and the Harvard students donated 10,000 armbands for protest marches and sold 50,000 peace buttons for mass distribution. (133) The Columbia students donated and the Harvard students sold 10,000 armbands to protesters and 50,000 peace buttons to bystanders. (134) [[DPThe Columbia students] [donatedj Δδ]] and [[DPThe Harvard students] [soldk [δ[℩Ej,k : proj,k][℩El,m : prol,m]O[Ej,k, El,m] ∃El(10K armbands tol protesters) and ∃Em(50K buttons tom bystanders)]]] Conjunction Reduction, Right-Node Raising, and Nonconstituent Coordination are all the same, varying only in the size of the clauses conjoined (and what is pronounced) and with that, the descriptive content of the event pronouns anaphoric to them. Returning to the deletion site of Right-Node Raising, recall from the discussion of (19) that there is an equal consideration that the assertion in the first clause that no Columbia student participated applies only to their participation in conspiracies
102
Chapter 2
with Harvard students to overthrow the government, to which end the deletion Δ in (135) wants interpretation if the quantificational claim is to be so limited: (135) [[DP No Columbia students] [Wi-ed Δδ]] and [[DP No Harvard students] [Wi-ed [δ [℩Ei,i : proi,i] conspired to overthrow the government]]] A spoken analog to the backward event pronoun denoting the events described by the Right-Node Raised phrase goes something like in (136): (136) a. No Columbia students participated in (any) such, and no Harvard students participated in—a conspiracy of theirs to overthrow the government. b. No Columbia students participated in it/that, and no Harvard students participated in—a conspiracy of theirs to overthrow the government. Showing first the relation of Δδ to the clause in which it is embedded, (137) develops the Eventish (see appendix 2) for the first clause of (135) (suppressing adverbialization and the relativization to spatiotemporal address and frame of reference): (137) [No X: ∃E C[E, X]](∃Ei W[Ei, X] [℩Ei : proi]Δδ[Ei]) and … i.e., ‘No Columbia students are such that they participate & it [their participation] is Δδ.’ The pronoun shown is just a token of those that have replaced bare variables as the basic device for cross-reference to events, such as would occur in like position to relate participation to drafting in No Columbia students drafted a press release, absent any deletion. It is the events that such a pronoun refers to that are to be related to events that the deleted content denotes. It suffices to say that they are among them: (138) [No X: ∃E C[E, X]](∃Ei W[Ei, X] [℩Ei : proi][∀e: Eie][℩Eδ : Δδ]Eδe) and … ‘No Columbia student are such that they participate and these events of participation are events that Δδ.’ In (138), the deleted content is taken up into a plural event description—that is, as translation for a descriptive pronoun anaphoric to the Right-Node Raised phrase. The logical form shows exactly the relation of the events thereby referred to and the events previously described within the first clause. The thought is that the meaning of deletion in Right-Node Raising is (i) (139) plural reference to—and (ii) (140) second-order predication of—the events that the Right-Node Raised phrase denotes: (139) … [∀e: Eie][℩Eδ : Δδ]Eδe) and … (140) … [∀e: Eie][℩Eδ : Δδ]Eδe) and …
DP and DP
103
The logical form for (7) is then as in (141). Note that the antecedent for the deleted content, the Right-Node Raised phrase, contains within it the token of the plural event pronoun first discussed referring to the events of the Columbia student’s and the Harvard student’s participation: (141) [No X: ∃E C[E, X]](∃Ei W[Ei, X] [℩Ei : proi][∀e: Eie][℩Eδ : Δδ]Eδ e) and [No Y: ∃E H[E, Y]](∃Ei W[Ei, Y] (δ [℩Ei,i : proi,i] conspired to overthrow the government[Ei,i])) (142) [No X: ∃E C[E, X]] (∃Ei W[Ei, X] [℩Ei : proi][∀e: Eie] [℩Eδ : [∀e: Eδ e][℩Ei,i : proi,i]Ei,ie & conspired … government[Eδ]] Eδ e) and [No Y: ∃E H[E, Y]](∃Ei W[Ei, Y] [℩Ei,i : proi,i] conspired to overthrow the government[Ei,i]) Because ‘proi,i’ within ‘[℩Ei,i : proi,i]’ is the only description of their collective action, it must appear one way or another in both conjuncts for them to be understood as comments on their collective action and only that. It is copied in (142) along with the rest of the Right-Node Raised phrase into the first clause as the deleted content (compare the paraphrase in (143) to (19)):13 (143) [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ei,i : proi,i] Ωδ) ⇒ [DPm D NP](Φ [℩Ei:Φ][∀e: Eie][℩Eδ: [∀e: Eδ e][℩Ei,i: Φ;Ψ] Ei,ie & Ωδ] Eδ e) and [DPn D NP] (Ψ [℩Ei,i:Φ;Ψ] Ω) ‘[DPm D NP] Φ in the events of NPm-Φ-ing and NPn-Ψ-ing that Ω, and [DPn D NP](Ψ & the events of NPm-Φ-ing and NPn-Ψ-ing Ω)’ In Right-Node Raising from deeply embedded position—that is, for very large Φ and Ψ, as in (144)—the Right-Node Raised phrase describes events to be related within each conjunct to values of an event variable itself embedded deeply within: (144) [[DP Not many a Columbia student] [proposed to the strike committee to adopt a resolution to encourage violent resistance toi Δδ]] and [[DP not many a Harvard student] [accepted to endorse to the strike committee a call for illegal action againsti [δ [℩Ei,i : proi,i] attempts by police to prevent them from chaining each other to the White House fence]]] ‘Not many a Columbia student proposed to the strike committee to adopt a resolution to encourage violent resistance to—and not many a Harvard student accepted to endorse to the strike committee a call for illegal action against—attempts by police to prevent them from chaining each other to the White House fence.’
104
Chapter 2
It is events of resistance and illegal action that are directed at attempts by police to prevent guerrilla theater at the White House. Not any such events, just the events of a Columbia student proposing to the strike committee to adopt a resolution to encourage such as violent resistance and events of a Harvard student accepting to endorse to the strike committee a call for such as illegal action. Right-Node Raising does not disrupt what for the Davidsonian reflects the structure of complementation, the variables used, and the relations they appear with. The events said to be directed at the police attempts are the same, Right-Node Raising or not, and the event variable abstracted on is the same. The effect of Right-Node Raising from a deeply embedded position is just that that abstraction is over very large Φ and Ψ providing the description to a descriptive anaphor, again, like they in (145): (145) After the dust settled—and nothing is certain given source confidentiality— not many a Columbia student related to the New York Times a report of a police action to subdue a demonstrator at Columbia, and not many a Harvard student delivered to the Boston Globe an eyewitness account of an FBI attempt to silence a demonstrator at Harvard. They, in fact, were not more than a couple suspects, if any, and they must have been among those whom the NYPD and FBI had on a watch list. In Eventish, given the supermonadicity of basic clauses, one does not need embedded clauses to illustrate abstraction of an embedded event variable. Sentence (12), repeated as the gloss to (146), has the rough parse shown: (146) [[DP No Columbia students] [draftedi Δδ]] and [[DP no Harvard students] [redactedi [δ [℩Ei,i : proi,i] their joint manifesto]]] ‘No Columbia students drafted and no Harvard students redacted their joint manifesto for the revolution.’ Prior to the interpretation of ‘[℩Ei,i : proi,i]’ and ‘Δδ’ according to (143), the Eventish for (146) is (147), suppressing adverbialization and relativization to spatiotemporal address and frame of reference and leaving the application of (143) as an exercise:14 (147) [No X1: ∃E C[E, X1]] (∃E0 W[E0, X1] [℩E0: pro0][℩E1: pro1]O[E0,E1] ∃E1∃X Agent[E1,X] [℩E1: pro1][℩E2: pro2]Cause[E1,E2] ∃E2 draft[E2] [℩E2: pro2][∀e: E2e][℩Eδ : Δδ]Eδe) and [No X1: ∃E H[E, X1]] ( ∃E0 W[E0, X1] [℩E0: pro0][℩E1: pro1]O[E0,E1] ∃E1∃X Agent[E1, X] [℩E1: pro1][℩E2: pro2]Cause[E1,E2] ∃E2 redact[E2] [℩E2,2: pro2,2][℩E3: pro3]O[E2,E3] [the X3: ∃E M[E, X3]] ∃E0 W[E0, X3][℩E0: pro0][℩E1: pro1]O[E0,E1] ∃E3∃X Patient[E3,X])
DP and DP
2.1.4
105
Translation myths and rules
At the end of this section are collected translation schemas for event pronouns with split antecedents and for Right-Node Raising, as if it were the literal truth that descriptive anaphora contained copies of their antecedent clauses, as if recited sotto voce: (148) No more than a few Deadheads could have all been at Magoo’s Pizza in Menlo Park on May 5, 1965, …, at Ken Kesey’s Acid Test in San Jose on December 4, 1965, …, at the Fillmore West in San Francisco on January 8, 1966, …, at the Monterey Pop Festival 1967, …, at Woodstock 1969, …, and at the Grateful Dead’s final concert July 9, 1995. They must all be no younger than 30. Despite their pilgrim fervor, no more than a few of the Grateful Dead’s fans could have all attended their over 2300 concerts, and so a fan laments truly in (148), piously invoking the concerts by name in over 2300 conjoined phrases. According to translation, he recites them all yet again in uttering the pronoun they. The longer the saga, the longer the retelling. Logically, it cannot be otherwise. On the one hand, the pronoun is plainly descriptive since it may turn out without harm to what is said that there is no Deadhead with perfect attendance. On the other hand, its descriptive content may not, under pain of referential failure, fail to mention any of the over 2300 concerts, for the Deadheads have been so assiduous in their devotions that for any proper subset of the over 2300 concerts, more than a few attended them all. Any descriptive pronoun may depend for its reference on anything and everything antecedently said about its intended referents. This is what the plodding translation is meant to codify, and of course it misses entirely the point that no thought deploying descriptive anaphora is so encumbered. It suffices for the pronoun to capture this semantics that the speaker have the anaphoric intention that it denote whatever its antecedent would denote qua description. This thought does not get heavier when an antecedent has droned on, and in its glibness, it affords the speaker occasion to speak without knowing what of, a sweet trick denied the speaker who has to spell out a description. Even I, hearing about concerts spanning 30 years and without recognizing them to be the complete list of the Grateful Dead’s concerts, might have said the second sentence in (148) in a dreaded conversation with a Deadhead uttering the first, from which I can recollect only the reminder that they played Woodstock and it was 1969 rather than 1970. Nevertheless, what I say commits me to an assertion equivalent to what would be said with the description at issue, although I could supply neither that description with over 2300 phrases nor any other description of the few Deadheads I refer to with they. Any more subtle theory of the route from anaphoric intentions to truth conditions would be a welcome improvement on translation that only reproduces the logic and semantics of the pronouns without fidelity to the grammar or structure of the thought.
106
Chapter 2
Translation of the event pronoun ‘[℩Ej,k : proj,k]’ is in (149), where Φ is the remnant of Right-Node Raising in the first clause, Δ, the deletion site, and Ω, the Right-Node Raised phrase. An abbreviation deployed in (149) for split antecedents is defined in (150). A final translation schema for Right-Node Raising exploiting the above is given in (151): (149) Event descriptive anaphora [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : proj,k] Ωδ) ⇒ [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : ([℩Xm : NPm]Φ,[℩Xn : NPn]Ψ)[Ej,k]] Ωδ) ‘[DPm D NP](Φ Δδ), and [DPn D NP](Ψ & the events of NPm-Φ-ing and NPn-Ψ-ing Ω)’ (150) Description with split antecedents ⌜(Φ,Ψ)[Ej,k]⌝ for ⌜[℩Ej: Φ][∀e: Eje]Ej,ke & [℩Ek: Ψ][∀e: Eke]Ej,ke] & ∀E(([℩Ej: Φ][∀e: Eje]Ee & [℩Ek: Ψ][∀e: Eke]Ee])→ [∀e: Ej,ke]Ee)⌝ ‘events Ej,k are the Φ-ings Ej & the Ψ-ings Ek’ (151) Translation for Right-Node Raising [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : proj,k] Ωδ) ⇒ [DPm D NP](Φ [∀e: Eje][℩Ej,k : ([℩Xm : NPm]Φ, [℩Xn : NPn]Ψ)[Ej,k] Ω] Ej,ke) and [DPn D NP](Ψ [℩Ej,k : ([℩Xm : NPm]Φ, [℩Xn : NPn]Ψ)[Ej,k]] Ω) ‘[DPm D NP] Φ in the events of NPm-Φ-ing and NPn-Ψ-ing that Ω, and [DPn D NP](Ψ & the events of NPm-Φ-ing and NPn-Ψ-ing Ω)’ Schemas (149)–(151) do not preclude instantiation of the subscripts so that j = k. Suppose the language were regimented so that the first token of the participation relation W in a clause occurs with ‘E0’ and so on, so that the clausal architecture itself fixes the variables used,15 or that a syntactic parallelism imposed on conjuncts in general or those hosting Right-Node Raising implied like variables in like position, or that syntactic copying was exact enough to demand that variables and their subscripts be copied exactly. If the language were so, we should often come across ‘[℩Ei,i : proi,i]’, since typically the antecedent descriptions will abstract on event variables in entirely parallel positions in their respective clauses. Later, this chapter will in fact impose alphabetic identity on copied variables. 2.1.5
Phrasing
The account on offer derives collectivized Right-Node Raising from two plural descriptive event pronouns. Collective reference here is an event pronoun with split antecedents, the pronoun prefixed to the Right-Node Raised phrase and referring
DP and DP
107
back to the events antecedently described in the preceding two clauses. Another pronoun cross-referring to the collectivized events the Right-Node Raised phrase thus describes occurs at the deletion site in the first clause and correctly confines that clause to comment about these collectivized events. There is a point of syntax about all this to be highlighted in concluding section 2.1: (152) [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ei,i : proi,i] Ωδ) In the logical form for Right-Node Raising, (152) and its expansion interpreting pronouns and deletion, Ωδ, the Right-Node Raised constituent so-called, remains entirely within the second conjunct. Its collectivized interpretation proves to offer no basis for concluding that it must appear outside the scope of the coordination, as in (153): (153) *([DPm D NP](Φ Δδ) and [DPn D NP]Ψ) Ωδ While it is certain that (154) is unrelated to (155), its interpretation can nevertheless be rendered without parsing (154) as (156), contrary to what is sometimes assumed: (154) John whistled and Mary hummed at equal volumes. (Jackendoff 1977, 193) (155) *John whistled at equal volumes, and Mary hummed at equal volumes. (156) (John whistled and Mary hummed)—at equal volumes. Rather, the solution proposed here to this classic semantic problem comports well with independent syntactic evidence for the parse ((DP Φ) (and (DP ΨΩ))) (see Wexler and Culicover 1980, 298–303; McCawley 1982; Levine 1984, 1985; McCloskey 1986; Kayne 1994;Wilder 1997, 1999; Camacho 1997, 2003).16 2.2
Number agreement
As to what went on, (157) and (158) report the same event. The plural definite description and the conjoined subject turn out to agree in what they say the students did: (157) The two Harvard students cooked up the Harvard beets. (158) Biff and Tiff cooked up the Harvard beets. Their equivalence is an inference between two rather different logical forms. The logical form for (157) asserts that the two Harvard students were Agents (or participants, W-ers) and the only Agents in e, and the logical form for (158), a RightNode Raised structure, says that Bill was the Agent in an e and Tiff was the Agent in another e, and those events in which they were the only Agents were a cooking up of the Harvard beets.
108
Chapter 2
The rejected alternative says that (157) and (158) say the same thing because the two Harvard students and Biff and Tiff refer to the same thing. Number agreement is sometimes taken to hold out an argument for this alternative. (159) The NYPD officer’s counsel and his union delegate has submitted an affidavit to Internal Affairs. (160) The NYPD officer’s counsel and his union delegate have submitted an affidavit to Internal Affairs. As (159) and (160) show, number agreement depends on semantic reference. The sentences are identical in every other detail, and yet the number agreement varies just in case the NYPD officer’s counsel and his union delegate are the same person, in which case it is singular, and it is plural if they are not. If number agreement depends on semantic reference and agreement is agreement with the subject, then— so the argument goes—the reference of the subject, the NYPD officer’s counsel and his union delegate, is to the one in (159) and to the two in (160). If the subject does refer to the one or the two, then indeed agreement fixes which. But the argument begs the question it is taken to answer. Pronominal number agreement is also semantic and agrees with the reference of its antecedents, as (161) and (162) show: (161) Later that evening somewhere in Metropolis, … a. A superhero stepped into a phone booth, and a mild-mannered civilian stepped out. He had held the forces of evil at bay for yet another day. b. The superhero stepped into a phone booth, and a mild-mannered civilian stepped out. He had held the forces of evil at bay for yet another day. c. A superhero stepped into a phone booth, and the mild-mannered civilian that Lois loved stepped out. He had held the forces of evil at bay for yet another day. d. The superhero stepped into a phone booth, and the mild-mannered civilian that Lois loved stepped out. He had held the forces of evil at bay for yet another day. (162) Later that evening somewhere in Metropolis, a/the superhero stepped out of a phone booth, and a/the mild-mannered civilian (that Lois loved) stepped in. They were both trying to reach the mayor on secret hotlines that only they knew about. Yet this observation provides no argument that the conjoined clauses refer to persons or that and should be construed so as to deliver such a referent from the clauses it conjoins. In fact, (162) contains no phrase other than the pronoun itself that refers to the superhero and the civilian. Directly relevant to the present
DP and DP
109
discussion, structures analogous to (161) and (162) involving verbal number agreement occur in SOV languages such as Basque (Sjoblom 1980, 117ff.; McCawley 1988, 532ff.; Camacho 1997, 184ff.): (163) Lindak ardau eta Anderek esnea edaten dabez. Linda wine and Ander milk drink be (3pl) ‘Linda will drink wine and Ander milk.’ (Sjoblom 1980) The plural number agreement in (163) refers to Linda and Ander, but no other phrase does, and the conjunction plainly conjoins sentences here too. Closer to home, English makes the same point with (164): (164) a. Robin earlier today and Hillary yesterday were drinking more bordeaux between them than the region produces in a week. b. Robin by this morning and Hillary by last night have drunk more bordeaux between them than the region produces in a week. There is no phrase referring to Robin and Hillary and providing a plural antecedent for number agreement. The conjunction plainly coordinates clausal descriptions of events, Robin’s action of today and Hillary’s from yesterday, and the Right-Node Raised constituent measures the amount of bordeaux they consume collectively. Such considerations defuse the argument that number agreement is evidence that Biff and Tiff in (158) refers to the two Harvard students. To the contrary, a closer reckoning will include number agreement among the arguments that support the sentential, Right-Node Raised analysis of (158) with its subatomic pronouns referring to events. Animating what follows is the thought that number agreement is just voicing one of the event pronouns that supermonadicity proliferates. In a so-called DP-coordination, it is identified with the event pronoun prefixed to the Right-Node Raised phrase. Number in number agreement is further comment about the number of participants in the events denoted. Agreement thus denotes whatever events the subject is antecedently said to participate in. Singular, it describes them, in effect, as those events there such that something participates in them or those events there all with the same one participant; and plural, as those events there with participants:17 (165) sg[E] ↔df ∀X(W[E,X] → ∀x∀y((Xx & Xy) → x = y)) (166) pl[E] ↔df ∀X(W[E,X] → ∃x∃y(Xx & Xy & ¬x = y)) Now, if singular agreement is of events with only one participant, then with more than one participant at issue—Columbia student and Harvard student in (167)– (168)—it must be that singular agreement occurs “sloppily” or “lazily,” the spoken
110
Chapter 2
morpheme and the one deleted under Right-Node Raising falling within the scope of different event quantifiers: (167) No Columbia student (secretly) and no Harvard student (openly) was on a strike committee in which the Columbia student led and the Harvard student followed. (168) Not many a Columbia student (secretly) and not many a Harvard student (openly) was on a strike committee in which the Columbia student led and the Harvard student followed. The different event quantifiers derive at least from the adverbialization18 of no Columbia student and no Harvard student as in (169) (see (170)):19 (169) No Columbia student anywhere anywhere a Columbia student is does she participate (‘W’) (secretly) Δ, and no Harvard student anywhere anywhere a Harvard student is does she participate (openly): &, the sg.event of her participating there was being on a strike committee in which the Columbia student there led and the Harvard student there followed. Note that in the translation of singular number agreement, the content of the event pronoun must be quite thin, the sg.event of her participating there or the sg.event of someone participating there, so as to describe both Columbia student and Harvard student indifferently, since exact copies under Right-Node Raising are tokened in both conjuncts. With plural number agreement (see (20), (22), (24), (26), (28)–(34)), the content is that much richer so that both its tokens describe what the Columbia and Harvard students do, as in (45). In realizing the “sloppy” or “lazy” interpretation of the sg.event of her participating there or the sg.event of someone participating there, singular number agreement in (169) falls within the scope of the boldface adverbs, anywhere or their deleted tokens, which include within their scope much of a conjunct—at least the RightNode Raised phrase or its deleted copy. Adverbs, singular number agreement, and their scope, in turn, all fall inside the scope of one of the allegedly conjoined DPs, as in (170) or alternatively, as in (171): (170) [D NP][℩E : NP] [[∃E0 : there[E,E0]]Φ Δδ] and [D NP][℩E : NP] [[∃E0 : there[E,E0]]Ψ sg.Ωδ] (171) [D NP][℩E : NP] [∃E0 : there[E,E0]] [Φ Δδ] and [D NP][℩E : NP] [∃E0 : there[E,E0]] [Ψ sg.Ωδ] Singular number agreement in (167)–(168) thus drives the singular quantifiers to positions quite high in their conjuncts from which they command a scope that is a rather large clause.
DP and DP
2.2.0
111
(In)definite descriptions vs. distributive quantifiers
This syntax for singular number agreement plays a role distinguishing two classes of DPs. The singular agreement allowed to a conjunction containing singular quantifiers is denied to (in)definite descriptions (Hoeksema 1983, 1988; Lasersohn 1995, 109ff.):20 (172) *A Columbia student and a Harvard student was on a strike committee. *A Columbia student and a Harvard student has joined a strike committee. (173) *A Columbia student secretly and a Harvard student openly was on a strike committee. *A Columbia student secretly and a Harvard student openly has joined a strike committee. With (in)definite descriptions, singular agreement only becomes possible and then it is obligatory when they are contained within a coordination of fully tensed clauses: (174) A Columbia student was and a Harvard student was (too) on a strike committee to divest from the war. A Columbia student was and a Harvard student still is on a strike committee to divest from the war. A Columbia student has and a Harvard student has (too) joined a strike committee to divest from the war. A Columbia student has and a Harvard student will have joined a strike committee to divest from the war. (175) *A Columbia student were and a Harvard student were (too) on a strike committee to divest from the war. *A Columbia student were and a Harvard student still were on a strike committee to divest from the war. *A Columbia student have and a Harvard student have (too) joined a strike committee to divest from the war. *A Columbia student have and a Harvard student will have joined a strike committee to divest from the war. Examples (176)–(179) distill singular number agreement for (in)definite descriptions and the contrast with other quantifiers: (176) *A Columbia student secretly and a Harvard student openly was on a strike committee. (177) Every Columbia student secretly and every Harvard student openly was on a strike committee. (178) A Columbia student secretly was and a Harvard student openly was on a strike committee.
112
Chapter 2
(179) Every Columbia student secretly was and every Harvard student openly was on a strike committee. Singular number agreement via subatomic event pronouns in (176)–(179) recalls singular overt pronouns. Whether from grammar or the rigors of pronoun disambiguation, indefinite antecedents, a Columbia student and a Harvard student, occurring in clauses preceding the pronouns demand plural pronouns, (180) vs. (181), even if the plural pronouns are understood distributively, in contrast to other quantifiers, (182): (180) *At graduation dinners in New York and Cambridge, a Columbia student (has) praised and a Harvard student (has) toasted his (own) mother(s). (181) At graduation dinners in New York and Cambridge, a Columbia student (has) praised and a Harvard student (has) toasted their (own) mother(s). (182) At graduation dinners in New York and Cambridge, every Columbia student (has) praised and every Harvard student (has) toasted his (own) mother(s). Suppose that in such parallel structures the antecedent for the pronoun must be both Q Columbia student and Q Harvard student throughout (180)–(182). If the content inherited from its antecedents goes long—for example, the Columbia student praising and Harvard student toasting[’s (own) mother(s)] or the praiser and toaster[’s (own) mother(s)]—no one is referred to and singular reference fails (opposite successful plural reference in (181)). If it goes short—for instance, the student there[’s (own) mother(s)] or the participant there[’s (own) mother(s)]21—Columbia student and Harvard student are described alike, and singular reference fails again to refer to one of them provided that the pronoun and its deleted copy in (180) do not fall within the scope of different event quantifiers to obtain a “sloppy” interpretation for the pronoun (neither within the scope of quantifiers that Tense introduces nor of those introduced by adverbialization). Granted that the independent tenses in (180) are occasion for two existential event quantifiers, these do not include within their scope his (own) mother(s). Recall that under supermonadicity, the events of which it is said that they are of his (own) mother(s), ‘[℩E : pro][℩X : ∃E′ his(own) mother(s)[E′,X] there[E,X]] ∃E″ O[E,E″] Theme[E″, X]’, are themselves referred to by a descriptive event pronoun. There is again no way to spell out its content so that copied into the first conjunct of (180) it denotes the praising or the time of it, and in the second conjunct it denotes the toasting or its time. Likewise, since singular reference fails in (180), the adverbialization of the singular indefinite descriptions must not capture much within its scope, confined to its local thematic or participation relations. In contrast, the singular quantifiers in (182), as above, perch high enough that their adverbialization includes within their scope the event quanti-
DP and DP
113
fier associated with Tense and licenses the “sloppy” interpretation of the pronouns, which as copies of each other, are otherwise identical in content. Thus, (182) parses as in (183) (see (170)), and the confined indefinite quantification in (180) or (181) as in (184): (183) [D NP] [℩E : NP] [∃E0 : there[E,E0]] [Φ Δδ] and [D NP] [℩E : NP] [∃E0 : there[E,E0]] [Ψ Ωδ] (184) [A(n) NP] [[℩E : NP] [∃E0 : there[E,E0]]Φ Δδ] and [A(n) NP] [[℩E : NP] [∃E0 : there[E,E0]] Ψ Ωδ]22 The singular reference denied to the Right-Node Raised pronoun in (180) and the one pronoun Right-Node Raised in (185) is available to the independent singular pronouns in (186): (185) *At graduation dinners in New York and Cambridge, a Columbia student (has) praised to his roommate’s parents, and a Harvard student (has) toasted to his roommate’s whole family, his (own) roommate’s graduation at the head of the class. (186) At graduation dinners in New York and Cambridge, a Columbia student (has) praised to his roommate’s parents, and a Harvard student (has) toasted to his roommate’s whole family, their roommates’ graduation at the head of the class. (187) At graduation dinners in New York and Cambridge, every Columbia student (has) praised to his roommate’s parents, and every Harvard student (has) toasted to his roommate’s whole family, his (own) roommate’s graduation at the head of the class. These singular pronouns in (186) are under no constraint to resemble each other, and they may separately take on whatever descriptive content it takes to secure reference to their antecedents, to a Columbia student praising or to a Harvard student toasting. There is not more to be said about the singular subatomic event pronouns of singular number agreement that distinguishes them from overt singular pronouns. The two independent tokens of singular number agreement in (178) are interpreted independently, like the singular pronouns in (186), to denote what a Columbia student participates in secretly and what a Harvard student participates in openly. In contrast, the two identical copies of singular agreement in (176), denoting what a single student participates in, fail—like the singular Right-Node Raised pronouns in (180) and (185)—to denote anything. The narrow scope of the indefinite descriptions and their adverbialization does not license the “sloppy” interpretation of singular number agreement or singular pronoun, which is reserved for the singular quantifiers in (179), (182), and (187) occupying higher position.23
114
Chapter 2
2.2.1 Number agreement under a frame of reference
Granted that coordinated quantifiers support singular agreement only from on high, nothing in these remarks yet implies that much else tells apart singular and plural agreement in (188) and (189) except for the descriptive content of the event pronoun containing number morphology: (188) No Columbia student (secretly) and no Harvard student (openly) was on a strike committee in which the Columbia student led and the Harvard student followed. (189) No Columbia student (secretly) and no Harvard student (openly) were a strike committee in which the Columbia student led and the Harvard student followed. In (190), in the singular, tokens of ‘[℩E0 : sg.pro0]’ spoken and deleted, describing events of solitary participation near E, manage well enough to denote the Columbia student’s participation in the first conjunct and the Harvard student’s in the second: (190) [No X: ∃E Columbia student[E,X]] [℩E : Columbia student[E,X]] [∃E0 : there[E,E0]] [W[E0, X] (secretly[E0]) Δ] and [No X: ∃E Harvard student[E,X]] [℩E : Harvard student[E,X]] [∃E0 : there[E,E0]] [W[E0, X] (openly[E0]) [℩E0 : there[E,E0] sg.W[E0, X]] …] The singular number agreement falling within the scope of adverbialization is interpreted ‘sloppily’, as desired, merely relativizing its descriptive content to the events that the adverbial denotes. In the plural, ‘[℩E0 : pl.pro0]’, varying the descriptive content as in sections 2.0.0 and 2.1.3, instead denotes the collective participation of Columbia and Harvard student. Yet the empirical finding from section 1.5.1, to which I return below—the contrast between (191) and (192) and between (193) and (194)—suggests that there is more to tell apart singular and plural number agreement than merely the descriptive content of an event pronoun: (191) *No rocker and any roller has ever grooved to a funky disco beat. (192) No rocker and any roller have ever grooved to a funky disco beat. (193) *No Columbia student and any Harvard student was ever on a strike committee in which the Columbia student led and the Harvard student followed. (194) No Columbia student and any Harvard student were ever on a strike committee in which the Columbia student led and the Harvard student followed.
DP and DP
115
What accounts for the contrast must involve more than just this: number is expressed by a morpheme within an event pronoun, identical tokens of this pronoun occur within both conjuncts, and these tokens denote different events only if they fall within the scope of different event quantifiers. What more is involved attends to the frames of reference to which all NPs are addressed, which, recall (section 2.0.0), are essential to the meaning of sentences like (188) and (189). The illusion of quantifying over pairs of a Columbia student and a Harvard student is an effect of a quantification over frames of reference and subsequent reference by definite description in (188) and (189) to the Columbia student in f and the Harvard student in f: (42) [DP Q X : ∃E ∃f (Address[E,f] NP[E,X])] [DP Det [AdrP Adr NP]] NPs occur in DPs with a morphology that describes location in a spatiotemporal frame of reference as in (42), and a sequencing of spatiotemporal address, like sequence of tense, relates successive addresses to antecedent addresses (cf. (43)): (195) No Columbia student at any address in a frame of reference (secretly)—and no Harvard student at another address in such a frame of reference (openly)—was/were … Now if all NPs are addressed to a frame of reference (sections 1.6–1.8; chapters 8–15), counting is a protocol always under a frame of reference, and sentences are all reports of events under frames of reference, it could be that singular or plural reference is also such only under a frame of reference. The event pronoun translating number agreement, ‘[℩E0 : sg.pro0]’ or ‘[℩E0 : pl.pro0]’, is to be thought of as demonstrative to a frame of reference, referring to those events there in all of which participates the same one participant, in the singular, or to those events there of manifold participation. In (195), without further comment, the only frames of reference in hand contain a Columbia student and a Harvard student, and thus an event pronoun in the singular fails to refer in any such f: the events in f such that the same one participant participates in them. An event pronoun in the plural, however, succeeds directly in referring to the events in f such that participants participate in them, referring in this case to the events in which the Columbia student of f and the Harvard student of f participate. To felicitously use an event pronoun in the singular requires requantification over frames of reference to locate a frame of reference f′ within f that frames the participation of a single student: (196) No Columbia student at any address in a frame of reference f (secretly)—and no Harvard student at another address in such a frame of reference f (openly)— [∃f′: within(f′,f)] [℩E0 : sg.pro0] was/were …
116
Chapter 2
Accordingly, the syntax of (188) and (189) (and that of (191) and (192)) differs further in that the singular (188) (and (191)) demands requantification, an extra layer of adverbial event quantification: (197) No Columbia student (secretly) and no Harvard student (openly) somewhere there was on a strike committee in which the Columbia student led and the Harvard student followed. The implementation offered below, although provisional, still imports from later chapters some apparatus and vocabulary for adverbialization and frames of reference that I would rather have deferred. It should not obscure a simple insight into the cinema of number agreement. Any scene or frame of reference with a cast of more than one, say, a Columbia student and a Harvard student, may be continuous to scenes with the same cast without cue, but there is no continuity to solo scenes without cueing a close-up, a reframing. The thoughts that (198) and (199) express are therefore distinct as is the structure of these thoughts, despite their equivalent truth conditions: (198) No Columbia student and no Harvard student was on a strike committee in which the Columbia student led and the Harvard student followed. (199) No Columbia student and no Harvard student were on a strike committee in which the Columbia student led and the Harvard student followed. Singular number agreement here has a modal flavor, as it were, with frames of reference for worlds. The current frame of reference does not support it without requantification to introduce a narrower one, and thus (198) has an extra operator that (199) does not. As remarked above, it did not have to be this way, it being easy enough to treat the sentences hosting singular and plural number agreement as entirely parallel in their structure. The warrant for the cinematography is the absence of such parallelism in the contrast between (191) and (192) and between (193) and (194): (193) *No Columbia student and any Harvard student was ever on a strike committee in which the Columbia student led and the Harvard student followed. (194) No Columbia student and any Harvard student were ever on a strike committee in which the Columbia student led and the Harvard student followed.
DP and DP
117
The entertained logical form for (188) with singular agreement is (200) with the boldface operator the one required and absent from the logical form (201) for (189) with plural agreement: (200) [No X : ∃E ∃f(Address[E,f] Columbia student[E,X])] [℩E : ∃f(Address[E,f] Columbia student[E,X])] [[∃E0 : N[E,E0] there[E,E0]] W[E0, X] (secretly[E0]) Δδ] and [No X : ∃E ∃f(Address[E,f] Harvard student[E,X])] [℩E : ∃f(Address[E,f] Harvard student[E,X])] [[∃E0 : N[E,E0] there [E,E0]] W[E0, X] (openly[E0]) (δ [∃Ef : ∃E0(O[E0,E] there[E0,Ef] there[Ef,E0])] [℩E0 : ∃X (sg[E0] N[Ef,E0] there[Ef,E0] W[E0, X] )] … δ)] (201) [No X : ∃E ∃f(Address[E,f] Columbia student[E,X])] [℩E : ∃f(Address[E,f] Columbia student[E,X])] [[∃E0 : N[E,E0] there[E,E0]] W[E0, X] (secretly[E0]) Δδ] and [No X : ∃E ∃f(Address[E,f] Harvard student[E,X])] [℩E : ∃f(Address[E,f] Harvard student[E,X])] [[∃E0 : N[E,E0] there[E,E0]] W[E0, X] (openly[E0]) (δ [℩E0 : ∃X (pl[E0] N[E,E0] there[E,E0] W[E0, X] …)] … δ)] These logical forms short-change the reference to frames of reference explicit in the earlier paraphrases. First, it is left implicit that the frames of reference initially described within the DPs no Columbia student and no Harvard student are such that any of them contains a Columbia student (and no other) at one address and a Harvard student (and no other) at another address, such a frame of reference f being required for the Columbia student and the Harvard student in (198) and (199) to refer felicitously to the Columbia student in f and the Harvard student in f, as already discussed (section 2.0.0). Second, the pronouns or definite descriptions in the paraphrases referring directly to frames of reference are omitted for brevity in favor of an alternative understanding of the events quantified over. For (200) and (201), events are taken to occur under a frame of reference, which includes at least some coordinate system for spatiotemporal orientation. Moreover, a frame of reference designates zero or more landmarks, assigning them addresses to highlight their location. Events are thus taken to be fine-grained. Occupying the same spatiotemporal region, distinct events may differ only in their frames of reference, and these in turn may differ only in the spatiotemporal regions (landmarked, i.e., addressed). In (200) and (201), ‘Address[E,f] Columbia student[E,X]’ means that X’s states E of being a Columbia student are landmarks in f. The relation ‘there[E,E0]’ says E’s and E0’s frames of reference are of the same spatiotemporal region, orientation, and coordinate system—that is, E0 are where E are; moreover, all of E’s landmarks are also landmarks in E0’s frame of reference, which is in turn extended to include as
118
Chapter 2
landmarks the spatiotemporal regions that E0 occupy. As elsewhere, ‘N[E,E0]’ is a spatiotemporal or causal relation that mediates adverbial modification throughout. The last line of (200) shows singular number agreement ‘[℩E0 : pro0]’ expanded with the content describing the events that the descriptive pronoun refers to, the events there all with the same one participant. Absent the boldface operator, singular number agreement fails, since the events there are the events under the initial f, which includes states of being a Columbia student and being a Harvard student, with more than one participant. Thus, singular number agreement by design requires the intervention of an operator to rescue it. That operator prefixed to the RightNode Raised phrase contains the free variable E and its tokens come to be bound by different event quantifiers, denoting in the first conjunct a state of being a Columbia student and in the second, a state of being a Harvard student. The boldface operator introduces some events Ef that are of the same spatiotemporal region as this state of being one student, landmarking exactly that state and free to omit from its frame of reference the state of being the other student. Within such a frame of reference singular number agreement is licensed. In the logical form (201) for the sentence with plural agreement, plural number agreement refers to the events there, where the Columbia student and the Harvard student are, with more than one participant. This, by itself, may be too broad a description if that frame of reference frames other states or events, but here, as before (see section 2.1.3), the descriptive content of plural number agreement may go on to spell out the participation of a Columbia student and a Harvard student. In answer to an empirical contrast ((191) vs. (192), (193) vs. (194)), singular agreement, by design in contrast to plural agreement, invokes an extra layer of quantification within the clauses coordinated. Since number agreement is just one event pronoun among many, that extra layer of quantification should affect in equal measure the other event pronouns introducing subatomic clauses: (202) No rocker and no roller has ever grooved to a funky disco beat. (203) No rocker and no roller have ever grooved to a funky disco beat. (204) ‘[℩Ei: proi] [A(n) x: ∃E funky disco beat[E,x]]To[Ei,x]’24 The Right-Node Raised phrase in (202) or (203) contains (204), as does the phrase deleted from the first conjunct. Whether or not rocker and roller are permitted different funky disco beats depends on whether or not the two tokens of ‘[℩Ei: proi]’ fall within the scope of different event quantifiers. As above, in the singular (202), different event quantifiers do intervene and the sentence readily allows rocker and roller her own beat. In the plural (203), their grooving is said to be to a funky disco beat, unless there is a further intervention from tacit each: (205) No rocker and no roller have ever grooved each to a funky disco beat.
DP and DP
119
Later sections sort out these different sources for a dependent interpretation of a funky disco beat. In section 2.3.0, the distribution of “lazy” or “sloppy” interpretations of the contents of Right-Node Raised phrases further probes the syntactic distribution of the event quantifiers that populate the clausal architecture; section 2.3.2.3 takes up tacit each. 2.2.2
Thematic relations and number agreement
In Lebanese Arabic, (206) and (207) may be reports of Alia and Marwan joining hands in the same circle. If singular number agreement in (206) as elsewhere describes events with only one participant, the events there with the same one participant, and any tokens of singular agreement in (206) occur within the scope of all the same event quantifiers, then tout court singular agreement does not describe an event or events in which both Alia and Marwan participate: (206) ʕamlit alia w marwaan Hal’a Hawl š-šažra formed.3fs Alia and Marwan circle around the-tree ‘Alia and Marwan formed a circle around the tree.’ (Munn 1999) (207) ʕamalou alia w marwaan Hal’a Hawl š-šažra formed.3mp Alia and Marwan circle around the-tree ‘Alia and Marwan formed a circle around the tree.’ But for singular agreement to locate an event in which Alia and Marwan do not both participate, they must do something different. Say there is an event in which Alia is the figured participant (‘W’) and an event in which Marwan is merely an accomplice (‘Cum’). Definite description denoting the events of participation, the W-ing, denotes events with only the one participant, Alia, to which Marwan joins his events of complicity (‘Cum’). If this is not to be entirely self-serving, this should be a distinction with a difference, to which end section 2.5 is about the difference in meaning between (206) and (207). Because (206) and (207) may report the same action in the world, it is a difference of perspective. The asymmetry in (206) is that of Figure and Ground (Talmy 1978, 1983; Gleitman et al. 1996), which makes itself felt in (208) and (209) even where a symmetry is being reported: (208) The humblest citizen is equal to the President. (209) The President is equal to the humblest citizen. (210) The President and the humblest citizen are equal(s). The humblest citizen and the President are equal(s). Asymmetric perspectival events that are different from each other and different from the action or symmetric state they are perspectives on again expose the handiwork of supermonadicity and its army corps of events.
120
Chapter 2
Number agreement is a description of events, with an aside that their participants are one or many. It is thus divorced from the reference of the subject, and therefore a language in which comitative phrases contribute to number agreement as in (211) is no cause for despair that the plural object Marta con María and the plural object Marta y María might not be the same if con ‘with’ and y ‘and’ are not synonyms:25 (211) Marta con María levantaron el piano Martha with Maria lifted.pl the piano (Camacho 1997, 206)
[Spanish]
(212) Marta y María levantaron el piano Martha and Maria lifted.pl the piano It is rather an occasion to discover what in the syntax of such a language allows the plural number agreement to describe in (211) an event or events in which both Marta and Maria participate (section 2.6). But already there is an argument for supermonadicity. Fair enough that (211) and (212) are not synonyms, one expressing comitativity (“con”) and the other not. But since both sentences entail that levantaron el piano, it cannot be that Martha is the Agent but Maria is not, if there is any sense at all to what relates agency, cause, and effect. So then what is it that Martha does according to (211) that Maria does not, if it is not to participate in a way (‘W’) that Martha doesn’t (‘con’) in events of theirs that amount to some Agents’ action causing the piano to lift, as only supermonadic logical form can express? 2.3
Number agreement in Lebanese Arabic
Number agreement is one of the arguments that support the sentential, Right-Node Raised analysis of (213) with its subatomic pronouns referring to events: (213) Biff and Tiff cooked up the Harvard beets. The argument turns to the widespread phenomenon of first-conjunct agreement, the case in point being Lebanese Arabic, which has been the subject of an illuminating exchange (Aoun, Benmamoun, and Sportiche 1994, 1999; Munn 1999; Camacho 1997, 2003; Doron 1999).26 Like English, number agreement in Lebanese27 is semantic in that it depends in (214) and (215) on whether the producer and the director are the same person: (214) l-muntijeh w l-muxrijeh maDyet contra ma’ paramount the-producer.F and the-director.F signed.3FS contract with Paramount ‘The producer and the director has signed the contract with Paramount.’ (215) l-muntijeh w l-muxrijeh maDuu contra ma’ paramount the-producer.F and the-director.F signed.3MP contract with Paramount ‘The producer and the director have signed the contract with Paramount.’
DP and DP
121
But as in many languages, conjoined nominals also afford a further occasion for singular agreement. When postverbal, and only when postverbal, agreement can pick out the first conjunct: (216) DeHk-et alia w marwaan laughed.3fs Alia and Marwan ‘Alia and Marwan laughed.’ (217) DeHk-u alia w marwaan laughed.3mp Alia and Marwan ‘Alia and Marwan laughed.’ (218) *DeHek alia w marwaan *laughed.3ms Alia and Marwan ‘Alia and Marwan laughed.’ (219) alia w marwaan DeHk-u Alia and Marwan laughed.3mp ‘Alia and Marwan laughed.’ (220) *alia w marwaan DeHk-et *Alia and Marwan laughed.3fs ‘Alia and Marwan laughed.’ (221) *alia w marwaan DeHek *Alia and Marwan laughed.3ms ‘Alia and Marwan laughed.’ Already there comes an objection to a view that would have every instance of Alia and Marwan refer to the two Harvard students. In Lebanese, the two Harvard students can never support singular agreement: (222)*DeHek telmizen Harvard *laughed.3ms student.dual Harvard ‘The two Harvard students laughed.’ Thus it cannot be held both that a verb agrees with the reference of its subject and that the reference of Alia and Marwan in (216) is to the two Harvard students. One way out denies the first premise, the rule of grammar (Munn 1999; Doron 1999): under certain conditions exemplified in (216), a verb is said to agree not with its subject but with the first conjunct of its subject. First-conjunct agreement is itself semantic—singular or plural in (223)–(224) according to whether or not the same woman is the producer and director.
122
Chapter 2
(223) Yesterday, we managed in a stroke to hire the principals for both the production staff and cast … maDyet l-muntijeh w l-muxrijeh w l-baTal w l-baTaleh signed.3fs the-producer.f and the-director.f and the-star.m and the-star.f contra ma’ paramount contract with Paramount ‘The producer and the director and the leading man and the leading lady signed a contract with Paramount.’ (224) maDuu l-muntijeh w l-muxrijeh w l-baTal w l-baTaleh signed.3mp the-producer.f and the-director.f and the-star.m and the-star.f contra ma’ paramount contract with Paramount ‘The producer and the director and the leading man and the leading lady signed a contract with Paramount.’ Despite the lurking threat to semantic compositionality this implies, I will not deny that such a rule of first-conjunct agreement could be rationalized within an appropriately chosen theory of syntax (see Munn 1999; Doron 1999).28 As a consequence of this liberalized rule, there is then no need to find a difference in structure between (216) and (217) or to depart from the idea that Alia and Marwan refers on both occasions to the two Harvard students. The alternative way out (Aoun, Benmamoun, and Sportiche 1994, 1999; Camacho 1997, 2003) holds to the simple rule that verbs agree with their subjects but denies instead the second premise, that Alia and Marwan is a subject in (216), and concomitantly denies that it is a DP referring to the two students. Rather, there is Conjunction Reduction: the conjoined DPs belong to separate clauses, where the overt verb in the first clause agrees with the simple DP that is its subject. (225) DeHket alia, w (DeHek) marwaan laughed.3fs Alia, and (laughed) Marwan ‘Alia laughed and Marwan laughed.’ With both ways out, the reported asymmetry between pre- and postverbal conjuncts is a point of syntax. Here the ungrammaticality of (226) amounts to the fact that the tacit and overt verb cannot occur in the reverse order as in the parse shown. As in many languages, the verb deletion process is found to be unidirectional. (226) *marwaan w alia DeHket *Marwan and Alia laughed.3fs ‘Marwan (laughed), and Alia laughed.3fs.’
DP and DP
123
As is to be expected, Conjunction Reduction here as elsewhere confronts the usual objections arising from collective predicates. Despite several cases, to which I return in note 80, where first-conjunct agreement excludes collectivized phrases (Aoun, Benmamoun, and Sportiche 1994, 1999), it only takes a few good examples to locate the problem that sentences such as (227)–(249) that report collective actions with conjoined DPs find no equivalent in a conjunction of full clauses. (227) ʕamlit alia w marwaan Hal’a Hawl š-šažra formed.3fs Alia and Marwan circle around the-tree ‘Alia and Marwan formed a circle around the tree.’ (Munn 1999) (228) fallet alia w marwaan sekraneen left.3fs Alia and Marwan drunk.3mp (cf. (163)) ‘Alia and Marwan left drunk.’ (229) kenet hiyyeh w iyyeh ʕam yelʕabo (230) was.3fs she and him prog playing.3mp ‘She and he were playing.’ *kenet henneh ʕam yelʕabo *was.3fs they prog playing.3mp (231) atlet alya w marwaan l-bsayne killed.3fs Alia and Marwan the cat29 ‘Alia and Marwan killed the cat.’ (232) kasaret alia w marwaan š-šebbek broke.3fs Alia and Marwan the-window ‘Alia and Marwan broke the window.’ (233) kasaret r-riH w l-gheSn š-šebbek broke.3fs the-wind and the-branch the-window ‘The wind and the branch broke the window.’ (234) fayya’et alia w marwaan l-bebe woke.3fs Alia and Marwan the-baby ‘Alia and Marwan woke the baby.’ (235) fayya’ l-alarm w l-etfe’iyyeh l-bebe woke.3ms the-alarm and the-firecar the baby ‘The car alarm and the fire engine woke the baby.’ (236) Hamlet alia w marwaan l-piaano lifted.3fs Alia and Marwan the piano ‘Alia and Marwan lifted the piano.’ (237) fatHet alia w marwaan beb l-xazneh opened.3fs Alia and Marwan door the-vault ‘Alia and Marwan opened the vault doors.’
124
Chapter 2
(238) waSlet alia w marwaan š-šraayeT connected.3fs Alia and Marwan the wires ‘Alia and Marwan connected the wires.’ (239) waSlet alia w marwaan muxaTTaTaatun (bibaʕD) connected.3fs Alia and Marwan plans-their (to each other) ‘Alia and Marwan connected their ideas/plans.’ (240) xalTet alia w marwaan d-dheen l-azra’ blended.3fs Alia and Marwan the-paint the-blue ‘Alia and Marwan blended the blue paint.’ (241) waffa’et tatsher w shmidt bein briitania w l-FRG allied.3fs Thatcher and Schmidt between the UK and the-FRG (242) SaalaHet maliket n-naruuj w klinton isra’iil w s-selTa reconciled.3fs Queen Norway and Clinton Israel and the-authority l-felsTiiniyyeh the-Palestinian ‘The Queen of Norway and Clinton reconciled Israel and the Palestinian Authority.’ (243) tweeznet alia w marwaan ʕa l-marjouHa balanced.3fs Alia and Marwan on the seesaw. ‘Alia and Marwan balanced on the seesaw.’ (244) weeznet alia w marwaan l-wleed ʕa l-marjouHa balanced.3fs Alia and Marwan the-children on the-seesaw ‘Alia and Marwan balanced the children on the seesaw.’ (245) tjammaʕet emma w l-anarkiyiin be l-haymarket gathered.3fs Emma and the anarchists in Haymarket. ‘Emma and the anarchists gathered in Haymarket.’ (246) ttaHdet emma w l-anarkiyiin taHt bayeen waahad united.3fs Emma and the anarchists under manifesto one ‘Emma and the anarchists united under one manifesto.’ (247) ttaHdet chrysler w mercedes-benz taHt mudiir waahad united.3fs Chrysler and Mercedes-Benz under CEO one ‘Chrysler and Mercedes-Benz united under one CEO.’ (248) ttaHdet chrysler w mercedes-benz ka šerkeh waHdeh united.3fs Chrysler and Mercedes-Benz as corporation one ‘Chrysler and Mercedes-Benz united as one corporation.’ ttaHad obama w biden ʕa lista intixabiyyeh waHdeh united.3ms Obama and Biden on list electoral one ‘Obama and Biden united on one electoral list.’
DP and DP
125
(249) Saaret ’eT ʕet z-zebdeh w T-THiin ʕajiineh became.3fs stick the-butter and the-flour dough ‘The stick of butter and the flour became dough.’ If one were inclined with Aristotle to dismiss the sentential conjunction when faced with the contrast between Donca and Mirçea met and *Donca met and Mirçea met, then (227)–(249) tell against separate clauses in the treatment of first-conjunct agreement. Disabused of Aristotle’s fallacy, it suffices to find Eventish logical forms for (227)–(249) in which plural event pronouns denote the several events the participants separately participate in, which are then said to amount to the collective event described. In so doing, I will introduce and suggest an analysis for some of the semantic conditions on first-conjunct agreement. It should be noted that the existence of such conditions is prima facie an argument against the first account liberalizing the rule of number agreement. As Aoun, Benmamoun, and Sportiche (1999, 677) remark, any interpretive difference between first-conjunct agreement and total agreement is unexpected under an account where the structures are identical and differ only in the calculation of number agreement. Thus number agreement, being semantically conditioned, turns out to argue in favor of the sentential conjunction rather than against it. Among its semantic conditions, first-conjunct agreement contrasts eventive (250) and stative (251):30 (250) Saret (maʕle’t) z-zebdeh w l-THiin ʕajiineh became.3fs (tbsp. of) the-butter and the flour dough ‘The (tbsp. of) butter and the flour became dough.’ Sar l-Haliib w l-xalTah gateau became.3ms the-milk and the-mix cake ‘The milk and the mix became cake.’ (251) *keenet (maʕle’t) z-zebdeh w l-THiin ’ajiin was.3fs (tbsp. of) the-butter and the-flour dough ‘The (tbsp. of) of butter and the flour were dough.’ (252) Saro (maʕle’t) z-zebdeh w l-THiin ʕajiineh became.3mp (tbsp. of) the-butter and the flour dough ‘The (tbsp. of) butter and the flour became dough.’ (253) *keeno (maʕle’t) z-zebdeh w l-THiin ’ajiin were.3mp (tbsp. of) the-butter and the-flour dough ‘The (tbsp. of) of butter and the flour were dough.’ Among reports of events, the first-conjunct agreement in (254) and (255) evokes a spatiotemporal proximity between Alia and Marwan, as in the report of a scene
126
Chapter 2
observed from a single vantage point, which is absent from reports with full agreement, (256) and (257). (254) meetet alia w marwaan died.3fs Alia and Marwan ‘Alia and Marwan died.’ (255) deHket alia w marwaan laughed.3fs Alia and Marwan ‘Alia and Marwan laughed.’ (256) meetou alia w marwaan died.3mp Alia and Marwan ‘Alia and Marwan died.’ (257) deHkou alia w marwaan laughed.3mp Alia and Marwan ‘Alia and Marwan laughed.’ Spatiotemporal proximity appears to be necessary and sufficient for (254) and (255). Sentence (254) could report deaths in the same hospital room by different causes but not deaths by the same cause, disease or natural disaster, in far-flung locations. Similarly, (255) suggests a single occasion in the presence of the observer at which Alia and Marwan laugh, perhaps at their own jokes, and despite prompting by the same joke or person, (255) would be an inappropriate report of Alia and Marwan laughing apart. No such conditions intrude on the sentences with full agreement, (256) and (257). Thus, more than a rule of number agreement is at stake. 2.3.0
The distribution and interpretation of event quantifiers and descriptions
The content of the semantic conditions that discriminate first-conjunct agreement and full agreement is not always a matter of spatiotemporal proximity, and I return to a fuller sketch below. There are some interesting quirks merely in the distribution of these conditions, which I turn to now, and these derive from the elementary features of the logical forms so far assumed. First, note that the spatiotemporal proximity felt in (255) disappears with the quantificational subjects in (258)31 and then returns with a vengeance in (259) when the quantifier is taken to bind the pronoun: (258) DeHket kell bent w kell Sabi laughed.3fs every girl and every boy ‘Every girl and every boy laughed.’ (259) DeHket kell mara w ebn-a laughed.3fs every woman and child-her ‘Every woman and her child laughed.’
DP and DP
127
In this case, a stronger, comitative condition, as Aoun, Benmamoun, and Sportiche (1999, 677) call it, has the woman and her child laughing with each other at the same thing.32 The conditions again disappear entirely with full clausal conjunctions and with overt Right-Node Raising: (260) DeHket alia w DeHek marwaan laughed.3fs Alia and laughed.3ms Marwan ‘Alia laughed and Marwan laughed.’ (261) DeHket alia w DeHek marwaan ʕa esteez linguistics laughed.3fs Alia and laughed.3ms Marwan at professor linguistics ‘Alia laughed and Marwan laughed at a linguistics professor.’ (262) DeHket kell bent w DeHek kell Sabi (ʕa esteez lingustics) laughed.3fs every girl and laughed.3ms every boy (at professor linguistics) ‘Every girl and every boy laughed at a linguistics professor.’33 A paradigm closely related to the above, according to the account to be given, concerns the interpretation of the Right-Node Raised constituents. The sentences in (263), with full number agreement, report that Alia and Marwan laugh at the same professor and the same joke: (263) DeHkou alia w marwaan ʕa esteez linguistics laughed.3mp Alia and Marwan at professor linguistics DeHkou alia w marwaan ʕa nekteh la Woody Allen laughed.3mp Alia and Marwan at joke of Woody Allen The DP-coordination thus differs from the overt Right-Node Raising in (264), which can be understood to be reports of them laughing at different professors and different jokes: (264) DeHket alia w DeHek marwaan ʕa esteez linguistics laughed.3fs Alia and laughed.3ms Marwan at professor linguistics DeHket alia w DeHek marwaan ʕa nekteh la woody allen laughed.3fs Alia and laughed.3ms Marwan at joke of Woody Allen First-conjunct agreement, as in (265), forces it to be the same, one professor or joke, unless the subjects are quantificational, as in (266). Here each girl may laugh at her own favorite professor or joke and similarly for each boy. (265) DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics DeHket alia w marwaan ʕa nekteh la Woody Allen laughed.3fs Alia and Marwan at joke of Woody Allen
128
Chapter 2
(266) DeHket kell bent w kell Sabi ʕa esteez linguistics laughed.3fs every girl and every boy at professor linguistics DeHket kell bent w kell Sabi ʕa nekteh la Woody Allen laughed.3fs every girl and every boy at joke of Woody Allen If however a pronoun covaries with the quantifier in the first conjunct, as in (267) and (268), the woman and the child are again reported to laugh with each other at the very same thing: (267) DeHket kell mara w ebna ʕa esteez linguistics laughed.3fs every woman and child-her at professor linguistics DeHket kell mara w ebna ʕa nekteh la Woody Allen laughed.3fs every woman and child-her at joke of Woody Allen (268) DeHket kell mara w kell ebn men wleed-a ʕa esteez laughed.3fs every woman and every child of children-her at prof linguistics linguistics DeHket kell mara w kell ebn men wleed-a ʕa laughed.3fs every woman and every child of children-her at joke nekteh la Woody Allen of Woody Allen The above pattern reflects the distribution of event quantifiers. Recall from chapter 1 that the occurrence of a tensed verb introduces anew a quantifier over events, leading (269) to its absurd reading: (269) *At Acropolis Pizza, the sauce is a perfect marriage of two rivals, and the cheese is a perfect marriage of two rivals. (270) At Acropolis Pizza, the sauce and the cheese are a perfect marriage of two rivals. In (264), the occurrence of two event quantifiers supports the “sloppy” interpretation of the Right-Node Raised constituent when it is copied into the first conjunct. The independently tensed clauses are as exempt from any special spatiotemporal or causal conditions between them as any other sequence of tensed clauses. (264) ∃e(Laughed.3fs(e) & Agent(e,Alia) & at(e, a linguistics professor)), and ∃e(laughed.3ms(e) & Agent(e,Marwan) & at(e, a linguistics professor)). ∃e(Laughed.3fs(e) & Agent(e,Alia) & at(e, a Woody Allen joke)), and ∃e(laughed.3ms(e) & Agent(e,Marwan) & at(e, a Woody Allen joke)).34 In (270) and in both (271) and (272), there is only the one tensed verb to introduce a quantifier over events. The phrases ‘at(e, a linguistics professor)’ and
DP and DP
129
‘at(e, a Woody Allen joke)’ apply to these events. Any further occurrence of these phrases would apply (redundantly) to the same events, without differentiating the objects of Alia’s and Marwan’s laughter: (271) ∃e(Laughed.3fs(e) & Agent(e,Alia) and Agent(e,Marwan) & at(e, a linguistics professor)). ∃e(Laughed.3fs(e) & Agent(e,Alia) and Agent(e,Marwan) & at(e, a Woody Allen joke)). (272) ∃e(Laughed.3mp(e) & Agent(e,Alia) and Agent(e,Marwan) & at(e, a linguistics professor)). ∃e(Laughed.3mp(e) & Agent(e,Alia) and Agent(e,Marwan) & at(e, a Woody Allen joke)).35 Below I return to the spatiotemporal condition that tells apart (271) from (272), and continue now to survey those effects that derive from the distribution of event quantifiers. In contrast to (263) and (265), it was remarked above that the quantificational subjects in (266) restore to every girl and boy the possibility of laughing at their own professors and jokes, despite the absence of a second tensed verb: (263) DeHkou alia w marwaan ʕa esteez linguistics laughed.3mp Alia and Marwan at professor linguistics DeHkou alia w marwaan ʕa nekteh la Woody Allen laughed.3mp Alia and Marwan at joke of Woody Allen (265) DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics DeHket alia w marwaan ʕa nekteh la Woody Allen laughed.3fs Alia and Marwan at joke of Woody Allen (266) DeHket kell bent w kell Sabi ʕa esteez linguistics laughed.3fs every girl and every boy at professor linguistics DeHket kell bent w kell Sabi ʕa nekteh la Woody Allen laughed.3fs every girl and every boy at joke of Woody Allen Here, it is the quantifiers themselves, in contrast to the names in (263) and (265) and the (in)definite descriptions in (273)–(274), that introduce the event quantifiers supporting the “sloppy” interpretation: (273) DeHkou shi bent w shi Sabi ʕa esteez linguistics laughed.3mp some girl and some boy at professor linguistics DeHket shi bent w shi Sabi ʕa esteez linguistics laughed.3fs some girl and some boy at professor linguistics
130
Chapter 2
(274) DeHkou bentein w tlat Sebyen ʕa esteez linguistics laughed.3mp girl.dual and three boys at professor linguistics As in the contrast from section 2.2 between (275) and (277), proper names and (in) definite descriptions do not license a following event quantifier that includes within its scope the remainder of the sentence (see section 2.2, (184)): (275) *A Columbia student (secretly) and a Harvard student (openly) was on a strike committee that met in her dorm room. (276) A Columbia student (secretly) and a Harvard student (openly) were on a strike committee that met in their dorm room(s).36 (277) Every Columbia student (secretly) and every Harvard student (openly) was on a strike committee that met in her dorm room. With Davidsonian logical forms, there can be no distributive quantification over objects without distributive quantification over events (sections 2.2 and 1.5.0; Schein 1993, 139ff.). The logical form for (266) must be as in (278), where the Right-Node Raised constituent and the elided phrase are captured by different event quantifiers, resulting in the “sloppy” interpretation. (278) ∃e(Laughed.3fs(e) & [Every x : girl(x)][∃e′ : e′ ≤ e](Agent(e′,x) & at(e′, a linguistics professor)), and [Every x : boy(x)][∃e′ : e′ ≤ e](Agent(e′,x) & at(e′, a linguistics professor))). ∃e(Laughed.3fs(e) & [Every x : girl(x)][∃e′ : e′ ≤ e](Agent(e′,x) & at(e′, a Woody Allen joke)), and [Every x : boy(x)][∃e′ : e′ ≤ e](Agent(e′,x) & at(e′, a Woody Allen joke))). (In)definite descriptions do not license a following event quantifier that includes within its scope the remainder of the sentence (see section 2.2.0, (184)): (279) a. *∃e(Laughed(e) & [Some x : girl(x)][∃e′ : e′ ≤ e](Agent(e′,x) & at(e′, a linguistics professor)), and [Some x : boy(x)][∃e′ : e′ ≤ e](Agent(e′,x) & at(e′, a linguistics professor))) b. ∃e(Laughed(e) & [Some x : girl(x)]Agenti(e,x) and [Some x : boy(x)]Agenti(e,x) & at(proi,i, a linguistics professor))37
DP and DP
131
(280) a. *∃e(Laughed(e) & [∃X : 2(X) & … girl(x) …][∃e′ : e′ ≤ e](Agent(e′,X) & at(e′, a linguistics professor)), and [∃X : 3(X) & … boy(x) …][∃e′ : e′ ≤ e](Agent(e′,X) & at(e′, a linguistics professor))) b. ∃e(Laughed(e) & [∃X : 2(X) & … girl(x) …]Agenti(e,X) and [∃X : 3(X) & … boy(x) …]Agenti(e,X) & at(proi,i, a linguistics professor)) The distribution of the “sloppy” interpretation of Right-Node Raised constituents coincides with the distribution of event quantifiers, reflecting the division of natural language DPs into distributive quantifiers and (in)definite descriptions (i.e., into first- and second-order quantifiers in Schein 1993, 118).38 Absent a second tensed verb or a distributive quantifier, the action described in the second clause must be related to the event of the first clause, thereby excluding “sloppy” interpretations with both full and first-conjunct agreement. Absent a second tensed verb, both of the second conjuncts in (281) and (282) relate to the same events as their respective first conjuncts, whether agreement is partial or full: (281) DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics (282) DeHkou alia w marwaan ʕa esteez linguistics laughed.3mpAlia and Marwan at professor linguistics Yet, following Aoun, Benmamoun, and Sportiche 1994, 1999, some difference in the configuration underlies the difference in number agreement. Although Alia and Marwan are related to the same laughing, (281) does not relate them to it in the same way that (282) does. When there are many in the same event, singular agreement imposes spatiotemporal proximity and a restriction to eventive predicates, as noted above. This is not much more than to remark that singular agreement arises only under special conditions where plural agreement should otherwise have been expected. If the many are related to different events, each with just a single participant, singular agreement is canonical and these events escape from any special conditions. Thus, singular agreement with the stative predicate in (283) is unacceptable, where Alia and Marwan are related to the same event. In contrast, since either the occurrence of a second tensed verb or a distributive quantifier licenses an additional event quantifier, singular agreement and the stative predicate is acceptable
132
Chapter 2
in (284) and (285) without implying a spatiotemporal proximity between Alia and Marwan or between teacher and student: (283) *keenet alia w marwaan bi ouda *was.3fs Alia and Marwan in room *ken marwaan w alia bi ouda was.3ms Marwan and Alia in room (284) keenet Alia w ken marwaan bi ouda Was Alia and was Marwan in room (285) ken kell esteez w kell telmiiz bi saff imtiHaneet Was every teacher and every student in class exams The contrast in Lebanese Arabic between the coordinations containing (in)definite descriptions in (263), (265), (273)–(274), and (283) and those containing distributive quantifiers in (266) and (285) resumes, as already remarked, the contrast between (275) and (277), where only distributive quantification licenses the “sloppy” interpretation of singular pronouns and singular number agreement, itself an event pronoun denoting events of a single participant: (275) *A Columbia student (secretly) and a Harvard student (openly) was on a strike committee that met in her dorm room. (277) Every Columbia student (secretly) and every Harvard student (openly) was on a strike committee that met in her dorm room. Given the logical syntax, it is a point of logic that the singular reference of pronoun or number agreement could not succeed unless the singular referring expressions fall within the scope of different event quantifiers: (286) [Every NP] [℩E : NP] [[∃E0 : there[E,E0]] W0 (secretly) Δδ] and [Every NP] [℩E : NP] [[∃E0 : there[E,E0]] W0 (openly) [℩E0 : sg.pro0[E0]].Ωδ] The Right-Node Raised phrase in (286) (see (183)), ‘[℩E0 : sg.pro0[E0]].Ωδ,’ copied into the antecedent deletion site Δδ (section 2.1.3), locates tokens of singular number agreement ‘[℩E0 : sg.pro0[E0]]’ within the scope of distinct tokens of ‘[℩E : NP],’39 where the descriptive anaphor denotes what events the same one participant participates in there: (287) [℩E0 : sg.(there[E,E0] W0)[E0]]40 No point of logic but in response to (126) and similar facts, a different syntax is stipulated for (in)definite descriptions denying the coordination the event quantification necessary for a “sloppy” interpretation of singular agreement:
DP and DP
133
(288) [A(n) NP] [[℩E : NP] [∃E0 : there[E,E0]]W (secretly) Δδ] and [A(n) NP] [[℩E : NP] [∃E0 : there[E,E0]] W (openly) [℩E0 : sg.pro0[E0]].Ωδ] Recall that the tokens of singular agreement, pronounced and unpronounced, are exact copies, and thus denote the same events unless within the scope of different event quantifiers, but there are no events that a Columbia student participates in alone and a Harvard student does too. Despite the empirical evidence distinguishing distributive quantification and (in)definite description, here and elsewhere (see appendix 1, sections 12.2–12.3 and 13.0, and especially Szabolcsi 2009), nothing in the meaning of (in)definite description recognized here precludes scope assignment analogous to distributive quantification. This is just as well since a mixed coordination licenses singular number agreement and an interpretation of (289) and (290) that affords Alia and the teachers each their own room: (289) ken kell estez w alia bi ouda was.3ms every teacher and Alia in room (290) kenet alia w kell estez bi ouda was.3fs Alia and every teacher in room It appears that the larger structure required for distributive quantification in one clause licenses the parallel structure in a coordinated clause:41 (291) [∃E Was.3ms[E]42 [℩E : pro[E]][℩E0,0 : pro0,0[E0,0]] O[E,E0,0] [ [Every teacher] [℩E : teacher] [[∃E0 : there[E,E0]] W0 Δδ] and [℩ Alia] [℩E : Alia] [[∃E0 : there[E,E0]] W0 [℩E0 : pro0[E0]] in roomδ]]] (292) [∃E Was.3fs[E] [℩E : pro[E]][℩E0,0 : pro0,0[E0,0]] O[E,E0,0] [ [℩ Alia] [℩E : Alia] [[∃E0 : there[E,E0]] W0 Δδ] and [every teacher] [℩E : teacher] [[∃E0 : there[E,E0]] W0 [℩E0 : pro0[E0]] in roomδ]]] The logical forms derived for (266), (285), (289), and (290), wherever at least one subject is a distributive quantifier, are nearly the equivalent of a conjunction of fully tensed clauses with their own independent event quantification. The structures are thus expected to show singular agreement without falling under the special semantic conditions for first-conjunct agreement. That is, a distributive quantifier in either conjunct is sufficient to escape the spatiotemporal and eventive restrictions that arise when (in)definite descriptions alone are conjoined under singular agreement.
134
Chapter 2
2.3.0.0 Summary
Davidsonian logical form commits itself to the thesis that every token of a distributive quantifier introduces its own partitive event quantifier and (in)definite descriptions do not. Thus a fundamental distinction between distributive quantifiers and (in)definite descriptions is provided for. Providing a semantics for cross-reference to events suffices to explain the facts of number agreement illustrated by the examples below: (281) DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics (282) DeHkou alia w marwaan ʕa esteez linguistics laughed.3mp Alia and Marwan at professor linguistics (293) keenou alia w marwaan bi ouda were.3mp Alia and Marwan in room (183) *keenet alia w marwaan bi ouda was.3fs Alia and Marwan in room *ken marwaan w alia bi ouda was.3ms Marwan and Alia in room (285) ken kell esteez w kell telmiiz bi ouda was.3ms every teacher and every student in room (289) ken kell estez w alia bi ouda was.3ms every teacher and Alia in room (290) kenet alia w kell estez bi ouda was.3fs Alia and every teacher in room The conjuncts within each of (281)–(283) describe the same event, of which it would be incoherent to say that Alia is its unique Agent/Theme and Marwan is too. It is similarly incoherent to allow that the same event took place in two different rooms, (293)–(283), or to allow that two different linguistics professors are each its unique target, (281)–(282). The “sloppy” interpretation of the Right-Node Raised constituent is therefore excluded: Alia and Marwan laugh at the same professor, and they are in the same room. Singular agreement implies that the event has a single participant bearing the thematic relation so modified. In the general case, the coordination of DPs reflects a coordination of similar structures, and thus overt singular agreement in one conjunct parallels tacit singular agreement in the other. But then, in the general case, singular agreement with conjoined (in)definite descriptions will be excluded, as in (283), implying that the same event has different but unique Themes. Some other configuration in which the DPs are treated asymmetrically, a comitative structure of some kind, will have to be introduced for those cases like (281) that present singular agreement with conjoined (in)definite descriptions. As
DP and DP
135
we have seen, the special structure brings with it conditions of spatiotemporal proximity and eventivity (ruling it out in (283)). In contrast, the partitive event quantification licensed by a distributive quantifier in (285)–(290) allows the separate conjuncts to speak of distinct events. In that case, singular agreement is expected, since in each of these events there is either only Alia or only a teacher. No condition of eventivity is imposed, and the events described by distinct conjuncts need not be any more proximate than the events described by any other sequence of sentences. Furthermore, since the events are distinct, it becomes possible to understand the Right-Node Raised constituent under a “sloppy” interpretation, describing each as taking place in a perhaps different room.43 2.3.1
Number agreement and subatomic “telescoping”
In (294), Alia and Marwan participate in the same event and so their laughter targets the same professor. As remarked earlier, first-conjunct agreement with conjoined (in)definite descriptions imposes further a spatiotemporal proximity between Alia and Marwan as of the report of a scene observed from a single vantage point. The spatiotemporal proximity felt in (294) disappears when either of the conjuncts is replaced by a distributive quantifier, (295)–(297), and now reckoning with many events, their laughter may target many different professors: (294) DeHket alia w marwaan laughed.3fs Alia and Marwan DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics (295) DeHket kell bent w kell Sabi laughed.3fs every girl and every boy DeHket kell bent w kell Sabi ʕa esteez linguistics laughed.3fs every girl and every boy at professor linguistics (296) DeHket kell bent w marwaan laughed.3fs every girl and Marwan DeHket kell bent w marwaan ʕa esteez linguistics laughed.3fs every girl and marwaan at professor linguistics (297) DeHket alia w kell Sabi laughed.3fs Alia and every boy DeHket alia w kell Sabi ʕa esteez linguistics laughed.3fs alia and every boy at professor linguistics But if an element within the second conjunct covaries with the first, (298)–(300), a stronger comitative condition takes over, according to which the woman and child
136
Chapter 2
act together in some way and again their laughter is directed at the same professor. They laugh with each other at the very same thing. (298) DeHket kell mara w ebn-a laughed.3fs every woman and child-her DeHket kell mara w ebn-a ʕa estez linguistics laughed.3fs every woman and child-her at professor linguistics (299) DeHket kell mara w kell walad men wleed-a laughed.3fs every woman and every child from children-her DeHket kell mara w kell walad men wleed-a laughed.3fs every woman and every child from children-her ʕa esteez linguistics at professor linguistics (300) DeHket kell mara w walad men wleed-i laughed.3fs every woman and child from children-me DeHket kell mara w kell walad men wleed-i laughed.3fs every woman and every child from children-me ʕa esteez linguistics at professor linguistics (301) DeHek walad men wleed-i w kell mara laughed.3ms child from children-me and every woman DeHek walad men wleed-i w kell mara laughed.3ms child from children-me and every woman ʕa esteez linguistics at professor linguistics This observation takes in not only the covariance that the pronouns induce in (298) and (299) but also the apparent scope interactions in (300). That is, if it is intended that the same child of mine laughs no matter which girl is considered, then (300) is like (296)—neither spatiotemporal proximity nor comitativity is implied and each of the women and the child may be laughing at someone else. Once, however, it is intended that for every woman there may be a different child of mine, then comitativity and a shared target between the woman and the child kick in. Note that in (301) there is no possibility for the indefinite to covary with the distributive quantifier, as if it were in its scope, and thus (301) is only understood to be like (297), without the semantic conditions and with perhaps different professors targeted. Suspicious of tacit structure, you pray that a compound, distributive quantifier in (295)–(297) include within its scope a linguistics professor, evaluated with respect to the different agents and events distributed over. If answered, what prayer will then explain how the internal structure of this compound quantifier, should it
DP and DP
137
express a covariance, leads to such global effects on the interpretation of (298)– (290)? If the alleged compound quantifier includes within its scope a linguistics professor in (295)–(297), why not also in (298)–(300)? Why should one compound quantifier, but not the other, imply that the objects quantified over act together with respect to events that are later described by the predicate to which the compound quantifier is applied? This is the Lebanese Arabic version of the dilemma concluding chapter 1 (see section 1.5.1) and rehearsed in section 2.0.1. Rather, all the sentences (294)–(301) conjoin clauses with tacit Right-Node Raising. Any theory of Right-Node Raising suffices for reconstructing the RightNode Raised constituent into each conjunct to derive the dependent interpretation of a linguistics professor, and of course it does so without implying that one remnant of the conjunction, such as every woman, includes within its scope the other. On the contrary, since it represents a coordination of clauses, it requires some special pleading for the second of the conjoined DPs to be understood as if it were within the scope of the first. The covariance that sometimes occurs across clauses derives from the intervention of an adverbial phrase, “telescoping,” subatomic examples of which were argued to occur in (264)–(267). Here again the covariance in (298)–(300) derives from an intervening adverbial, imposing a structure that strikingly distinguishes these sentences from those in (295)–(297): (302) ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and whenever a woman W-ed, her child W-ed (at a linguistics professor)). ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and whenever a woman W-ed, every child of her children W-ed (at a linguistics professor)). ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and whenever a woman W-ed, a child of mine W-ed (at a linguistics professor)). In deploying an adverbial, the speaker intends some connection between what the woman does and what the child does strong enough for the former to locate the latter. With a causal or comitative relation between them, the woman provokes or is joined by the child. Every woman’s laughter is infectious to some particular child, so it seems.44 In any case, it is not surprising, and even necessary, that a speaker— because she uses an adverbial phrase in (298)–(300)—intends that there is more going on between woman and child than she would intend uttering (295)–(297) or (301) without an adverbial phrase and so without having in mind any generalization from what a woman does to how a child responds.45 The intervention of an adverbial phrase also explains an otherwise puzzling reversal in the semantic conditions contrasting (in)definite descriptions and distributive
138
Chapter 2
quantifiers. As observed earlier, plural agreement in (317) and singular agreement in (318) imply no particular spatiotemporal proximity or any other condition among the events described. (317) kenou alia w marwaan bi ouda were.3mp Alia and Marwan in room (318) ken kell esteez w alia bi ouda was.3ms every teacher and Alia in room kenet alia w kell esteez bi ouda was.3fs Alia and every teacher in room But plural agreement when a distributive quantifier is conjoined, as in (319), comes to imply pairwise consideration of Alia and each teacher such that Alia and the teacher were in a room acting together in some sense: (319) kenou kell esteez w alia bi ouda was.3mp every teacher and Alia in room kenou alia w kell esteez bi ouda was.3mp Alia and every teacher in room Plural agreement in this case becomes anomalous if there is no understanding of the togetherness implied, as in the report (321) of a blackout one sultry day on the Corniche:46 (320) ba’d noss se’a, kenet kell seHfet bouza w l-gato glace berak After half hour, was.3fs each bowl ice-cream and the-cake iced puddles ‘After a half hour, every bowl of ice cream and the ice cream cake were puddles.’ ba’d noss se’a, ken l-gato glace w kell seHfet bouza berak After half hour, was.3ms the-cake iced and each bowl ice-cream puddles ‘After a half hour, the ice cream cake and every bowl of ice cream were puddles.’ (321) #ba’d noss se’a, kenou kell seHfet bouza w l-gato glace berak After half hour, was.3mp each bowl ice-cream and the-cake iced puddles ‘After a half hour, every bowl of ice cream and the ice cream cake were puddles.’ #ba’d noss se’a, kenou l-gato glace w kell seHfet bouza berak After half hour, was.3mp the-cake iced and each bowl ice-cream puddles ‘After a half hour, the ice cream cake and every bowl of ice cream were puddles.’
DP and DP
139
Plural agreement itself denotes what some did, and is covariant with the universal quantifier in the first clause, referring to Alia’s and the teacher’s events in (319) or to those of the ice cream cake and the bowl of ice cream in (321). As we have seen, such covariance is achieved only with a “telescoping” adverbial phrase, which imposes the comitative condition that excludes (321). The reversal in the semantic implications of singular and plural agreement thus confirms that the comitative and eventive conditions emerge from the interaction of syntactic configuration with number agreement as a denoting expression. The discovery of a tacit adverbial phrase in (298)–(300), (319), and (321) explains the occasion for a comitative condition. Still unexplained is the restriction to the same target: Why can’t it be allowed that a woman’s laughter at one professor prompts or is accompanied by her child’s laughter at perhaps another professor? Despite some efforts, varying the example and the situation, neither the Choueiri sisters nor Sarah Ouwayda acquiesce to such a case. The restriction derives, I believe, from the logical form that spells out the tacit adverbial in (302). To begin, let ‘C(e′,e″)’ be the causal or comitative relation that is the basis for the adverbial modification.47 The content of the “telescoping” adverbial is fixed by the antecedent clause and thus the logical form for (298)–(300) will be some further expansion of (322), which also shows the partitive event quantifier characteristic of distributive quantification and the deletion site Δ of Right-Node Raising:48 (322) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) & Δδ and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x : her49 child(x)] … W(e″,x) … & (δ at(e″, a linguistics professor) δ) ) The point now is to show that no expansion of (323) will represent the distributive or “sloppy” interpretation that allows woman and child to target different professors. (323) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) & Δδ and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x : her child(x)] … Wi(e″, x) … & (δ [℩e″ : proi] at(e″, a linguistics professor) δ) ) Recall that a distributive interpretation obtains when the terms cross-referring to events contain a pronoun, whatever there was done, whose tokens are bound by different antecedents for there.50 The second conjunct should mean that whatever was done there, the child’s event, was at a linguistics professor, and the first conjunct should mean that whatever was done in the woman’s event was at a linguistics professor, too. Distinct events, what the woman did and what the child did, may have
140
Chapter 2
distinct targets. To this end, the event pronoun in (323) should at minimum take the form in (324) to refer to whatever is done there, which, as the adverbial dictates, was comitative to e′.51 (324) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) & Δδ, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x : her child(x)] … W(e″,x) … & (δ [℩e″ : C(e′,e″) & ∃y W(e″, y)] at(e″, a linguistics professor) δ) ) Turning next to the first conjunct, what Right-Node Raising has deleted is at the very least a comment on what the woman does (see section 2.1.3):52 (325) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′ : proi] Δδ[e′], and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x : her child(x)] … W(e″,x) … & (δ [℩e″ : C(e′,e″) & ∃y W(e″, y)] at(e″, a linguistics professor) δ) ) That ought to be to say that what the woman does is among the events that the Right-Node Raised phrase describes, which is exactly what the Eventish logical form says in (288)–(289) of section 2.1.3. Adapted for the simpler logical forms here, at the very least it amounts to: (326) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′ : proi] [∀e: e ≤ e′] [℩eδ : Δδ] e ≤ eδ, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x : her child(x)] … W(e″,x) … & (δ [℩e : C(e′,e″) & ∃y W(e″, y)] at(e″, a linguistics professor) δ) ) What Right-Node Raising deletes is an exact copy of the Right-Node Raised phrase, and so (326) cashes out as (327), with the hope that the two event pronouns may denote events with different linguistics professors: (327) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′ : proi] [∀e: e ≤ e′] [℩e″ : C(e′,e″) & ∃y W(e″, y) & at(e″, a linguistics professor)] e ≤ e″, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x : her child(x)] … W(e″,x) … & [℩e″ : C(e′,e″) & ∃y W(e″, y)] at(e″, a linguistics professor))
DP and DP
141
But (327), with this little structure, is defective and ends up contradictory. The first clause says that the woman’s action is part of the action comitative with it that is at a linguistics professor. If it is not already to end in contradiction with the first clause, it must be assumed that ‘C(e′,e″)’ is not irreflexive, to mean better that e′ is within the spatiotemporal neighborhood of itself and events comitative to it. So it must be, anyway, for (328) that none of its neighborhood relations are irreflexive in relating the events denoted by modifer to events denoted by the modified: (328) Laughing, every laugher laughs laughing. But the second clause goes on to say that an action comitative to the woman’s action is her child’s action and the action comitative to the woman’s action is directed at a linguistics professor, entailing the contradiction that the child is the agent and the woman, also the agent of the same action. What has gone wrong is that the second clause does not say, what it should, that the child’s action comitative to the woman’s action is at a linguistics professor. To sidestep the contradiction, the event pronoun prefixed to at a linguistics professor might be emended to mention the child— ‘[℩e″: C(e′,e″) & [℩x: her child(x)]W(e″, x)]’—but mentioning the child forgoes the “sloppy” interpretation of at a linguistics professor when the phrase is copied into the first clause, since both tokens now comment only on the child. If the definite description her child were as free as a distributive quantifier to take scope higher, parallel to every woman in the first clause, it could be suggested that the prefixed event pronoun denotes x’s comitative action—‘[℩e″: C(e′,e″) & [℩y: y = x] W(e″, y)]’—where with ‘x’ bound in the first clause by every woman, it would denote the woman’s action, and in the second clause, the child’s action, ‘x’ being bound by her child qua distributive quantifier. I will return to this suggestion when considering the coordination of distributive quantifiers proper, every woman … and … every child of hers. Here given the syntax of (in)definite description as it is, which precludes the “sloppy” interpretation of at a linguistics professor, as desired, it remains to find a logical form for (298) that represents the comitativity condition without contradiction. To represent (298)’s meaning, it should be recognized for an instance of collectivized Right-Node Raising with a plural event pronoun ‘[℩E″: pl.pro[E″]]’, which says of the actions in which woman and child participate that they were directed at a linguistics professor. But that engages the same one linguistics professor for them both, and thus the comitativity condition is correctly represented along with the entailment that woman and child laugh at the same professor:
142
Chapter 2
(329) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′: proi][∀e: e ≤ e′] [℩E″: C[e′,E″] & ∃Y W[E″, Y]] & at[E″, a linguistics professor]] e ≤ E″, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [℩x: her child(x)]j … W(e″,x) … & [℩E″: C[e′,E″] & ∃Y W[E″, Y]] at[E″, a linguistics professor])) The first clause now says that the woman’s action is part of those actions comitative with hers that are at a linguistics professor, and the second says that an action comitative with the woman’s action is her child’s action and the actions comitative with the woman’s action are at a linguistics professor. The covariation in (298) of her child with every woman has required adverbialization, imposing the comitativity condition. Then, if her child conforms to the syntax of (in)definite descriptions seen earlier, the only coherent interpretation entails, as desired, comitative action against the same one linguistics professor. The comitativity condition and the restriction to the same one linguistics professor observed in (298) are imposed whenever the DP of the second clause is covariant with that of the first, even if it is itself a distributive quantifier as in (299): (299) DeHket kell mara w kell walad men wleed-a laughed.3fs every woman and every child from children-her ʕa esteez linguistics at professor linguistics ‘Every woman and every child of hers laughed at a linguistics professor.’ Within their respective clauses, every woman and every child of hers have a scope that is parallel. The distributive quantification involves concurrent partitive event quantification in both conjuncts. In the second, it should relate to the events the ‘telescoped’ adverbial introduces if this adverbial like all others frames event quantification in the clause modified: (330) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′: proi][∀e: e ≤ e′] [℩e‴: proj & at(e‴, a linguistics professor)] e ≤ e‴, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [[every x : child of hers(x)] [∃e‴ : e‴ ≤ e″] (… Wi(e‴,x) … & [℩e‴: proj] at(e‴, a linguistics professor)))
DP and DP
143
The event pronoun ‘[℩e‴: proj]’ in the second conjunct ought to refer to the events e‴ that x participates in at an e″ comitative with the participation of the woman of e′: (331) [℩e‴: ∃e″(C(e′,e″) & e‴ ≤ e″) & [℩y: y = x] W(e‴, y)] Spelling out the interpretation of Right-Node Raising, copying its content from its antecedent, obtains (332) from (330): (332) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′: proi][∀e: e ≤ e′] [℩e‴: ∃e″(C(e′,e″) & e‴ ≤ e″) & [℩y: y = x] W(e‴, y) & at(e‴, a linguistics professor)] e ≤ e‴, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [[every x : child of hers(x)] [∃e‴ : e‴ ≤ e″] (… W(e‴,x) … & [℩e‴: ∃e″(C(e′,e″) & e‴ ≤ e″) & [℩y: y = x] W(e‴, y)] at(e‴, a linguistics professor))) The event pronoun prefixed to at a linguistics professor denotes the events comitative to the woman’s action that are x’s. With the event pronoun copied into the first clause, the first clause now says that the woman’s action is part of the action of hers comitative with it that is at a linguistics professor, and the second clause says that for every child, an action comitative to the woman’s action is her child’s action and the child’s action comitative to the woman’s action is also directed at a linguistics professor. There is no contradiction here, and comitativity holds between the woman’s and the child’s actions. But at the same time, it also represents at first glance the meaning for (299) it does not have: woman and child are allowed different linguistics professors to laugh at. The scent has been lost of a simpleminded explanation. Recall that the covariation derives from “telescoping,” the intervention of an adverbial phrase, so that logical forms for both (298) and (299) must depart from something like (333) ( = (302)): (333) ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and whenever a woman W-ed, her child W-ed (at a linguistics professor)). ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and whenever a woman W-ed, every child of her children W-ed (at a linguistics professor)).
144
Chapter 2
If the “telescoped” content itself copies from the antecedent clause including the alphabetic identity of the variable, (333) is (334): (334) ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and [whenever e′: e′ ≤ e & [a x : woman(x)] … W(e′,x) …, her child W-ed (at a linguistics professor)). ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and [whenever e′: e′ ≤ e & [a x : woman(x)] … W(e′,x) …, every child of her children W-ed (at a linguistics professor)). Adverbial modification next deploys some relation, ‘C(e′,e″)’, to relate the events the adverbial describes to those events the modified clause describes, for which the use of distinct variables is essential, *‘C(e′,e′)’. Thus (334) is in turn (335): (335) ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and [whenever e′: e′ ≤ e & [a x : woman(x)] … W(e′,x) …, [∃e″: C(e′,e″)] (her child W-ed (at a linguistics professor)[e″])). ∃e(Laugh(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … W(e′,x) …, and [whenever e′: e′ ≤ e & [a x : woman(x)] … W(e′,x) …, [∃e″: C(e′,e″)] (every child of her children W-ed (at a linguistics professor)[e″])). In the second clause, at a linguistics professor ought to be about the event e″ that that clause describes, what the child does. But then so is the copy of this Right-Node Raised phrase in the first clause (and also the copy “telescoped” in the adverbial), where it should rather be about e′, what the woman does. In short, the intervention of the relation ‘C(e′,e″)’ disrupts the parallel use of variables that the distributive, “sloppy” interpretation would require, and thus this interpretation is absent when covariance, as in (298) and (299), implies such a relation between the events described by the two conjuncts. An analogous effect is felt in English. In (336)–(338), Sam and Joe can be taken to work on different houses unless a temporal relation between their work intrudes, in which case they are understood to work on the same house: (336) a. Sam repainted and Joe re-sided a new house. b. Sam repainted and then Joe re-sided a new house. (337) a. What Sam repainted and Joe re-sided was a new house. b. What Sam repainted and then Joe re-sided was a new house. (338) a. What Sam repainted and what Joe re-sided was a new house. b. What Sam repainted and what Joe then re-sided was a new house. The dilemma according to the simpleminded account is that the canonical logical form for Right-Node Raising fits the template in (339) with first conjunct, second
DP and DP
145
conjunct, and Right-Node Raised phrase all about e′. And yet no logical form also squeezing in a relation between the conjuncts fits (see (340)): (339) … [∃e′: ξ[e′]] (Φ[e′] & Δδ[e′]) and … [∃e′: ζ[e′]](Ψ[e′] & Ωδ[e′]) (340) *… [∃e′: ξ[e′]] (iΦ[e′] & Δδ[e′]i) and [∀e′: proi][∃e″: C(e′,e″)](Ψ[e″] & Ωδ[e″]) *… [∃e′: ξ[e′]] (iΦ[e′] & Δδ[e′]i) and [∀e′: proi][∃e″: C(e′,e″)](Ψ[e″] & Ωδ[e′]) *… [∃e′: ξ[e′]] (iΦ[e′] & Δδ[e″]i) and [∀e′: proi][∃e″: C(e′,e″)](Ψ[e″] & Ωδ[e″]) There is something compelling in the simpleminded account, vivid in (336)–(338): the ‘sloppy’ interpretation of a new house does demand a parallelism between the two conjuncts, which a temporal relation between them gets in the way of. But no true account of this effect can in fact appeal to the template in (339) since it denies the possibility of collectivized Right-Node Raising, the point of which is not to say of both an event e′ of Φ-ing and an event e′ of Ψ-ing that it is an event e′ of Ω-ing. Collectivized, it rather says that those events are among the events E of Ω-ing, and everything about the syntax and semantics of Right-Node Raising has been contrived toward that end. The template from sections 2.1.3 and 2.1.4 is (341) where i and j are the indices of the event variable to be abstracted on in the indicated conjunct, and as before, Ωδ is the Right-Node Raised phrase and Δδ, its deletion in the first conjunct: (341) … Φi Δδ and … Ψj Ωδ ⇒ … Φi [℩Ei: Φi][∀ei: Eiei][℩Ei,j : Ω]Ei,jei and … Ψj [℩Ej: Ψj][∀ej: Ejej][℩Ei,j : Ω]Ei,jej In the citation examples of collectivized Right-Node Raising, where the conjuncts have been entirely parallel in structure, without any “telescoping” between them, the Right-Node Raised term ‘[℩Ei,i : Ω]’ has collected events described in antecedent clauses with the same variable ei. This is not always the case, and it cannot be a point of grammar to exclude terms ‘[℩Ei,j : Ω]’ with different indices i and j, since this is exactly what the acceptable and collectivized interpretations of (298), (299), and (336b)–(338b) do in saying that the events e′ and e″ of mother (e′) and child (e″) laughing are at a linguistics professor, or that the events e′ and e″ of Sam repainting (e′) and Joe re-siding (e″) are of a new house. It can, however, still be left a point of grammar that the variables exposed in the remnants of Right-Node Raising, ei in Φi and ej in Ψj, be among the variables abstracted on in the Right-Node Raised phrase ‘[℩Ei,j : Ω]’. But if so, any attempt at a “sloppy” interpretation for a linguistics professor or a new house, which is not collectivized, remains rather an attempt to say about each that what she alone did was at a linguistics professor or what he alone did was of a new house, relying on an event description in a single variable. Different variables ei and ej cannot both be that variable, excluding then the “sloppy” interpretation exactly when “telescoping” has forced the use of different variables,
146
Chapter 2
as desired. The simpleminded account survives in spirit in an account adapted to a logical language with collectivized Right-Node Raising as its raison d’être.53 The sentence (342) is just like (298): mother and child must read the same story although the stories that the mothers choose may differ. As before, the adverbial telescoping necessary to relate each woman to her child disrupts the parallel use of event variables that the distributive, “sloppy” interpretation of a story would require given that copied expressions use the same variables as their antecedents. (342) aryet kell mara w ebn-a essa read.3fs each woman and child-her story A similar account will explain the contrast between (343) and (344)–(345) (see note 33). Without the Right-Node Raising observed in (342), these appear to coordinate larger constituents: (343) kell mara aryet essa baʕdein ebn-a eri-a baʕd-a each woman read.3fs story (&) then child-her read.3ms-it after-her (344) *aryet kell mara essa baʕdein eri-a ebn-a baʕd-a read.3fs each woman story then (it)read.3ms child-her after-her (345) *aryet kell mara essa baʕdein eri-a Sami baʕd-a read.3fs each woman story then (it) read.3ms Sami after-her (346) aryet kell mara essa read.3fs each woman story (see (383)) (347) kell mara aryet essa each woman read.3fs story Despite distributive scope in (346) with a postverbal subject, “telescoping” and descriptive anaphora are acceptable only where, as in (343), the subject each woman is preverbal, occupying a superior position within its conjunct. Singular number agreement, if semantic, must fall within the scope of the subject in the logical form for (346) as well as the logical form for (347) (see note 31). In this respect (346) and (347) are the same despite appearances. Yet, if an account of the contrast between (343) and (344)–(345) is forthcoming, subject inversion must have some effect on logical form. Recall that number agreement, tense, and the verb all sponsor their own event quantifier as well. I take it that inversion fixes the relative scope of tense and the subject, while (346) and (347) are evidence that the location of number agreement is unaffected. The logical forms for (346) and (347) should be along the lines of (348) and (349), respectively: (348) [∃e : Past(e)](Read(e) & [Each x : Fx][∃e′ : e′ ≤ e](W(e′,x) & [℩e′: SG.W W(e′,x)] [∃y : Sy]of[e′,y]) ‘There was reading in which every woman was such that she read a story.’
DP and DP
147
(349) [Each x : Fx][∃e : Past(e)](W(e,x) & [℩e: SG.W W(e,x)](Read(e) & [∃y : Sy]of[e,y]) ‘Every woman was such that she reads a story.’ Note that in (348), the event quantifier ‘[∃e′ : e′ ≤ e]’ licensed by the distributive quantifier each woman ‘[Each x : Fx]’ quantifies over parts of the antecedently given event e. In contrast, in (349) the events of what each woman did are not related to a larger event.54 Subject inversion thus introduces in (348) a further relation between events and, along with it, a use of event variables that makes the “telescoping” and descriptive anaphora in (344)–(345) impossible, as we will now see. When the subject is preverbal as in (349), we find an example of the standard, molecular “telescoping” similar in all relevant respects to its English counterpart. The tacit adverbial that introduces the second clause inherits its content from the first: (350) [Each x : Fx][∃e : Past(e)](W(e,x) & [℩e′: SG.W W(e′,x)](Read(e) & [∃y : Sy]of[e,y]), and [∀e : [∃x : Fx](Past(e) & W(e,x) …)][∃e′ : C(e,e′)], [℩x : her child(x)][∃e″ : e″ ≤ e′ & Past(e′)] W(e″,x) & [℩e″: SG.W W(e″,x)] (Read(e″) & …). ‘Every woman was such that she reads a story, and whenever a woman read a story, that woman’s child read that story.’ The adverbial quantifier simply abstracts on the highest event variable used in the description of the individual events of what a woman did. Using its descriptive anaphora felicitously, the second conjunct says that in any of those events, the child of the woman in that event read the story in that event. To do the same for the inverted structure in (344)–(345) derives (351): (351) [∃e : Past(e)](Read(e) & [Each x : Fx][∃e′ : e′ ≤ e](W(e′,x) & [℩e′: SG.W W(e′,x)] [∃y : Sy]of[e,y]), and [∀e′ : [∃x : Fx](e′ ≤ e & W(e′,x) …)][∃e″ : C(e′,e″)], [∃e″ : Past(e″)](Read(e″) & [℩x : her child(x)]W(e″,x) & [℩e″: SG.W W(e″,x)] …). ‘There was reading in which every woman reads a story, and whenever within it a woman reads a story, that woman’s child reads that story.’ Observe, however, that the variable e (boldfaced and underlined) is stranded unbound in (351), and thus the only representation of the intended interpretation is ungrammatical, deriving the contrast between (343) and (344)–(345). One could
148
Chapter 2
try a freer hand at copying the adverbial’s content, perhaps closing off the free variable as in (352): (352) … and [∀e′ : [∃e : Past(e)][∃x : Fx]( e′ ≤ e & W(e′,x) …)][∃e″ : C(e′,e″)], … But this would entail that at any and all events in the past where a woman reads a story, her child reads it too, which is plainly not any meaning of (344)–(345). The events that are to be universally quantified over are just those that are part of some event in the past of women reading stories. In addition, there could very well be many past events of women reading stories in which their children do not. What defeats any logical form for (344)–(345) that attempts this interpretation is that ‘[∃e : Past(e)]’, remaining in the first conjunct, does not include the second in its scope, and the offending variable e in (351) cannot, I assume, itself be replaced by a descriptive anaphor. The section began with the puzzle of why a condition of spatiotemporal proximity present in (294) and suspended by the quantification in (296) should return as an even stronger comitative condition when a pronoun or quantifier covaries in the second conjunct as in (298) and (300): (294) DeHket alia w marwaan laughed.3fs Alia and Marwan DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics (296) DeHket kell bent w marwaan laughed.3fs every girl and Marwan DeHket kell bent w marwaan ʕa esteez linguistics laughed.3fs every girl and marwaan at professor linguistics (298) DeHket kell mara w ebn-a laughed.3fs every woman and child-her DeHket kell mara w ebn-a ʕa estez linguistics laughed.3fs every woman and child-her at professor linguistics (300) DeHket kell mara w walad men wleed-i laughed.3fs every woman and child from children-me DeHket kell mara w kell walad men wleed-i laughed.3fs every woman and every child from children-me ʕa esteez linguistics at professor linguistics
DP and DP
149
The answer was that such covariation can only be derived by “telescoping”—that is, by interpolating an adverbial quantifying over events described by the first conjunct. The comitative condition that then emerges is just that expressed by adverbial modification. Whenever there is covariation, we find the structure in (353), where ‘C(e,e′)’ expresses the relation between antecedent and consequent events. (353) … [∀x : NP][∃e : Φ]Ψ, and [whenever e : Φ & Ψ][∃e′ : C(e,e′)](… Γ[e′] …) Given this structure, the first conjunct describes events using one variable, e, and the second conjunct uses another, e′. The Right-Node Raised phrase must abstract on both variables, and if the Right-Node Raised phrase is itself an abstraction on only one variable, the variables of the antecedent conjuncts must be the same, which ‘telescoping’ in ‘C(e,e′)’ makes impossible. This observation delivers the explanation offered earlier for why mother and child in (298) and (299) cannot be understood to laugh at their own, perhaps different linguistics professors together. Finally, subject inversion is found to oppose the two structures in (354), but ‘telescoping’ in the adverbial structure in (354) strands a variable ‘e’, thereby excluding descriptive anaphora in (344)–(345) and explaining the contrast with their uninverted counterpart in (343): (354) [∃e : Φ][∀x : NP][∃e′ : Γ[e]] Ψ, and [whenever e′ : Γ[e] & Ψ][∃e″ : C(e′,e″)] … 2.3.2
Vagaries of number agreement amid the layers of event quantification
As remarked earlier, Basque and English make the same point with (312) and (313) respectively: (312) Lindak ardau eta Anderek esnea edaten dabez. Linda wine and Ander milk drink be (3PL) ‘Linda will drink wine and Ander milk.’ (Sjoblom 1980)
[Basque]
(313) a. Robin earlier today and Hillary yesterday were drinking more bordeaux (between them) than the region produces in a week. b. Robin by this morning and Hillary by last night have drunk more bordeaux (between them) than the region produces in a week. There is no phrase referring to Robin and Hillary and providing a plural antecedent for number agreement. The conjunction plainly coordinates clausal descriptions of events, Robin’s action of today and Hillary’s from yesterday, and the Right-Node Raised constituent measures the amount of bordeaux they consume collectively.
150
Chapter 2
2.3.2.0 Interaction with the linear order of subjects and adverbs
The same coordination with parenthetical intonation also supports singular agreement, although now the Right-Node Raised constituent can neither be overtly collective nor be so understood: (355) a. Robin (earlier today), and Hillary yesterday, was drinking more bordeaux (*between them) than the region produces in a week. b. Robin (as of this morning), and Hillary as of last night, has drunk more bordeaux (*between them) than the region produces in a week. Robin and Hillary must have each drunk more bordeaux than the region produces in a week. None of this is surprising given the semantics of Right-Node Raising and number agreement. What is curious, however, is that inverting the name and the adverb within either conjunct makes plural agreement unacceptable and also therefore the collective interpretation: (356) a. * (Earlier today) Robin (earlier today) and yesterday Hillary were drinking more bordeaux (between them) than the region produces in a week. b. * (As of this morning) Robin (as of this morning) and as of last night Hillary have drunk more bordeaux (between them) than the region produces in a week. (357) a. (Earlier today) Robin (earlier today), and yesterday Hillary, was drinking more bordeaux (*between them) than the region produces in a week. b. (As of this morning) Robin (as of this morning), and as of last night Hillary, has drunk more bordeaux (*between them) than the region produces in a week. (358) a. * Earlier today Robin and (yesterday) Hillary (yesterday) were drinking more bordeaux (between them) than the region produces in a week. b. * As of this morning Robin and (as of last night) Hillary (as of last night) have drunk more bordeaux (between them) than the region produces in a week. (359) a. Earlier today Robin, and (yesterday) Hillary (yesterday), was drinking more bordeaux (*between them) than the region produces in a week. b. As of this morning Robin, and (as of last night) Hillary (as of last night), has drunk more bordeaux (*between them) than the region produces in a week.
DP and DP
151
Why can Hillary yesterday but not yesterday Hillary serve as an antecedent to a collective description of events? Surely to think of their respective logical forms simply as ‘… W(e,h) & yesterday(e) …’ and ‘… yesterday(e) & W(e,h) …’ is a blind alley with little in it on which to hang the apparent difference. There must be more to the syntax, and I think the explanation for the difference in number agreement starts with the observation that Yesterday Hillary BE drinking is, in effect, a larger clause than Hillary yesterday BE drinking, with the initial adverb appearing in a higher projection. Recall that whatever event quantification is already given by the sentence every Ninja turtle ate a whole pizza, its modification by the frame adverbial in (360) introduces a yet higher event quantification that locates the turtles’ swallowing a pizza, each in a New York minute, throughout the course of a whole day: (360) In the course of a day, every Ninja turtle ate a whole pizza. [∃e: In(e, a day)][every x : Tx][∃e′ : Past(e′) & e′ ≤ e](Agent(e,x) & …)) If this modification amounts to a larger clause, we can appeal to its presence to account for why the initial adverb blocks plural number agreement in (356)–(360). The reach of number agreement is, after all, not unbounded. As in (362), both of the coordinated clauses have tense, and number agreement in the second clause cannot include events described by the first. (361) a. Yesterday Robin was, and today Hillary is, drinking more bordeaux (*between them) than the region produces in a week. b. As of last night Robin has, and as of this morning Hillary (also) has, drunk more bordeaux (*between them) than the region produces in a week. (362) a. *Yesterday Robin was, and today Hillary are, drinking more bordeaux (between them) than the region produces in a week. b. *As of last night Robin has, and as of this morning Hillary (also) have, drunk more bordeaux (between them) than the region produces in a week. Perhaps Yesterday Hillary BE drinking is also large enough to confine its number agreement and in this respect differs from Hillary yesterday BE drinking. Let us suppose that adverbial phrases are always quantificational, as shown in (360), and so is Tense as I have been assuming all along. The logical form (360) illustrates one assignment of scope, where the adverbial quantifier includes within its scope the event quantifier introduced by Tense. The logical form for (313) will, in effect, realize the remaining possibility that reverses their relative scope. Recall that number agreement is plural in (313) in virtue of Robin and Hillary participating in the same event, with the one participating earlier today and the other yesterday. Although I
152
Chapter 2
return shortly to consider more carefully the meaning of (313) and the like, this observation is enough to suggest the skeleton in (363), reversing the scope relation found between the adverbial quantifier and the event quantifier introduced by Tense. What is common to both (360) and (363) and, I should think, constant across any sequence of event quantifiers is that an outer quantifier restricts the events in the domain of the next one in its scope: (363) [∃e : Past(e)] ([℩x : Rx][∃e′: earlier today(e′) & e′ ≤ e]W(e′,x) … and … [℩x : Hx][∃e′: yesterday(e′) & e′ ≤ e]W(e′,x) …) Fleshing out (363) will support the plural agreement in (313) in that Robin and Hillary both participate in the larger event e. Note that the second conjunct of (363) translates Hillary yesterday BE drinking. Now to account for the failure of plural number agreement in (356) and (358), it must be that these sentences do not have a logical form according to which Robin and Hillary participate in the same event. That is, neither (363) nor (364) translates (356) and (358): (364) *[∃e : Past(e)] ( [∃e′: earlier today(e′) & e′ ≤ e][℩x : Rx]W(e′,x) … and … [∃e′: yesterday(e′) & e′ ≤ e] [℩x : Hx]W(e′,x) …) Instead, preposing the adverb in either conjunct—for example, Yesterday Hillary BE drinking—locates it (as in (360)) in a position superior to Tense within that conjunct. With an adverb preposed in either conjunct, both conjuncts must then contain their own Tense quantifiers, and thus no event is described where both Robin and Hillary participate: (265) [∃e′: earlier today(e′)][℩x : Rx][∃e : Past(e) & e ≤ e′]W(e,x) … and … [∃e′: yesterday(e′)] [℩x : Hx][∃e : Past(e) & e ≤ e′]W(e,x) … Translation as in (265) assimilates (356) and (358) to (362), where the events described by the first conjunct are plainly beyond the reach of number agreement in the second. 2.3.2.1 Molecular event quantifiers licensing distributivity
In offering this account, I assumed on the strength of (360) that adverbial modification introduces a new layer of event quantifiers.55 Reflecting on the meaning of coordinations like (313), we soon discover that it would make no sense to do otherwise. For, although Robin and Hillary can participate in the same drinking, as can Alia and Marwan in (266) and (267), it cannot be that the same events are both yesterday and today, as the discussion of supermonadicity in chapter 1 has already observed:
DP and DP
153
(266) šerbou alia w marwaan anninet pepsi drank.pl Alia and Marwan bottle Pepsi (267) šerbou alia mbeereH w marwaan lyom anninet pepsi drank.pl Alia yesterday and Marwan today bottle Pepsi The intended meaning of (313) is clear enough—some of what Robin did earlier today and some of what Hillary did yesterday amounted to their drinking more bordeaux between them than the region produces. It does not issue in the contradiction that some event was today and it was also yesterday, as would a logical form like (268): (268) *∃e(W(e,r) & today(e) and W(e,h) & yesterday(e) …) The paraphrase, if accurate, betrays further event quantifiers within the conjuncts. As further indication of the richer structure, recall from section 2.3.0 that the obligatory plural number agreement in (293)–(283) and concomitantly the absence of a “sloppy,” distributive interpretation in (293), (369), and (370) derives from the premise that clauses coordinating (in)definite descriptions, which do not license wide-scope event quantifiers on their own, end up redescribing the very same events. (293) keenou alia w marwaan bi ouda were.3mp Alia and Marwan in room kenou marwaanw alia bi ouda were.3mp Marwan and Alia in room (283) *keenet alia w marwaan bi ouda *was.3fs Alia and Marwan in room *ken marwaanw alia bi ouda was.3ms Marwan and Alia in room56 (369) šerbou alia w marwaan anninet pepsi drank.3mp Alia and Marwan bottle Pepsi (370) šerbet alia w marwaan anninet pepsi drank.3fs Alia and Marwan bottle Pepsi It has, however, just been observed in (313) and (267) that the intrusion of adverbs makes it impossible for the conjuncts to describe the very same event. So, with the intrusion of these adverbs, a “sloppy,” distributive interpretation is made available. Thus (267) is ambiguous, in contrast to (266), allowing the reading that gives to Alia and Marwan a bottle of Pepsi each. While the meaning of the adverb modifiers and their occasioning a distributive interpretation intimate richer structure, the facts of number agreement still seem to tell that the conjuncts describe the same events. Despite the adverbial modifiers,
154
Chapter 2
plural agreement remains obligatory when no comitative conditions are implicated, and in fact the comitative construction drops out altogether ((370) vs. (373)): (371) alia mbereH w marwaan Alia yesterday and Marwan (272) *alia mbereH w marwaan Alia yesterday and Marwan
lyom šerbou anninet pepsi today drank.3mp bottle Pepsi lyom šerbet anninet pepsi today drank.3fs bottle Pepsi
*alia mbereH w marwaan lyom šereb anninet pepsi Alia yesterday and Marwan today drank.3ms bottle Pepsi (373) *šerbet alia mbeereH w marwaan lyom anninet pepsi drank.3fs Alia yesterday and Marwan today bottle Pepsi In English too, absent parenthetical intonation, the plural agreement in (313), with postnominal adverbs, is obligatory and for what should be the same reason, namely, that Robin and Hillary participate in the same events, like Alia and Marwan. (374) a. *Robin (earlier today) and Hillary yesterday was drinking more bordeaux than the region produces in a week. b. *Robin (as of this morning) and Hillary as of last night has drunk more bordeaux than the region produces in a week. The postnominal modification may be extended indefinitely without affecting number agreement, which remains obligatorily plural when the subjects are (in)definite descriptions and the conjuncts lack both tense and parenthetical intonation: (375) Marvin this afternoon from Great Neck and Bernice this evening from Syosset are arriving at Leonard’s in a rented Mercedes. (376) *Marvin this afternoon from Great Neck and Bernice this evening from Syosset is arriving at Leonard’s in a rented Mercedes. Again the extended modifiers serve only to elaborate what each has done separately. The layer of event quantifiers that adverbs introduce avoids the near collision between a plural agreement describing the same event and adverbs speaking of different events and, in (367), possibly of different Pepsis. Treating the adverbs as clauses with their own event quantifiers, the proposal is to treat (367) akin to (377), with ‘Δi’ and italics indicating the deletion site and Right-Node Raised constituent: (367) šerbou alia lyom w marwaan mbeereH anninet pepsi drank.3mp Alia today and Marwan yesterday bottle Pepsi (377) drank.pl Alia [pro participating today] Δi, and Marwan [pro participating yesterday] a bottle of Pepsii
DP and DP
155
The phrase a bottle of Pepsi receives a distributive, “sloppy” interpretation when its tokens fall within the scope of the different adverbial event quantifiers, one at Δ within the scope of today and the other within the scope of yesterday. Number agreement, as it finds itself outside the scope of the modifiers, falls only within the scope of the main event quantifier, which the conjuncts share, forcing plural agreement. Consider the logical form (378):57 (378) [∃E: Past(E)]([℩E″ : [∀e: E″e]Ee & pl.proi,i] drink(E″) & [℩x : Ax][℩E′ : W(E′,x) & today(E′)][∃E″ : C(E′,E″)](Wi(E″,x) & Δ) and [℩x : Mx][℩E′ : W(E′,x) & yesterday(E′)][∃E″ : C(E′,E″)] Wi(E″,x)) & [℩E″ : pl.proi,i] [∃y : Py] Patient(E″,y))58 It says that there is some past event where whatever two or more do there is drinking, and Alia’s participation there is today and Marwan’s participation there is yesterday, and the drinking is of a bottle of Pepsi, which delivers the reading that reports a shared Pepsi. As with all collectivized Right-Node Raised constituents, the second pronoun, ‘[℩E″ : pl.proi,i],’ denotes collected actions, whatever there was whatever Alia participated in today and of whatever Marwan participated in yesterday. Whatever they did, they acted on a single bottle of Pepsi, and both the RightNode Raised constituent and its exact copy at Δ say as much and nothing more. Alternatively, a “sloppy,” distributive interpretation of the Right-Node Raised constituent relates it to the event quantifiers that the adverbs have introduced: (379) [∃E: Past(E)]([℩E″ : [∀e: E″e]Ee & pl.proi,i] drink(E″) & [℩x : Ax][℩E′ : W(E′,x) & today(E′)][∃E″ : C(E′,E″)](Wi(E″,x) & Δ) and [℩x : Mx][℩E′ : W(E′,x) & yesterday(E′)][∃E″ : C(E′,E″)] Wi(E″,x)) & [℩E″ : sg.proi] [∃y : Py] Patient(E″,y)) The token of ‘[℩E″ : sg.proi],’ that occurs in the second conjunct denotes what Marwan did yesterday but its copy in the first conjunct at Δ denotes what Alia did today. Thus the conjuncts come to make different assertions: that Alia drank a bottle of Pepsi today and Marwan drank a bottle of Pepsi yesterday.59 Either way, number agreement, ‘[℩E″ : pl.pro]’, remains plural since the higher events have two participants whether the constituent events are described according to (378) or according to (379). Thus simple plural or conjoined (in)definite descriptions require plural number agreement (except of course under the conditions for partial agreement). As for distributivity, the interest of Lebanese is the singular circumstance under which it is licensed—the contrast between (366) and (367). In general, (in)definite descriptions in postverbal positions, subject or object, never license of their own
156
Chapter 2
accord distributive interpretations. Thus (380)–(382) contrast with the overt expression of a distributive quantifier, as in (383)–(385): (380) šerbou t-telmizeen (mbeereH) anninet pepsi drank.pl the-students.dual (yesterday) bottle Pepsi (381) baʕtet liina alia w marwaan (mbeereH) ʕa ouda sent.3fs Lina Alia and Marwan (yesterday) to room liina baʕtet alia w marwaan (mbeereH) ʕa ouda Lina sent.3fs Alia and Marwan (yesterday) to room baʕtet liina telmiizeen (mbeereH) ʕa ouda sent.3fs Lina student.dual (yesterday) to room liina baʕtet telmiizeen (mbeereH) ʕa ouda Lina sent.3fs student.dual (yesterday) to room baʕtet liina t-tleemiiz (mbeereH) ʕa ouda sent.3fs Lina the-students (yesterday) to room liina baʕtet t-tleemiiz (mbeereH) ʕa ouda Lina sent.3fs the-students (yesterday) to room (382) ʕarrafet liina alia w marwaan ʕa esteez linguistics introduced.3fs Lina Alia and Marwan to professor linguistics liina ʕarrafet alia w marwaan ʕa esteez linguistics Lina introduced.3fs Alia and Marwan to professor linguistics ʕarrafet liina telmiizeen introduced.3fs Lina student.dual liina ʕarrafet telmiizeen Lina introduced.3fs student.dual
ʕa esteez linguistics to professor linguistics ʕa esteez linguistics to professor linguistics
ʕarrafet liina t-tleemiiz ʕa esteez linguistics introduced.3fs Lina the-students to professor linguistics liina ʕarrafet t-tleemiiz ʕa esteez linguistics Lina introduced.3fs the-students to professor linguistics (383) šereb kell telmiiz (mbeereH) anniinet pepsi drank.3ms every student (yesterday) bottle Pepsi (384) baʕtet liina kell telmiiz (mbeereH) ʕa ouda sent.3fs Lina every student (yesterday) to room liina baʕtet kell telmiiz (mbeereH) ʕa ouda Lina sent.3fs every student (yesterday) to room
DP and DP
157
(385) ʕarrafet liina kell telmiiz (mbeereH) ʕa esteez linguistics introduced.3fs Lina every student (yesterday) to professor linguistics liina ʕarrafet kell telmiiz (mbeereH) ʕa esteez linguistics Lina introduced.3fs every student (yesterday) to professor linguistics In this paradigm with plural number agreement throughout, (366), (367), and (381), distributivity appears only where conjuncts contain adverbs. Tense and adverbs introduce event quantifiers, but (in)definite descriptions do not. Number agreement and distributivity directly reflect the distribution of event quantifiers. In the absence of tense within each conjunct, the conjuncts share the same events, and thus agreement is plural, the participants being more than one. And, absent tense within each conjunct, only when adverbs intrude does further quantification come to describe constituent events that are separate, solitary actions, within the scope of which tokens of the Right-Node Raised phrase receive different evaluations. Basque (312) presents similar circumstances. Linda alone drinks the wine, and Ander the milk. Thus there is an event in which Linda is the exhaustive agent and wine the exhaustive patient, and another event in which Ander and milk fill these roles. (312) Lindak ardau eta Anderek esnea edaten dabez. Linda wine and Ander milk drink be (3PL) ‘Linda will drink wine and Ander milk.’ (Sjoblom 1980)
[Basque]
In this case too, then, existential event quantifiers occur within each conjunct, along the lines of (386), and plural number agreement will arise as it does in English:60 (386) … ∃e([℩x : Lx]Agent(e,x) & ∃e′(C(e,e′) & [℩x : Wx]Patient(e′,x))) and ∃e([℩x : Ax]Agent(e,x) & ∃e′(C(e,e′) & [℩x : Mx]Patient(e′,x))) … Although an (in)definite description will not license an event quantifier on its own, a tense or an adverb does, and now (312) and (386) attest that one is also introduced in a largish, small clause where more than one thematic relation is made to cohere into an event description. 2.3.2.2 Subatomic event quantifiers: Plural reference to events without phrasal distributivity
It can be shown further that the introduction of event quantifiers generalizes to all small clauses, including those that consist of only a single thematic relation. To this end, consider the following sentences: (387) keenou l-snoubraat w l-sendyeeneet xfeef bi mantʕa were.3mp the-pines and the-oaks sparse in region Hadd mantaʕa bašariyyeh near region human ‘The pines and the oaks were sparse in a region near human settlement.’
158
Chapter 2
(388) keenou l-snoubraat men 5000 seneh w l-sendyeeneet were.3mp the-pines from 5000 year and the-oaks men 1000 seneh xfeef bi mantʕa Hadd mantaʕa bašariyyeh from 1000 year sparsein region near region human ‘The pines 5000 years ago and the oaks 1000 years ago were sparse in a region near human settlement.’ (389) ghaTTou l-snoubraat w l-sendyeeneet mantʕa covered.3mp the-pines and the-oaks region Hadd mantaʕa bašariyyeh near region human ‘The pines and the oaks covered a region near human settlement.’ (390) ghaTTou l-snoubraat men 5000 seneh w l-sendyeeneet men covered.3mp the-pines from 5000 years and the-oaks from 1000 seneh mantʕa Hadd mantaʕa bašariyyeh 1000 years region near region human ‘The pines 5000 years ago and the oaks 1000 years ago covered a region near human settlement.’ As before, the pines can be understood to occupy a region distinct from the one with the oaks when the adverbs are present ((388) and (390)) but not in (387) or (389) absent the adverbs; and so, as before, under no analysis of (387) or (389) should tokens of “(in) a region near human settlement” be applied to events that are different from one conjunct to the next. The interest of these examples lies in a residual ambiguity and its lack of interaction with the facts just noted. The predicate, sparse or cover, can be taken to apply either to the trees as one piney, oaken woods or to the pines and the oaks separately, as it would under Conjunction Reduction. For (387), one can imagine a context where the woods are dense enough so that the sentence is true only when the predicate sparse is applied separately, and yet even then (387) can only be taken to describe a single region near human settlement, containing both the pines and the oaks. The pines and the oaks occupy different states of being sparse, and the sentence can be understood to assert as much, but every other feature described must, by force of syntax, be applied to the collected states: (391) [∃e : Past(e)]([℩E : pro]Be(e) & [℩X : … Px …][∃e′ : e′ ≤ e]W[e′,X] & Δ and ([℩X : … Ox …][∃e′ : e′ ≤ e]W[e′,X] & [℩E : pl.pro](sparse[E]61 & [∃y : Ry] … In(E,y) …))) The solitary thematic relation ‘W[e′,X]’ is sufficient license to introduce an event quantifier within each conjunct,62 but—and this is crucial—because it is licensed by
DP and DP
159
the thematic relation, it remains local to it. Both tokens of ‘[∃e′ : e′ ≤ e]’ in (391) include within their scope only the thematic relation. Thus the reconstruction of the Right-Node Raised constituent Δ does not apply it to distinct events. Analogous to what was seen earlier with (366), the same events cannot be uniquely located in distinct regions and it is therefore understood that a single region is under discussion. The event quantifiers that the thematic role licenses differ in their placement from those introduced by adverbs modifying clauses, as in (367), (388), and (389), and from those that distributive quantifiers license to include within their scope the entire scope of the distributive quantifier. In contrast, these event quantifiers do include Δ within their scope, reflecting the syntactic position of the elements that license them, and afford an alternative evaluation of the reconstructed Right-Node Raised constituent.63 2.3.2.3 Pre- and postverbal subjects, the distribution of tacit each in Lebanese and Slavic, and further evidence of Conjunction Reduction
Recall now that the discussion of distributivity and the location of event quantifiers has largely concerned sentences with postverbal subjects. With preverbal subjects in Lebanese, distributivity shakes off its shackles—to each his own Pepsi—with or without adverbs and with or without coordination, plural number agreement remaining obligatory: (392) t-telmiizeen šerbou (mbeereH) anninet pepsi the-student.dual drank.pl (yesterday) bottle Pepsi (393) alia w marwaan šerbou (mbeereH) anninet pepsi Alia and Marwan drank.pl (yesterday) bottle Pepsi Likewise, both the pines and the oaks may have a region of their own when the subject is preverbal: (394) l-snoubrat w l-sendyeeneet keenou xfeef bi mantaʕ Hadd the-pines and the-oaks were.pl sparse in region near mantaʕa bašariyyeh region human (395) l-snoubrat w l-sendyeeneet ghaTTou mantaʕa Hadd mantaʕa bašariyyeh the-pines and the-oaks covered.pl region near region human Commenting on the English counterparts to (392) and (393), Hoeksema (1983, 1988) and Lasersohn (1995) conclude from the invariant number agreement that the analysis of the (in)definite descriptions is likewise invariant. Distributivity instead derives from some further modification of VP, in effect, from an occurrence of tacit each:64
160
Chapter 2
(396) t-telmiizeen šerbou (mbeereH) kell waaHad anninet pepsi student.dual drank.pl (yesterday) each bottle Pepsi ‘The two students drank.pl (yesterday) each a bottle of Pepsi.’ (397) alia w marwaan šerbou (mbeereH) kell waaHad anninet pepsi Alia and Marwaan drank.pl (yesterday) each bottle Pepsi ‘Alia and Marwan drank.pl (yesterday) each a bottle of Pepsi.’ It follows from the above that preverbal coordinations containing adverbs, such as (398) and (399), present alternative but equivalent means for deriving a distributive interpretation—either by different evaluations of the Right-Node Raised constituent’s tokens as in (377), or by an occurrence of tacit each (‘[∀x : [∃e′ :E′e′]W(e′,x))]’) in (400)): (398) Alia yesterday and Marwan today drank a bottle of Pepsi. (399) Marvin this afternoon from Great Neck and Bernice this evening from Syosset are arriving at Leonard’s in a rented Mercedes. (400) [℩x : Mx][℩E : afternoon(E) & from(E,g)][∃e′ : ∃e(Ee & C(e,e′))](Wi(e′,x) & Δ) and [℩x : Bx] [℩E : evening(E) & from(E,s)] [∃e′ : ∃e(Ee & C(e,e′))](Wi(e′,x) & [℩E′ : proi,i.pl[E′]]](arrive(E′) & [∀x : [∃e′ :E′e′]W(e′,x))][∃e′ :E′e′](W(e′,x) & at(e′,l) & in(e′,m)))) Inside VP, Lebanese and English part ways. Lebanese excludes tacit each from any position that can be found within the scope of a postverbal argument. Given the invariance of the (in)definite descriptions themselves, I take the failure of distributivity in (380)–(382) to reflect the ungrammaticality of its only possible sources: (401) *šerbou t-telmiizeen (mbeereH) kell waaHad anninet pepsi *drank.pl the-student.dual (yesterday) each bottle Pepsi (402) *baʕtet lina alia w marwaan (mbeereH) kell waHad ʕa ouda *sent.3fs Lina Alia and Marwan (yesterday) each to room *lina baʕtet alia w marwaan (mbeereH) kell waHad ʕa ouda *Lina sent.3fs Alia and Marwan (yesterday) each to room *baʕtet lina telmiizeen (mbeereH) kell waHad ʕa ouda *sent.3fs Lina students.dual (yesterday) each to room (mbeereH) kell waHad ʕa ouda *lina baʕtet telmiizeen *Lina sent.3fs students.dual (yesterday) each to room *baʕtet lina t-tleemiiz (mbeereH) kell waHad ʕa ouda *sent.3fs Lina the-student.pl (yesterday) each to room *lina baʕtet t-tleemiiz (mbeereH) kell waHad ʕa ouda *Lina sent.3fs the-student.pl (yesterday) each to room
DP and DP
161
(403) *ʕarrafet lina alia w marwaan kell waHad ʕa esteez linguistics *introduced.3fs Lina Alia and Marwan each to professor linguistics *lina ʕarrafet alia w marwaan kell waHad ʕa esteez linguistics *Lina *introduced.3fs Alia and Marwan each to professor linguistics *ʕarrafet lina telmiizeen kell waHad ʕa esteez linguistics *introduced.3fs Lina student.dual each to professor linguistics *lina ʕarrafet telmiizeen kell waHad ʕa esteez linguistics *Lina introduced.3fs student.dual each to professor linguistics *ʕarrafet lina t-tleemiiz kell waHad ʕa esteez linguistics *introduced.3fs Lina the-student.pl each to professor linguistics *lina ʕarrafet t-tleemiiz kell waHad ʕa esteez linguistics *Lina introduced.3fs the-student.pl each to professor linguistics Lebanese confines tacit each to a position above those in which postverbal nominals surface. Thus, only by raising its antecedent will tacit each fall within its scope. For (396), I have in mind a derivation along the following lines: (404) Tense—each—the two students—drink ⇒ the two studentsj—drinki+tense —each—tj ti In contrast, the distributive interpretations of English (405) and (406) indicate either a freer distribution of tacit each within the VP or more extensive movements of all complements to positions that include tokens of tacit each in their scope.65 (405) Lina sent Alia and Marwan to a room. Lina sent two students to a room. Lina sent the students to a room. (406) Lina introduced Alia and Marwan to a linguistics professor. Lina introduced two students to a linguistics professor. Lina introduced the students to a linguistics professor. In yet another family of languages, Slavic, tacit each seems to be proscribed entirely. Dalrymple, Hayrapetian, and King (1998a) report that neither simple plurals nor comitative plurals support a distributive reading, which requires an overt marker po as in the (b) examples: (407) a. mužčiny vyigrali $100. men.nom won $100 b. mužčiny vyigrali po $100. (408) a. mužčiny s ženščinami vyigrali $100. men.nom with women.inst won $100 b. mužčiny s ženščinami vyigrali po $100.
162
Chapter 2
The absence of tacit each does not, however, impede an understanding that applies the predicate to each conjunct separately: (409) mužčiny i ženščiny vyigrali $100. men.nom and women.nom won $100 (410) Boris i Natasha vyigrali $100. Thus the men are taken to have won $100 altogether and the women too, and similarly Boris won $100, as did Natasha. It should be noted that there is no interpretation for (409) according to which each man or woman won $100. Plainly, Conjunction Reduction is indicated in order to distinguish the coordination from other phrases that appear to refer to the same participants and, in the absence of distributors, to derive what distributivity there is. The reduction analysis is subject to the following two constraints: that the predicate be applied to each conjunct separately without appeal to a tacit operator such as each—at least for Slavic—and that plural number agreement remain obligatory, as in (410), even when the predicate is understood to apply to each conjunct separately. To the left of the verb, it must be that an (in)definite description finds itself in a position from which it now includes within its scope an event quantifier with scope over the entire VP. But, in contrast to a distributive quantifier (see section 2.3.0 and especially note 43), the (in)definite description is not high enough for number agreement to be captured by an event whose sole participants are given by that (in)definite description. Crucially, the translations of number agreement, ‘[℩E′ : …pl.proi,i]’ in (411) and its copy inside Δ, are evaluated with respect to the same events introduced by ‘[∃E : Past(E)]’: 66 (411) [∃E : Past(E)]( [℩x : Bx] [∃E′ : [∀e: E′e]Ee](Wi(E′,x) & Δ) and [℩x : Nx] [℩E′ : [∀e: E′e]Ee & pl.proi,i] win(E′) & [∃E′ : [∀e: E′e]Ee] (Wi(E′,x) & [℩E′ : sg.proi] [∃Y: $100(Y)] Theme(E′,Y))) Because number agreement and the cross-referring pronouns that do not surface as agreement are themselves definite descriptions of events, the distribution of event quantifiers throughout a clause, unsurprisingly, affects their interpretation and reference. Adverbs are event quantifiers as is Tense, and small clauses license event quantifiers locally as do distributive quantifiers. This section began with what appears to be a locality condition on fixing the content of number agreement: in Today Robin and yesterday Hillary was/*were …, the coordinated clauses are too large for number agreement in the second conjunct to denote events described by the first, and therefore only singular number agreement surfaces, in contrast to Robin today and Hillary yesterday were. … Next we observed some respects in which the placement
DP and DP
163
of event quantifiers licenses the “sloppy” interpretation of Right-Node Raised constituents, an effect of Conjunction Reduction putting tokens of the Right-Node Raised constituent within the scope of different event quantifiers. Although the adverbial event quantifiers occur properly contained within the conjoined phrases in (367), Conjunction Reduction puts tokens of a bottle of Pepsi within their scope, and thus the sentence acquires a distributive interpretation that is otherwise not available to (366). (366) šerbou alia w marwaan anninet pepsi drank.pl Alia and Marwan bottle Pepsi (367) šerbou alia mbeereH w marwaan lyom anninet pepsi drank.pl Alia yesterday and Marwan today bottle Pepsi Adverbs within the conjuncts prompt alternative evaluations of the Right-Node Raised constituent, but number agreement itself answers only to a higher event quantifier shared by both conjuncts. It remains obligatorily plural, denoting what both Alia and Marwan did. Tacit distributive operators, such as each or both, threaten to obscure the interaction between Conjunction Reduction and event quantifiers. In Lebanese and Slavic, we are fortunate to find languages that quiet this threat. Slavic appears to have no such tacit operators and Lebanese to allow none to occur in postverbal positions. Absent such operators, it seems to me that (387)–(388) and (409)–(410) can tell us only one thing—the distribution of event quantifiers fixing their meanings: (387) keenou s-snoubraat Δ w l-sendyeeneet xfeef bi mantaʕa Hadd mantaʕa were.pl the-pines Δ and the-oaks sparse in region near region bašariyyeh human were.pl the pines Δ and the oaks sparse in a region near human settlement (388) keenou s-snoubraat Δ men 5000 seneh w l-sendyeeneet men 1000 were.pl the-pines Δ from 5000 year and the-oaks from 1000 seneh xfeef bi mantaʕa Hadd mantaʕa bašariyyeh year sparse in region near region human were.pl the pines 5000 years ago Δ and the oaks 1000 years ago sparse in a region near human settlement (409) mužčiny i ženščiny vyigrali $100. men.nom Δ and women.nom won.pl $100 (410) Boris i Natasha vyigrali $100. Boris.nom Δ and Natasha.nom won.pl $100 In (387), both tokens of “in a region near human settlement,” the one Right-Node Raised and the other reconstructed at Δ, fall within the scope of the same event
164
Chapter 2
quantifiers, and thus the pines and the oaks must be in the same region. The adverbs in (388) do however provide distinct event quantifiers for independently evaluating the tokens of the Right-Node Raised phrase that fall within their scope. Along another dimension of contrast with (387), the preverbal subjects in (409)–(410) (and in (395) above) find themselves in a position so high that the Right-Node Raised constituent itself contains an event quantifier, as in (411), the two tokens of which again allow independent evaluation of “win $100” (and “in a region near human settlement” in (395)). 2.4
Coordination and subordination in Davidsonian logical form
Despite the plural agreement in both (409)/(410) and (408)/(412) (see section 2.6), the failure of the latter to support a distributive interpretation exposes a difference between coordination and subordination: (408) mužčiny s ženščinami vyigrali $100. men.nom with women.inst won.pl $100 (412) Boris s Natashej vyigrali $100. Boris.nom with Natasha.inst won.pl $100 Whether the comitative phrase is an adverbial clause or merely expresses another thematic relation (or, as in McNally 1993 and Dalrymple, Hayrapetian, and King 1998a, forms a complex DP with the subject), the subordinate phrase does not implicate Right-Node Raising and the reconstruction of some Δ. The presence of Δ, distinguishing coordination in “Boris Δ i Natasha …” from subordination in “Boris s Natashej …,” is felt twice here. First, as noted and exactly as the proposed distribution of Δ implies, the (in)definite descriptions in (408)/(412) do not distribute without an overt distributor po, and the distributive interpretations in (409)–(410) are only those that the occurrence of Δ predicts—those that Conjunction Reduction rather than plural distributivity would derive—thereby excluding the individual men and women of (408) from each winning $100. The second observation that reveals the presence of Δ concerns conditions on anaphora. As McNally (1993) notes, a possessive reflexive can relate a comitative phrase to a subject ((414b), (416b)), in contrast to an apparent coordination of DPs, (413b), (415b) (see also note 11) (thanks to Denis Paperno for corrections to typographical errors): (413) a. povar i ego pomoščniki prigotovili obed. chef.nom and his assistants.nom prepared.pl dinner b. *povar i svoj pomoščniki prigotovili obed chef.nom and self’s assistants.nom prepared.pl dinner
DP and DP
165
(414) a. povar s ego pomoščnikami prigotovili obed. chef.nom with his assistants.instr prepared.pl dinner b. povar so svoim pomoščnikami prigotovili obed. chef.nom and self’s assistants.instr prepared.pl dinner (415) a. Maša i eë sestra ušli. Masha.nom and her sister.nom left.pl b. *Maša i svoja sestra ušli. Masha.nom and self’s sister.nom left.pl (416) a. Maša s eë sestroj videli film. Masha.nom with her sister.instr saw.pl film b. Maša so svoej sestroj videli film. Masha.nom with self’s sister.instr saw.pl film The coordination of DPs conceals a clause ‘DP Δ’, which prevents the DP from serving as the antecedent to the reflexive. When a possessive pronoun shows quantificational covariance, it has been shown in section 2.3.1 to be an effect of subatomic “telescoping”, the content of which Δ provides. Absent Δ, a reflexive pronoun finds its antecedent among superior arguments in just the way that any reflexive embedded in a subordinate phrase would. In another setting, this presumed contrast between coordination and subordination would be uncontroversial, but some further rationalization is demanded now that nearly all phrasal composition is conjunction according to the Davidsonian analysis. After all, the difference between coordination in (410) and subordination in (412) is alleged to be just that schematized in (417) and (418) (cf. (411)): (417) … ([℩x: Boris(x)] Δ) (and ([℩x: Natasha(x)] … W[e,x] …)) (418) … [℩x: Boris(x)]W[e, x] & ([℩x: Natasha(x)] … with[e,x] …) The contrast between coordination and subordination can be recast as one between, say, molecular coordination in (417) and subatomic coordination in (418). Failing at deeper explanation, it would be no worse than the traditional stipulations—that Right-Node Raising occurs only with coordination, and that only subordination produces the configuration (c-command) required for reflexivization. To do better requires some account of why Δ is obligatory in (417),67 why only with it present does that first small-clause conjunct disrupt reflexivization (cf. (414b) and (416b) parsed as (418)), and why a stray Δ does not intrude in the first conjunct of (418) to license a distributive interpretation for (408a). Since and means “&,” a point of grammar rather than meaning singles out this lexical item. Meaning alone demands some proposition or other, and it will not explain why the very small first clause of (418) (consisting only of the thematic relation) will not suffice for and in (417) without further reconstruction of the Right-Node Raised constituent.
166
Chapter 2
There is no semantic explanation for why ‘[∃e : Past(e)][[℩x: Boris(x)]W[e, x]’ is not a sentence with the meaning that Boris was in some event. Rather, bound morphemes such as Past and W are stipulated to occur only in construction with other phrases, their complements. Let us suppose then that W must occur in construction with a complement Δ, ‘W & Δ’.68 The point then about and is that for some reason it is opaque to complementation, *(W (and Δ)), compelling the reconstruction of the thematic relation’s complement within Δ in (417). Note that if opaque to complementation, any apparent coordination of verbs, for example, as in (419), must instead be parsed as in (420) with complements occurring within both conjuncts, a conclusion that falls in with the rather different theoretical and empirical considerations that the syntax literature has mustered in support of (420) against (421) (see Kayne 1994; Wilder 1994, 1995a, 1995b, 1997, 1999; Zoerner 1995). (419) (420) (421) (422)
The Strand Bookshop bought and sold used books. The Strand Bookshop ((bought Δ) (and (sold used books))) *The Strand Bookshop ((bought and sold) used books). *The Strand Bookshop (bought (and (sold used books)))
If something along these lines is on track, the grammar of complementation provides the account for why the first conjunct in (417) requires more than the isolated thematic relation of (418) and so reconstructs the Right-Node Raised constituent, with the result that (417) is the only analysis of “Boris i Natasha …,” as desired. With “Boris s Natashej …,” the worry is that a stray reconstruction of Δ into the first conjunct of its logical form (418) should not license a distributive (i.e., “sloppy”) interpretation. Recall that the preverbal subjects in (409)–(410) (and in (394) above) find themselves in a position so high that the Right-Node Raised constituent itself contains an event quantifier that includes within its scope the entire VP, ‘[∃e′ : …] (W[e′,x]…)’ in (423) (see (411)69), which, when copied at Δ, allows Boris and Natasha to each win their own $100. (423) … [℩x : Boris(x)] Δ and [℩x : Natasha(x)] … [∃e′ : …](W[e′,x] …) If the preverbal comitative phrase were high enough to trigger number agreement, also as high as the coordinated DPs, a stray Δ might be equally expected to license a distributive interpretation: (424) … [℩x: Boris(x)]W[e, x] Δ & ([℩x: Natasha(x)] … with[e,x] … [∃e′ : …](W[e′,x] …)) Yet, to the extent that (418) just is the logical form of complementation in (263), complete and fully interpreted, a minimalist constraint against unmotivated copying would suffice against Δ in (424). My worry, however, is perhaps too general to be
DP and DP
167
addressed only by constraints on movement and copying since it takes in cases where movement or copying appear not to be at issue: (425) a. Boris with Natasha and with Rocky won $100. b. #Boris with Natasha and with Natasha won $100. (426) *Boris with Natasha with Rocky won $100. (427) Boris and Natasha have won $100. (428) *Boris Natasha have won $100. It will of course be said that (426) in contrast to (425) is incoherent, violating a Principle of Full Interpretation for which it is assumed that with Natasha is interpreted in this sentence if and only if with Rocky is not. But in a Davidsonian setting, this assumption begs the question to the extent that (425a) and (426) are taken to say the same thing. Similarly, the incoherence of (428) is not transparent if it is taken to mean the same thing as (427). A difference of meaning is, however, provided. Although coordination and complementation both involve logical conjunction, they differ in the arrangement of the event quantifiers holding the sentence together. So recall that (427) has the logical form (411) when asserting that each has won $10070 and the logical form (429), swapping ‘[℩E′: pl.proi,i]’ for ‘[℩E′: sg.proi]’ when a collective attribution is asserted:71 (411) (∃E)([℩x : Bx] [∃E′ : [∀e: E′e]Ee](Wi(E′,x) & Δ) and [℩x : Nx](Past(E) & [℩E′ : [∀e: E′e]Ee & pl.proi,i] win(E′) & [∃E′ : [∀e: E′e]Ee] (Wi(E′,x) & [℩E′ : sg.proi] [∃Y: $100(Y)] Theme(E′,Y))) (429) (∃E)([℩x : Bx] [∃E′ : [∀e: E′e]Ee](Wi(E′,x) & Δ) and [℩x : Nx](Past(E) & [℩E′ : [∀e: E′e]Ee & pl.proi,i] win(E′) & [∃E′ : [∀e: E′e]Ee] (Wi(E′,x) & [℩E′ : pl.proi,i] [∃Y: $100(Y)] Theme(E′,Y))) Only distributive quantifiers, adverbial phrases, and tensed clauses introduce (molecular72) existential or partitive event quantifiers. All other event quantifiers, number agreement, cross-reference, and so on, are (second-order) definite descriptions. Thus a simple sentence without adverbs and with only (in)definite descriptions can count on only a single existential event quantifier.73 Preverbal and postverbal subjects flank this quantifier, showing a scope interaction the effects of which have been noted throughout. In (411) and (429), the subject, tense morpheme, number agreement, and the raised verb line up before the existential event quantifier, and so it is at Δ too.
168
Chapter 2
Now, to rule it out, (428) should be denied logical forms that are identical to (411) and (429) but for the substitution of & for and. Given that any syntactic structure for (428) will agree with the parse (Boris (Natasha (have won $100))), what subordination indicates is that the subordinate phrase, with all of its event quantifiers, falls entirely within the scope of the event quantifiers introduced with the superior phrase. In particular, although (411) and (429) instantiate the schema (430), the logical form for (428) is at best an instance of (432) and not of (431): (430) [℩x : Bx][Q1 e1 : Φ[e1]] … [Qn en : Φ[en]]Ψ and [℩x : Nx]Δ (431) *[℩x : Bx][Q1 e1 : Φ[e1]] … [Qn en : Φ[en]]Ψ & [℩x : Nx]Δ (432) [℩x : Bx][Q1 e1 : Φ[e1]] … [Qn en : Φ[en]](Ψ & [℩x : Nx]Δ) This aspect of subordination is sufficient to doom any interpretation for (428). Across any sequence of event quantifiers, an outer quantifier restricts the events in the domain of the next one in its scope.74 That is, a logical form for (428) will say, in effect, that (433) Boris participated (‘W’), and in some of this Natasha participated (‘W’), and that was winning $100. Recall from chapter 1 that the basic device of cross-reference to events, from one phrase to another, is not a naked event variable (although I have frequently abbreviated it as such) but a pronominal, definite description of events, pro, which, to be exhaustive, refers exactly to whatever α and only α did. If so, the pronominal this in (433) refers to what Boris and only Boris participated in, and it is analytically contradictory to say of any of this that Natasha participated there as well.75 This characterization of (428)’s flaws can be strengthened if it is further supposed that the phrase structure of a simple clause stipulates that there is no more than a single token of a thematic relation such as ‘Agent’, each of which projects no more than one token of ‘W’: W-Agent-V, * W-W-Agent-V. If so, in (433) and in any other candidate logical form for (428), one token of ‘W’ must be the result of copying the other. If also, as suggested earlier, event domain restrictions are explicit, then distinct variables are required to express the relation between what Boris did and what Natasha did, so that (433) translates along the lines of (434) or (435): (434) [℩x : Bx] … W[e,x] … & [∃e′ : [this℩E : [℩x : Bx]W[e,x]] Ee′] … [℩x : Nx]W[e′,x] & [that℩E′ : …] was(e′) winning $100. (435) [℩x : Bx] … W[e,x] … & [∃e′ : [this℩E : [ιx : Bx]W[e,x]] e′ ≤ e] … [℩x : Nx]W[e′,x] & [that℩E′ : …] was(e′) winning $100. In (434) and (435), the pronoun copies its content from its antecedent clause and, by hypothesis, one of the remaining tokens of ‘W’ is a copy of the other. All the tokens of ‘W’ are constrained to be copies of one another, and yet because copying
DP and DP
169
is exact including the alphabetic identity of variables, it is impossible for (434) and (435) to meet this condition while representing the domain restriction of one event quantifier by another. Literally, no logical form for (428) is coherent, since no candidate is well formed. It is thus possible to account for the robust judgment that (427) and (428) contrast, recasting the notions implicated—coordination, subordination, and complementation—as reflections in a Davidsonian setting of the scope relations and interactions among event quantifiers.76 The stray copying of Δ into the first clause of (424), which threatened to license a distributive interpretation of (412), proves to be as incoherent as copying ‘W’ in the attempt to save (228). (412) Boris s Natashej vyigrali $100. Boris.nom with Natasha.inst won.pl $100 (424) … [℩x: Boris(x)]W[e, x] Δ & ([℩x: Natasha(x)] … with[e,x] … [∃e′ : …](W[e′,x] …)) The parse for (412) is one of subordination, (Boris W (with Natasha (won $100))), with the scope relations among event quantifiers and their domain restrictions that that is taken to imply: (436) Boris participated (‘W’), and this was with Natasha, and some of that was winning $100.77 (437) [℩x : Bx] … W[e,x] … & [this ℩E′ : e′ ≤ e & [℩x : Bx]W[e,x]] … [℩x : Nx]With[e′,x] & [∃e″ : [that℩E′ : …] e″ ≤ e′] was(e″) winning $100 Now copying the Right-Node Raised constituent into the first-clause will strand an event variable, e′. Moreover, such copying absent any overt token of and to deliver coordination, implies further subordination, (Boris W (Δ (with Natasha (won $100)))). Yet the logical form that results fails to express that certain events of winning are both restricted to those in which Boris participates and restrict the events with Natasha: (438) *[℩x : Bx] … W[e,x] … [∃e″ : [that ℩E′ : …] e″ ≤ e′] was(e″) winning $100 & [this℩E′ : e′ ≤ e & [℩x : Bx]W[e,x]] … [℩x : Nx]With[e′,x] & [∃e″ : [that℩E′ : …] e″ ≤ e′] was(e″) winning $100 Because these event quantifiers are all nested, with any one restricting the domain of its successor, we cannot nest a copy of one higher in the tree and expect it to be well formed with respect to both positions. In sum, the logical syntax of subordination precludes the stray copying of Δ in the comitative construction *‘Boris Δ s Natashej …,’ while that of coordination, as in “Boris Δ i Natasha …,” requires a copy of the Right-Node Raised constitutent at Δ, thus deriving the crucial contrast between these two constructions.
170
Chapter 2
It now remains to the syntax of these two constructions to account for their differences with respect to anaphora, namely, that only subordination licenses the possessive reflexives in (415b) and (416b): (415) b. *Maša i svoja sestra ušli. Masha.nom and self’s sister.nom left.pl (416) b. Maša so svoej sestroj videli film. Masha.nom with self’s sister.instr saw.pl film What will matter is that no event quantifier within Δ includes the possessive reflexive within its scope in (415b), but the possessive reflexive in (416b) falls within the scope of all the (molecular) event quantifiers between the subject and comitative phrase. So let us begin by acknowledging that an intimation of reflexivity separates The barber shaves himself from its coarse rendering as The barber shaves the barber (see Reinhart and Reuland 1993). It prompts an impulse to treat reflexives as predicate operators of some sort—The barber self-shaves—rather than as expressions referring to the same thing as their antecedents. The point (pace its formalization) is to interpret the reflexive by invoking neither barbering nor direct reference to the barber. I will take it that what the barber does is to shave the shaver, as it were, drawing only on content that the predicate itself can provide to spell out the reflexive. The reflexive is a pronoun different from others in that it copies its content from an antecedent that must be a thematic relation, which will of course be a relation to some e—for example, ‘[℩x : W[e,x]]’. In (415b), although Masha is the intended antecedent, the reflexive token of ‘[℩x : W[e,x]]’ (svoja) fails to denote the W-er in Masha’s event, since its event variable e cannot be bound by any event quantifier in Δ, given the opacity of coordination. In contrast, subordination in (416b) puts both the antecedent thematic relation and its copy in the reflexive ‘[℩x : W[e,x]]’ (svoej) within the scope of the same event quantifier binding e and thus the reflexive comes to denote the W-er in the event in which Masha is the W-er, as intended.78 2.5
First-conjunct agreement (with conjoined (in)definite descriptions)
Conjoined (in)definite descriptions speak of the same events, from which has followed that Alia and Marwan must laugh at the same professor and be in the same room, (439)–(444), whether agreement is full or partial. (439) meetet alia w marwaan died.3fs Alia and Marwan (440) DeHket alia w marwaan ʕa esteez linguistics laughed.3fs Alia and Marwan at professor linguistics (441) *keenet alia w marwaan bi ouda *was.3fs Alia and Marwan in room
DP and DP
(442) meetou alia w marwaan died.3mp Alia and Marwan (443) DeHkou alia w marwaan ʕa esteez laughed.3mp Alia and Marwan at professor (444) keenou alia w marwaan bi ouda was.3mp Alia and Marwan in room
171
linguistics linguistics
Speaking of the same event, it is also possible, whether agreement is full or partial, for the Right-Node Raised constituent to describe a collective action, as in (445)– (465), that would not truthfully fit either subject alone. (445) ʕamlit alia w marwaan Hal’a Hawl š-šažra formed.3fs Alia and Marwan circle around the-tree ‘Alia and Marwan formed a circle around the tree.’ (Munn 1999) (446) fallet alia w marwaan sekraaniin left.3fs Alia and Marwan drunk.3mp (cf. (163)) (447) atlet alia w marwaan l-bsayneh killed.3fs Alia and Marwan the-cat (448) kasret alia w marwaan š-šebbeek broke.3fs Alia and Marwan the-window (449) kasret r-riiH w l-gheSn š-šebbeek broke.3fs the-wind and thebranch the-window (450) waʕʕet alia w marwaan l-bebe woke.3fs Alia and Marwan the-baby (451) waʕʕet l-’eTee’iyyeh w alarm s-sayyaara l-bebe woke.3fs the-fire-engine and alarm the-car the baby (452) Hemlet alia w marwaan l-piano lifted.3fs Alia and Marwan the-piano (453) fatHet alia w marwaan bweeb l-xazneh opened.3fs Alia and Marwan doors the-vault (454) waSalet alia w marwaan l-waSleet connected.3fs Alia and Marwan the-wires (455) waSalet alia w marwaan afkaar-un connected.3fs Alia and Marwan ideas-their (456) xalaTet alia w marwaan d-dheen l-azra’ blended.3fs Alia and Marwan the-paint the-blue (457) waffa’et tatsher w shmidt briitaniaw l-FRG allied.3fs Thatcher and Schmidt Britain and the-FRG
172
Chapter 2
(458) SaalaHet maliket n-naruuj w klinton isra’iil w s-selTa reconciled.3fs the Queen of Norway and Clinton Israel and the Palestinian l-felsTiiniyyeh Authority (459) tweezanet alia w marwaan ʕa l-marjuuHa balanced.3fs Alia and Marwan on the-seesaw (460) weeznet alia w marwaan l-wleed ʕa l-marjouHa balanced.3fs Alia and Marwan the-children on the-seesaw (461) tjammaʕet emma w l-anarkiyiin be l-haymarket gathered.3fs Emma and the anarchists in Haymarket (462) ttaHdet emma w l-anarkiyiin taHt bayeen waahad united.3fs Emma and the anarchists under manifesto one (463) ttaHdet chrysler w mercedes-benz taHt mudiir waahad united.3fs Chrysler and Mercedes-Benz under CEO one (464) ttaHdet chrysler w mercedes-benz ka šerkeh waHdeh united.3fs Chrysler and Mercedes-Benz as corporation one ttaHad obama w biden ʕa lista intixabiyyeh waHdeh united.3ms Obama and Biden on list electoral one (465) Saaret ’eTʕet z-zebdeh w T-THiin ʕajiineh became.3fs stick the-butter and the-flour dough Within the constraint of speaking of the same event, there remains, however, a difference of meaning between full and partial agreement. First, let’s consider how such a difference should be reflected in logical form. 2.5.0
The logical form of first-conjunct agreement
Since coordination is always a coordination of clauses, and, in particular, an apparent coordination of singular DPs does not itself refer to a plurality, it must be said that plural number agreement in such cases—that is, in cases of full agreement—arises when the conjoined DPs stand in the same relation to, say, the verb. The logical form for (466) with full agreement puts Alia and Marwan in the same relation to the event. The coordination in (467) is the smallest one that can be countenanced given that and is a sentential connective, and recall that plural number agreement is itself to be glossed as whatever they, Alia and Marwan, W-ed. (466) DeHkou alia w marwaan laughed.3mp Alia and Marwan (467) ∃e(Laughed.3mp(e) & W(e,Alia) and W(e,Marwan)) With the difference of meaning that partial agreement occasions, it will not do to say that (466) and (468) have the same logical form, such as (467), and differ only
DP and DP
173
in some syntactic convention for recording number agreement, as Aoun, Benmamoun, and Sportiche (1999, 677) point out. The logical forms of the coordinations must themselves differ in some way. (468) DeHket alia w marwaan laughed.3fs Alia and Marwan (469) *∃e(Laughed.3fs(e) & W(e,Alia) and W(e,Marwan)) We cannot agree with the naive suggestion that (468) differs in containing a second, tacit token of the verb if that is also understood to introduce a new event quantifier, as in (470). Partial agreement with collective action, (446)–(465), precludes talk of a new event of the kind denoted by the verb every time agreement is partial. (470) *∃e(Laughed.3fs(e) & W(e,Alia) and ∃e(Laugh(e) & W(e,Marwan))) On the other hand, omitting the second event quantifier, as in (471), is also rather lame. It is not clear that (471) means anything different from (467). The tacit verb token looks as if it serves only to mark that agreement is partial. (471) *∃e(Laughed.3fs(e) & W(e,Alia) and Laugh(e) & W(e,Marwan)) In fact, (471) is worse than lame if the singular agreement refers, as suggested, to whatever she, Alia, W-ed. Recall from chapter 1 that this reference is intended to be exhaustive, referring to what Alia and only Alia W-ed, which the first clause of (471) says is a laughing. If so, the second clause then issues in a contradiction, saying of that where only Alia W-ed that Marwan W-ed there too. Some difference of logical form underwrites the difference in meaning between (466) and (468). Whatever it is, it should preserve that the coordinated phrases speak of the same event, whether the agreement is full or partial (section 2.3.0 and (446)– (365)). Now, if in addition agreement means what I say and singular agreement means that only Alia W-ed there, it follows that Marwan did not W there and must have done something else there, call it ‘Cum(e,x)’. This last point now delivers a distinction in logical form between partial and full agreement: (472) DeHket alia w marwaan laughed.3fs Alia and Marwan ∃e(Laughed.3fs(e) & W(e,Alia) and Cum(e,Marwan)) (473) DeHkou alia w marwaan laughed.3mp Alia and Marwan ∃e(Laughed.3mp(e) & W(e,Alia) and W(e,Marwan)) Whether Alia and Marwan have lifted the piano or Alia with Marwan has lifted the piano (and similarly for the rest of (446)–(365)), a collective effort is required to move the piano.79 However it is accomplished, both Alia and Marwan are its
174
Chapter 2
agents. To describe it as Alia and Marwan acting or as Alia with Marwan acting implies no difference in the mechanics of what passed in the world. Chapter 1 already provides for this outcome, in arguing that ‘W(e,x)’ is a thematic relation distinct from Agent, Theme, Patient, and so on, in examples like (474), which leaves it vague as to who among the Columbia students and the Harvard students was at which demonstration and coordinates predicate phrases assigning different thematic relations to their subjects, Agent and Theme, respectively. (474) The Columbia students and the Harvard students surrounded the Pentagon and were crowded into the Mall. (475) ∃ei∃e′j ∃e″(W(ei,c) and W(ei,h) & O(proi,proj) & [∃xAgent(ej′, x) & C(e′j,e″) & surround(e″) & Theme(e″,p)] and [∃xTheme(e′j, x) & C(e′j,e″) & crowd(e″) & Into(e″,m)]) The argument for recognizing ‘W(e,x)’ as distinct from the more familiar thematic relations may rest on multiple coordinations such as (476), but the expectation has to be that it will carry over elsewhere. The simpler description of a collective action in (476) is analyzed along the lines of (477) or as in (478), which more accurately represents the cross-reference to events within each conjunct: (476) The Columbia students and the Harvard students surrounded the Pentagon. (477) ∃ei∃e′j ∃e″(W(ei,c) and W(ei,h) & O(proi,proj) & [∃xAgent(ej′, x) & C(e′j,e″) & surround(e″) & at(e″,p)]) (478) ∃e([The X: … Cx …][℩E : prok] Wi(E, X) and [The Y : … Hy …](Wi(e, Y) & (k [℩E : proi][℩E′ : proj]O(E, E′) & (j ∃e′∃e″[∃xAgent(e′, x) & C(e′,e″) & surround(e″) & Theme(e″,p)]))) When Alia and Marwan have lifted the piano and Alia with Marwan has lifted the piano, they are said to do the same thing because it is the Right-Node Raised constituent that says they are Agents: (479) Lifted.3mp Alia and Marwan the piano. (480) ∃e∃e″(lift.3mp(e″) & [℩x: Ax][℩E : prok] Wi(E, x) and [℩y: My](Wi(e, y) & (k [℩E : proi][℩E′ : proj]O(E, E′) & (j ∃e′[∃xAgent(e′, x) & C(e′,e″) & Theme(e″,p)]))) (481) Lifted.3fs Alia and Marwan the piano. (482) ∃e∃e″(lift.3fs(e″) & [℩x: Ax][℩E : prok] Wi(E, x) and [℩y: My](Cumi(e, y) & (k [℩E : proi][℩E′ : proj]O(E, E′) & (j ∃e′[∃xAgent(e′, x) & C(e′,e″) & Theme(e″,p)]))) If number agreement is agreement with the highest argument, with those that W, then in (480), it must be plural, denoting what they, Alia and Marwan, W-ed. On the
DP and DP
175
other hand, only Alia W-s in (482), and agreement is singular, denoting what she, Alia, W-ed. In (480), ‘[℩E : proi]’ then denotes exactly what Alia and Marwan W-ed, and this is said to exactly overlap an event of lifting the piano. In (482), ‘[℩E : proi]’ denotes exactly what Alia W-ed and Marwan Cum-ed. This reference to diverse sorts of situations, W-ing and Cum-ing, is just another instance of what has already been seen earlier in (483) and (484), where, as in (483) for example, the event pronoun denotes what goings-on were a student proposing and a professor accepting. Similarly, in (484), for some thirty authors and ten publishers, it denotes what was the authors donating and the publishers selling: (483) Not many a student proposed and not many a professor (of his) accepted that they should collaborate even more than they already have. (484) Thirty authors donated and ten publishers sold 1500 textbooks to inner-city schools and 200 recordings to inner-city libraries. In (482), what Alia W-ed and Marwan Cum-ed exactly overlaps an event of lifting the piano just as what Alia W-ed and Marwan W-ed does in (480).80 2.5.1
The meaning of first-conjunct agreement
For reasons of syntax and logical form, it has been argued that when agreement is partial, as in (485) and (486), Alia W-s and Marwan does not W but Cum-s instead. (485) DeHket alia w marwaan laughed.3fs Alia and Marwan ∃e(Laughed.3fs(e) & W(e,Alia) and Cum(e,Marwan)) (486) Hamlet alia w marwaan l-piano lifted.3fs Alia and Marwan the-piano ∃e(lift.3fs(e) & (W(e,Alia) and (Cum(e, Marwan) & VP[e]))) (487) DeHkou alia w marwaan laughed.3mp Alia and Marwan ∃e(Laughed.3mp(e) & W(e,Alia) and W(e,Marwan)) (488) Hamlou alia w marwaan l-piano lifted.3mp Alia and Marwan the-piano ∃e(lift.3mp(e) & (W(e,Alia) and (W(e, Marwan) & VP[e]))) Thus singular agreement predicts some asymmetry in how one regards the conjoined subjects. In reporting an earthquake, (489) puts the reporter on the street viewing the building and inferring its effect on the furniture. In contrast, (490) puts her inside the building observing the furniture and inferring the cause of its shaking. The plural agreement in (491) and (492) takes a neutral stance:
176
Chapter 2
(489) nhazzet l-bineeyeh w l-fareš shook.fs the-building and the-furniture (390) nhazz l-fareš w l-bineeyeh shook.ms the-furniture and the-building (491) nhazzou l-bineeyeh w l-fareš shook.mp the-building and the-furniture (492) nhazzou l-fareš w l-bineeyeh shook.mp the-furniture and the-building The casual skyward observer is inclined toward (493) while she who tugs at the string to guide the kite may say (494), and again plural agreement is neutral between them. (493) tHarrket t-tayyara w l-xeit maʕ l-riiH moved.3fs the-kite and the-string with the-wind (494) tHarrak l-xeit w t-tayyara maʕ l-riiH moved.3ms the-string and the-kite with the-wind (495) tHarrakou t-tayyara w l-xeit maʕ l-riiH moved.3mp the-kite and the-string with the-wind (496) tHarrakou l-xeit w t-tayyara maʕ l-riiH moved.3mp the-string and the-kite with the-wind It better fits the objects’ construction or development to say (497) and (501): (497) ghaTTa l-ghaTa l-’mesh w l-aber d-dahabi l-muumia covered.ms the-shroud the-linen and the-sarcophagus the-gold the-mummy (498) ghaTTa l-aber d-dahabi w l-ghaTa l-’mesh l-muumia covered.ms the-sarcophagus the-gold and the-shroud the-linen the-mummy (499) ghaTTou l-ghaTa l-’mesh w l-aber d-dahabi l-muumia covered.mp the-shroud the-linen and the-sarcophagus the-gold the-mummy (500) ghaTTou l-aber d-dahabi w l-ghaTa l-’mesh l-muumia covered.mp the-sarcophagus the-gold and the-shroud the-linen the-mummy (501) ghaTTou l-Hajra w l-’eshre bezret l-fekha sheath.mp the-stone and the-peel seed the fruit (502) ghaTTet l-’eshre w l-Hajra bezret l-fekha sheath.fs the-peel and the-stone seed the fruit (503) ghaTTou l-Hajra w l-’eshre bezret l-fekha sheath.mp the-stone and the-peel seed the fruit (504) ghaTTou l-’eshre w l-Hajra bezret l-fekha sheath.mp the-peel and the-stone seed the fruit
DP and DP
177
Yet some contexts, such as a dissection from the outside in, favor the opposing perspective: (505) Howard Carter, muktašef ’abr Tutankhamen, ʕam yektešef ennou Howard Carter, discoverer tomb Tutankhamen prog discovering that ghaTTa l-aber d-dahabi w l-ghaTa l-’mesh l-muumia covered.ms the-sarcophagus the gold and the-shroud the-linen the-mummy ‘Howard Carter, the discoverer of Tutankhamen’s tomb, is discovering that the gold sarcophagus and the linen shroud covered the mummy.’ (506) #Howard Carter, muktašef ’abr Tutankhamen, ʕam yektešef ennou Howard Carter, discoverer tomb Tutankhamen prog discovering that ghaTTa l-ghaT l-’mesh w l-aber d-dahabi l-muumia Covered.ms the-shroud the-linen and the-sarcophagus the-gold the-mummy ‘Howard Carter, the discoverer of Tutankhamen’s tomb, is discovering that the linen shroud and the gold sarcophagus covered the mummy.’ (507) Howard Carter, muktašef ’abr Tutankhamen, ʕam yektešef ennou Howard Carter, discoverer tomb Tutankhamen prog discovering that ghaTTou l-aber d-dahabi w l-ghaTa l-’mesh l-muumia covered.mp the-sarcophagus the gold and the-shroud the-linen the-mummy ‘Howard Carter, the discoverer of Tutankhamen’s tomb, is discovering that the gold sarcophagus and the linen shroud covered the mummy.’ (508) Howard Carter, muktašef ’abr Tutankhamen, ʕam yektešef ennou Howard Carter, discoverer tomb Tutankhamen prog discovering that ghaTTou l-ghaT l-’mesh w l-aber d-dahabi l-muumia Covered.mp the-shroud the-linen and the-sarcophagus the-gold the-mummy ‘Howard Carter, the discoverer of Tutankhamen’s tomb, is discovering that the linen shroud and the gold sarcophagus covered the mummy.’ (509) Alia am tekol w ’am tekteshef ennou mghaTTayeh l-’eshreh w Alia is eating and discovering that covering.fs the peel and l-’ajweh bezret l-fekha the stone the fruit’s seed. (510) #Alia am tekol w ’am tekteshef ennou mghaTTayeh l-’ajweh w #Alia is eating and discovering that covering.fs the stone and l-’eshreh bezret l-fekha the peel the fruit’s seed. (511) Alia am tekol w ’am tekteshef ennou mghaTTayiin l-’eshreh w Alia is eating and discovering that covering.pl the peel and l-’ajweh bezret l-fekha the stone the fruit’s seed.
178
Chapter 2
(512) Alia am tekol w ’am tekteshef ennou mghaTTayiin l-’ajweh w Alia is eating and discovering that covering.pl the stone and l-’eshreh bezret l-fekha the peel the fruit’s seed. Furthermore, failing superior imagination, there are sentences where no context seems to afford a natural perspective to locate the asymmetry expressed: (513) raH l-lansh w s-skiyyeur ʕa š-šaTT went.ms the-motorboat and the-water-skier to the-shore (514) #raHet s-skiyyeur w l-lansh ʕa š-šaTT #went.fs the-water-skier and the-motorboat to the-shore (515) raHou l-lansh w s-skiyyeet ʕa š-šaTT went.mp the-motorboat and the-water-skier to the-shore (516) raHou s-skiyyeet w l-lansh ʕa š-šaTT went.mp the-water-skier and the-motorboat to the-shore (517) xtafet disappeared.fs (518) #xtafa disappeared.fs
Tayyara w xaTT dexxaan abyaD bi l-maghreb jet and streak smoke white into the-sunset xaTT dexxaan abyaD w Tayyara bi l-maghreb streak smoke white and jet into the-sunset
(519) xtafou Tayyara w xaTT dexxaan abyaD bi l-maghreb disappeared.mp jet and streak smoke white into the-sunset (520) xtafou xaTT dexxaan abyaD w Tayyara bi l-maghreb disappeared.mp streak smoke white and jet into the-sunset When the selfish divorce the infirm to escape the burden of a spouse’s infirmity, (522) is infelicitous:81 (521) T-Talla’ l-aeeni w l-mariiDa divorced.ms the-selfish and the-infirm (522) #T-Talla’et l-mariiDa w l-aneeni #divorced.ms the-infirm and the-selfish (523) T-Talla’ou divorced.mp (524) T-Talla’ou divorced.mp
l-aeeni the-selfish l-mariiDa the-infirm
w l-mariiDa and the-infirm w l-aneeni and the-selfish
Since water bears the infection and water is the visible agent, (525) is felicitous and (526) is not. Again, plural agreement is neutral: (525) ntašret l-mayy w t-talawwos bi l-madiineh spread.fs the-water and the-infection in the-city
DP and DP
179
(526) #ntašar t-talawwos w l-mayy bi l-madiineh spread.msthe-infection and the-water in the-city (527) ntašarou l-mayy w t-talawwos bi l-madiineh spread.mp the-water and the-infection in the-city (528) ntašarou t-talawwos w l-mayy bi l-madiineh spread.mp the-infection and the-water in the-city Note that in this last example, the water is both what is directly perceived and the first in the causal chain. Their divergence however affords two perspectives, as in (529)–(532), differing minimally from (525)–(528). Sentence (528) is felt to adopt a perspective that traces out causation, and (530), a perspective that mirrors direct perception.82 (529) Teleʕ s-samm w xaTT d-damm l-aHmar men maHall climbed.sg the poison and streak the-blood the-red from location el 3aDDa la ketf-ou the bite to his-shoulder ‘The toxin and the red blood streak traveled from the bite to his shoulder.’ (530) Teleʕ xaTT d-damm l-aHmar w s-samm men maHall el 3aDDa la climbed.sg streak the-blood the-red from location the bite to ketf-ou his-shoulder ‘The red blood streak and the toxin traveled from the bite to his shoulder.’ (531) Telʕou s-samm w xaTT d-damm l-aHmar men maHall el 3aDDa climbed.pl the-poison and streak the-blood the-red from location the-bite la ketf-ou to his shoulder (532) Telʕou xaTT d-damm l-aHmar w s-samm men maHall el 3aDDa la climbed.pl streak the-blood the-red from location the bite to ketf-ou his-shoulder The spatiotemporal proximity felt to distinguish partial agreement ((254), (255)) from full agreement ((256), (257)) derives from a relation to an (asymmetric) perspective on the same scene and from the grounds one may have for regarding the events reported as a scene under the perspective given: (254) meetet alia w marwaan died.3fs Alia and Marwan (255) DeHket alia w marwaan laughed.3fs Alia and Marwan
180
Chapter 2
(256) meetou alia w marwaan died.3mp Alia and Marwan (257) DeHkou alia w marwaan laughed.3mp Alia and Marwan Of course, whatever will prove to be invariant among scenes and perspectives to render them coherent, it will not include an absolute spatiotemporal scale: (533) nfajret andromeda w s-surayya exploded.fs Andromeda and the-Pleiades An astronomical scene is scaled to astronomical distances. Nevertheless, from what we understand about death and laughter and the ordinary circumstances under which these come to pass, the speaker, failing unforeseen circumstances, has grounds to say that Alia succumbs with Marwan only if they are in the same room on the same occasion. That they are judged near each other (merely a symmetric relation, after all) is a corollary of the organizing perspective (expressed by W and Cum) that puts them in the same scene and space. The conditions of use for partial agreement appear to parallel in many respects (with divergences noted below) the conditions on overt comitative phrases, such as with NP and together, surveyed in the literature (see, e.g., McNally 1993; Camacho 1995, 1997, 2000, 2003; Dalrymple, Hayrapetian, and King 1998a; Moltmann 1994, 1997; Lasersohn 1995; Schwarzschild 1994).83 As with partial agreement, the togetherness is sometimes causally effective and sometimes merely spatiotemporal: (534) alia w marwaan Hamalou l-piano sawa Alia and Marwan held-up the-piano together alia Hamlet l-piano maʕ marwaan Alia held up the-piano with Marwan (535) alia w marwaan Hamalou souret hawiyyeh sawa Alia and Marwan held up photo ID together ‘Alia and Marwan held up an ID photo together.’ alia Hamlet souret hawiyyeh maʕ marwaan Alia held up photo ID with Marwan (536) alia w marwaan ʕaTaSou sawa Alia and Marwan sneezed together alia ʕaTset maʕ marwaan Alia sneezed with Marwan
DP and DP
181
(537) ashum Dow Jones w ashum NASDAQ ʕelyou sawa stocks Dow Jones and stocks NASDAQ rose together ashum Dow Jones ʕelyou maʕ ashum NASDAQ stocks Dow Jones rose with stocks NASDAQ Although these phrases may mean nothing more than spatiotemporal proximity, as in (538), their use as modifiers in (539) amounts to more than an assertion that Alia and Marwan have blue eyes and they are together: (538) alia w marwaan sawa Alia and Marwan (are) together alia maʕ marwaan alia (is) with marwaan (539) #alia w marwaan ʕandon ʕyoun zer’ sawa #Alia and Marwan at-them eyes blue together ‘Alia and Marwan have blue eyes together.’ #alia ʕanda ʕyoun zer’ maʕ marwaan #Alia at-her eyes blue with Marwan. ‘Alia has blue eyes with Marwan.’ (540) #alia w marwaan amerkeen sawa #Alia and Marwan Americans together #alia amerkeniyyeh maʕ marwaan #Alia American with Marwan Rather, their being together must in some way locate or circumscribe their having blue eyes, and one stumbles at the thought that their having blue eyes has a definite location (except maybe in their eyes, where it stays whether they are together or not). As we have seen, partial agreement in Lebanese is subject to similar constraints.84 Most of the literature, assuming naturally enough that Alia is with Marwan if and only if Marwan is with Alia, is concerned with the spatiotemporal or dynamic constraints that their being together imposes on the events so modified, with slight regard for what might tell apart (541) and (542): (541) alia w marwaan sawa Alia and Marwan (are) together alia w marwaan maʕ baʕDon Alia and Marwaan (are) with each-other
182
Chapter 2
(542) alia maʕ marwaan Alia (is) with Marwaan marwaan maʕ alia Marwaan (is) with Alia Yet, while (481) puts Alia and Marwan together lifting the same piano, and some notion of togetherness constrains all the examples of partial agreement in (485)– (532), we have seen that the syntax and logical form of number agreement point to a difference for Alia and Marwan, who, as a result, must not bear exactly the same thematic relations to the events described—a difference of meaning that becomes manifest in (485)–(532). The difference in meaning goes hand in hand with the difference in structure, distinguishing Alia and Marwan have lifted the piano from Alia with Marwan has lifted the piano. Such a difference in meaning should also tell apart (531) and (542), despite the rough paraphrase that each of the sentences is for any of the others. Such a correlation of meaning and structure has been observed in (543)–(552) (Tversky 1977; Tversky and Gati 1978; Talmy 1978, 1983; Gleitman et al. 1996). (543) a. The humblest citizen is equal to the President. b. The President is equal to the humblest citizen. (544) a. The humblest citizen and the President are equal. b. The President and the humblest citizen are equal. (545) a. North Korea is similar to Red China. b. Red China is similar to North Korea. (546) a. The button matches the shirt. b. The shirt matches the button. (547) a. The waitress resembles the senator. b. The senator resembles the waitress. (548) a. Meryl Streep met my sister. b. My sister met Meryl Streep. (549) a. The bicycle is near the garage. b. The garage is near the bicycle. (550) a. A horse is tied to a tree. b. A tree is tied to a horse. (551) a. The drunk collided with the lamppost. b. The lamppost collided with the drunk. c. The drunk and the lamppost collided.
DP and DP
183
(552) a. The critic compares Ecuador with the United States. b. The critic compares the United States with Ecuador. The intent of (543a) is to elevate the humblest citizen, and (543b), to diminish the President. Now, if ever there was a symmetric relation, it is is equal to. As Gleitman et al. (1996, 324) remark, to conclude that the asymmetry observed in (543) is intrinsic to the lexical concept is to hold that there are no symmetric concepts at all. Rather, the asymmetry must derive from a difference in the meaning of the surrounding constructions, as Talmy (1978, 1983) has suggested. The difference hangs on the syntactic position of the compared terms. A task eliciting judgments of synonymy (Gleitman et al. 1996, Experiment 3) showed robust discrimination between (543a) and (543b), while (544a) and (544b) were judged synonymous pace the effect of a slight preference, occurring with all predicates, to name the more salient character first (The Pope and Sam prayed for peace vs. Sam and the Pope prayed for peace). What induces the asymmetry of meaning found in (543) is that one of the terms is a subject and the other is not, and it persists even where the terms offer no other cues, as in the nonce “The ZUM is identical to the GAX” (Gleitman et al. 1996, Experiment 5).85 The difference in meaning observed in (543)–(552) is not directly about how events are in the world but rather a difference that plays itself out in how to construct the scene or perspective under which some events are to be thought of. Of course events in the world are not open to arbitrary depiction, and thus aspects of the scene’s construction have implications for how the events depicted are, spatiotemporally and causally. It is not surprising that most of the time one’s awareness of perspective is submerged while attending to how things are out there, especially if one hopes to understand what features of the environment are invariant from one observation to the next. If the sentences of (542) seem to say the same thing, it is that no particular scene or perspective is presented, and one knows that if there is an event that is Alia with Marwan under some perspective, then there is another perspective under which it is Marwan with Alia.86 (553) JFK is with Marilyn Monroe, and Marilyn Monroe is with JFK. If some perspective or other verifies the first conjunct of (553), then some perspective or other verifies the second and vice versa. This a priori knowledge about symmetric predicates makes (553) an assertion of propositions known to be effectively equivalent. Normally this would sound tedious and unprovoked, unless there is enough context to justify the same state of affairs being redescribed under a different formulation, as there would be if the preacher, swearing by (554), intends to damn both JFK and Marilyn Monroe with his assertion of (553). (554) If x is with a starlet, then x is damned. If y is with an adulterer, then y is damned.
184
Chapter 2
The proofs of their damnation from (554) justify saying (more or less) the same thing twice, with only a difference in syntax. The intent for (553) to enter into proof makes it felicitous here, whether or not the speaker is conscious of any difference in meaning between the conjuncts (and whether or not such a difference exists). Other contexts however make plain how a difference in perspective conveys a difference in circumstances: (555) At every moment, day or night, during the next fortnight, JFK and Marilyn Monroe are with each other exclusively. (More specifically, I can volunteer that) this week JFK is with Marilyn, and next week Marilyn is with JFK. This week is at the Beverly Hills Hotel, and next week is at the White House. The spokesman’s evident purpose is to contrast in some way the two weeks and so implicates that the description offered of one does not also fit the other. Of course, it cannot be that the first week JFK is with Marilyn and the second he is not, unless the spokesman means to contradict his opening sentence. Contradiction aside, one should query how the spokesman manages with the follow-up sentence to say anything that he has not already said in his opening. Let ‘With(x,y,e)’ abbreviate a symmetric relation (i.e., ‘With(x,y,e) ↔ With(y,x,e)’), which is not to be confused with the (asymmetric) sentence construction x is(e) with y. The latter can be taken as ‘W(x,e) & Cum(y,e) & ∃e′(O(e,e′) & With(x,y,e′)).’87 Gloss the spokesman’s speech as follows: (556) At every moment, day or night, during the next fortnight, JFK and Marilyn Monroe [under a perspective that holds them in equal regard] are [under that perspective] with each other [i.e., ∃e(With(jfk,mm,e) & With(mm,jfk,e))] exclusively. (More specifically, I can volunteer that) this week [there is an (asymmetric) perspective e under which] JFK is with Marilyn [W(jfk,e) & Cum(mm,e) & ∃e′(O(e,e′) & With(jfk,mm,e′))], and next week [there is an (asymmetric) perspective e under which] Marilyn is with JFK [W(mm,e) & Cum(jfk,e) & ∃e′(O(e,e′) & With(mm,jfk,e′))]. Their being together is symmetric, but the surrounding circumstances that bring them to that state are not. The first week JFK goes to Marilyn Monroe at her place, and the second week she goes to him at his. Notice that this accords with Talmy’s (1978, 1983) assignment of Figure and Ground: the oblique argument provides the fixed point and frame of reference, the Ground, for the Figure’s transition. The spokesman would have misspoken—and if you reflect on it, rather strikingly so— had Marilyn Monroe gone to the White House in the first week and JFK, to the Beverly Hills Hotel in the second.88 The shift in perspective need not entail a shift in venue, although a move from one coast to the other provides obvious grounds for a shift. Consider some newsreel segment of the First Family at the beach in Hyannis Port:
DP and DP
185
(557) Here on a sunny morning near the Kennedy compound, Jack and Jackie are with each other. Now Jack is with Jackie, and now Jackie is with Jack. The First Couple, just like couples everywhere in America, … The narrator may be inclined toward (557) if, for example, the first clip shows Jack draping himself on Jackie and the second, Jackie draped on Jack. Or, perhaps, first Jack focuses his attention on Jackie while she gazes absently out to sea and then she fixes her gaze and thoughts on him while he studies a sailboat. Clearly diverse circumstances may condition a shift in perspective, but again, whether it is posture or thought, it is the Figure’s posture or thought that undergoes some transition with respect to the Ground. The narration in (557) would be incoherent if the scenes were reversed—if Jackie first drapes herself or focuses on Jack. It would also be incoherent if the newsreel were to show Jack and Jackie sitting on the beach reading and giving no indication of anything else. As if there were a failure of reference, one could not in that case discern the two perspectives intended. The assigned perspectival relations thus bear real teeth, and it will not do to imagine that they contribute to a lexical ambiguity in with, where an asymmetric witha occurs in x is with y in (555), and a symmetric withs in are with each other. As Gleitman et al. and Talmy emphasize, the latent ambiguity generalizes across the lexicon and correlates with the difference in syntactic structure. Besides, if there were two withs, why not two reciprocals, withs each other and witha each other? Nothing bars reciprocal expressions based on genuine asymmetric relations, love each other, hit each other, be against each other, and the like, but there is arguably no witha each other. Note that (558)–(560) can be taken to assert that the time the truck hit the bus was different from the time the bus hit the truck: (558) The truck hit the bus, and the bus hit the truck, at different times. (559) The truck and the bus hit each other (—but) at different times. (560) Each of the truck and the bus hit the other—(but) at different times. (561) The truck collided with the bus, and the bus collided with the truck, at different times. (562) #The truck and the bus collided (with each other) (—but) at different times. (563) Each of the truck and the bus collided with the other—(but) at different times. (564) JFK was against RFK, and RFK was against JFK—at different times. (565) JFK and RFK were against each other, at different times. (566) Each of JFK and RFK was against the other—(but) at different times. (567) JFK was with Marilyn Monroe, and Marilyn Monroe was with JFK—at different times.
186
Chapter 2
(568) #JFK and Marilyn Monroe were with each other(—but) at different times. (569) Each of JFK and Marilyn Monroe were with the other—(but) at different times. Sentences (561) and (563) can express the analogous thought that the time the truck collided with the bus was different from the time the bus collided with the truck. But the comparison fails in (562), which means only that they collided at least twice, without implying a difference between a collision of truck with bus and of bus with truck. The contrast between (559) and (562) is unexpected if there is a lexical, asymmetric witha. On the other hand, if with is univocal and symmetric, then every event e at which “With(x,y,e)” is also an event at which “With(y,x,e)” and it is absurd to think that the time at which the truck is with the bus is different from the time at which the bus is with the truck, hence the unacceptability of (562). As the contrast between (562) and (563) shows, when the scope of the reciprocal includes the perspectival relations occurring elsewhere in the sentence, it then includes within its scope the expression of an asymmetric relation that can support the temporal comparison. The same argument mutatis mutandis can be made from (564)–(569): against intrinsically expresses an asymmetric relation, and with never does.89 The arguments of Gleitman et al. and Talmy as well as the evidence of (558)–(569) show that expressions denoting symmetric events (e.g., “With(x,y,e)”), and expressions denoting collective events (e.g., collide, lift the piano) that do not themselves discriminate among the collective participants, nevertheless occur in sentences that regard the participants asymmetrically. The expression of this asymmetry must come from somewhere other than the lexical items themselves. These findings suggest that sentences contain some (asymmetric) perspectival relations in addition to the lexical content of their predicates. The logical forms I should wish for are arrived at here from the meaning, from observations of a stray asymmetry that lingers despite the symmetric and collective predicates. This argument thus converges with the opening one (section 2.2.2) that the syntax and logical form of partial agreement require that conjoined subjects be treated separately, as a W-er and a Cum-er, their collective participation being viewed under an asymmetric lens, the effects of which become transparent in (489)–(532). Now recall again that the argument in chapter 1 for higher thematic relations such as ‘W(e,x)’ concludes at the same time that the events so described coincide or cause the events described as having Agents, Themes, Patients, etc., as in (571), without asserting an identity between the higher and lower events: (570) The Columbia students and the Harvard students surrounded the Pentagon and were crowded into the Mall. (571) ∃ei∃e′j ∃e″(W(ei,c) and W(ei,h) & O(proi,proj) & [∃xAgent(ej′, x) & C(e′j,e″) & surround(e″) & Theme(e″,p)] and [∃xTheme(e′j, x) & C(e′j,e″) & crowd(e″) & Into(e″,m)])
DP and DP
187
So, in addition to there being perspectival relations, W and Cum, distinct from the more familiar thematic relations, a sentence may apply the two classes to different events. The discovery of an exact meaning for ‘O’ laying down how the events proi and proj in (571) and the like are to be related remains an empirical question. The logical form does, however, afford that there may be some distance between the perspectival events and the events to which they are related by overlap or coincidence, or whatever. Lebanese appears to hold collective subjects and the thematic relations of the collective on an especially tight leash. In (572) and (573), both members of the collective, the ceiling and the floor, must be understood to be in motion: (572) Beneath Dr. No’s laboratory, a. S-Sa’ef w l-’areD maʕasou James Bond the ceiling and the floor crushed/squashed.3mpl 007 b. nmaʕasou S-Sa’ef w l-’areD James Bond crushed/squashed.3mpl the ceiling and the floor 007 c. nmaʕas S-Sa’ef w l-’areD James Bond crushed/squashed.3sg the ceiling and the floor 007 (573) Beneath Dr. No’s laboratory, a. S-Sa’ef w l-’areD lta’u the ceiling and the floor met.3mpl, b. lta’u S-Sa’ef w l-’areD met.3mpl the ceiling and the floor, c. lta’a S-Sa’ef w l-’areD met.3sg the ceiling and the floor, but 007 had escaped unharmed90 None of these sentences, whether agreement is partial or full, felicitously describes a drop ceiling and a stationary floor, which better fit a (di)transitive construction, to crush against the floor or to meet the floor (see (551c)). I assume that the perspectives on Dr. No’s death chamber accessible to us do not include relativizing motion: what is stationary remains so from wherever one thinks one is standing when judging these sentences.91 Although the perspectival relations vary within (572) and within (573), it remains constant throughout (572) that any member of the collective subject is a Theme and Agent of 007’s (the Patient’s) destruction, and throughout (573), any member of the collective subject is also a Theme. To be a Theme here is to be in motion, and thus it cannot be that either the ceiling or the floor is different from the other in this respect. Even with every Theme in motion, we can still take the measure of the distance between the perspectival events and the events that realize the absolute conditions reported to have been observed:
188
Chapter 2
(574) lta’et alia w marwaan met.3fs Alia and Marwan (575) lta’ou alia w marwaan met.3mp Alia and Marwan (576) byelta’a l-baaS w t-treen bi mHaTTet grandcentral meet.3ms the-bus and the-train in Station Grand Central (577) byelta’ou l-baaS w t-treen bi mHaTTet grandcentral meet.3mp the-bus and the-train in Station Grand Central (578) a. lta’et l-’arD w l-marriix met.3fs Earth and Mars b. lta’a l-muzannab w l-kuwaykeb met.3ms the-comet and the-asteroid (579) a. lta’ou l-’ard w l-marriix met.3mp Earth and Mars b. lta’ou l-muzannab w l-kuwaykeb met.3mp the-comet and the-asteroid Sentence (574) with partial agreement, according to which Alia W-s but Marwan Cum-s, may describe their meeting in conference but not their chance encounter on the street. It matters little that their meeting has them both in motion, doing the same thing as far as the absolute conditions on meetings go. In fact, the symmetry of their movements removes obvious grounds for the asymmetric perspective that partial agreement presents, and leaves us to search for what is intended. There are of course many things that are true of Alia or Marwan and not true of the other, although these might not be constitutive of the event of meeting narrowly construed. In the case of a conference, there are the plans for the meeting, the issues to be resolved, the mandates and charges of the meeting’s various participants, and the meeting’s effects on future action, any of which may put Alia and Marwan in a different light for us. Perhaps it is just that Alia was the instigator—although they arrive at the meeting simultaneously and take turns speaking in an entirely balanced and equitable fashion—which prompts us to concede that Alia with Marwan met. The meeting itself, their being there together and their behavior during, may be entirely symmetric, but the surrounding circumstances that bring them to that event are not. For (574), it suffices that the events grounding an asymmetric perspective locate, frame, or bring about (‘O(e,e′)’) the symmetric event at which Alia and Marwan are equal participants. As for the chance encounter on the street, it may be that Alia wished and plotted for it to happen, but wishing and plotting do not get a fix on a meeting unless it was no accident. There are thus no grounds for an asymmetric perspective, given the fact of a chance encounter and that we cannot think of one
DP and DP
189
participant as in motion and the other as stationary, both being Themes. As a consequence of these considerations relating perspectival events and the events reported, partial agreement manages to tell apart conferences from chance encounters. The remaining sentences (576)–(579) cover the same ground. Partial agreement (576) cannot be used in a report of a chance encounter or collision between the bus and the train. If however they meet at Grand Central according to schedule, and the schedule is held to account for their meeting, then partial agreement is again acceptable and the asymmetric perspective is assumed to emerge from the schedule’s organization of bus and train routes. As a further example, partial agreement is inappropriate as a report of a chance encounter between a comet and an asteroid (578b). The laws of celestial mechanics may, however, provide an asymmetric perspective on the Earth and Mars in an event from which their movement into conjunction, their meeting, is a known consequence. Lina Choueiri has also remarked to me that (578a), in contrast to (579b), implicates a temporary conjunction as against, say, the steady state of double stars. Recall the early observation that partial number agreement also requires eventivity, illustrated in the contrast between (250) and (386): (250) Saret (maʕle’t) z-zebdeh w l-THiin ʕajiineh became.3fs (tbsp. of) the-butter and the flour dough (386) *keenet (maʕle’t) z-zebdeh w l-THiin ’ajiin was.3FS (tbsp. of)the-butter and the-flour dough Some absent degree of eventivity also seems to spoil (580): (580) *bteb’a alia w marwaan aghbiya *remain.3fs Alia and Marwan stupid.3mp (581) byb’ou alia w marwaan aghbiya remain.3mp Alia and Marwan stupid.3mp Yet what is at stake is not the intrinsic stativity of remaining such-and-such but whether or not there is a perspectival event that is relevantly asymmetric and that would, in a way understood by speaker and hearer, fix or circumscribe the event of remaining such-and-such, an event that by itself affords no basis to regard Alia and Marwan any differently. If an appropriate perspective is provided, as when Alia with Marwan conspire to obstruct justice, then the stativity of remaining silent does not obstruct partial agreement: (582) lamma stajwabet-on š-šerTa, be’yet alia w marwaan when questioned-them the-police, remained.3fs Alia and Marwan seektiin silent.3mp92
190
Chapter 2
An observer of planetary motion understands that it is lawful and can well imagine that within the frame of reference of her observations, there are prior forces, configurations, and velocities with respect to which the Earth and Mars are different but that end with them behaving very much alike in a meeting. Within the same frame of reference, the double stars are immovable and their state is a meeting. The meeting is as before symmetric, but here celestial mechanics are irrelevant and one fails to grasp where the asymmetric perspective could have come from. Of course, if the discussion and frame of reference were the formation of the universe, one could well imagine surrounding circumstances that end in the meeting of a double star, but that would again be a transitional state, only on a different scale. The contrast between (250) and (386) deserves further comment. There is no reason to think that becoming dough itself affords an asymmetric perspective on its ingredients. One turns instead to the recipe or to the action in the bakery. But if such affords an asymmetric perspective on the symmetric event of becoming dough, why not also on the being dough, the baker’s end result? Except for laying out some boundary conditions, I have obviously been equivocating on the meaning of ‘O(e,e′)’, especially on the point of whether it is a full-blooded causal relation or a more topographical one (‘overlap’). In chapter 1, I allowed that there may be a class of relations that take the place of ‘O(e,e′)’ depending on the complement verb. If it should turn out instead that decomposition yields only the one relation ‘O(e,e′)’, then the contrast between (250) and (386) shows, at the least, that if e′ is a becoming, the events e that ‘O(e,e′)’ are more far-flung than if e′ is a being. There just aren’t all that many ways to look at being dough. Such, in effect, is also the conclusion of Gleitman et al. (1996), who discover with a task that elicits judgments of synonymy (Experiment 3) that there is a more robust discrimination of the (a) and (b) alternatives when the predicate is eventive ((583), (584)) than when it is stative ((585), (586)). (583) a. Meryl Streep met my sister. b. My sister met Meryl Streep. (584) a. The bus collides with the scooter. b. The scooter collides with the bus. (585) a. North Korea is similar to Red China. b. Red China is similar to North Korea. (586) a. The button matches the shirt. b. The shirt matches the button. 2.5.2
Summary remarks on the logical form and meaning of first-conjunct agreement
Section 2.3.0 showed that postverbal conjoined (in)definite descriptions speak of the same events, which the Right-Node Raised constituent must therefore describe
DP and DP
191
collectively rather than separately, as it would under a “sloppy” or distributive construal. This aspect of the meaning and logical form of conjoined (in)definites remains in place whether number agreement is full or partial. Yet the meaning of full and partial agreement also tells them apart. As Aoun, Benmamoun, and Sportiche (1999, 677) urge, for number agreement to hook up with a difference in meaning, there must be a difference in structure (broadly understood to include differences in the enumeration of tacit or overt lexical items). For me, logical form is an exercise in Davidsonian decomposition. Now, whatever the purported difference in meaning, whether Alia and Marwan have lifted the piano or Alia with Marwan has lifted the piano, the piano is lifted and both are agents. The Right-Node Raised constituent will have to say as much, it being a collective remark true of what Alia did and Marwan did, whether agreement is partial or full—and similarly for meetings, where the participants are all Themes, and so on. The differences of meaning should not obscure representation of what is the same between sentences that are identical except for number agreement. The difference in meaning, as argued, amounts only to a difference in perspective. With respect to this perspective, if ‘W(e,Alia) and W(e, Marwan)’ is the logical form of Alia and Marwan under full agreement (as something like it must be, given that and is always sentential), then the requirement that number agreement reflect both a difference of structure and a difference of meaning enjoins that the logical form under partial agreement be something else. There isn’t much else it could be other than, say, ‘W(e,Alia) and Cum(e,Marwan)’, where Cum, despite the mnemonic, could stand for anything that isn’t W. It was also argued earlier that singular agreement denotes whatever (s)he did and if what she did at e was to W then only she W-ed at e. It follows that Alia W-ed at e, but not Marwan, who Cum-ed there instead. Suppose that there is just one disturbance in the field that is a lifting of the piano, and we judge it true both that Alia and Marwan have lifted the piano and Alia with Marwan has lifted the piano. By hypothesis, there is just the one lifting, e′, with Alia and Marwan its Agents. But there is also an event at which both Alia and Marwan W (Alia and Marwan) and an event where Alia W-s but Marwan does not (Alia with Marwan). These obviously cannot be the same event, and yet they are both related to the lifting. That puts some distance between the lifting and at least one of these, which ‘O(e,e′)’ records. This is just as well, since chapter 1 argues on independent grounds that the logic of multiple conjunctions multiplies the events talked about and the causal or topographical relations between them, like ‘O(e,e′)’.93 That speakers discern some or another difference of meaning between full and partial agreement while firmly grasping that a difference of only number agreement leaves much the same meaning intact has determined, even before we explore the difference in meaning, several aspects of logical form. In particular, there is a ‘higher’ and a ‘lower’ event, the latter described by the Right-Node Raised constituent and
192
Chapter 2
the former by the content of the conjoined propositions. In the case of partial agreement, that content necessarily describes the participants as doing different things in the higher event. So much derives from just the fact of a meaning difference and general considerations about a Davidsonian logical form and the syntax and semantics of number agreement. The arguments of Gleitman et al. and Talmy, and the evidence of (558)–(569), show that there are perspectival asymmetries not occasioned by the choice of lexical items but expressed elsewhere in the sentence. Partial agreement in Lebanese bears this out, and we have also seen that there can be some distance between the higher, perspectival event and the lower, reported event, as the logical form proposed anticipates. 2.6
Comitative phrases and number agreement in other languages
With partial agreement, Alia W-s and Marwan Cum-s, as in (482): (481) Hamlet alia w marwaan l-piano Lifted.3fs Alia and Marwan the-piano (482) ∃e∃e″(lift.3fs(e″) & [℩x: Ax][℩E: prok] Wi(E, x) and [℩y: My](Cumi(e, y) & (k [℩E : proi][℩E’ : proj]O(E, E′) & (j ∃e′[∃xAgent(e′, x) & C(e′,e″) & Theme(e″,p)]))) Despite one’s W-ing and the other’s Cum-ing, they nevertheless manage collectively to lift the piano. Recall that their collective action derives, following the general analysis of Right-Node Raised constituents as collectivized phrases, from the intervention of an event pronoun. ‘[℩E : proi]’ refers to whatever events there are of Alia W-ing and Marwan Cum-ing, and these are said to overlap or coincide with the lifting. I take it that ‘[℩E : proi]’ would be a plural expression in this case although it is unpronounced like most event pronouns. Verbal number agreement, however, is also an expression of this kind, and recall that under analogous Right-Node Raising structures, both the plural number agreement and its clausal antecedents can be overt: (312) Lindak ardau eta Anderek esnea edatendabez. Linda wine and Ander milk drink be (3PL) ‘Linda will drink wine and Ander milk.’ (Sjoblom 1980)
[Basque]
If Linda acting on wine and Ander acting on milk can induce plural number agreement in some language, why not Alia W-ing and Marwan Cum-ing? A comitative phrase in some language should be able to induce plural agreement if phrase struc-
DP and DP
193
ture locates it in a position for it to serve as an antecedent for the verbal agreement. For all I know, Lebanese could be such a language, but it is difficult to tell whether a preverbal occurrence of Alia and Marwan is tacitly Alia W-s and Marwan W-s or Alia W-s and Marwan Cum-s, since either parse is expected to induce plural number agreement on the following verb and Alia W-s and Marwan W-s is felicitous anywhere that Alia W-s and Marwan Cum-s is. This difficulty stands in the way of discovering direct evidence in any language that Alia W-ing and Marwan Cum-ing yields plural number agreement. Still, the larger point has been that verbal number agreement engages clausal antecedents, whether they are overt as in (312) or hidden as in Alia and Marwan, denying that Alia and Marwan is a phrase that refers to two children. This point should embrace any phrase that manages to induce plural number agreement in virtue of Alia and Marwan being two (see (312)) without its meaning the same thing as Alia and Marwan. In the comitative construction of a variety of languages, both Alia and Marwan and Alia with Marwan deliver plural agreement when preceding the verb (see McNally 1993; Camacho 1995, 1997, 2000, 2003; Dalrymple, Hayrapetian, and King 1998a): (587) Marek z Piotrem podniesli pianino M.nom with P.instr lifted.pl piano ‘Mark with Peter lifted the piano.’ (McNally 1993, 376)
[Polish]
(588) Boris s Petej podnjal rojal’ B.nom with P.instr lifted.pl piano ‘Boris with Peter lifted the piano.’ (McNally 1993, 373; typographical correction, Denis Paperno)
[Russian]
(589) Marta con María levantaron el piano Martha with Maria lifted.pl the piano (Camacho 1997, 206)
[Spanish]
As expected, the comitative construction supports plural agreement only preceding the verb. Elsewhere agreement is singular—that is, ‘partial’, so to speak. Throughout, when asked to volunteer a paraphrase, I have been using ‘with’ to gloss ‘Cum’. In light of the above, there is a temptation to let Lebanese coordination with a Cum-phrase and the comitative construction found in other languages differ just in what is pronounced: (590) … W(e, Alia) and Cum(e, Marwan) … (591) … W(e, Alia) & Cum(e, Marwan) …
[Semitic] [Slavic, Romance, Turkish, etc.]
194
Chapter 2
Lebanese would pronounce the conjunction and but not the thematic relation Cum, while languages with the comitative construction pronounce the thematic relation and not the conjunction. We would soon rue such a facile identification—Why can’t one ever say “Alia is (*and) Marwan” parsed with tacit Cum to mean that Alia is with Marwan? Translation between (590) and (591) suggests that one should be able to substitute for ‘with Marwan’ ‘and Marwan’, parsing the latter with tacit Cum. Conversely, wherever analysis reveals a parse of “and Marwan” containing tacit Cum, it should be possible to substitute “with Marwan.” If (590) should occur preverbally in Lebanese and license plural agreement, then why can’t its alleged counterpart ‘Alia with Marwan’ do the same, as it does in languages with the comitative construction: (592) *alia ma’ marwan Hamalo l-piano Alia with Marwan lifted.pl the piano The translation falters if ever comitative phrases and coordination with a Cumphrase diverge, unless syntax steps up to tell us where and/& and with/Cum are licensed pronounced and unpronounced. Whatever could be said to prop up the translation between (590) and (591), section 2.5.1 has in fact already decided against it. Recall that Cum and other perspectival relations were enlisted to give expression in a Davidsonian setting to the conclusion by Talmy and Gleitman et al. that structure itself conveys meaning, taken here to be a frame of perspectival or aspectual relations surrounding the overt lexical items. By hypothesis, Cum and With should not be intersubstitutable. Furthermore, it was argued that they are not synonyms. The lexical item by itself denotes symmetric events that are then subjected to the asymmetric perspective that Cum imposes. Recall that the asymmetric perspective intrinsic to ‘x is with y’ (see (553), (555)) is meaning from a structure that includes occurrences of both Cum and With: ‘W(e,x) & Cum(e,x) & ∃e′(O(e,e′) & With(x,y,e′))’. Thus, even within Lebanese, there is a difference of logical form between Alia and Marwan, coordination with a Cum-phrase, and Alia with Marwan, a comitative phrase: (593) Lifted.3fs Alia … [℩x: Ax]W(e, x)…
and Marwan and … [℩x: Mx]Cum(e, x) …
the piano.
(594) Lifted.3fs Alia with Marwan the piano. … [℩x: Ax]W(e, x) & [℩x: Mx]∃e′(Cum(e,x)& O(e,e′) & ∃yWith(y,x,e′)) … The thought is that with Marwan, if not a clause as full-blooded as, say, being with Marwan, is nevertheless a reduced clause with at least some of the perspectival projections characteristic of any phrase containing a lexical predicate and complements. The difference between coordination and subordination also intrudes on the syntax of Cum- and comitative phrases. Recall from section 2.4 that all coordination
DP and DP
195
that falls short of coordinating tensed clauses involves Right-Node Raising to some extent. That is, (593) is parsed as in (595), with a backward-looking Δ: (595) (Lifted Alia Δ (and (Marwan the piano)))) Whether a comitative phrase is adverbial or merely expresses another thematic relation, subordination precludes Right-Node Raising and the reconstruction of some Δ: (596) (Lifted Alia ((with Marwan) the piano)) The sentences in (593) and (594) have been fair paraphrases of one another: with appears to be so pliable94 that it contributes little to the meaning of the phrase, making the contribution of Cum transparent with slight dilution. What is common to the meaning of comitative constructions and coordination with Cum-phrases should derive from their common meaning and from the similar way they modify their host sentences.95 The divergences between these constructions will emerge, it can be hoped, from the differences in structure suggested by (593) and (595) on the one hand and (594) and (596) on the other. As with Lebanese partial agreement, the togetherness implied by comitative constructions depends on the nature of the events the sentence describes and the context of use. Despite its contextual dependence, the condition of togetherness has real teeth. McNally (1993, 376f.) and Camacho (2000, 368f.) observe that the togetherness is ultimately not defeasible and conclude that it must be a semantic entailment rather than a pragmatic implicature: Polish (McNally 1993) (597) Marek z Piotrem podniesli’ pianino, ale Marek podniosl je zanim zrobil M.nom with P.instr lifted.pl piano but ark lifted before did to Piotrek. it Peter ‘Mark with Peter lifted the piano, but Mark did it before Peter did.’ (598) *Marek z Piotrem podniesli’ pianino, ale naprawde každy zrobil to M.nom with P.instr lifted.pl piano but really each did it osobno. separately ‘Mark with Peter lifted the piano; in fact, each did it separately.’ Spanish (Camacho 2000) (599) *Con Marta fuimos al cine, pero no fuimos juntos ni a la misma película. ‘I with Marta went.1pl to the movies, but we didn’t go there together and not to the same picture.’
196
Chapter 2
(600) *Montagnier con Gallo descubrieron el virus del SIDA al mismo tiempo.96 ‘Montagnier with Gallo discovered.3pl the AIDS virus at the same time.’ If the speaker denies the hearer any understanding of the sense in which togetherness is intended, the sentences become anomalous. Expressing therefore a semantic condition on events, the comitative phrase merits a logical form, as in (594), where it applies to events (as Camacho concludes); in this respect, it is the same as the Cum-phrase of partial agreement (593). The semantic conditions that the comitative construction and Lebanese partial agreement impose are remarkably similar. With one striking, apparent exception, I have not found any divergence between them in their conditions of spatiotemporal proximity (togetherness) or eventivity (see note 84). The exception is the following: (601) Anna z Piotrem mają brązowe oczy. A.nom with P.instr have.pl brown eyes Out of the blue and without prompting, the comitative construction in (601) is highly anomalous, as McNally remarks (1993, 369). Brown-eyedness is, after all, neither eventive nor cooperative. It improves however if the context provides a relevant criterion for grouping. Imagine that the speaker comments on an arrangement of siblings or performs a sorting task the result of which puts Anna and Peter together and (601) is used to report that Anna with Peter were the only ones with brown eyes. It appears that surrounding circumstances including the speaker’s epistemic grounds for the assertion of brown-eyedness may be called on to satisfy the comitative conditions. In contrast, Lebanese partial agreement tolerates none of this despite desperate efforts to elicit some measure of approval for the likes of (601) in a variety of contexts: (602) *kent ’am fatteš be l-’ayleh ’ala l-mušerikin l-munesibin. kenet alia w marwan berS, fa ’melt t-tajrubeh ’layon. ‘I was searching through the family for the right subjects. Was.fs Alia and Marwan albinos. So I experimented with them.’ (603) kent ’am fatteš be l-’ayleh ’ala l-mušerikin l-munesibin. kenou alia w marwan berS, fa ’melt t-tajrubeh ’layon. ‘I was searching through the family for the right subjects. Were.pl Alia and Marwan albinos. So I experimented with them.’ If genuine, the contrast exposes a problem of acquisition, difficult as it is to imagine the relevant experience guaranteed to all speakers that would discriminate these contexts for the contrasting languages.
DP and DP
197
The problem is not so grim, however, since the juxtaposition of (601) and (602) overlooks a confounding difference in number agreement. As with all comitative and Cum-phrases in postverbal position, agreement in (602) is singular, but there is plural agreement with the preverbal phrase in (601). If, however, the preverbal comitative phrase also occurs with singular agreement, as in (604), then it too bars surrounding circumstances from satisfying the comitative conditions. The sentence, like (602), is unacceptable in all contexts, brown-eyedness being neither eventive nor cooperative: (604) *Anna z Piotrem ma brązowe oczy. A.nom with P.instr has.sg brown eyes 97 (Maria Bittner, p.c. ) Lebanese partial agreement and languages with the comitative construction agree ((602) and (604)) that the circumstances surrounding the reported event cannot satisfy the comitative conditions unless number agreement is plural.98 Moreover, since none of these languages allows plural agreement with a postverbal comitative or Cum-phrase, they also all agree that the remote circumstances cannot satisfy the comitative conditions imposed by these postverbal phrases. We advance on the acquisition problem if the differences between the languages can be reduced to more overt differences in the distribution of number agreement, comitative, and Cum-phrases. In Arabic, we have seen that it never permits partial agreement in its SVO word order (Aoun, Benmamoun, and Sportiche 1994): (481) *alia w marwaan Hamlet l-piano. *Alia and Marwan lifted.3fs the-piano That is, a preverbal Cum-phrase contributes obligatorily to number agreement. As hard as it is to tell apart preverbal coordination with a Cum-phrase from coordination with like thematic relations, it could just as well be universal that Cum-phrases may occur in preverbal position with obligatory number agreement. Of course Arabic has an overt comitative, too, and here is where the languages certainly differ. From preverbal position, the languages with the comitative construction allow a contribution to number agreement as in (587)–(589) or not, as in (605)–(607): (605) Marek z Piotrem podniosl pianino M.nom with P.instr lifted.sg piano ‘Mark with Peter lifted the piano.’
[Polish]
(607) Boris s Petej podnjal’[?] rojal’ B.nom with P.instr lifted.sg piano ‘Boris with Peter lifted the piano.’
[Russian]
(607) Marta con María levantó el piano ‘Martha with Maria lifted.sg the piano.’
[Spanish]
198
Chapter 2
The overt comitative phrases in Arabic never contribute to number agreement: (608) alia ma’ marwan Hamlet l-piano Alia with Marwan lifted.3fs the piano (609) *alia ma’ marwan Hamalo l-piano Alia with Marwan lifted.3pl the piano Beyond such gross differences, subtleties such as the contrasting contexts for (601) and (602) derive from a relation between grammar and meaning that is in fact invariant. The next section turns toward explaining the connection between number agreement and satisfaction of the comitative conditions. 2.6.0
Counting participants and their accomplices at the scene
It has so far been remarked that the comitative With-phrases, projected as they are from lexical morphemes, are not to be confused with the subatomic Cum-phrases, instances of which they may very well include. Furthermore, despite equivalent paraphrase and similar conditions of use, the comitative construction in (594) and the partial agreement construction (593) differ on all those points of logical syntax ((596) vs. (595)) that tell apart subordination and coordination as laid out earlier in section 2.4. But, as articulated as these structures have become, they are meant to be fully general: subordination with a comitative phrase is the same as subordination elsewhere. In particular, it should be recalled from chapter 1 that all prepositional phrases and their unpronounced kin, thematic roles such as Agent, Patient, and so on, describe events in which to participate at all is to participate in the manner described. Such phrases therefore describe different events, to be related to one another by yet other relations. Recall that the sentence (610) is false if no Columbia student sneered at any Columbia student. (610) The Harvard students and the Columbia students sneered at the Columbia students. Given the need demonstrated in chapter 1 to deploy a vague participation relation (“W”), these truth conditions are represented only if the sentence is taken to say that some event in which the Harvard students and the Columbia students particpated caused a different event, a being sneered at, in which the Columbia students particpated. To participate in an action is to participate in a different event from its effect, and thus the Columbia students’ participation in the latter does not count toward their participation in the former, which would have gone wrong if all that mattered was their participation in some way or another in the larger event comprising both action and effect. From such considerations, it was concluded that the logical form for (610) involves ‘W[e, the Havard students and the Columbia students]’ about an e and ‘At[e′, the Columbia students]’ about an e′. The formal resemblance to (594), with ‘W[e, Alia]’ about an e and ‘With[e′, Marwan]’ about an e′, follows
DP and DP
199
from the nature of subordination itself. We have yet to discover an effective difference beyond the lexical items themselves to tell apart comitative phrases from any other subordinate, oblique phrase. This formal resemblance among oblique phrases prompts a question made more poignant by the existence of languages where comitative phrases contribute to number agreement. Why doesn’t (611), analogous to (587)–(589), support either plural agreement or an anomalous interpretation that Echo and Narcissus whispered sweet nothings to Narcissus—that is, Echo and Narcissus, doing whatever they did in their different events, caused by their participating in these events sweet nothings to be whispered (to Narcissus)? (611) Echo to Narcissus whispered sweet nothings. Short on explanation, I will offer correlations. 2.6.0.0 The chains that bind accomplices
There are two other respects in which comitative phrases differ from other phrases. A fully general and universal property of their logic, whether or not they contribute to number agreement and whether or not they occur in construction with the subject, is that unlike other phrases, including their closest kin, instrumental phrases,99 they do not drop salva veritate. That is, (612), in contrast to (613), does not entail (614). (612) a. Russell wrote the whole of Principia Mathematica with Whitehead.100 b. Russell, with Whitehead, wrote the whole of Principia Mathematica. (613) a. Russell wrote the whole of Principia Mathematica with a Whitehead pencil. b. Russell, with a Whitehead pencil, wrote the whole of Principia Mathematica. (614) Russell wrote the whole of Principia Mathematica. This is unexpected if the occurrence of a comitative phrase contributes nothing more to logical form than a conjunct, “… & With(e, Whitehead) &. …”101 An unexpected scope interaction, exemplified in (615) and (616),102 also belies a simple conjunction for comitative phrases, the second respect in which they differ from other phrases: (615) a. Nora melted the chocolate with her lens with Willy Wonka. b. Nora, with Willy Wonka, melted the chocolate with her lens. (616) a. Nora melted the chocolate with Willy Wonka with her lens. b. Nora, with her lens, melted the chocolate with Willy Wonka.
200
Chapter 2
Nora uses her lens to focus sunlight on the chocolate, and in (615), Willy Wonka is an accomplice to melting the chocolate with her lens. Perhaps he steadied her hand. In (616), he is an accomplice to melting the chocolate but without any necessary connection to the lens. Perhaps he used a burner to heat the chocolate from below. It is as if With Willy Wonka scopes ‘melt chocolate with her lens’ in (615) but only ‘melt chocolate’ in (616). The contrast between (615) and (616) again defies the analysis of comitative phrases as simple conjuncts, and in fact Parsons (1990) offers the absence of such scope effects among locative and temporal adverbials as justification for the treatment of these other phrases as simple conjuncts. The meaning of a comitative phrase is to intervene in the causal or topological chain that holds together the subatomic events that a sentence describes. It is its universal characteristic to break a causal chain when dropped. Thus, according to the logical forms (617)–(620), no event in which Nora participates alone causes the chocolate to melt. Rather, a comitative event, joining her participation to Willy Wonka’s, intervenes between what she does and what happens to the chocolate, and only this larger, comitative event is said to melt the chocolate. As desired, no such logical form entails the sentence that omits the comitative phrase and reports Nora doing it on her own. Among these logical forms, the instrumental phrase with her lens may now be understood to apply to the larger, comitative event, as in (617) and (619), putting both Nora and Willy with the lens, or only to what Nora does, as in (618) and (620), leaving her alone with it.103 (617) ∃e∃e′∃e″∃e‴([℩x: Nx]Wi(e, x) & [℩E′ : [℩E : proi](O(e,e′) & [℩x: Wx]Cum(e′,x) & With(e,e′))] (∃xWj(e′,x) & [PL℩E′ : proj][℩E″ : prok](O(e′,e″) & (k∃xAgent(e″,x) & Cause(e″,e‴) & melt(e‴) & [℩x: Cx]Patient[e‴,x] & [℩x: Lx]Instrument[e″,x])))) ‘Nora participates, and her participation with Willy overlaps a melting of the chocolate using her lens.’ (618) ∃e∃e′∃e″∃e‴([ιx: Nx]Wi(e, x) & [℩E′ : [℩E : proi](O(e,e′) & [℩x: Wx]Cum(e′,x) & With(e,e′))] (∃xWj(e′,x) & [PL℩E′ : proj][℩E″ : prok](O(e′,e″) & (k ∃xAgent(e″,x) & Cause(e″,e‴) & melt(e‴) & [℩x: Cx]Patient[e‴,x] & [℩x: Lx]Instrument[e,x])))) ‘Nora participates, and her participation with Willy overlaps a melting of the chocolate with her using her lens.’
DP and DP
201
(619) ∃e∃e′∃e″∃e‴([℩x: Nx]Wi(e, x) & [SG℩E : proi][℩E′ : proj](O(e,e′) & (j ∃xAgentk(e′,x) & [℩E″ : [℩E′ : prok](O(e′,e″) & [℩x: Wx]Cum(e″,x) & With(e′,e″))] Cause(e″,e‴) & melt(e‴) & [℩x: Cx]Patient[e‴,x] & [℩x: Lx]Instrument[e″,x])))) ‘Nora participates, and it overlaps an action with Willy that melts the chocolate using her lens.’ (620) ∃e∃e′∃e″∃e‴([℩x: Nx]Wi(e, x) & [SG℩E : proi][℩E′ : proj](O(e,e′) & (j ∃xAgentk(e′,x) & [℩E″ : [℩E′ : prok](O(e′,e″) & [℩x: Wx]Cum(e″,x) & With(e′,e″))] Cause(e″,e‴) & melt(e‴) & [℩x: Cx]Patient[e‴,x] & [℩x: Lx]Instrument[e,x])))) ‘Nora participates, and it overlaps an action with Willy that melts the chocolate with her using her lens.’ Recall that the overt, verbal number agreement refers to what W-ers did in the highest event. If the comitative phrase should occur high enough (i.e., preceding the number morphology) so that Willy Wonka intervenes in what W-ing there was, the number agreement is plural, as in (617) and (618). Elsewhere number agreement will not reflect his presence, (619) and (620), although his intervention in the action is needed just the same to melt the chocolate.104 The meaning of the comitative phrase is in effect invariant, with number agreement recording what is a difference of scope. To understand the comitative phrase at all is to understand that some participant does not participate alone (with respect to whatever thematic relation is being modified) and thus to understand that what can be said of one with the other might not be true of the one without the other. If Nora with Willy melted the chocolate (and so she alone did not), it must be that only their joint action is the event causing the chocolate to melt, as in (617): (611) Nora with Willy have melted the chocolate. (617) ∃e∃e′∃e″∃e‴([℩x: Nx]Wi(e, x) & [℩E′ : [℩E : proi](O(e,e′) & [℩x: Wx]Cum(e′,x) & With(e,e′))] (∃xWj(e′,x) & [PL℩E′ : proj][℩E″ : prok](O(e′,e″) & (k∃xAgent(e″,x) & Cause(e″,e‴) & melt(e‴) …)))) If prior to modification, one had in mind a structure along the lines of (612), the result (613) still asserts that Nora participated but only describes effects accomplished with Willy’s mediation:
202
Chapter 2
(612) … Nora Wi(e, x) … [sg℩E : proi] … (O(e,e′) & ∃xAgent(e′,x) & Cause(e′,e″) …) (613) … Nora Wi(e, x) … with Willy … (∃xWj(e′,x) & [pl℩E : proj] … (O(e,e′) & ∃xAgent(e′,x) & Cause(e′,e″) …) For the comitative phrase to modify any thematic relation is, as it were, to reduplicate it,105 in which respect it resembles coordination. If W is modified, as it is in (613), the result of reduplication fixes number agreement. Other oblique phrases do not express causal or topological chaining. That is to say that a speaker knows that they do not subscribe to the logic of comitative phrases. Thus, if one should intrude between the subject and number agreement as in (614), it nevertheless fails to join Narcissus’s action to Echo’s; the only W-er in the causal action remains Echo alone and number agreement remains singular (615). (614) *Echo to Narcissus have whispered sweet nothings. (615) ∃e∃e′∃e″([℩x: Ex]Wi(e, x) & Echo [℩E″ : [℩E : proi](O(e,e″) & [℩x: Nx]To[e″,x]) To Narcissus [SG℩E : proi][℩E′ : proj](O(e,e′) & (j∃xAgent(e′,x) & Cause(e′,e″) & whisper(e″) …))) Whispers 2.6.0.1 A locality condition on the comitative phrases that count for number agreement
About the syntax of comitative phrases, my assumption has been that its internal syntax and semantics is invariant wherever it may appear in phrase structure, but there is some correspondence between phrasal position and the thematic relation modified. McNally 1993 observes that any phrase intruding between the subject and the comitative blocks the latter’s contribution to number agreement: (616) Anna segodnja s Petej ušla A.nom today with P.instr left.fs ‘Anna left today with Peter.’ (McNally 1993, 353)
[Russian]
(617) *Anna segodnja s Petej pridut A.nom today with P.instr come.3pl ‘Anna came today with Peter.’ (McNally 1993, 354)
[Russian]
I should like to think that the comitative in (617) is too far removed from the subject to modify the thematic relation, W, and thus in no position to affect who the W-ers are when number agreement comes to refer to what they did. To put on display a particular execution of the implied locality condition, let me first suppose that comitative phrases are rather more like adverbial clauses, with a
DP and DP
203
logical syntax that modifies via an accessibility relation (see section 2.3.1). That is, if if p,q is really ‘[∀e : if p(e)][∃e′ : R(e,e′)]q(e′)’ for some accessibility relation R,106 the logical syntax of ‘with DP XP’ should draw a parallel ‘[℩E : with-DP(e)] [℩E′ : R(e,e′)]XP(e′)’. If so, the logical form (617) on display for (611) is deficient, omitting an accessiblity relation: (611) Nora with Willy have melted the chocolate. (617) ∃e∃e′∃e″∃e‴([℩x: Nx]Wi(e, x) & [℩E′ : [℩E : proi](O(e,e′) & [℩x: Wx]Cum(e′,x) & With(e,e′))] (∃xWj(e′,x) & [pl℩E′ : proj][℩E″ : prok](O(e′,e″) & (k∃xAgent(e″,x) & Cause(e″,e‴) & melt(e‴) …)))) The second line is the event quantifier restricted by with Willy, and the third translates the XP have melted the chocolate. A formally correct logical form in (618) supplies O as the accessibility relation: (618) ∃e∃e′∃e″∃e‴([℩x: Nx]Wi(e, x) & [℩E′ : [℩E : proi](O(e,e′) & [℩x: Wx]Cum(e′,x) & With(e,e′))][℩E: O(e′,e)] (∃xWj(e,x) & [pl℩E : proj][℩E″ : prok](O(e,e″) & (k∃xAgent(e″,x) & Cause(e″,e‴) & melt(e‴) …)))) But honestly, the accessibility relation could just as well be identity, “… [℩E: e′=e] …,” my purpose being syntactic. In (619), there are three identical copies of the modified thematic relation, an identity embracing the event variables used: (619) ∃e∃e′∃e″∃e‴ ([℩x: Nx]W(e, x) & [℩E′ : [℩E : [℩x: Nx]W(e,x)](O(e,e′) & with Willy)][℩E: O(e′,e)] (∃xW(e,x) & have melted the chocolate))) This is as it should be. It was suggested in section 2.4 in a discussion of (328) that the phrase structure of a simple clause such as (328) or (611) stipulates that the enumeration of its lexical items contains no more than a single token of each thematic relation such as ‘Agent’, and each such thematic relation projects no more than one token of ‘W’. (328) *Boris Natasha have won $100. If so, two of the tokens in (619) duplicate exact copies of one original, and copying must preserve the alphabetic identity of event variables. Recall that section 2.3.1 invoked exact copying to explain why no interpretation of (298) allows that a woman’s laughter at one professor prompts or is accompanied by her child’s laughter at perhaps another professor. (298) DeHket kell mara w ebn-a ʕa estez linguistics laughed.3fs every woman and child-her at professor linguistics
204
Chapter 2
If the logic of a comitative phrase (i.e., its causal chaining) compels three tokens of the modified thematic relation, as I have argued, syntax steps in to demand their token identity. Introducing the accessibility relation in (618) renders a logical form that complies with the syntax of copying (cf. (617)), which in itself recommends this particular formulation. Now what happens to logical form if an adverb such as yesterday should intervene as it does in (620) between the subject and the comitative phrase? All the candidate logical forms for (620) prove to be flawed in one way or another: (620) *Nora yesterday with Willy have melted the chocolate (see (617)) (621) ∃e∃e′∃e″∃e‴([℩x: Nx]W(e, x) & [∃E′ : (i e′ ≤ e & yesterday(e′))] [℩E″ : [℩E′ : proi(e′)](O(e′,e″) & with Willy)][℩E: O(e″,e)] (∃xW(e,x) & have melted the chocolate))) (622) ∃e∃e′∃e″∃e‴([℩x: Nx]W(e, x) & [∃E′ : (i e′ ≤ e & yesterday(e′))] [℩E″ : [℩E : [℩x: Nx]W(e,x)](O(e′,e″) & with Willy)][℩E: O(e″,e)] (∃xW(e,x) & have melted the chocolate))) (623) ∃e∃e′∃e″∃e‴([℩x: Nx]W(e, x) & [∃E′ : (i e′ ≤ e & yesterday(e′))] [℩E″ : [℩E : e ≤ e′ & [℩x: Nx]W(e, x)](O(e,e″) & with Willy)][℩E: O(e″,e)] (∃xW(e,x) & have melted the chocolate))) In accord with the logical syntax of subordination (section 2.4), any nested event quantifiers restrict in some fashion the domains of their immediate successors. Thus, the events in the domain of yesterday are restricted to those antecedently introduced and restrict those in the domain of subsequent event quantifiers, with Willy, in particular. In (621), where proi is anaphoric to the events yesterday, the comitative phrase is not represented as modifying a thematic relation, although some such modification is independently required: (624) *Yesterday the chocolate melted with Willy. (625) *Yesterday the chocolate was melted with Willy. Cf. Yesterday the chocolate was melted by Nora with Willy. (626) *Yesterday it rained with snow. In contrast, (622) represents the modification and respects exact copying, but now yesterday fails to restrict the domain of the next event quantifier. The logical form (623) manages all of these constraints—with Willy modifies a thematic relation, the tokens of the thematic relations are exact copies of one another, and event quantifiers restrict their successors—but deviates elsewhere. The content of the pronoun ‘[℩E : pro(e)]’ connecting with Willy to antecedent events is ‘[℩E : e ≤ e′ & [℩x: Nx]W(e, x)]’. The pronoun is thus anaphoric to both the immediately superior event quantifier, yesterday, and more remote structure. So far, it has sufficed for an
DP and DP
205
event quantifier to inherit its domain restriction only from its immediate superior; (623) would breach what could otherwise be considered a regular, sytematic property of nested event quantifiers. The intervention of yesterday between the subject and the comitative phrase thus appears to be inconsistent on independent grounds with the plural number agreement. Number agreement refers to what the W-ers did. When it is singular (as in (616)), the comitative accomplice is not among them, and it must be that the comitative phrase modifies the Agent thematic relation or a lower one:107 (627) Nora yesterday with Willy has melted the chocolate (cf. (616)). (628) ∃e∃e″ ∃e″″([℩x: Nx]Wi(e, x) & [SG℩E : proi] [∃E′ : (i e′ ≤ e & yesterday(e′))] [℩E″ : proj](O(e′,e″) & (j ∃xAgentk(e″,x) & [℩E‴ : [℩E″ : prok](O(e″,e‴) & with Willy)][℩E″: O(e‴,e″)] (∃xAgentk(e″,x) & Cause(e″,e″″) & melt(e″″) …))))
Nora.sg108 yesterday with Willy melted …
However the locality condition is derived, it remains a fact that comitative phrases contribute to number agreement only when placed high in the phrase structure. A comitative phrase, like any other PP, shares the logical syntax that broadly characterizes subordination (section 2.4), but comitative phrases modify thematic relations, intervening in the description of a causal or topological chain, in a way that clearly tells apart Nora with Willy from Echo to Narcissus. Only if the comitative phrase modifies the highest ‘W’ does it affect number agreement, and it must be close to that ‘W’, so it appears, to modify it. Elsewhere in phrase structure, the comitative phrase will modify some other thematic relation (see note 104). Modifying Agent, as in (628), derives an interpretation that is almost identical (616), without letting the accomplice count for number agreement. 2.6.1
Number agreement and comitativity
We can now make quick work of the connection between number agreement and satisfaction of the comitative conditions. Recall that remote circumstances surrounding the reported event cannot satisfy the comitative conditions unless number agreement is plural. Out of the blue, the comitative construction in (601) and (604) is anomalous, brown-eyedness being neither eventive nor cooperative: (601) Anna z Piotrem mają brązowe oczy. A.nom with P.instr have.pl brown eyes (604) *Anna z Piotrem ma brązowe oczy. A.nom with P.instr has.sg brown eyes The construction (601) in the plural improves, however, if the context provides a relevant criterion for grouping Anna with Peter, such as a sorting task that has put
206
Chapter 2
them together as the only ones with brown eyes. In this case, surrounding circumstances such as the speaker’s epistemic grounds for the assertion of brown-eyedness are called on to satisfy the comitative conditions. But surrounding circumstances cannot rescue the singular construction (604), which is unacceptable in all contexts where brown-eyedness is neither eventive nor cooperative. Because it contributes to number agreement in (601), the comitative phrase modifies ‘W’ (cf. (518)), and because it does not in (604), it modifies not ‘W’ but some lower thematic relation instead (e.g., Possessor, cf. (628)). That is, number agreement locates the comitative phrase on opposite sides of the causal/topological relation O: (629) … W(e,x) & … & O(e,e′) & … & Possessor(e′,x) … In the plural (601), the comitative phrase is applied to e in (629), and it is applied to e′ in the singular (604). Now recall that section 2.5.1 (and section 2.5.2) found that O puts some distance between the perspectival event e and the event e′ that it is a presentation of. Although partial agreement in (574) expresses some asymmetry between Alia and Marwan, it was observed that the meeting and their behavior during the meeting can be entirely symmetric, provided that the asymmetry can be located in the events that frame or bring about the meeting: (574) lta’yet alia w marwaan Met.3fs Alia and Marwan Likewise, if the comitative phrase is applied to the higher, perspectival event e in (629), it is this event that meets the comitative conditions, as it does when it is the presentation of brown-eyedness rather than brown-eyedness itself that puts Anna and Peter together. If, instead, the comitative phrase applies to the state of brown-eyedness, it describes an impossible condition and the singular (604) is unacceptable. This connection between number agreement and the comitativity conditions is, or so I claim, invariant. To know the distribution of number agreement and comitative phrases in one’s language is to know what circumstances will satisfy the comitative phrase. There remains, however, a brute fact to be learned: some languages such as Slavic and Spanish allow the construction where a comitative phrase contributes to number agreement and some, such as English and Arabic, do not (cf. note 98). Some languages, in present terms, appear to bar comitative phrases from positions in phrase structure where they can both modify W and antecede the overt number agreement morphology. This point of grammar may have to be learned as such if it does not follow from other differences in phrase structure (such as verb placement). Either way everyone knows a priori what such phrases would mean both when they contribute and when they do not contribute to number agreement, with such differences reduced to a matter of scope.
DP and DP
2.7
207
Summary
Argument and data have run thick, but to the grammar of and, which means “&,” I have only added that the lexical item is opaque to complementation, and in this way coordination is distinct from subordination (section 2.4). Any bound morpheme such as the thematic relation W finds its complements only within its coordinate, within Δ: DP W Δ and DP. … Since and is univocal and sentential, every conjunct contains at least a thematic relation, but being also opaque to complementation, each conjunct must contain quite a bit more, and thus Right-Node Raising is the rule. The semantics of Right-Node Raising derives from the subatomic pronouns that according to chapter 1 refer to events and are the stuff that holds together sentences quite generally. These pronouns, like other descriptive anaphora, copy their content from their antecedents. Even so, identical tokens of an event description, if they fall within the scope of different event quantifiers, may describe different events and thus derive a “sloppy” or distributive interpretation. As we have seen throughout section 2.3.0, the distribution of “sloppy”, distributive and “strict” collective interpretations for Right-Node Raised constituents probes the locations of event quantifiers throughout the clausal structure and shows the effects of variation in both word order (pre- vs. postverbal subjects, adverb placement, and so on) and the choice of quantifier (distributive quantifiers vs. (in)definite descriptions).
3
PredP and PredP: Of Subjects and Ancient Grievance
Chapter 2 rescues the logical syntax of DP coordinations, as in Donca and Mirçea met, from Aristotle’s fallacy, which takes the conjunction to be other than the sentential ‘&’. Although DPs happily serve up citation forms for puzzling out the meaning of and, the more populous field of grammatical study lies in predicative coordination, where we find gapping and other elliptical constructions, nonconstituent coordination, symmetric and asymmetric extractions and reconstruction effects, and crosslinguistic variation throughout. If a fog does not roll in, there should be plenty of beacons by which to navigate the structure of coordination among PredPs. The thesis that and is always and only a sentential connective precludes coordination of unsaturated expressions, anything less than a sentence. Appearances to the contrary, the coordination of PredPs must be an illusion of subjects going unpronounced, as it turns out, sometimes by movement (sections 3.1, 3.2.1.3, 3.5) and sometimes, null anaphora (sections 3.3, 3.4). But if, as supposed, an unspoken subject is ever a copy of its antecedent, runaway translation threatens (1) with a meaning equivalent to (2)—a threat that spells imminent doom if logical form is as impoverished as (3) and (4) pretend: (1) A rocker shimmied and shook. (2) A rocker shimmied, and a rocker shook. (3) [A(n) x: rocker(x)] ((x shimmied) and (x shook)) (4) [A(n) x: rocker(x)] ((x shimmied) and [A(n) x: rocker(x)](x shook)) There is in that case no way to copy the antecedent from the first conjunct into the second without (1) and (2) meaning the same. For the timid Davidsonian, however, or for anyone allowing meager place for events, copies of the subject in both conjuncts do not imply the equivalence of (1) and (2), provided their event quantification distinguishes them as in (5) and (6): (5) [A(n) x: rocker(x)] ∃e ((x shimmied(e)) and [A(n) x: rocker(x)] (x shook(e))) (6) [A(n) x: rocker(x)] ∃e ((x shimmied(e)) and [A(n) x: rocker(x)] ∃e (x shook(e)))
210
Chapter 3
Even so, it should be small comfort that (1) escapes meaning what (2) does if it ends up implying, like (5), that the shimmying is the shaking. It does not contradict any meaning for (1) that no sinuous movement—shimmying—is also its opposite, a saccadic movement, shaking. To recognize that a shimmy is no shake and still to reason as above requires further a decomposition at least as rich as in (7) and (8): (7) [A(n) x: rocker(x)] ∃e ((Agent(e, x) & ∃e′(Cause(e,e′) & shimmied(e′))) and [A(n) x: rocker(x)] (Agent(e, x) & ∃e′(Cause(e,e′) & shook(e)))) (8) [A(n) x: rocker(x)] ∃e ((Agent(e, x) & ∃e′(Cause(e,e′) & shimmied(e′))) and [A(n) x: rocker(x)] ∃e (Agent(e, x) & ∃e′(Cause(e,e′) & shook(e)))) According to (7), (1) says twice over—once in each clause—that a rocker was the agent in the same event, a cause with two different effects, a shimmy and a shake. The shimmier is indeed the shaker, as any meaning for (1) must imply, although her shimmies are not her shakes. In contrast, according to (8), (2) speaks of (perhaps) different causes, and to say that the agent of each was a rocker is to imply nothing about whether these rockers were two or the same one. It requires Eventish supermonadicity if the difference in meaning between (1) and (2) is to reduce to just a reflection of their event quantification. As a corollary, copying nominal content from an antecedent does end up with a different denotation, a “sloppy” (aka “nonrigid”) interpretation, if as in chapter 2 and below in the discussion of reconstruction (sections 3.1, 3.2.1.3, 3.5), the copies end up in the scope of different event quantifiers. It’s all about the grammar of event quantification and the distribution of event quantifiers. Supermonadicity is essential to this perspective, so as not to confound shimmies and shakes and more generally to allow it to be asked of any and every nominal copied from within one conjunct to the next, as in (9), whether the θ-ing is the same and thus has the same NP θ-ing in it, without confound for any other events described in the sentence: (9) [A(n) x: NPi(x)] … θj(e, x) … and … [A(n) x: NPi(x)] … θj(e, x) … Now if event quantification is all there is to tell apart (1) and (2), what about it is contingent on the mere pronunciation of the second subject? As advertised in chapter 1, deployed passim and argued for in chapters 8–16, adverbialization interposes a denominal adverbial between a DP and its scope: (10) The man surpassed the boy of his childhood in regrets and not much else. ‘The man while a man surpassed the boy while a boy …’ In all the sentences alleged to exemplify adverbialization, it is seen applied to pronounced DPs, without comment on those unpronounced. This chapter subjects adverbialization to the thesis in (11) so that (1) and (2) come to be read respectively as in (12) and (13): (11) Pronunciation cues adverbialization; silence—resumption.
PredP and PredP: Of Subjects and Ancient Grievance
211
(12) A rocker while a rocker shimmied, and still there a rocker shook. (13) A rocker while a rocker shimmied, and a rocker while a rocker shimmied. To leave a subject unspoken signals the intention to continue commentary on the action already under discussion. The spoken subject cues an adverbial, event quantification that shifts attention elsewhere.1 The fact that (1) and (2) do not mean the same leads in short order from the thesis that and is always a sentential connective to supermonadicity and a translation into Eventish regimented according to (11). Yet, (11) should not be dismissed as an artifact of an unwelcome thesis about and. It will stand on its own once it is recognized that (14) and (15) do not mean the same, with little of logical form to tell them apart except what translation into Eventish according to (11) affords: (14) Kunstler is sitting and is standing. (15) Kunstler is sitting and Kunstler is standing. (16) Kunstler while Kunstler is sitting and still there is standing. (17) Kunstler while Kunstler is sitting and Kunstler while Kunstler is standing. It can then be no surprise that the apparatus—event quantification, Eventish, and (11)—called on independently to make sense of (14) and (15) forges the difference it does between (1) and (2). As just remarked, and, always the sentential connective, precludes coordination of unsaturated expressions, anything less than a sentence. Appearances to the contrary are an illusion of subjects going unpronounced: (18) No rocker will have ever been shimmying and shaking (to any Mozart minuet). (19) a. *No rocker will have ever been shimmying and he will have ever been shaking (to any Mozart minuet). b. *No rocker will have ever been shimmying and any of his mamas will have ever been shaking (to any Mozart minuet). It therefore cannot be said in explanation of the contrast between (18) and (19) that unsaturated predicates, shimmying and shaking, are coordinated in (18) and fall within the scope of no rocker. Rather, whatever puts the pronounced subject in (20), any of his mamas, in the scope of no rocker does the same for the unpronounced subject in (18): (20) No rocker will have ever been shimmying and any of his mamas shaking (to any Mozart minuet). (21) *No rocker will have ever been shimmying and will have ever been shaking (to any Mozart minuet). And, moreover, whatever keeps the pronounced subjects in (19) outside the scope of no rocker does the same for the unpronounced subject in (21). Always the
212
Chapter 3
sentential connective, and nevertheless connects sentences varying in size, from the smallest participial clauses, (shimmying(e) and shaking(e)) in (18) to the fully tensed clauses in (19) and some sizes in between. The variation in size puts the squeeze on various grammatical processes, quantifier scope among them (section 3.2). With so small a coordination as that of the participial phrases in (18) and (20), there is room enough outside it for no rocker to find a perch that includes within its scope both conjuncts. In contrast, the largest coordinations of fully tensed clauses in (19) and (21) cage no rocker within the first conjunct. In between, there is another grade of scopal involvement. The smallest coordinations allow both the pronounced first subject and its adverbialization to include within their scope the entire coordination, the largest confine both the subject and its adverbialization to the first conjunct. The intermediate allow the nominal quantifier the higher scope while confining its adverbialization to the first conjunct: (22) No rocker while a rocker (sometimes Φ and sometimes Ψ) (23) No rocker (while a rocker sometimes Φ and sometimes then Ψ) (24) No rocker while a rocker sometimes Φ, and sometimes then he Ψ The unspoken event pronoun, then in (23) or (24), or subject pronoun, he in (24), that fails to fall within the scope of its antecedent is then subject to the vagaries and conditions that attach to unbound descriptive anaphora to rescue them under certain conditions (see discussion of “telescoping,” passim). These cursory remarks about the size of coordination and adverbialization according to (11) already suffice to determine that no two sentences of (25)–(28) have the same structure: (25) (26) (27) (28)
Kunstler Kunstler Kunstler Kunstler
is is is is
sitting sitting sitting sitting
and standing. and Weinglass standing. and is standing. and Kunstler/he is standing.
Just as well, since it will be seen that no two of these sentences are synonymous. But it leads straightaway to a violation in (26) of the Coordinate Structure Constraint in the asymmetric movement of the subject out of the first conjunct, given that it properly contains a coordination of participial clauses: (29) [Kunstleri is [[ti sitting] and [Weinglass standing]]. The analysis thus corroborates Johnson’s (2009) discovery of yet more apparent exceptions to this beleaguered constraint. A further empirical finding is also already implied, in the contrast between (27) and (21): (30) *[Kunstleri [[ti [is sitting and is standing]]. (31) *[Kunstleri [[ti is sitting] and [ti is standing]]. (32) [Kunstleri is sitting] and [proi is standing].
PredP and PredP: Of Subjects and Ancient Grievance
213
Sentence (27) is derived neither from a nonsentential predicative coordination (30) nor by an across-the-board movement of its subject from within both conjuncts (31), since either would license (21). It must rather be that an unspoken pronoun (32)— not a trace left behind by movement—refers to Kunstler, like the overt pronoun in (28), without falling within the scope of the first subject, safely precluding (21) and (19). This same will be said about an unspoken event pronoun in the second clause. These are unspoken, coordinative pronouns (section 3.3) that exist only in coordinative structures—‘is standing’ is not itself a sentence of English—which corroborates earlier findings in English and German (Schwarz 1999; 2000, chap. 3), in Icelandic (Rögnvaldsson 1982, 1990; Bresnan and Thràinnson 1990), and in Japanese (Tomioka 1993). The size of coordination squeezes quantifier scope, as above in the contrast between (18)/(20) and (19)/(21). It squeezes other interpretive processes too (section 3.4): (33) A rocker has shimmied on every bump and shaken on every grind. [A rockeri has [[ti shimmied on every bump] and [ti shaken on every grind]]. (34) A rocker has shimmied on every bump and has shaken on every grind. [A rockeri has shimmied on every bump] and [proi has shaken on every grind]. The reconstruction (Fox 1995, 2000) that affords (33) an interpretation in which every bump and every grind has its own rocker performing it undoes an acrossthe-board movement for which there must be room enough outside the coordination. A coordination so large that it blocks movement and resorts to a null coordinative pronoun, as in (34), means only that the same rocker executes all dance steps. Recall from chapter 1 that the disjunctive interpretation of the coordination in (35) dividing the rockers between shimmiers and shakers is the effect of two unspoken plural event pronouns—one to refer to the shimmying and shaking, another to refer to the twenty rockers’ action—and a relation, causative or another, to relate them, for which there must be room enough to host it all outside the coordination: (35) Twenty rockers will have been shimmying and shaking. [∃X : 20 rockers(X)] … ∃Ei Agent[Ei,X] … [℩Ei: proi][ ℩Ej,k: proj,k] Cause[Ei,Ej,k] … [∃Ej shimmying[Ej] and ∃Ek shaking[Ek]] (36) Twenty rockersi will have been shimmying and proi will have been shaking. If the coordination is too large to leave enough room outside it, the disjunctive interpretation is lost as in (36), which says that twenty rockers shimmy and the twenty also shake, as expected if a null coordinative pronoun occupies the subject position of the second clause (section 3.4 and chapter 6).
214
Chapter 3
In the course of this chapter and empirical investigation, I take on board a grammar of coordination that goes beyond the thesis that and is always a sentential connective and logical form, an expression in Eventish characterized by supermonadicity, descriptive event anaphora, and adverbialization. As above, it includes (11), a principle of translation into logical form, grammar for the distribution of event quantifiers, the postulated null coordinative pronouns, Johnson’s (2009) asymmetric extraction from coordinate structures, and constraints on movement and reconstruction that relate quantifier scope to the size of the clause host. Also, copying, that engine spelling out the content of descriptive anaphora and Right-Node Raising (see chapter 2), proves here to be exact down to the detail of the alphabetic identity of variables, hinted at in a pair of sentences: (37) Sam repainted and Joe re-sided a new house. (38) Sam repainted and then Joe re-sided a new house. The temporal relation, then, in (38) creates a contrast with (37), requiring of (38) alone that Sam repaint and Joe re-side the same new house. The interpretation of (37) that allows Sam and Joe a different house involves the “sloppy” interpretation of identical copies of the same phrase, ‘[A(n) x : new house(x)]Patient(e,x),’ falling within the scope of different event quantifiers, to describe a repainting in the first conjunct and a re-siding in the second. But to assert a temporal relation, ‘then(e,e′)’, between them, that the re-siding is after the repainting, requires their description in distinct variables, e and e′, which precludes the “sloppy” interpretation of exact copies in the same variable. Given clauses of arbitrary size and internal structure, the above foreshadows a grammar for the quantification, scope, ellipsis, and descriptive anaphora that adorn them. There is also something to be said about the internal phrase structure of clauses and the morphosyntax of translation into logical form. With Eventish enlarged with unpronounced vocabulary and Conjunction Reduction driving translation helter-skelter, the simple sentence (39) has an unknown parentage among (40): (39) James Brown will have been shimmying and shaking. (40) i. ii. iii. iv.
[James Brown W- O- will W- O- have W- O- been W- O- -ing W- OAgent- Cause- [shimmy and shake]]. [James Brown W- O- will W- O- have W- O- been W- O- -ing W- OAgent- [Cause- shimmy and Cause- shake]]. [James Brown W- O- will W- O- have W- O- been W- O- -ing W- O[Agent-Cause- shimmy and Agent- Cause- shake]]. [James Brown W- O- will W- O- have W- O- been W- O- -ing W[O- Agent-Cause- shimmy and O- Agent- Cause- shake]].
PredP and PredP: Of Subjects and Ancient Grievance
215
v.
[James Brown W- O- will W- O- have W- O- been W- O- -ing [W- O- Agent-Cause- shimmy and W- O- Agent- Cause- shake]]. vi. [James Brown W- O- will W- O- have W- O- been W- O- [-ing W- OAgent-Cause- shimmy and -ing W- O- Agent- Cause- shake]]. vii. [James Brown W- O- will W- O- have W- O- been W- [O- -ing W- OAgent-Cause- shimmy and O- -ing W- O- Agent- Cause- shake]]. … xvii. [James Brown [W- O- will W- O- have W- O- been W- O- -ing W- OAgent-Cause- shimmy and W- O- will W- O- have W- O- been W- O-ing W- O- Agent- Cause- shake]] xviii. [[James Brown W- O- will W- O- have W- O- been W- O- -ing W- OAgent-Cause- shimmy] and [James Brown W- O- will W- O- have W- O- been W- O- -ing W- O- Agent- Cause- shake]] Similarly, consistent with the thesis that and is always a sentential connective, (41) may descend from any of (42): (41) Ike and Tina will have been shimmying. (42) i.
[[Ike W-] and [Tina W- O- will W- O- have W- O- been W- O-ing W- O- Agent- Cause- shimmy]. … xvii. [Ike W- O- will W- O- have W- O- been W- O- -ing W- O- AgentCause- shimmy] and [Tina W- O- will W- O- have W- O- been WO- -ing W- O- Agent- Cause- shimmy].
Although it is not native to Eventish (chapter 4), I will import the traditional notion of complementation said to hold between morphemes and their phrasal complements. Whether principle or merely robust empirical generalization, it is hardly controversial that if there is anything to the notion that YP is the grammatical complement to X, it means that X embedded within one conjunct cannot find its so-called complement in the next conjunct over: (43) *[… X …] and YP *[… X …] and [… YP …] (44) [… X YP …] and YP [… X YP …] and [… YP …] Next, whatever skepticism attaches to the existence of the unspoken vocabulary of thematic, causal, and topological relations, if they do exist, there is not much controversy that, say, ‘Cause’ would enter into a relation of grammatical complementation with its complement. If so, as already observed in chapter 2, the least expression, ‘W’ in (42), introduced so that and coordinates sentences, will itself demand to find its complements within the first conjunct, thus implying the Right-Node Raising discussed throughout chapter 2, implying in effect that (41) must be parsed as in
216
Chapter 3
(42xvii). Given that and is always a sentential connective, grammatical complementation largely determines the structure of an alleged DP-coordination. The variation in PredP-coordination seen in (40) falls into two cases. The relation between freestanding auxiliary verb and participial phrase, be+shimmying, is as canonical an instance of grammatical complementation as there is. Complementation in this case does not seem to be disrupted under coordination of the latter, an observation commonly encoded in the category of the resulting coordination: ‘[be [PartP[PartP shimmying] and [PartP shaking]]]’. In response to the empirical findings of this chapter, I will suppose that parsing is minimized, so that the smallest parse consistent with the requirements of grammatical complementation is the only parse. What is pronounced “… been shimmying and shaking” is therefore always ‘[been [PartP[PartP shimmying]’ and ‘[PartP shaking]]]’and never ‘*[been [PartP shimmying]] and [been [PartP shaking]]’ with an unpronounced token of the auxiliary verb in the second conjunct. In the second case seen in (40), the morpheme to be related to its phrasal complements is a bound, verbal morpheme, -ing, -en, Cause-, O-, Agent-, or Tense, which coordination cannot separate from its complements, a general fact about bound morphology: (45) *The surgeons will co-(laborate and operate) in the OR. As Lin (2000) notes, this consideration implies that (51) must be parsed with two tokens of the past-tense morpheme, within the conjuncts so that none stands outside the coordination. Note that the morpheme ‘W’ is not classified as a bound verbal morpheme.2 Given its role in deriving the disjunctive interpretation of coordinated PredPs (see chapter 1 and below), it must stand outside. Grammatical complementation, parsing minimization, and the classification of freestanding vs. bound verbal morphology render deterministic the parses for (39) and (41), even with Conjunction Reduction run amok. These aspects of verbal grammar join (11) about adverbialization to determine, as remarked earlier, that no two sentences of (46)–(49) have the same structure, differing in at least the presence of a denominal adverbial or an auxiliary verb, with the result, as will be seen, that no two are in fact synonymous: (46) (47) (48) (49)
Kunstler Kunstler Kunstler Kunstler
is is is is
sitting sitting sitting sitting
and standing. and Weinglass standing. and is standing. and Kunstler/he is standing.
These aspects of verbal grammar also determine that no two sentences of (50)–(52) share a logical form: (50) James Brown did shimmy and shake. (51) James Brown shimmied and shook. (52) James Brown did shimmy and did shake.
PredP and PredP: Of Subjects and Ancient Grievance
217
The logical form for (50) contains one temporal quantifier, corresponding to the solitary tense morpheme appearing there, and (51) and (52) in contrast each contain two temporal quantifiers, one in each conjunct, the two tense morphemes. The conjuncts in (52) swell further with auxiliary verbs. The coordinations in (50)–(52) are a series strictly increasing in size: the conjoined bare verbs are clauses smaller than the simple tensed verbs, which are in turn clauses smaller than the conjoined tensed auxiliary verb phrases. The chapter surveys the effects on meaning of such variation in the logical syntax of coordination. “Generative Semantics Meets Minimalism” names the relationship implicit above between logical form and the syntax of natural language. There is no meaning without representation—no operators that are “in the semantics” but not in logical form—and no representation without morphological weight.3 The syntactic dilemma from section 1.5.1 (see also section 2.0.1) puzzles over a relationship between the scope dependency in (53)–(54) and number agreement: (53) *No rocker and any roller has ever grooved to a funky disco beat. (54) No rocker and any roller have ever grooved to a funky disco beat. Because there are two, rocker and roller, the semantics requires for the singular agreement in (53) the intervention of a distributive operator. If so, the clauses conjoined in (53) are strictly larger than those conjoined in (54) absent the distributive operator—large enough then that no rocker is caged within (53)’s first conjunct, unable to include any roller within its scope. Similarly, given the syntactic and morphological weight of all that is represented in logical form, the Participial Phrase in (56), with subject pronounced in a Case position and with a denominal adverbial absent from (55), is strictly larger than the Participial Phrase with unpronounced subject: (55) Diego has stirred a mojito and shaken a margarita. (56) Diego has stirred a mojito and Frida shaken a margarita. The strictly larger Participial Phrase is indicated by the scope of simultaneously in the contrast between (57) and (58) (see Jackendoff 1971; Johnson [1996] 2003): (57) Diego has simultaneously stirred a mojito and shaken a margarita. DPi Tense Aux W simultaneously O [[-en V] and [-en V]] (58) * Diego has simultaneously stirred a mojito and Frida shaken a margarita. DPi Tense Aux [[ti W simultaneously O -en V] and [DP W O -en V]] The Participial Phrase with a pronounced subject is so much larger that it entraps the adverb, which then cannot express that Diego’s stirring and Frida’s shaking were simultaneous. Throughout, clauses with pronounced subjects are found to be systematically larger than their counterparts without them and are thus vulnerable to
218
Chapter 3
restrictions on their meaning that the clauses without pronounced subjects are immune to (section 3.2). From the slippery slope to Conjunction Reduction in chapter 1 to its analysis in chapter 2, all apparent DP-coordination proves to be extreme Right-Node Raising from conjuncts joined by a sentential and. Collective reference is just an illusion of collectivized Right-Node Raising and the embedded plural event pronouns cloaking their plural reference to events in silence. It is the same for the disjunctive or collective interpretation of predicative coordination. The burden of collective reference is divorced from the narrow semantics of the connective and and from the evaluation of the truth-functional Φ and Ψ . It rests on plural event pronouns referring to the Φ-ing and Ψ-ing. This introduction ends with two advertisements that plural event pronouns are indeed the effective devices for collective reference. The first (see section 3.5) is just the syntactic puzzle posed by reconstruction with collectivized Right-Node Raising, where (59) comes to mean what (60) does: (59) Some libertarians to hisses & boos and some conservatives to wild applause were likely to clash at every state Republican caucus. (60) It is likely for there to be at every state Republican caucus some libertarians to hisses & boos and some conservatives to wild applause clashing. If some libertarians to hisses & boos and some conservatives to wild applause is to lower into the scope of likely and every state Republican caucus, what constituent is this that moves thus, let alone seeming to lower into the phrase that has been RightNode Raised from inside it? There is no way around an illicit movement if the coordinated phrase must be the subject of the collective clash. But, according to chapter 2, collectivized Right-Node Raising allows for two tokens of the Right-Node Raised phrase, one in each conjunct, without loss to the collective meaning of the tokens so instantiated. If so, entirely within the first conjunct of (59), some libertarians and to hisses & boos may lower into the scope of likely and every state Republican caucus as these phrases are tokened there, and similarly for some conservatives and to wild applause in the second conjunct. The second advertisement is as much a warning that the plural event pronouns that hold the Eventish sentence together are so good at being pronouns that they undergo movement—lowering (section 3.2.1.4) and clitic climbing (chapter 6). 3.0
Of subjects and ancient grievance
Before turning to novel considerations, let me plant a flag where the structure of coordinated PredPs has already been fixed: because and is always a sentential connective, it cannot be flanked by unsaturated expressions, and so any predicate that occurs within a coordination occurs there with its full complement of arguments,
PredP and PredP: Of Subjects and Ancient Grievance
219
pronounced or unpronounced. As far as subjects are concerned, this amounts to the VP-internal subject hypothesis4 in a Davidsonian setting, at least where the conjuncts involve their subjects in distinct thematic relations: (61) Alvy did exercise and faint after complaining of chest pain. … ((… Agent(e,x) exercise(e) …) and (… Experiencer(e,x) faint(e) …))) The special case where the thematic relation is the same leaves open the possibility that factoring it out of the coordination may remove the subject altogether from within, as in (62a): (62) James Brown did shimmy and shake. a. … Agent(e,x) (shimmy(e) and shake(e)) b. … ((… Agent(e,x) shimmy(e)) and (… Agent(e,x) shake(e))) Note that (62a) does not coordinate unsaturated expressions. Although there happens to be no place for the subject, the coordination is fully sentential. Grammar will, however, shortly insist on a uniform structure so that (62a) gives way to (62b) in complete compliance with the VP-internal subject hypothesis. This special case and point of grammar aside, it remains that the logical syntax of and on its own entails that coordinated PredPs almost always contain their subjects. Any evidence to the contrary reaching beyond the special case of a shared thematic relation speaks directly against the thesis sine qua non.5 Where conjuncts without pronounced subjects differ from those with, ancient grievance against Conjunction Reduction has asserted there is no subject and and is therefore not sentential, joining predicates instead. So, let’s canvass what happens in subject positions to subjects pronounced and unpronounced. 3.1
Scope and reconstruction into subject position
Start with the observation that (63) and (64) do not mean the same thing, nor do (65) and (66): (63) All night long, a rocker has shimmied and (has) shaken. (64) All night long, a rocker has shimmied and a rocker (has) shaken. (65) Ernest has stirred and shaken a mojito. (66) Ernest has stirred a mojito and shaken a mojito. Failing synonymy, it follows without qualification that (63) and (64) differ in their logical syntax. There is between them more than a mere difference in what is spoken and unspoken, and similarly, more than pronunciation must tell apart (65) and (66). One approach to the difference in meaning leaps to articulate it as a difference in the type of coordination. It looks as if the phrases coordinated in (63) and (65)
220
Chapter 3
are smaller than their counterparts in (64) and (66). It could be that there simply isn’t enough room in (63)’s second conjunct for another (tacit) token of a rocker, and not enough in (65)’s first conjunct for another (tacit) token of a mojito. If all there is room for is a variable, then perhaps this observation betrays what is a coordination of predicates rather than of clauses. The gross structure of what is coordinated cannot however be what tells (63) and (64) apart, as Fox (1995, 2000) demonstrates with a direct argument comparing the likes of (67) and (68): (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. Note first that it is given prima facie that whatever is coordinated (maybe even as little as a pair of participial phrases), it is the same in both sentences. The sentences, after all, differ only in exchanging the and every. Observe next that (68) has an interpretation that plainly betrays the presence of a token of a rocker in the second conjunct, namely, that interpretation where the two tokens of a rocker fall within the scope of the universal quantifiers: for every bump, there was a rocker who shimmied through it, and on every grind, a rocker who shook. But, if coordination is the same in (67) and (68), there must then be room enough in (67) as well for a second, unspoken, token of a rocker as in (69): (69) All night long, a rocker has shimmied through the bumps and a rocker shaken on the grinds. Granted that, as before, (67) does not mean the same thing as (70), the force of Fox’s argument is that there is no difference in the type of these coordinations to explain their difference of meaning, for which we must look elsewhere. (70) All night long, a rocker has shimmied through the bumps and a rocker shaken on the grinds. The contrast in meaning illustrated in (63)–(66) and the larger class of empirical problems from which it is drawn, concerned with scope reconstruction effects and “sloppy” interpretations, dictate that there be some difference in logical syntax, and any study of coordinate predicative phrases must attend to it. Yet it need not be a difference of phrasal type. A logical form is not yet fixed until it has been fitted with event quantification, cross-reference to events, and the relations that mediate between them. The variation in meaning is resolved as variation among these.
PredP and PredP: Of Subjects and Ancient Grievance
3.1.0
221
Predicative coordination and the distribution of event quantifiers
In chapter 2, Right-Node Raising traffics in unspoken copies of spoken phrases. The copies are exact, including, as argued there, the alphabetic identity of the variables used. Recall that, despite the Right-Node Raising, (71) can only be taken to report that a single Pepsi was shared and not that Alia drank a Pepsi and Marwan did too. (71) šerbou alia w marwaan anninet pepsi. drank.pl Alia and Marwan bottle Pepsi However many tokens, spoken or unspoken, there may be of ‘& [Ax : bottle(x) of Pepsi]Theme(e,x)’, they merely repeat—if every token is applied to the same event e—that what was consumed at e was the one bottle of Pepsi. Recall also that in sentence (71), the solitary tensed verb and the (in)definite descriptions as subjects provide only a single existential event quantifier so that indeed the tokens in the first and second conjunct of ‘& [Ax : bottle(x) of Pepsi]Theme(e,x)’ describe the same events, the collective effect of Alia and Marwan drinking, their drinking a bottle of Pepsi. With this example, there is still room to equivocate on the reference to their drinking, whether it is reference to a single event of drinking in which they both participate or plural reference to an event of Alia drinking and an event of Marwan drinking such that their drinkings were altogether of a bottle of Pepsi. Recall from section 2.3.2.2 that an example such as (72) plainly indicates plural reference (which I assume is carried over to (71) without harm). There need not be a sparsity of pine and oak, which taken together may be more dense than sparse. Rather, as in the logical form (73), there are some events, one of pine and one of oak, and these are sparsities, and these were in a region near human settlement, the same region being related to these several events. Again, it is the same region because both the spoken and unspoken tokens of ‘[Ax : region(x) near human settlement]In(e,x)’ fall within the scope of the same (token-identical) event quantifier, which happens, as this example shows, to be a plural. (72) keenou l-snoubraat w l-sendyeeneet xfeef bi mantʕa were.3mp the-pines and the-oaks sparse in region Hadd mantaʕa bašariyyeh. near region human ‘The pines and the oaks were sparse in a region near human settlement.’ (73) [∃E : Past(E)] (pl.be(E) & [℩X : pines(X)] [∃e′ : Ee] Wi(e′,X) Δ and [℩X : oaks(X)] [∃e′ : Ee] Wi(E′,X) & [℩E′ : pl.proi,i] (sparse(E′) & [∃y : region(y)] in(E′,y))) Note that because there are some events of pine and other events that are of oak, the logical form (73) must introduce existential event quantifiers within both
222
Chapter 3
conjuncts. But it is stipulated, without comment for the moment, that these existential event quantifiers include within their scope only the thematic relation, and thus neither includes within its scope a token of the Right-Node Raised phrase in a region near human settlement. In contrast to (71) and (72), (74)–(76) distribute bottles of Pepsi all around and (77) allows the pines and the oaks different regions: (74) šerbet alia w šereb marwaan anninet pepsi. drank.3fs Alia and drank.3ms Marwan bottle Pepsi (75) šerbou alia mbeereH w marwaan lyom anninet pepsi. drank.pl Alia yesterday and Marwan today bottle Pepsi (76) šerbet kell bent w kell Sabi anninet pepsi. drank.3fs every girl and every boy bottle Pepsi (77) keenou l-snoubraat men 5000 seneh w l-sendyeeneet were.3mp the-pines from 5000 year and the-oaks men 1000 seneh xfeef bi mantʕa Hadd mantaʕa bašariyyeh. from 1000 year sparse in region near region human ‘The pines 5000 years ago and the oaks 1000 years ago were sparse in a region near human settlement.’ The tokens of a bottle of Pepsi, although exact copies, and the tokens of a region near human settlement now fall within the scope of distinct event quantifiers within the conjuncts: (78) ∃e(drank Alia & [Ax : bottle(x) of Pepsi]Theme(e,x)), and ∃e(drank Marwan & [Ax : bottle(x) of Pepsi]Theme(e,x)) (79) ∃e′(Drank(e′) Alia [∃e : e ≤ e′ & yesterday(e)](… & [Ax : bottle(x) of Pepsi]Theme(e,x)), and Marwan [∃e : e ≤ e′ & today(e)](… & [Ax : bottle(x) of Pepsi]Theme(e,x))) (80) ∃e′(Drank(e′) [Every y : Gy][∃e : e ≤ e′](… & [Ax : bottle(x) of Pepsi]Theme(e,x)), and [Every y : By][∃e : e ≤ e′](… & [Ax : bottle(x) of Pepsi]Theme(e,x))) Tensed verbs introduce their own event quantifiers (74), as do adverbs (75) and (77). Distributive quantification over objects in Davidsonian logical form is also a concurrent quantification over events (76). Thus the sentences (74)–(76) come to distribute a bottle of Pepsi to many different events. The absence of a “sloppy” or distributed interpretation from (71) and from (72) does not indicate a prohibition to reconstruct a bottle of Pepsi and a region near human settlement into their respective first conjuncts. Rather, logical syntax dictates such a reconstruction, and it turns out not to matter when speaking of the same
PredP and PredP: Of Subjects and Ancient Grievance
223
events. On this view, the distribution of “sloppy” or distributed interpretations describes instead the scope and distribution of event quantifiers.6 The same considerations inform predicative coordinations and their instances of Right-Node Raising: (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. We will find that the contrast between (67) and (68), repeated here, is just the one already encountered in (71) and (76). In (67), the event quantifiers are distributed so that both tokens of a rocker happen to apply to the same events. The same remark, that the participants in these events were a rocker, is thus asserted twice over. But in (68), the pronominal description of these events falls within the scope of different event quantifiers, those concurrent with the universal quantifiers, every bump and every grind, from which the “sloppy” or distributed interpretation of these pronouns emerges. Returning to (63)–(64), Fox’s argument from reconstruction reveals the presence of subjects, pronounced or unpronounced, in all of these conjuncts.7 The reconstruction effect itself reflects the distribution of event quantifiers, and at the end of the day, we will see that the overt pronunciation of the subject in (64) signals some difference in event quantification sufficient to distinguish its meaning from its counterpart in (63) with subject unpronounced. 3.2
Case’s place
The phrases flanking and are compelled to be clauses, and reconstruction confirms the constant presence of subjects. When a contrast surfaces between conjuncts with unpronounced subjects and those with subjects pronounced, it therefore cannot be a contrast between small predicates and big clauses. If it is a difference of size, it must instead be between big clauses and even bigger ones when subjects are pronounced. Something in syntactic structure prompted by the subject’s pronunciation distinguishes (81) and (82) in that only the former, with the conjuncts’ subjects unpronounced, allows the coordination to be fronted, (83) vs. (84): (81) The rockers have shimmied and shaken. (82) The rockers have shimmied and the rollers shaken. (83) Though shimmied and shaken the rockers have for sure, not an ankle was twisted nor an elbow sprained. (84) *Though shimmied and the rollers shaken the rockers have for sure, not an ankle was twisted nor an elbow sprained.
224
Chapter 3
Similarly, the contrasting ellipses in (85) and (86) attest to a difference in syntactic structure: (85) The rockers have shimmied and shaken, and the rollers have, too. (86) *The rockers have shimmied and the twisters shaken, and the rollers have too. The elided phrase in (85), where the subjects of the antecedent phrase are unpronounced, says that the rollers have shimmied and shaken. But, in (86), pronunciation of the subject within the antecedent phrase rules out the ellipsis and blocks the meaning that the rollers have both shimmied and shaken.8 Adverb scope (see Jackendoff 1971; Johnson [1996] 2003) makes it out as a difference of size. The adverb simultaneously includes within its scope the entire coordination in (87) and thus the sentence reports that the stirring and the shaking are simultaneous: (87) Diego has simultaneously stirred a mojito and shaken a margarita. (88) *Diego has simultaneously stirred a mojito and Frida shaken a margarita. (89) Diego has simultaneously stirred a mojito and a martini and Frida shaken a margarita (while shuffling in a conga line). In contrast, (88), with conjunct subject pronounced, fails as a report that Diego’s stirring and Frida’s shaking are simultaneous. The adverb is trapped within the first conjunct, where there are not two events reported, leaving (88) incoherent. In (89), the adverb is interpreted entirely within the first conjunct, synchronizing mojito and martini. So it appears that a coordination with subject pronounced is too big to be fronted or elided and too big to fall within the scope of the adverb as positioned in (87)–(89), while a coordination without pronounced subject is just the right size for these constructions. If the smaller coordination, with subjects unpronounced, is already clausal as argued, then one may be tempted to regard the larger coordinations in (82) and (88) as reductions from fully tensed clauses as in (92) and (93), given the nominative subjects (see (90) vs. (91)) and the association of nominative Case with tensed clauses. (90) He has stirred and she shaken. (91) *He has stirred and her shaken. (92) The rockers have shimmied and the rollers have shaken. (93) *Diego has simultaneously stirred a mojito and Frida has shaken a margarita. Section 4.1 and chapter 5 will, however, discredit the reduction from a Tensed clause. No apparent coordination of participial phrases, subjects pronounced or not, con-
PredP and PredP: Of Subjects and Ancient Grievance
225
ceals unspoken tokens of tensed auxiliary verbs. Throughout (81), (82), and (87)– (89), conjoined participial phrases share the single token of a tensed auxiliary verb.9 As a result, the verbal morphology in (81) and (82) is exactly the same and so is the syntax it projects there; likewise for (87) and (88). If, as argued, the clauses coordinated in (82) and (88) are larger than those in (81) and (87), the difference must be located elsewhere and appears to coincide merely with the incidence of a pronounced subject—“the rollers shaken” is larger than “x shaken.” This difference of structure becomes manifest as a striking difference of meaning in (94) and (95), where it is beyond dispute that they are identical in their syntax except for what hangs on the overt pronunciation of the subject: (94) #Kunstler is sitting and is standing. (95) Kunstler is sitting, and Kunstler is standing. Kunstler is sitting, and he is standing. Consider that during trial, counsel sometimes stands before the court and sometimes remains seated: (96) Kunstler is sitting and standing. These postures cannot be simultaneous, but that does not preclude (96) from asserting that Kunstler’s action in progress will ceteris paribus amount to some sitting and some standing. Perhaps his peculiar courtroom antics, jumping up and down and thus alternating the postures, is what has warranted the assertion. In contrast, (94) describes a scene that is impossible but for contortions imagined to be simultaneous sitting and standing. What (94) demands is that what is now in progress be a sitting, and it also be a standing. At first glance, (95) describes the same contradictory scene. A coherent scene can, however, be imagined if one consciously allows that the coordination in (95), like the narrative in (97), advances the moment of evaluation, the now, from one sentence to the next: (97) Kunstler is sitting. Kunstler is standing. In contrast, (94) is impervious to such manipulation and cannot thereby escape contradictory description of the very same moment.10 It is as if pronouncing the second subject—and only by doing so—provides occasion to recalibrate the temporal anchor in the second clause. Otherwise, the tensed auxiliary verb in the second clause of (94) merely invokes reference to the time of the first. These observations implicate a difference of logical form between (94) and (95). Yet a difference of logical form leading to a difference of meaning surprises most views offering little beyond tokens of the subject DP to tell (94) and (95) apart. If that is the end of it, (94) and (95) should mean the same, no matter whether (94) contains a second, unpronounced token of Kunstler as in (98) making it identical in
226
Chapter 3
logical form to (95), or it contains a variable in subject position bound to the overt name (treated as a quantifier phrase) (99), or a generalized coordination of predicate phrases derives a predicate applied to a single subject (100): (98) Kunstler is sitting, and Kunstler is standing. (99) [℩x : Kunstler(x)](x is sitting and x is standing) (100) λy. y is sitting and y is standing (Kunstler) [℩x : Kunstler(x)] λy. y is sitting and y is standing (x) Adverbialization (see chapters 1, 8–15) steps in to rescue (94) and (95) from synonymy (and provide itself further proof). In chapter 10, the lifetime effects in (101)– (103) derive from denominal adverbials, such as Lenin-izing, that interact with Tense and intervening adverbial modifiers bald and dead: (101) Lenin resembled Queen Victoria. (102) Lenin bald resembled Queen Victoria. (103) Lenin dead resembled Queen Victoria. Adverbialization applies with full generality to any token of nominal quantification, (104), so that if (105) translates the Tensed clause resembled Queen Victoria occurring in each of (101)–(103), articulated logical form delivers (106)–(108): (104) … [Dα : ∃ENP[E,α]] (… θ[Ej,β] Ψ) ⇒ … [Dα : ∃ENP[E,α]][℩E : NP[E,α]][∃Ei : N[E,Ei]](… θ[Ej,β] Ψ) (105) … W[E0,α] [℩E0:pro0] Past[E0] O[E0,E1] ∃XTheme[E1,X] … resemble[Ei] … (106) [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]][∃E0 : N[E,E0]] ((105) … Past[E0] …) ‘Lenin Lenin-izing resembled Queen Victoria.’ (107) [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]][℩E-1 : N[E,E-1] bald[E-1]] [∃E0 : N[E-1,E0]] ((105) … Past[E0] …) ‘Lenin Lenin-izing bald resembled Queen Victoria.’ (108) [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]] [℩E-1 : N[E,E-1] dead[E-1]] [∃E0 : N[E-1,E0]] ((105) … Past[E0] …) ‘Lenin Lenin-izing dead resembled Queen Victoria.’ With the pronunciation of both subjects in (95), adverbialization applies in both conjuncts so that (95) says that Kunstler Kunstler-izing is sitting, and Kunstler Kunstler-izing is standing: (109) [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … sitting[Ei] …), and [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … standing[Ei] …) It follows that the sitting and the standing must both occur in the neighborhood of Kunstler-izing and that they are both in the present, but nothing here enforces that
PredP and PredP: Of Subjects and Ancient Grievance
227
the standing coincides with the sitting. If one is furthermore made aware that the speech time has advanced from one conjunct to the next, shifting with it what counts as present, then (95) articulated as in (109) serves as a report that Kunstler first sits and then stands. In arguing that adverbialization applies to all tokens of nominal quantification, chapters 8–15 survey only pronounced tokens, and the subjects in (95) simply join the like examples found there. Now that meaning indicates a difference of logical form between (95) and (94), it had better be that adverbialization does not also apply in (94) to make their logical forms identical once again. Pronunciation cues articulation. It signals a phrase made larger by the denominal adverbial phrase and invites one to recalibrate temporal reference. In contrast, silence indicates an intention to continue with those events under discussion that have been antecedently described, to which end logical form will be seen in section 3.3 to supply a tacit definite description as in (111), referring to the earlier events and deriving for (94) the implication that the standing coincided with the sitting.11 Moreover, since (94) cannot escape this implication, no analysis allows it to share the logical form (110) that the adverbialization of two pronounced subjects assigns to (95). (110) Kunstler is sitting, and Kunstler is standing. [℩x: ∃E Kunstler[E,x]][℩E: Kunstler[E,x]][∃E0: N[E,E0]] (… Present[E0] … sitting[Ei] …), and [℩x: ∃E Kunstler[E,x]][℩E: Kunstler[E,x]][∃E0: N[E,E0]] (… Present[E0] … standing[Ei] …) (111) Kunstler is sitting and is standing. [℩x: ∃E Kunstler[E,x]][℩E: Kunstler[E,x]][∃E0: N[E,E0]] (… Present[E0] … sitting[Ei] …), and [℩E0: (… Present[E0] … sitting[Ei] …)][∃E1: N[E0,E1]] (… Present[E1] … standing[Ei] …) 3.2.0
Tailoring coordination to size
Recognizing that a conjunct with a pronounced subject is larger than one without defuses an objection to unpronounced subjects (Godard 1989; Lasersohn 1995, 103f.):12 (112) *No congressmeni admire Kennedy, and theyi are very junior. (113) Only a few students have read this book, and they remember it poorly. (114) No congressmeni admire Kennedy and are very junior. (115) Only a few students have read this book and remember it poorly. In (112) and (113), the coordination of fully tensed clauses with the conjuncts’ subjects pronounced confines the quantifier’s scope to the first clause, as reflected in
228
Chapter 3
both pronominal binding (112) and (113)’s entailment that only a few students have read this book. Absent a pronounced subject, in contrast, the quantifier in (114) and (115) manages to include within its scope the entire coordination. Indeed predicative coordination would derive the facts of (114) and (115), but the facts do not conversely imply predicative coordination, else (116) and (117) should imply the absence of the second subjects (see McCawley 1993; Lin 2001, 2002; Johnson 2000a, 2009, [1996] 2003). (116) a. No rocker has shimmied and (any of) his mama(s) (ever) shaken to a throbbing disco beat. b. No child has (ever) watched Sesame Street and her teddy bear gone to bed unhappy. (117) a. No rocker is shimmying and his mama shaking to that funky disco music. b. No child is watching Sesame Street and her teddy bear falling asleep. Because participial phrases are coordinated in (116) and (117), the coordination is small enough to fall entirely within the scope of the quantifier sitting beside the tensed auxiliary verb—no matter that the conjuncts with pronounced subjects are somewhat larger than their counterparts with unpronounced subjects, as movement, ellipsis, and adverb scope have indicated in (83)–(88). In (118) and (119), the coordination of fully tensed clauses with pronounced subjects once again puts the second subject outside the scope of the quantifier: (118) a. *No rocker has shimmied, and his mama has shaken to that funky disco music. b. *No child has (ever) watched Sesame Street, and her teddy bear has gone to bed unhappy. (119) a. *No rocker is shimmying, and his mama is shaking to that funky disco music. b. *No child is watching Sesame Street, and her teddy bear is falling asleep. It now becomes an empirical question whether or not a coordination of fully tensed clauses with subjects unpronounced, although smaller than its counterpart with pronounced subjects, is still large enough to confine the quantifier within the first conjunct. On their own, the facts of (114) and (115) suggest that only the very largest coordination of tensed clause with pronounced subject is confining. Often this remains so when at least one of the conjuncts is a simple tensed clause as in (114) and (115) and (120)–(122): (120) No rocker shimmied and shook (to that funky disco music). (121) No student of mine studies syntax and publishes semantics. Not any student of mine studies syntax and publishes semantics. (122) No race car skidded and headed for the viewing stands.
PredP and PredP: Of Subjects and Ancient Grievance
229
Tensed auxiliary phrases however behave differently. A coordination of participial phrases, as before, falls easily within the scope of a subject quantifier. But a coordination of tensed auxiliary phrases at least sometimes confines the quantifier to the first clause: (123) No rocker was shimmying and shaking (to that funky disco music). (124) *No rocker was shimmying and was shaking (to that funky disco music). (125) No rocker has shimmied and shaken (to that funky disco music). (126) *No rocker has shimmied and has shaken (to that funky disco music). (127) No student of mine is studying syntax and publishing semantics. Not any student of mine is studying syntax and publishing semantics. (128) *No student of mine is studying syntax and is publishing semantics. *Not any student of mine is studying syntax and is publishing semantics. (129) No race car was skidding and heading for the viewing stands. (130) *No race car was skidding and was heading for the viewing stands. (131) No race car has skidded and headed for the viewing stands. (132) *No race car has skidded and has headed for the viewing stands. The observed effect concerns only nonincreasing quantifiers.13 The universal quantifiers in (133)–(137) appear to include felicitously the entire coordination within their scope, and the name in (138)–(142) is of course the subject of both conjuncts: (133) (134) (135) (136) (137)
Every Every Every Every Every
rocker was shimmying and was shaking (to that funky disco music). rocker has shimmied and has shaken (to that funky disco music). student of mine is studying syntax and is publishing semantics. race car was skidding and was heading for the viewing stands. race car has skidded and has headed for the viewing stands.
(138) (139) (140) (141) (142)
James Brown was shimmying and was shaking (to that funky disco music). James Brown has shimmied and has shaken (to that funky disco music). James Brown is studying syntax and is publishing semantics. James Brown was skidding and was heading for the viewing stands. James Brown has skidded and has headed for the viewing stands.
However the empirical questions are resolved, note that discrimination between quantifier types suffices to preclude treatment of the coordination in (133)–(142) as predicative. For, if the coordination here represented a derived predicate, there in principle could be no reason why that predicate should combine with a name or universal quantifier but not with a decreasing quantifier as in (124)–(132) or with any other type of quantification. On the other hand, the presence of tacit descriptive pronouns in the second clause falls in with their overt counterparts in discriminating
230
Chapter 3
among quantifier types in the configuration for “telescoping” (Roberts 1989; Ludlow 1994; Poesio and Zucchi 1992; chapters 1 and 2 of the present book): (143) Every rocker was shimmying. He was also shaking to that funky disco music. (144) *No rocker was shimmying. He was also shaking to that funky disco music. Even with a universal quantifier, “telescoping” is not a sure thing. It is not merely a formal configuration but depends on the coherent interpolation of an adverb with content fixed by the antecedent clause. Kyle Johnson (p.c., 2003) caught a contrast between (145) and (146), which on reflection shadows a contrast in the interpolated modification made explicit in (147) and (148): (145) Every race car was skidding and (then) (it) ate up track heading for the viewing stands. (146) ?*Every race car has skidded and (then) (it) ate up track heading for the viewing stands. (147) Every race car was skidding and when a race car was skidding (it) ate up track heading for the viewing stands. Every race car was skidding and (when) skidding (it) ate up track heading for the viewing stands. (148) ?*Every race car has skidded and when a race car has skidded (it) ate up track heading for the viewing stands. ?*Every race car has skidded and (when) having skidded (it) ate up track heading for the viewing stands. In (146) and (148), the interpolated adverbial underwriting the descriptive pronoun relies incoherently on a (perfective) state to locate an episodic event. Thus (146) joins (124)–(132) in showing the unpronounced subject of a tensed clause subject to the vagaries of covarying pronouns. All of this strongly suggests that a coordination of tensed auxiliary phrases is never predicative, contrary to first appearances in (112)–(115), and that a Tensed auxiliary phrase without pronounced subject nevertheless contains an unpronounced one in need of some account of how it comes to covary with an antecedent quantifier. With what has been seen so far, a first approximation could remark that a Tensed auxiliary phrase is simply and always opaque to quantification so that none of the quantifiers, decreasing or universal, in the subject position of the first clauses of (123)–(142) ever escapes to properly include the coordination within its scope. Appearances to the contrary in (133)–(137) are the effect of “telescoping,” which favors universal quantifiers in creating its illusion.14
PredP and PredP: Of Subjects and Ancient Grievance
231
This approximation misses a systematic elicitation that in plain contrast to (123)– (132) shows decreasing quantifiers—no less—achieving scope over a coordination of tensed auxiliary phrases: (149) No child is watching Sesame Street and looking unhappy. (150) ?No child is watching Sesame Street and is looking unhappy. (151) No child has watched Sesame Street and gone to bed unhappy. (152) ?No child has watched Sesame Street and has gone to bed unhappy. (153) No student is reading the text and taking notes on what she reads. (154) ?No student is reading the text and is taking notes on what she reads. The natural understanding of the sentences in (149)–(154) imposes an asymmetry on the conjuncts, the first conditioning the second, as made explicit in (155)–(157): (155) ?No child is watching Sesame Street and watching Sesame Street, is looking unhappy. (156) ?No child has watched Sesame Street and having watched Sesame Street, has gone to bed unhappy. (157) ?No student is reading the text and reading the text, is taking notes on what she reads. That is, the second conjunct is introduced by a tacit pronoun referring to the events described by the antecedent conjunct. The essential role of this pronoun is further exposed when a negative-polarity item contained within its antecedent clause induces, in effect, its referential failure and spoils the coordination of Tensed auxiliary phrases (and only this larger coordination): (158) No child is ever watching Sesame Street and looking unhappy. (159) *No child is ever watching Sesame Street and is looking unhappy. (160) No child has ever watched Sesame Street and gone to bed unhappy. (161) *No child has ever watched Sesame Street and has gone to bed unhappy. (162) No student is ever reading the text and taking notes on what she reads. (163) *No student is ever reading the text and is taking notes on what she reads. In (159) and (161), the event pronominal introducing the second clause attempts reference to an antecedent ‘ever watching Sesame Street’ or ‘ever having watched Sesame Street’, and similarly in (163), to ‘ever reading the text’. Note that the overt pronoun in (165) also fails when its antecedent is replaced with a negative-polarity item in (166): (164) No one saw anything suspicious and went to the police. No one saw a single suspect and went to the police.
232
Chapter 3
(165) No one saw something suspicious and reported it to the police. No one saw a suspect and reported him to the police. No one saw a suspect and reported it to the police. (166) *No one saw anything suspicious and reported it to the police. *No one saw a single suspect and reported him to the police. *No one saw a single suspect and reported it to the police. If the negative-polarity item is inherently modal (Dayal 1998)15 (and little else could tell it apart from its indefinite counterpart in (165)), (166) assimilates to cases of failed “modal subordination,” (168) and (170) (Roberts 1989): (167) A thief might break into the house. He would take all the silver. (168) *A thief might break into the house. He took all the silver. *A thief might break into the house. He is wanted in five states. (169) A wolf might come in. It would eat you first. (170) *A wolf might come in. It is fierce and gray. The pronoun in the second clause falls outside the modal context of the first in (166), (168), and (170), and it cannot secure definite reference within its own context to anything that has been antecedently described—as if anything suspicious means something suspicious wherever, and it attempts reference to the thing suspicious here although nothing has been introduced that answers to the narrower description. Except to see in (159), (161), and (163) an analogous failure of a tacit event pronoun to refer when its antecedent is a negative-polarity item, I cannot imagine how else the intrusion of ever should undermine the coordination of Tensed auxiliary phrases. The coordination is similarly undermined when a negative-polarity item is the antecedent for a tacit subject pronoun as in (172) and (173):16 (171) ?No student of mine is reading the text and is taking notes on what she reads. (172) *Not any student of mine is reading the text and is taking notes on what she reads. (173) *It is not the case that any student of mine is reading the text and is taking notes on what she reads. (174) Not any student of mine is reading the text and taking notes on what she reads. (175) It is not the case that any student of mine is reading the text and taking notes on what she reads. Let us now size up coordination. A coordination of Tensed phrases with pronounced subjects is opaque to quantification, (112), (113), (118), and (119). The quantifier in the subject position of the first clause remains within. A “telescoping” that tends to be incoherent with decreasing quantifiers, (143) vs. (144), may however
PredP and PredP: Of Subjects and Ancient Grievance
233
create an illusion of wide scope, (143), (145) and (147), if the interpolated adverb otherwise makes sense, (145) and (147) vs. (146) and (148). Stepping down to a coordination of tensed phrases with a subject unpronounced, it appears that quantifiers do raise to include such a coordination within their scope, (114), (115), (120)– (122), (150), (152), (154), (155)–(157), (164), and (165). For the moment, consider only such coordinations of tensed auxiliary phrases. By hypothesis, pronunciation cues articulation (i.e., adverbialization) only in that first conjunct where the subject has been pronounced, licensing the denominal adverb in the first conjunct but not in the second (176). Absent a pronounced second subject, the intention is to continue with those events under discussion that have been antecedently described, and the only way to close off the event variable exposed in the second conjunct, E1 in (177), is with an event pronoun denoting the antecedent events as in (177): (176) *[No x : NPi] ([℩E : NPi[E,x]][∃E0 : N[E,E0]](… Tense[E0] …) and [℩E : NPi[E,x]][∃E0 : N[E,E0]](… Tense[E0] …)) (Cf. (110)) (177) [No x : NPi]
([℩E : NPi[E,x]] [∃E0 : N[E,E0]](ΦI … Tense[E0] …) and [℩E0 : Φi] [∃E1 : N[E0,E1]](… Tense[E1] …)) (Cf. (111))
The thought expressed is felicitous as in (150), (152), (154), and (155)–(157) if it fits the asymmetric, conditioning relation that (177) imposes on the conjuncts—unless an error of modal subordination thwarts the event pronoun, (159), (161), and (163). If there is no thought that the events described by the second clause are conditioned by those of the first, (177) is not the logical form. The symmetric coordinations in (124), (126), (128), (130), (132), which (176) would represent, fail in that the unpronounced second subject rules out adverbialization of a denominal adverb and thus strands an event variable in the second conjunct. Return now to the coordinations with simple tensed verb phrases, (114), (115), (120)–(122), (164), (165), which readily fall within the scope of the first subject when the second is unpronounced ((120)-(132); (178)-(185) selectively repeated below). Surely it is not that the relation between the conjuncts of (120) is meant to be any different from what fails in (124) and (126), and thus there is no reason to suppose that (177) rescues (120) when its meaning prevents it from doing so for (124) and (126): (120) No rocker shimmied and shook (to that funky disco music). (124) *No rocker was shimmying and was shaking (to that funky disco music). (126) *No rocker has shimmied and has shaken (to that funky disco music). (121) No student of mine studies syntax and publishes semantics. Not any student of mine studies syntax and publishes semantics. (128) *No student of mine is studying syntax and is publishing semantics. *Not any student of mine is studying syntax and is publishing semantics.
234
Chapter 3
(122) No race car skidded and headed for the viewing stands. (130) *No race car was skidding and was heading for the viewing stands. (132) *No race car has skidded and has headed for the viewing stands. Similarly, (121) is close enough in meaning to (128), and (122) to (130) and (132), that one could not claim that these coordinations of simple tensed verb phrases express an asymmetric conditioning. Yet, given that the second subject remains unpronounced, neither does the articulation of a second denominal adverb come to the rescue, since the likes of (176) remains unavailable. Given that the second conjunct provides for itself neither denominal adverb nor pronoun denoting antecedent events, what finally does in (124), (126), (128), (130), and (132) is that the second tensed auxiliary phrase is just too large to fall within the scope of the denominal adverb that the first subject introduces, and thus strands its highest event variable unbound. If a simple tensed verb phrase escapes this fate, it must be that it is not so large. At one extreme, represented by (180), (120) assimilates in logical form to (178). At a very safe distance from (179), the position the subject is pronounced in sits beside a tacit tensed auxiliary verb, well outside the coordination falling within its scope: (120) No rocker shimmied and shook. (178) No rocker did shimmy and shake. (179) No rocker has shimmied and has shaken. No rocker did shimmy and did shake. (180) [No x: NPi][℩E: NPi[E,x]][∃E0: N[E,E0]]W[E0,x]Tense[E0] ((… V[Ei] …) and (… V[Ei] …)) But, less extreme and allowing each conjunct of (120) to hold onto its own expression of tense, any structure would also suffice to distinguish (120) from (179) that put the position of the pronounced subject outside the coordination of simple tensed verb phrases, perhaps even positioning outside it only the Case position licensing pronunciation, as in (181): (181) [No x : NPi][℩E : NPi[E,x]][∃E0 : N[E,E0]] ((W[E0,x] … Tense[Ei] …) and (W[E0,x] … Tense[Ei] …)) Attending to the contrasts discussed above, what must be said is that there is a fundamental difference between the phrase structure of a simple tensed verb phrase and that of a tensed auxiliary phrase. The latter is so large that it captures within it any position that might launch a pronounced subject, getting to be so large perhaps as the result of a movement to higher position special to auxiliary verbs, as Pollock (1989) and subsequent syntax literature have conjectured.17
PredP and PredP: Of Subjects and Ancient Grievance
235
Where it falls between (180) and (181), the structure for the smaller coordination of simple tensed verb phrases is further constrained by the distribution of the disjunctive interpretation (chapter 6). Recall from chapter 1 that a participation relation W is recruited to derive the disjunctive interpretation—to express the vague participation in events that the conjuncts collectively describe. Now any conjunct of coordinated simple tensed verb phrases can itself be a coordination of simple tensed verb phrases disjunctively interpreted: (182) Twenty thousand students surrounded the Pentagon and crowded into the Mall, and made it onto the evening news. Sentence (182) is taken to assert that all 20,000 made it onto the evening news but of course they were divided between those at the Pentagon and those on the Mall. That is, what they did (‘W’) coincides with some surrounding the Pentagon and some crowding into the Mall, and it coincides with a making it onto the evening news. Since any such conjunct may itself be expanded as a disjunctively interpreted coordination, every conjunct contains its own token of W. Note that these tokens sit outside the coordination to be disjunctively interpreted. But a topmost coordination is also subject to disjunctive interpretation in (183), dividing the students between those at the Pentagon and those on the Mall: (183) Twenty thousand students surrounded the Pentagon and made it onto the evening news and crowded into the Mall and didn’t make it onto the evening news. The facts of (182) and (183) prima facie recommend sprinkling ‘W’- all around as in (184). Rather than (181), the logical form for (120) in (185) coordinates phrases at least small enough to leave a token of ‘W’ outside the coordination: (184) [∃X : 20K(X) … i][℩E : NPi[E,X]][∃E0 : N[E,E0]]W[E0,X] ((O[E0, E1] ∃X W[E1,X] … Tense[Ei] …) and (O[E0, E1]∃XW[E1,X] … Tense[Ei] …)) (185) [No x : NPi][℩E : NPi[E,x]][∃E0 : N[E,E0]]W[E0,x] ((O[E0, E1]W[E1,x] … Tense[Ei] …) and (O[E0, E1]W[E1,x] … Tense[Ei] …)) With (185), the logical form for (120) thus positions itself between that of (178) in (180) and the logical form for (179) seen earlier in (177). 3.2.1
Small-clause sizes
Pronunciation as a cue to articulation and silence as resumption contrast the pronounced and unpronounced subjects of tensed clauses in (94) and (95) (repeated here):
236
Chapter 3
(94) #Kunstler is sitting and is standing. (95) Kunstler is sitting, and Kunstler is standing. Some difference of size, as adverbialization or its absence implies, has already been found to distinguish small clauses with pronounced and unpronounced subjects with respect to movement (83)–(84) and ellipsis (85)–(86), and perhaps most telling with respect to adverb scope (87)–(89) (repeated here): (87) Diego has simultaneously stirred a mojito and shaken a margarita. (88) *Diego has simultaneously stirred a mojito and Frida shaken a margarita. (89) Diego has simultaneously stirred a mojito and a martini and Frida shaken a margarita (while shuffling in a conga line). The adverb simultaneously includes within its scope the entire coordination in (87), where subjects are unpronounced, reporting that the stirring and the shaking are simultaneous. In contrast, (88), with conjunct subject pronounced, fails as a report that Diego’s stirring and Frida’s shaking are simultaneous. The adverb is trapped within the first conjunct, where there are not two events reported, leaving (88) incoherent.18 A difference of size between the small clauses of (186)–(187) with unpronounced subject and those of (188)–(189) with subject pronounced will provide the scaffold on which to hang differences of meaning: (186) A rocker has shimmied and shaken. (187) Twenty rockers have shimmied and shaken. (188) A rocker has shimmied and a rocker shaken. (189) Twenty rockers have shimmied and twenty rockers shaken. Recall that any such differences reflect the incidence of a pronounced subject, as section 4.1 promises to show that no apparent coordination of participial phrases, subjects pronounced or not, conceals unspoken tokens of tensed auxiliary verbs. Thus, the conjoined participial phrases in each of (186)–(189) share the single token of a tensed auxiliary verb. It follows immediately that the syntax of coordination with subject pronounced is eccentric—with conjoined phrases assumed to be syntactically identical, it must be that the subject of the first conjunct has raised alone outside the coordination (Johnson 2002, 2009, [1996] 2003): (189) Twenty rockersi have [[ti shimmied] and [twenty rockers shaken]] (190) Hei has [[ti (*simultaneously) stirred a mojito] and [she shaken a margarita]] Confirming this eccentricity, note that (191) and (192) put the second decreasing quantifier obligatorily within the scope of the first and thus deny that a tree can batter a structure without adverse effects elsewhere:
PredP and PredP: Of Subjects and Ancient Grievance
237
(191) No tree has battered a window and no glass at all broken. (192) No tree has battered a roof and no shingles shaken loose. In contrast, (193) and (194), coordinations of tensed clauses, just deny battering and deny structural damage: (193) No tree has battered a window and no glass at all has broken. (194) No tree has battered a roof and no shingles have shaken loose.19 Despite the clear indications of the resulting structure, one will need to lean quite a bit on syntax to account for the actual distribution of pronounced participial subjects. For, if it were enough to observe that conjuncts and conjoined phrase all belong to the same semantic type, why can’t the conjunct with pronounced subject complement the auxiliary verb, given that its parent coordination does? (195) *Twenty rockers have twenty rockers shaken. (196) *Twenty rockers have them(selves) shaken. (197) *Hei has she shaken a margarita. A meaning for (195) could be coherent if it meant, say, that from the viewpoint of twenty rockers, they have in mind twenty rockers shaken. Even if a viewpoint always had to be of oneself and only perhaps of some others too (as in (189)), it would not derive the ungrammaticality of (195)–(197). Rather, syntax seems to stipulate that the Case necessary for overt pronunciation is assigned by the conjunction and itself (Munn 1993) and hence is unavailable to the participial subjects in (195)–(197).20 Opposite (195)–(197), the pronunciation of a participial subject inside a coordination raises details of grammar that I would rather not see dilute an argument about logical syntax. With this last worry in mind, let me repeat that the eccentric syntax of this construction comes with the discovery that it is not a gapping structure, that the second conjunct does not conceal an unspoken token of the auxiliary verb. It is not a consequence or undue complication of the thesis that and is a sentential connective, and—thanks to the participial subject’s overt pronunciation—it comes where there can be no doubt that and appears there as such. 3.2.1.0 The Case position of small-clause subjects
Leaving syntax to fend for itself, let’s return to the meaning that tells apart (186)– (187) and (188)–(189). The first difference is the ancient grievance again that (188) has no interpretation entailing that the shimmier is the shaker, while (186)’s only interpretation entails that they are the same, an observation all the more acute if (186) has a logical form as argued in section 3.1 that comes to resemble (188), reconstructing an unspoken token of the subject in the second clause: (198) A rocker has shimmied and a rocker shaken. (188) A rocker has shimmied and a rocker shaken.
238
Chapter 3
Now, whatever compels the rockers to be the same when a rocker is unpronounced (see (186)/(198)) must not compel identity when the second subject is pronounced (188). Correlatively, whatever allows for different rockers when the second subject is pronounced must not enter into the analysis of its unpronounced counterpart. A second difference of meaning, closely related to the first, emerges when the subject is plural as in (187) and (189), inviting disjunctive interpretation of the coordinated predicate. Recall that the disjunctive interpretation, so-called in chapter 1 and discussed further in chapter 6, allows that the 20,000 students of (199) be divided between those who surrounded the Pentagon and those crowded into the Mall: (199) Twenty thousand students surrounded the Pentagon and were crowded into the Mall. So too do the coordinated participial phrases in (187), with subject unpronounced, allow twenty rockers to divide between shimmiers and shakers. In contrast, (189), with pronounced subject, asserts twenty of each. Twenty shimmy and twenty shake, even if it happens that the shimmiers are the shakers. The contrast is sharpened by the proposed analysis of disjunctive interpretation. Recall that twenty rockers are said to be the participants (‘W[E1,X]’) in that, referring to some shimmying (by some) and some shaking (by some). A backward event pronoun ‘[℩E2:ΦΨ],’ its content fixed by the conjoined phrases, describes shimmying and shaking that is then said to be the collective effect of what the rockers did: (200) Twenty rockers W[E1,X] have O[E1,E2] [℩E2: ΦΨ](Φ∃X shimmied) and (Ψ∃X shaken). Existential closure must dispose of the subject positions within the conjuncts, it remaining vague who shimmies and who shakes. To the extent then that (187) is alleged to resemble in its disjunctive interpretation something like (201), one may wonder how that interpretation is to be denied (202): (201) Twenty rockers have some shimmied and some shaken. (202) Twenty rockers have shimmied and some shaken. Moreover, the vagueness of who did what notwithstanding, the reported shimmying and shaking is of course confined by (187) to what twenty rockers did (‘W[E1,X]’): the shimmying and shaking referred to (‘[℩E2: ΦΨ]’) overlaps exactly that (‘O[E1,E2]’), and it is thus inferred that any of the shimmiers or shakers is among the twenty rockers. Were the participial phrases with subject pronounced in (202) subject to the same analysis, (202) should expect an interpretation entailing that the shakers were among the twenty rockers, which it plainly does not have. A similar pathology threatens (189), where disjunctive interpretation of something like (203) would
PredP and PredP: Of Subjects and Ancient Grievance
239
entail that the shimmiers and shakers were among the same twenty rockers, all of whom shake while perhaps only some shimmy: (203) Twenty rockers have some shimmied and twenty rockers shaken. The account of this second difference of meaning concerned with the disjunctive interpretation will subsume the first, so I begin there. (187) Twenty rockers have shimmied and shaken. (189) Twenty rockers have shimmied and twenty rockers shaken. With just a single auxiliary verb, both (187) and (189) report a having, so to speak, of some shimmying and some shaking. With this much structure common to both, it cannot be that to be the havers of some events is to be the participants in those events, or else the twenty rockers that are the subject of the auxiliary verb in (189) would be all of those who shimmy and shake. If, as before, to participate in some events that overlap some others (‘O[Ei,Ej]’) is also to participate in these others, this relation must not be the one bridging the having and the shimmying and shaking. Some other relation as in ‘Have[Ei] & R[Ei,Ej]’ is involved, or absent an analysis of perfect aspect, the auxiliary verb itself may be taken, as shown below, to express it, ‘Have[Ei,Ej]’.21 Either way, there is to be no inference from having to shimmying or shaking.22 An immediate consequence for (189) is that its first conjunct must itself contain a reference to the twenty rockers, absent an inference to their participation in what is described within. That is, the trace of the asymmetric movement in (189) must be interpreted either as a bound variable or as a definite description of some kind as in (204): (189) Twenty rockersi have [[ti shimmied] and [twenty rockers shaken]] (204) a. [∃X : 20(X) ∃E rocker[E,X]] have [[X shimmied] and [twenty rockers shaken]] b. [∃X : 20(X) ∃E rocker[E,X]] have [[[℩Y: Y=X] Y shimmied] and [twenty rockers shaken]] c. [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] …[[[℩X: W[E0,X]] X shimmied] and [twenty rockers shaken]] d. [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] …[[[℩X: 20(X) ∃E rocker[E,X]] W[E0,X]] X shimmied] and [twenty rockers shaken]] Sorting out the differences between (187) and (189) as to who shimmies and who shakes now rests on the internal structure of the coordination complementing the auxiliary verb. Suppose that ‘O[Ei,Ej]’ continues to bear the implication that the participants in coincident events must be the same. What distinguishes (189) is that it contains no token of ‘[℩Ej:proj] O[Ei,Ej]’ where ‘[℩Ej:proj]’ refers collectively to
240
Chapter 3
shimmying and shaking. Rather, there are tokens within each conjunct, with the event pronouns seeking their own proximate antecedents: (205) [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] (
… W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shimmied …)
and ([∃X : 20(X) ∃E rocker[E,X]] … W[E1,X][ιE1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shaken …) Within the first conjunct, ‘[℩E2:pro2]’ refers to some shimmying; in the second, to some shaking. Within each conjunct, ‘[℩E1:pro1]’ is to refer to whatever X did (‘W[E1,X]’), where the antecedents for X differ between the conjuncts. In (187) in contrast, the apparatus in boldface is located outside the coordination. Crucially, there is but a single token of ‘W[E1,X]’; ‘[℩E1:pro1]’ (however often it may be repeated) refers only to what the same twenty rockers did. The event pronoun ‘[℩E2:pro2]’, content fixed by the coordination in its scope, refers collectively to shimmying and shaking: (206) [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] … W[E1,X][℩E1:pro1][℩E2:ΦΨ]O[E1,E2] (Φ∃E2∃XAgent[E2,X] … shimmied …) and (Ψ∃E2∃XAgent[E2,X] … shaken …) Reasoning as above, if what these twenty do exactly overlaps some shimmying and shaking, then they are the shimmiers and shakers, without it being said who among them shimmies and who shakes. Notice that reconstruction into the conjuncts would, in effect, replace one existential quantifier ‘∃X’ with another: (207) [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] …W[E1,X][℩E1:pro1][℩E2:ΦΨ]O[E1,E2] (Φ∃E2[∃X : 20(X) ∃E rocker[E,X]]Agent[E2,X] … shimmied …) and (Ψ∃E2[∃X : 20(X) ∃E rocker[E,X]]Agent[E2,X] … shaken …) Provided that reconstruction does not change phrase structure—the boldface phrase remaining outside the coordination—it remains that the same twenty rockers are the shimmiers and shakers. Because it is now asserted, however, that twenty shimmy and twenty shake, the conjunctive interpretation is derived, entailing that the twenty rockers shimmy and the same twenty shake. The proposed difference of structure attending pronunciation of the participial subject carries over when it happens to be singular as in (186) and (188). If the second subject goes unpronounced, what the one rocker does is said to exactly overlap some shimmying and shaking. That rocker, given that no one else is said to do anything, is then both shimmier and shaker. It matters not if, by reconstruction,
PredP and PredP: Of Subjects and Ancient Grievance
241
it is further said that the shimmier is a rocker and so is the shaker, since this is already inferred. With the participial subject pronounced in (188), reference to the events described by the participles is properly contained within each conjunct. There is, as in (205), no collective reference to shimmying and shaking and no assertion that the shimmying and the shaking overlap what the same rocker does. As before, although the subject of the auxiliary is singular, there is no inference from having to shimmying or shaking. Pronunciation of the participial subject captures within its conjunct the phrasing that has appeared above in boldface. That is, the Case position for this subject apparently falls to the left. If the position of adverbs such as simultaneously falls somewhere to the right of the Case position and only phrases of like dimension are coordinated, then indeed pronunciation of the second participial subject is sufficient to trap the adverb in (88) within the first clause: (87) Diego has W[E1,X][℩E1:pro1]simultaneously[℩E2:ΦΨ]O[E1,E2] (Φstirred a mojito) and (Ψshaken a margarita). (88) *Diego has (… W[E1,X][℩E1:pro1] simultaneously [℩E2:pro2]O[E1,E2] stirred a mojito) and (Frida W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] shaken a margarita). The contrast is distilled in (208) and (209): (208) DPi Tense Aux [[ti W simultaneously O -en V] and [DP W O -en V]] (209) DPi Tense Aux W simultaneously O [[-en V] and [-en V]] 3.2.1.1 Right-Node Raising
Granted then that the participial phrase with pronounced subject is larger than one without, a further observation fixes the relative position of the pronounced subject within its conjunct. Recall from the discussion of Lebanese in section 2.3.2.2 that the overt subjects in (72) (repeated below) acquire their own events, one of piney sparsity and one oaken respectively, and yet their position below the tensed verb does not allow them to capture within their scope in a region near human settlement, which instead describes a single region containing both pine and oak: (72) keenou l-snoubraat w l-sendyeeneet xfeef bi mantʕa were.3mp the-pines and the-oaks sparse in region Hadd mantaʕa bašariyyeh. near region human ‘The pines and the oaks were sparse in a region near human settlement.’ Similarly, Diego and Frida in (210) and (212) act in their own events, without the implication that stirring is shaking or that inhaling is eating through. But the stirring and shaking are of the same mojito in (210), and (212) describes the anomaly that
242
Chapter 3
Diego and Frida consume the same plate of plantains. The participial coordinations contrast with the coordinations of tensed clauses in (211) and (213), which allow Diego and Frida each their own mojito and fried plantains: (210) Diego has stirred and Frida shaken a mojito furioso. (211) Diego has stirred and Frida has shaken a mojito furioso. (212) #By the last tango, Diego had inhaled and Frida eaten through a plate of fried plantains. (213) By the last tango, Diego had inhaled and Frida had eaten through a plate of fried plantains. The post-Tense subjects include within their scope only their own thematic relations. Adverbialization governs only the scope of the pronounced argument that cues it. The additional event quantifiers adverbialization supplies in (72), (210), and (212) confine their scope to the local thematic relations. Thus, in a region near human settlement, a mojito furioso, and a plate of fried plantains remain outside the scope of the event quantification introduced when pronouncing the subject within a conjunct. In such a configuration, for the reasons given in chapter 2, a Right-Node Raised phrase has only a collectivized interpretation—a shared region near human settlement, a shared mojito furioso, and a shared plate of plantains. Suppressing fewer details now, (210) has the logical form in (214) and (189) (here repeated) has the one in (215):23 (210) Diego has stirred and Frida shaken a mojito furioso. (214) [℩X : ∃E Diego[E,X]][℩E : ∃E Diego[E,X]] [∃E0 : N[E,E0]]W[E0,X] [℩E0:pro0]Present[E0] [℩E1:ΦΨ]Have[E0,E1] (Φ[℩X: W[E0,X]] [℩E0 : W[E0,X]] [∃E1 : N[E0,E1]] W[E1,X] [℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] [℩E2:pro2][℩E3:pro3]C[E2,E3] ∃E3stirred[E3] …) and (Ψ[℩X : ∃E Frida[E,X]][℩E : ∃E Frida[E,X]] [∃E1 : N[E,E1]] W[E1,X] [℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] [℩E2:pro2][℩E3:pro3]C[E2,E3] ∃E3shaken[E3] …) (189) Twenty rockers have shimmied and twenty rockers shaken. (215) [∃X : 20(X) ∃E rocker[E,X]][℩E : ∃E rocker[E,X]] [∃E0 : N[E,E0]]W[E0,X] [℩E0:pro0]Present[E0] [℩E1:ΦΨ]Have[E0,E1] (Φ[℩X: W[E0,X]] [℩E0 : W[E0,X]] [∃E1 : N[E0,E1]] W[E1,X] [℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] [℩E2:pro2][℩E3:pro3]C[E2,E3] ∃E3shimmied[E3]) and (Ψ[∃X : 20(X) ∃E rocker[E,X]][℩E : ∃E roller[E,X]] [∃E1 : N[E,E1]] W[E1,X] [℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] [℩E2:pro2][E3:pro3]C[E2,E3] ∃E3shaken[E3])
PredP and PredP: Of Subjects and Ancient Grievance
243
The same considerations mutatis mutandis explain that Ernest stirs and shakes the same mojito in (65) (here repeated): (65) Ernest has stirred and shaken a mojito. Ernest is stirring and shaking a mojito. What now crucially allows the stirrings and shakings of (66) and (216) their own mojitos, in contrast to (65) and (210), is that the first token of ‘… a mojito …’ along with the logical form of the phrase containing it is not constrained to be an exact copy of the second, unlike the tokens related by Right-Node Raising. (66)
E.rnest has stirred a mojito and shaken a mojito. Ernest is stirring a mojito and shaking a mojito. (216) Diego has stirred a mojito and Frida shaken a mojito. Absent Right-Node Raising, the first phrase contains an event pronoun referring back to the stirring and the second, a different one referring back to the shaking: (217) … ∃E stiri[E] [℩E: proi][A x : ∃E mojito[E,x]] … Theme … [E,x] and ∃E shakej[E] [℩E: proj][A x : ∃E mojito[E,x]] … Theme … [E,x] Recall again that (210) and (211) contrast in allowing different mojitos to be stirred and shaken, although both (210) and (211) derive from Right-Node Raising a mojito furioso: (210) Diego has stirred and Frida shaken a mojito furioso. (211) Diego has stirred and Frida has shaken a mojito furioso. It is not that Right-Node Raising neglects in (211) to insist on exactly the same content for the tokens it relates. Rather, only in (211), and not in (210), do these tokens fall within the scope of distinct event quantifiers, those associated with Tense. As observed in section 3.2.0, between (210) and (211), there is the structure that coordinates tensed auxiliary phrases while leaving the second subject unpronounced: (218) Ernest has stirred and has shaken a mojito. Ernest is stirring and is shaking a mojito. The stirring and shaking in (218) are once again of the same mojito, but it goes by a different account. Like (211), the spoken and unspoken tokens of a mojito fall within the scope of distinct event quantifiers associated with the separate tokens of Tense. Remember, however, that absent a pronounced subject in the second clause, the second clause is introduced by reference to the events described by the first. From this alone it does not yet follow that if the first events of (218) are of a mojito as are the second, they need be of the same mojito. For, in locating the events of the second clause, the first events of (17) (repeated here) (see section 5.0), for
244
Chapter 3
example, need not be identical to the second events—the glaring is not the formulating: (17) Kunstler is glaring at the judge and is formulating his rebuttal. In fact, a plausible scene is made available to (219)—hammering one frozen pipe and thereby loosening another. Still, the sentences can only report on a single pipe: (219) The plumber has hammered and has loosened a frozen pipe. The plumber is hammering and is loosening a frozen pipe. Yet the reference in the second clause to the events described by the first is essential to the account. The Lebanese fact that in (220) every woman and child read the same story was already explained in section 2.3.1. And it was explained that in English, the intrusion of a temporal relation, then, in (222) creates a contrast with (221), requiring of (222) alone that Sam repaint and Joe re-side the same new house. (220) aryet kell mara w ebn-a essa. read.3fs each woman and child-her story (221) Sam repainted and Joe re-sided a new house. (222) Sam repainted and then Joe re-sided a new house. Recall that a “telescoping” adverbial, mediating the covariation of her child / a child with every woman, in effect assimilates (220) to (222): the second conjunct is introduced by an adverbial referring to the events described by the first. Expressing a temporal or causal relation between distinct events, the adverbial contains a relation ‘C[E,E’]’ of course in distinct variables (*‘C[E,E]’), where the first conjunct describes the events E and the second, the events E′. The excluded interpretation, which would allow woman and child a different story and Sam and Joe a different house, would involve the “sloppy” interpretation of identical copies of the same phrase— for example, , e.g., ‘[A x : story(x)]Theme[E,x]’. But, precisely because the relation between the conjuncts has forced the one to be about events E and the other to be about events E′, a phrase in the event variable of one conjunct cannot be used coherently within the other. Only a collectivized reference to what woman and child read or to what Sam repaints and Joe re-sides can be repeated twice over in the exact same form in both conjuncts, in which case it is the same story and the same new house. 3.2.1.2 A free event variable in small clauses?
Sizing up coordination (section 3.2.0), it was proposed that the occurrence of a Tense morpheme exposes an event variable in the second conjunct of (94) (repeated below). Failing adverbialization of the unpronounced subject, the exposed variable can be closed off only by a pronoun denoting the events described by the antecedent
PredP and PredP: Of Subjects and Ancient Grievance
245
clause, with the anomalous result that the standing is said to coincide with the sitting. In (95), in contrast, the adverbialization of the pronounced second subject introduces event quantification that includes within its scope the second token of Tense and obviates the need to refer to events antecedently described. (94) #Kunstler is sitting and is standing. [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] . (… Present[E0] … sitting[Ei] …), and [℩E0 : (… Present[E0] … sitting[Ei] …)][∃E1 : N[E0,E1]] (… Present[E1] … standing[Ei] …) (95) Kunstler is sitting, and Kunstler is standing. Kunstler is sitting, and he is standing. [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … sitting[Ei] …), and [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … standing[Ei] …) Recall that for (94) and the like, the essential role of reference to events antecedently described is corroborated by its failure under “modal insurbordination”: (150) ?No child is watching Sesame Street and is looking unhappy. (159) *No child is ever watching Sesame Street and is looking unhappy. (152) ?No child has watched Sesame Street and has gone to bed unhappy. (161) *No child has ever watched Sesame Street and has gone to bed unhappy. (154) ?No student is reading the text and is taking notes on what she reads. (163) *No student is ever reading the text and is taking notes on what she reads. In (159) and (161), the event pronominal introducing the second clause attempts reference to an antecedent ever watching Sesame Street or ever having watched Sesame Street, and similarly in (163), to ever reading the text. The tacit event pronouns behave like the overt pronoun in (165), which also fails when its antecedent is replaced with a negative-polarity item in (166): (164) No one saw anything suspicious and went to the police. No one saw a single suspect and went to the police. (165) No one saw something suspicious and reported it to the police. No one saw a suspect and reported him to the police. No one saw a suspect and reported it to the police. (166) *No one saw anything suspicious and reported it to the police. *No one saw a single suspect and reported him to the police. *No one saw a single suspect and reported it to the police.
246
Chapter 3
It is uncontroversial that Tense is about events (or time). What is perhaps novel is the view that the morpheme expresses only a temporal relation and does not itself come complete with the necessary reference or quantification over events or times, which is instead accomplished elsewhere with the intervention of tacit adverbs and pronouns. The tense morpheme itself only exposes an event variable that the rest of the structure then has to trouble with. Based on analogy to tensed clauses, it can now be argued that participial clauses also contain a morpheme—obviously other than Tense—that exposes an event variable. This morpheme occupies a position high enough in the participial clause to itself sometimes require a pronoun referring to antecedently described events to close it off. Having conceived a meaning for a dependent clause expressly to overcome the absence of an auxiliary verb in (223) and (224), one must wonder what is left to keep Weinglass not standing dependent in (225) other than stipulating it so: (223) Kunstler is standing and Weinglass not standing. (224) (With) Weinglass not standing, Kunstler is standing. (225) *Weinglass not standing. There is, as far as I can see, no escape from some stipulation that the structure is one way rather than another. In a logical syntax held together only by cross-reference to events, it could be that a dependent phrase is just a formula that would otherwise strand an event variable unbound if it did not occur in its licensed contexts (see chapter 6 for further discussion of complementation). Consider the sentences (226)–(228), discussed again at greater leisure in section 5.0, and a contrast that holds between (226) and (228). The sentences describe the behavior of tires on a slalom course alternating between skid and slip. Sentence (226) describes the alternation of a single tire as it skids (greater-than-normal road contact) when it is inside a turn and slips (less-than-normal road contact) when it is outside. Sentence (227), in a now familiar contrast, describes a physical impossibility, a tire that skids and slips at the same time. The coordination of tensed auxiliary phrases without pronunciation of a second subject compels the second conjunct to refer to the events described by the first, as above. (226) The tire is skidding and slipping. (227) The tire is skidding and is slipping. (228) a. The left front tire is skidding and the right front tire slipping. b. The left front tire is skidding and the left rear tire slipping. In (228), overt pronunciation of a participial subject also proves sufficient to compel reference to the events described by the first conjunct. The sentences cannot describe an alternation in the states of two tires. Rather, the states described by the conjuncts are simultaneous. Sentence (228a) describes what is true in a controlled slalom at
PredP and PredP: Of Subjects and Ancient Grievance
247
every other turn where the left tires are in a skid and the right tires are in a slip. In contrast, sentence (228b) only describes catastrophe, a moment at which one left tire is in a skid and the other in a slip, and the car is out of control. A similar contrast involving the perfect construction (see section 5.1.2) again tells apart the coordination of participial phrases with pronounced and unpronounced subject. The sentences with subjects pronounced imply some causal or circumstantial connection between the events the conjuncts describe: (229) The wind has battered the window and the rain gotten in. (230) The clouds have parted and the sun come out. (231) The darkness has retreated and the sun come out. Failing to grasp what that connection could be makes the like structures in (232)– (235) infelicitous: (232) #The wind has battered the window and the doorbell rung. (233) #The wind has battered the window and the rain gotten in through the chimney. (234) #The waters have parted and the sun come out. (235) #The flood has retreated and the sun come out. It is implied by (232) that battering the window has somehow caused the doorbell to ring. Some such connection is to be expected if the second clause is modified by reference to the first, as if (229)–(235) were tacit counterparts to (236)–(242): (236) The wind has battered the window and the window battered, the rain gotten in. (237) The clouds have parted and the clouds parted, the sun come out. (238) The darkness has retreated and the darkness retreated, the sun come out. (239) #The wind has battered the window and the window battered, the doorbell rung. (240) #The wind has battered the window and the window battered, the rain gotten in through the chimney. (241) #The waters have parted and the waters parted, the sun come out. (242) #The flood has retreated and the waters retreated, the sun come out. No such causal connection is required when the participial subject is unpronounced, no matter what else may be implicated about the temporal sequence of the events described. Thus, (243), in a minimal pair with (232), does not imply that the events described by the first conjunct condition those of the second: (243) The wind has battered the window and rung the doorbell. Rather, the wind battered the window, perhaps first, and it also, later or perhaps concurrently, rang the doorbell.
248
Chapter 3
Consonant with what was found in the coordination of tensed auxiliary phrases, reference to antecedent events can fail (“modal insubordination”) if the antecedent description contains a negative-polarity item: (244) No one has lifted a finger and gotten the work done. No one has lifted a finger and gotten any work done. (245) *No one has lifted a finger and the work gotten done. ??No one has lifted a finger and any work gotten done. (Cf. (246) No one has lifted a magic wand and the work gotten done. No one has lifted a magic wand and any work gotten done (that way).) (247) No don has lifted a finger and broken up the plot against him. (248) *No don has lifted a finger and the plot against him broken up.24 (Cf. (249) No don has nodded and the plot against him broken up.) (250) No one has ever budged an inch and gotten anything. (251) *No one has (ever) budged an inch and compromise gotten him anything. (Cf. (252) No one has compromised and goodwill gotten him anything.) Pronouncing the participial subject in (245) forces reference to the antecedent events, in contrast to (244), which does not, and reference fails as it does in (253): (253) *No one has lifted a finger and lifting a finger, the work gotten done. Since these examples coordinate participial phrases, the cross-reference is to the events described by the antecedent participial phrase. The offending negative polarity must be contained within the antecedent description. Those, such as ever in (250), including within their scope the entire participial coordination do not undermine the cross-reference within.25,26 To assimilate the treatment of coordinated participial phrases to that of tensed auxiliary phrases—to uncover an exposed event variable when the participial subject is pronounced—amounts to parsing (223) as something like (254), and (229) as (255). This is as if the pronunciation of the participial subject were always to indicate an absolutive construction requiring a complementizer, like with or its null counterpart in (224), presumably to assign Case27 and relate the events described by the absolutive to those described by the matrix. (223) Kunstler is standing and Weinglass not standing. (254) Kunstler is standing and that with Weinglass not standing. Kunstler is standing and therewith Weinglass not standing.
PredP and PredP: Of Subjects and Ancient Grievance
249
(229) The wind has battered the window and the rain gotten in. (255) The wind has battered the window and with that the rain gotten in. The wind has battered the window and therewith the rain gotten in. (224) (With) Weinglass not standing, Kunstler is standing. This suggestion treats the coordination as asymmetric in its structure as in (256), where ‘therewith[E]’ is the term exposing an event variable (cf. ‘Tense[E]’ in coordinations of tensed auxiliary phrases):28 (256) DPi Tense Aux [[ti W O -en V] and [therewith[E] DP W O -en V]] What is important for the meaning of the construction is that the first conjunct not contain an event variable unbound within it, since the effect of referring to events described by the other conjunct is observed only in the second. To this end, the asymmetric structure (256) simply omits ‘therewith[E]’ from the first conjunct.29 The meaning of this element is as the gloss suggests, some causal or circumstantial relation. Participial clauses with pronounced subjects are always qualified by a relation and reference to the events described by the antecedent participial clauses. Participial phrases without pronounced subjects are exempt from this condition, absent ‘therewith[E]’ and the Case position of the absolutive construction. 3.2.1.3 Quantifier Lowering and scope inversion
The interpretation of (68) (repeated below) derived from inverting the object quantifiers with tokens of a rocker argues, following Fox (1995, 2000), for the constant presence of a position for subjects within the conjoined participial phrases, even in (67), from which (68) hardly differs: (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. (70) All night long, a rocker has shimmied through the bumps and a rocker shaken on the grinds. Why then, goes the ancient grievance against Conjunction Reduction, doesn’t (67) share an interpretation with (70)? One structure for (67), already encountered in section 3.2.1.0 in the discussion of (187), is the only one that can derive the collective, disjunctive interpretation for (187). Schematized in (209) and shown in (206), the smallest coordination of participial phrases serves as a (backward) antecedent for a pronoun referring collectively to some shimmying and some shaking: (187) Twenty rockers have shimmied and shaken. (209) DPi Tense Aux W O [[-en V] and [-en V]]
250
Chapter 3
(206) [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] … W[E1,X][℩E1:pro1][℩E2:ΦΨ]O[E1,E2] (Φ∃E2∃XAgent[E2,X] … shimmied …) and (Ψ∃E2∃XAgent[E2,X] … shaken …) Recall that the phrase in boldface asserts of this shimmying and shaking that it coincides with what twenty rockers and no others do. Because existential quantifiers are already the subjects of these conjuncts and the meaning derived is still of just some twenty rockers, it hardly matters that the subject itself might reconstruct into these positions as in (207): (207) [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] … W[E1,X][℩E1:pro1][℩E2:ΦΨ]O[E1,E2] (Φ∃E2[∃X : 20(X) ∃E rocker[E,X]]Agent[E2,X] … shimmied …) and (Ψ∃E2[∃X : 20(X) ∃E rocker[E,X]]Agent[E2,X] … shaken …) As remarked in section 3.2.1.0, such a reconstruction would insist that the shimmiers are twenty and so are the shakers—but still the same twenty rockers according to the boldface phrase, thereby deriving instead the equally good conjunctive interpretation of sentence (187). Neither (206) nor (207) mistakes (187) for (189): (189) Twenty rockers have shimmied and twenty rockers shaken. As in section 3.2.1.0, when the subject is pronounced in the second participial conjunct, the coordination is itself larger, capturing within it the boldface phrases: (189) Twenty rockersi have [[ti shimmied] and [twenty rockers shaken]] (208) DPi Tense Aux [[ti W O -en V] and [DP W O -en V]] (205) [∃X : 20(X) ∃E rocker[E,X]] … W[E0,X] … Present[E0] … Have[E0,E1] ( … W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shimmied …) and ([∃X : 20(X) ∃E rocker[E,X]] … W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shaken …) But now the analysis required by (189) affords an alternative parse for (187), where larger phrases are coordinated: (257) Twenty rockersi have [[ti shimmied] and [ti shaken]] DPi Tense Aux [[ti W O -en V] and [ti W O -en V]] Section 3.2.1.0 took pains to point out that the twenty rockers’ mere having in (189) is no license to infer that these twenty were both the shimmiers and shakers there, and correctly so since (189) allows the shakers to be some other twenty rockers. If so, reconstruction in (257) threatens (187) with the interpretation (258) that it should
PredP and PredP: Of Subjects and Ancient Grievance
251
not share with (189) (cf. (205)), where the strike-through in (258) records that the participial subjects are unspoken copies: (258)
… W[E0,X] … Present[E0] … Have[E0,E1] ([∃X : 20(X) ∃E rocker[E,X]] … W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shimmied …) and ([∃X : 20(X) ∃E rocker[E,X]] … W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shaken …)
What spares (187)/(258) from meaning what (189)/(205) means is highlighted. The two tokens of ‘[℩E1:pro1]’ refer to what twenty rockers do (‘W[E1,X]’). Since the two (unspoken) tokens of twenty rockers are themselves constrained to be identical copies, the descriptive content of these event pronouns is also identical. Moreover, neither event pronoun (nor their antecedent descriptions of events E1) falls within the scope of an event quantifier that does not also include the other. These event pronouns therefore refer to the same events, what some twenty rockers and no others do. These events are then said to coincide both with some shimmying and with some shaking, deriving again a conjunctive interpretation, applied to the same twenty rockers as desired. With what has accumulated to distinguish the structure of (189) with its second subject pronounced from that of (187) with it unpronounced, there is little danger that (187) and (189) will share an interpretation. Pronouncing twenty rockers once and then pronouncing it again exempts these tokens from being exact copies of one another, and indeed the presumption is to assign them different addresses in any given frame of reference.30 Thus, an event pronoun describing what the first twenty rockers did will not coincide with one describing what the second twenty rockers did. Moreover, the pronunciation of the second subject cues its articulation—that is, adverbialization frames the second conjunct and locates the second token of the event pronoun ‘[℩E1:pro1]’ within the scope of an event quantifier that does not include the first. The event pronoun’s two tokens denoting what some twenty rockers did, do so “sloppily,” each for perhaps a different twenty, as desired: (259)
… W[E0,X] … Present[E0] … Have[E0,E1] ([∃X : 20(X) ∃E rocker[E,X]] … W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shimmied …) and ([∃X : 20(X) ∃E rocker[E,X]][℩E: ∃X(20(X) rocker[E,X])] [∃E1 : N[E,E1]]W[E1,X][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] … shaken …)31
252
Chapter 3
There is yet more structure to tell apart (187) and (189) if, as argued in section 3.2.1.2, the participial clause (260) with subject pronounced projects at its left edge a relation, therewith, with that, thereupon, or as a result, demanding anaphoric reference to the events antecedently described: (260) The wind has battered the window and the rain gotten in. ‘The wind has battered the window and with that the rain gotten in.’ Twenty rockers have shimmied and twenty rockers shaken. ‘Twenty rockers have shimmied and with that twenty rockers shaken.’ (261) DPi Tense Aux [[ti W O -en V0] and [[℩E0:pro0][℩E1:pro1]With[E0, E1] [DP W1 O -en V]] The divergence in descriptive content and structure that serves the meaning of (189) is, however, exactly what is proscribed when the two tokens of the subject are identical copies, as is the case in (187) when one goes unpronounced. These observations taken together answer why (67) does not share an interpretation with (70). (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. (70) All night long, a rocker has shimmied through the bumps and a rocker shaken on the grinds. The conjuncts of (67) contain some definite descriptions that refer to events. The descriptive content of these terms is identical because the descriptive contents of the subjects of these conjuncts are themselves identical, being copies of one another. These descriptions of events all fall within the scope of the same quantifiers. Evaluated with respect to the same parameters, these descriptions refer to the same events. But these events are those that coincide both with the shimmying and with the shaking, and thus it is the same participant, a single rocker, who does it all. In contrast, in (70), something that distinguishes them from some point of view is understood to supplement the descriptive content of the two tokens of a rocker (see note 30), so that the descriptions of what they do are also different and describe different events, one of which is a rocker shimmying and the other of which, a rocker shaking. It now remains to be seen how Quantifier Lowering and scope inversion in (68) allow for different shimmiers and shakers despite lowering into the conjuncts copies of the same content just as in (67). Start with the observation that (262) is a fair report that the mojitos are stirred simultaneously: (262) Ernest has simultaneously stirred every mojito.
PredP and PredP: Of Subjects and Ancient Grievance
253
Equally, (263) may report of a single bartender that her mojitos are stirred simultaneously: (263) A bartender has simultaneously stirred every mojito. But (263) does not allow scope inversion within the scope of the adverb to obtain a report, as it were, that every mojito has been stirred by a (different) bartender and the mojitos stirred simultaneously. It is no fault of the adverb per se to refuse this interpretation, since this is exactly one of the meanings of (264) (see note 18): (264) Simultaneously has a bartender stirred every mojito. Scope inversion may itself occur outside the scope of the adverb in (265) to convey that for every pair of mojitos, there is a bartender who has simultaneously stirred that pair: (265) A bartender has simultaneously stirred every pair of mojitos. Recall that section 3.2.1.0 locates a Case position for participial subjects to the left of the adverb: (263) A bartender has simultaneously stirred every mojito. (266) A bartenderi Tense have [ti W simultaneously O -en stir every mojito] (cf. (208)) The interpretation denied (263) reflects a constraint on the syntax of lowering: lowering is to the Case position shown in (266) with no further trespass into the scope of the adverb. (When we return to coordination, it is structures like (187)/ (258) that will concern us, since only the larger coordinations provide within the conjuncts the Case positions that are the landing sites for lowering. The structures mentioned earlier, (209)/(206), combining reconstruction with the narrowest coordination of participial phrases are ruled out on syntactic grounds from the evidence of (263).) The lower bound imposed in (263) dovetails with an argument that all scope inversion involves at least some lowering. According to Johnson and Tomioka 1997 and Johnson 2000b (see also Hornstein 1995), scope inversion in a simple sentence such as (267) or (269) requires two movements: raising the quantifier from object position and, crucially, a lowering of the subject quantifier to a position low enough for negation to disrupt it, as in (268) and (270): (267) Some bartender has stirred every mojito. (∃∀, ∀∃) (268) Some bartender hasn’t stirred every mojito. Some bartender has not stirred every mojito. (∃¬∀, ∃∀¬, *¬∃∀, *¬∀∃, *∀∃¬, *∀¬∃)
254
Chapter 3
(269) Some linguist speaks some language from every family. [some linguist][every family][some language] [every family][some linguist][some language] (Johnson and Tomioka 1997) (270) Some linguist doesn’t speak some language from every family. [some linguist][every family][some language] not *[every family][some linguist][some language] not (271) Ken Hale doesn’t speak some language from every family. (∀∃¬) Note the freedom of the object quantifier to raise outside negation in (268) (“∃∀¬”), (270), and (271) shows that negation must instead constrain the subject’s lowering, and the position to which it must lower in the course of scope inversion occurs to the right of negation. Thus the position to which the subject is lowered is both to the left of simultaneously according to the interpretation of (263) and to the right of negation according to this last argument—at ti in (272): (272) A bartender hasn’t ti simultaneously stirred every mojito. Presumably such other positions for negation as might be imagined, as in (273), involve secondary or marginal occurrences that are not involved in the judgments elicited for (268) and (270): (273) Standing upright, even Harry hasn’t simultaneously not stood on his left leg and not stood on his right leg. With the landing site for lowering now fixed, scope inversion in (68) will move the object quantifiers to some superior position to include within their scope the lowered subjects, with the result in (274): (68) A rocker has shimmied through every bump and shaken on every grind. (274) … W[E0,X] … Present[E0] … [℩E0:pro0][℩E1:ΦΨ] Have[E0,E1] (Φ[every y : ∃E bump[E,y]][℩E : bump[E,y]][∃E0 : N[E,E0]] [A x : ∃E rocker[E,x]][℩E :rocker[E,x]][∃E1 : N[E,E1]] W[E1,x][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XAgent[E2,X] shimmied[E2] W[E0,y] O[E0,E3] ∃YGoal[E3,Y]through[E2,E3] ) and (Ψ[every y : ∃E grind[E,y]][℩E : grind[E,y]][∃E0 : N[E,E0]] [A x : ∃E rocker[E,x]] [℩E :rocker[E,x]][∃E1 : N[E,E1]] W[E1,x][℩E1:pro1][℩E2:pro2]O[E1,E2]∃E2∃XAgent[E2,X] shaken[E2] W[E0,y] O[E0,E3] ∃Y Location[E3,Y]on[E2,E3] ) As before, the highlighted tokens of ‘[℩E1:pro1]’ refer to what a rocker does (‘W[E1,X]’). Since the two (unspoken) tokens of a rocker are themselves
PredP and PredP: Of Subjects and Ancient Grievance
255
constrained to be identical copies, the descriptive content of these event pronouns is still identical. But now the event pronouns fall within the scope of different event quantifiers, those accompanying the distributive quantifiers, every bump and every grind. The first token of ‘[℩E1:pro1]’ is thus evaluated relative to an individual bump and refers to what a rocker does there, in the horizon of that bump, and similarly, the second token refers elsewhere, to what a rocker does there within the horizon of an individual grind. Referring to different events, these pronouns do not compel the shaking and the shimmying to coincide in the action of a single rocker. 3.2.1.4 Number agreement on the lam
The Case position to which the subject is lowered (section 3.2.1.3) is also a landing site for number agreement. If singular number agreement is semantic and therefore referring to the one, it must itself lower to within the scope of the universal quantifier in (275) if its reference is not to fail after Quantifier Lowering and scope inversion have applied, and similarly, in the Lebanese (276), for which it has been argued in chapter 2 that each woman does not raise to include within its scope the tensed verb. If number agreement is not lowered to fall within the scope of the universal quantifier, there will not be a one for it to refer to: (275) A senator is likely to attend every rally. (276) aryet kell mara essa read.3fs each woman story According to chapter 2, number agreement so-called is just another definite description and so its mobility should not be any more perplexing than lowering any other definite or indefinite description. At least, the logical syntax has cast number agreement in such a way that one can envision it participating in such a process. In section 6.3, I suggest that unspoken event pronouns undergo a process akin to clitic climbing in deriving the disjunctive interpretation when simple, tensed Verb Phrases are coordinated. Number agreement must arguably lower, provided that it is indeed semantic and remains so even after its agreeing subject has lowered, rather than defaulting in that case to, say, whatever the reference of it is in It rains. As often remarked, it appears to have semantic force at least in some contexts. Just as the singular pronoun in (277) succeeds only because Superman is Clark, singular number agreement appears to succeed in (278) on the same grounds: (277) Superman did the rescue. Clark wrote the story. And then, after a long day, he rested. (278) Her protector and her confidant watches over Lois Lane day and night.
256
Chapter 3
Note that nothing in the logical form of the antecedent sentences of (277) hints at the identity or the existence of only one actor, but unless one knows this, there will be no accommodation of the singular reference that is entirely the burden of the singular pronoun to have imposed. The conclusion that (278) is alike in containing only a single device of singular reference identifying Superman and Clark may be forestalled if (278) and its kind are translated to resemble (279) (Winter 2002): (279) Someone who is her protector and her confidant watches over Lois Lane day and night. If always so translated, it could be said that number agreement, unlike the singular pronoun in (277), holds to (280): (280) Number agreement is singular only if it has a unique antecedent and the antecedent refers to or describes a single thing. If so, it would vacate the conclusion that singular number agreement is semantic and itself lowers in (275) and (276). It could be said instead that it is merely a morphological reflex of the subject’s form, meaning nothing more than the expletive it in It rains. Certainly, a sentence such as (278) entails singular reference, but only because the subject itself says as much under the form in (279). If such a subject should need to lower because its singular reference can only be construed as such within the scope of lower quantifiers, there is no surprise there having already conceded as much, namely, Quantifier Lowering. An argument now follows that singular number agreement, contrary to (280), sometimes splits its antecedency, like the pronoun in (277). It remains that singular number agreement succeeds only if its antecedents refer to the same one thing, a condition that in this case could only reflect its own requirements as a singular definite description. In Moore’s (1999) Good Luck Bar (see section 10.1), Superman sits attired as the mild-mannered reporter Clark Kent. Despondent and hoping to double his chances for a date, Clark has decided to try men as he strikes up a feeble, halting conversation at the bar with someone handsome enough. At the same time, he is in conversation qua Superman on his mobile phone with a woman who will date Superman. That is, (281) Clark Kent fecklessly and Superman adroitly has been chatting up his date. Clark Kent fecklessly and Superman adroitly is chatting up his date. (282) A mild-mannered reporter fecklessly and a superhero adroitly has been chatting up his date. A mild-mannered reporter fecklessly and a superhero adroitly is chatting up his date.
PredP and PredP: Of Subjects and Ancient Grievance
257
The intruding adverbs foreclose on a unique antecedent, (283), as does any hope for a “sloppy” interpretation of his date to refer, as called for, to the man at the bar and the woman on the phone, (284).32 (283) *Someone who is Clark Kent fecklessly and Superman adroitly … *Someone who is a mild-mannered reporter fecklessly and a superhero adroitly … (284) #Someone who is Clark Kent and Superman has been chatting up his date. #Someone who is a mild-mannered reporter and a superhero is chatting up his date. The evidence of (281) and (282) by itself does not yet complete the argument against (280). For reasons to be reviewed shortly, it must also be judged that these same structures demonstrate the original effect underlying the claim that (singular) number agreement is semantic. So, suppose now that Lex Luthor, Superman’s archenemy, and Superman are both at the Good Luck Bar. Superman, as ever, is charming a woman, and Lex Luthor, evil genius, is not genius enough to realize that the woman he is hitting on—maybe or not the same one charmed with Superman—is too good for the likes of him. Knowing that Lex Luthor is not Superman and that no supervillain is a superhero, one should balk at the singular number agreement in (285) and (286): (285) *Lex Luthor fecklessly and Superman adroitly has been chatting up his date. *Lex Luthor fecklessly and Superman adroitly is chatting up his date. (286) *A supervillain fecklessly and a superhero adroitly has been chatting up his date. *A supervillain fecklessly and a superhero adroitly is chatting up his date. The contrast between (281)–(282) and (285)–(286), if one is recognized, secures the argument. For, if the contrastive judgment is a judgment about the same structure evaluated under different circumstances, there is nothing but the reference of singular agreement itself that varies with circumstance. Its antecedents are split and properly contained within separate phrases, and both antecedents themselves refer under all circumstances. The finding that judgments are contrastive has to be elicited with some care, however. Recall from section 2.3.2.0 that singular agreement is tolerated in (287) without intending Robin to be Hillary, if supplied, as the commas indicate, with parenthetical intonation:33 (287) a. Robin (earlier today), and Hillary yesterday, was drinking more bordeaux (*between them) than the region produces in a week.
258
Chapter 3
b. Robin (as of this morning), and Hillary as of last night, has drunk more bordeaux (*between them) than the region produces in a week. In contrasting (281)–(282) and (285)–(286), intonation must be calibrated. In particular, in judging (281)–(282) acceptable, one must take care not to slip into the parenthetical intonation that would also render (285)–(286) acceptable.34 It does not seem to me to be an especially subtle judgment, and examples (281)–(282) and (285)–(286) have been designed to help. They are all reports of a single scene under direct observation, and the manner adverbs, unlike the temporal adverbs in (287), do not easily commandeer Tense to put their events in different locations.35 To my ear, there is clearly a nonparenthetical intonation in the course of intoning which one goes along with the singular number, Clark being Superman, and balks at it because Lex Luthor is not. As further evidence, consider that the singular number agreement referring to Clark and Superman may be collective, referring to Clark and Superman, yet singular, because they is one, as it were (see note 32). Only under this collective use of the singular can it be said that two dates are approached: (288) Clark Kent fecklessly and Superman adroitly has been chatting up two dates. (289) A mild-mannered reporter fecklessly and a superhero adroitly has been chatting up two dates. (290) Clark Kent fecklessly and Superman adroitly has been chatting up their dates. (291) A mild-mannered reporter fecklessly and a superhero adroitly has been chatting up their dates. If they be not one (292)–(299), forcing a parenthetical intonation, or if the structure itself precludes collective, singular number (300)–(307), it is then implied that each character approaches two on his own:36 (292) Lex Luthor fecklessly, and Superman adroitly, has been chatting up two dates. (293) A supervillain fecklessly, and a superhero adroitly, has been chatting up two dates. (294) Lex Luthor fecklessly, and Superman adroitly, has been chatting up their dates. (295) A supervillain fecklessly, and a superhero adroitly, has been chatting up their dates. (296) Lex Luthor just now, and Superman earlier, was chatting up two dates. (297) A supervillain just now, and a superhero earlier, was chatting up two dates.
PredP and PredP: Of Subjects and Ancient Grievance
259
(298) Lex Luthor just now, and Superman earlier, was chatting up their dates. (299) A supervillain just now, and a superhero earlier, was chatting up their dates. (300) Clark Kent has fecklessly and Superman has adroitly been chatting up two dates. (301) A mild-mannered reporter has fecklessly and a superhero has adroitly been chatting up two dates. (302) Clark Kent has fecklessly and Superman has adroitly been chatting up their dates. (303) A mild-mannered reporter has fecklessly and a superhero has adroitly been chatting up their dates. (304) Clark Kent just now was, and Superman earlier was, chatting up two dates. (305) A mild-mannered reporter just now was, and a superhero earlier was, chatting up two dates. (306) Clark Kent just now was, and Superman earlier was, chatting up their dates. (307) A mild-mannered reporter just now was, and a superhero earlier was, chatting up their dates. The contrast between (288)–(291) and (292)–(307) is further proof that a single token of singular number agreement in (288)–(291) is the only effective instrument of singular reference taking two for one, thus falsifying (280). Must singular number agreement now lower, trailing after its subject(s)? Well, almost. Ever in retreat, it could still be advanced that singular agreement is sometimes expletive, as in (278) analyzed as (279), and, concessively, sometimes not—when the presence of adverbs indicates the contrary. It remains to be seen that singular number agreement in exactly these contexts where it has been found to be referential also participates in Quantifier Lowering. Thus, alongside the contrast between (308) and (309), we find the contrast between (310) and (311) in this political season and that between (312) and (313) when all of Caesar’s assassins were of two minds like Brutus, sincere friends of Caesar who loved Rome more: (308) A protector and a confidant is likely to watch over every comic book heroine. (309) A protector and a confidant are likely to watch over every comic book heroine. (310) A spouse cautiously and a colleague enthusiastically is likely to support every current senator’s bid for reelection. (311) A spouse cautiously and a colleague enthusiastically are likely to support every current senator’s bid for reelection.
260
Chapter 3
(312) A lover of Rome eagerly and a friend of Caesar’s reluctantly is likely to approach him from every corner of the Senate floor on the Ides of March. Some alleged lover of Rome eagerly and some so-called friend of Caesar’s reluctantly (so says he) is likely to approach him from every corner of the Senate floor on the Ides of March. (313) A lover of Rome eagerly and a friend of Caesar’s reluctantly are likely to approach him from every corner of the Senate floor on the Ides of March. Some alleged lover of Rome eagerly and some so-called friend of Caesar’s reluctantly (so says he) are likely to approach him from every corner of the Senate floor on the Ides of March. The argument finishes with these contrasts. Like the context for the singular pronoun in (277), singular number agreement is the only element in (310)–(313) whose success in referring depends on whether they whom it refers to be one. That condition survives, as reflected in these contrasts, even when its antecedents are lowered within the scope of other quantifiers, and that condition can only be satisfied there if number agreement follows suit and is also lowered.37 3.2.1.5 Focus and reconstruction into subject position
That (67) (repeated here) does not mean the same as (70) has been bedrock for the ancient grievance against Conjunction Reduction:38 (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. (70) All night long, a rocker has shimmied through the bumps and a rocker shaken on the grinds. In replying to it—or rather to the conclusion that subjects are absent and coordination, predicative in (67)—Fox (1995, 2000) and I have for the sake of argument accepted at face value what the contrast between (67) and (70) must be taken to represent if it is to serve as even the beginning of an objection to Conjunction Reduction. We have so far taken it for granted that structures like (67) never mean the same as their counterparts like (70). German is now famous for a contrast between (314) and (315) discussed by Höhle (1990, 1991) and Büring and Hartmann (1998), among others:39 (314) Eine Frau ist in Amerika Außenministerin und bekleidet in Deutschland sogar das zweithöchste Amt des Staates. ‘A woman is in America the Secretary of State and occupies in Germany even the second-highest office in the land.’
PredP and PredP: Of Subjects and Ancient Grievance
261
(315) In Amerika ist eine Frau Außenministerin und bekleidet in Deutschland sogar das zweithöchste Amt des Staates. ‘In America is a woman the Secretary of State and occupies in Germany even the second-highest office in the land.’ Sentence (315),40 not unlike (67), insists that the same woman is a government official in both America and Germany, but (314) allows them to be different, an effect according to Höhle (1990, 1991) of undoing topicalization per se to reconstruct the topicalized eine Frau into both conjuncts. The German prompts the observation that near translation into English (316) also happens to allow for the officials to be different, in contrast to expectations inherited from (67), and more interestingly in contrast to (317), which once again forces the officials to be the same: (316) A neoliberal technocrat governs Britain and presides in France. (And (in each country) he rules with a velvet glove over an iron fist.) (317) a. A neoliberal technocrat won a seat yesterday in the House of Commons and defeated an agrarian opponent in the National Assembly. (And (in both elections) he conducted an American-style campaign.) b. Last year, a neoliberal technocrat won a seat in the House of Commons and defeated an agrarian opponent in the National Assembly. (And (in both elections) he conducted an American-style campaign.) The shift in aspect from the stative/habitual in (316) to the episodic in (317) plays a role, mentioned below, but it is not decisive for the contrast, which can be reversed in other contexts. So, Uncle Walter, in high spirits at family gatherings, intones (318) and (319) with pride: (318) A Jew invented the polio vaccine and discovered relativity. (319) A Jew founded the first bank in the New World and organized its first labor union. The aspect of (318) and (319) is clearly episodic throughout, and yet the sentences convey Uncle Walter’s boast that the inventor of the polio vaccine was a Jew, the discoverer of relativity was a Jew, and so were the founders of the first bank and the first labor union, without implying that any of them were the same Jew. On the other hand, with the next example, a sentence in stative aspect nevertheless implies that the subjects are the same. We have only to imagine that we are hearing the traffic helicopter announce (320) for the first time: (320) A multicar pileup blocks the southbound Major Deegan Expressway at the Triborough Bridge and narrows the westbound Bruckner Expressway to a single lane.
262
Chapter 3
A single accident has obstructed both expressways; (320) cannot be taken as a report that one multicar pileup blocks one expressway and (perhaps) another narrows the other. In the sentences that escape implying that the subjects are the same, the predicate phrases describe what is presupposed. That is, in the elicited contexts for these sentences, it is known that Britain is governed, France presided over, the polio vaccine already invented, relativity discovered, the first New World bank founded, and its first labor union already organized, and the force of these sentences is to assert something about who did these familiar things. Where the subject is implied to be the same, the predicate phrase relates no such presupposition. Rather, the sentence as a whole reports some news; answering an implicit question “what happened?,” it asserts that news. We don’t know that a seat or anything else was won yesterday in the House of Commons or National Assembly. It would be infelicitous without a special context that invites accommodation to refer as if we knew otherwise— #What’s new? The win yesterday in Commons was by a neoliberal technocrat. It is often the case that the answer to “What happened?” or “What’s new?” will be something episodic, hence the ready contrast between (316) and (317), but this is not necessary. The normal force of (320) is to inform the radio audience that there is now a state of affairs in which the Major Deegan and Bruckner Expressways are in trouble. There is no presupposition that this is so—except perhaps among the traffic controllers sent to investigate the causes of previously reported accidents. They, unlike the radio audience, could take (320) to report their finding that blocking the southbound Major Deegan Expressway is a multicar pileup and narrowing the westbound Bruckner Expressway is one, too. If the above description is correct, I should join Jackendoff (1972) and more specifically Herburger (2000) in surmising that presupposition focus is reflected directly in logical form. The indefinite descriptions reconstructed into the lower subject positions come to describe different things just in case—as always, by now— they happen to fall within the scope of distinct event quantifiers, which here are those definite descriptions of events whose content is what is presupposed.41 Of course the presupposition-focus structure of any sentence is subject to the vagaries of context, and if any sentence can give rise to interpretation that reveals the reconstruction of its subject, then no sentence, not even (67), shelters predicative coordination. 3.2.2
Opacity in coordinate structures
At the top, there is no escape for a quantifier from its conjunct ((118), (119) (here repeated)), illusions to the contrary ((321), (322)) being an effect of “telescoping” (section 3.2.0):
PredP and PredP: Of Subjects and Ancient Grievance
263
(118) a. *No rocker has shimmied, and his mama has shaken to that funky disco music. b. *No child has (ever) watched Sesame Street, and her teddy bear has gone to bed unhappy. (119) a. *No rocker is shimmying, and his mama is shaking to that funky disco music. b. *No child is watching Sesame Street, and her teddy bear is falling asleep. (321) a. Every rocker has shimmied, and his mama has shaken to that funky disco music. b. Every child has watched Sesame Street, and her teddy bear has gone to bed happy. (322) a. Every rocker is shimmying, and his mama is shaking to that funky disco music. b. Every child is watching Sesame Street, and her teddy bear is falling asleep. Coordination at the top includes the coordination of tensed auxiliary verb phrases with pronounced subjects and all their derivatives by Right-Node Raising: (323) *No rocker has (ever) shimmied to that funky disco music and (any of) his mama(s) has (ever) shaken to it (too). (324) *No rocker has (ever) shimmied and (any of) his mama(s) has (ever) shaken to that funky disco music. (325) *No rocker has and (any of) his mama(s) has (ever) shimmied and shaken to that funky disco music. (326) *No rocker and (any of) his mama(s) has (ever) shimmied and shaken to that funky disco music. Thus, in addition to (118)–(119), (323)–(326) all confine their subjects’ scope to their conjuncts. At the top, both subjects are pronounced, both are therefore adverbializing, and in every other respect the conjuncts’ structures are symmetric and of equal dimension. All of (118)–(119) and (323)–(326) show singular number agreement. The two tokens in the conjuncts would accordingly have to be interpreted to denote different events. In contrast, when number agreement as pronounced within the Right-Node Raised phrase is plural and collectivized, the quantifier slips from the first conjunct to include the entire coordination within its scope, as in (327)–(331), (326) and (327) providing minimal pairs: (327) No rocker and (any of) his mama(s) have (ever) shimmied and shaken to that funky disco music.
264
Chapter 3
(328) No soldier and (any of) his weapon(s) have ever been separated. (329) No owner and her prize pooch have entered the judging ring tugging at opposite ends of the leash. (330) No wolf mother and any of her cubs have remained close for very long. (331) No Park Avenue therapist and any of her patients have invested in Wall Street together. Sentences (327)–(331) are instances of Right-Node Raising in the extreme explored in chapter 2, which leaves behind nothing spoken except for what can be mistaken for a coordination of DPs. In case it is forgotten that this is a mistake, the following sentences interpose adverbs (the slippery slope to Conjunction Reduction), and yet the contrast between singular and plural number agreement survives, (332) and (333) providing again a minimal pair: (332) Not any student of mine yesterday and any one of yours today have conjectured and proven (between them) p and not p. No student of mine in one instance and any one of yours in the next have conjectured and proven (between them) p and not p. (333) *Not any student of mine yesterday and any one of yours today has conjectured and proven (between them) p and not p. *No student of mine in one instance and any one of yours in the next has conjectured and proven (between them) p and not p. (334) *Not any student of mine (yesterday) has (yesterday) and any one of yours (today) has (today) conjectured and proven (between them) p and not p. *No student of mine (in one instance) has (in one instance) and any one of yours (in the next) has (in the next) conjectured and proven (between them) p and not p. (335) *Not any student of mine (yesterday) has (yesterday) conjectured and any one of yours (today) has (today) proven p and not p (between them). *No student of mine (in one instance) has (in one instance) conjectured and any one of yours (in the next) has (in the next) proven p and not p (between them). (336) *Not any student of mine (yesterday) has (yesterday) conjectured p and any one of yours (today) has (today) proven not p. *No student of mine (in one instance) has (in one instance) conjectured p and any one of yours (in the next) has (in the next) proven not p. Just below coordinations at the top, in a coordination of tensed auxiliary phrases with the second subject unpronounced, a quantifier may include the entire coordination within its scope ((150), (152), (154) (repeated below)), provided that the events
PredP and PredP: Of Subjects and Ancient Grievance
265
described by the first conjunct causally or circumstantially condition those described by the second conjunct ((124)–(128) (repeated here)): (150) ?No child is watching Sesame Street and is looking unhappy. (152) ?No child has watched Sesame Street and has gone to bed unhappy. (154) ?No student is reading the text and is taking notes on what she reads. (124) *No rocker was shimmying and was shaking (to that funky disco music). (126) *No rocker has shimmied and has shaken (to that funky disco music). (128) *No student of mine is studying syntax and is publishing semantics. *Not any student of mine is studying syntax and is publishing semantics. Recall that the event variable Tense exposes in the second conjunct can only be closed off by a pronoun referring to the events described by the first. The unpronounced subject is without adverbialization and the adverbialization of the first subject remains within the first conjunct adjacent to its Case position within. The asymmetric conditioning itself disappears if, once again, the seeming scope of the quantifier is an illusion of “telescoping”: (133) Every rocker was shimmying and was shaking (to that funky disco music). (134) Every rocker has shimmied and has shaken (to that funky disco music). (135) Every student of mine is studying syntax and is publishing semantics. Shrinking the coordination further, if it is small enough to fall entirely within the scope of both the raised quantifier and its adverbialization, the quantifier includes within its scope the entire coordination without any apparent conditioning between the conjuncts: (123) No rocker was shimmying and shaking (to that funky disco music). (125) No rocker has shimmied and shaken (to that funky disco music). (127) No student of mine is studying syntax and publishing semantics. Not any student of mine is studying syntax and publishing semantics. This is plainly true when the coordination is small enough to complement the auxiliary verb as in (123)–(127), and here even a pronounced second subject must be contained within (section 3.2.1): (191) No tree has battered a window and no glass at all broken. (192) No tree has battered a roof and no shingles shaken loose. It is also true in (120) and (121) (repeated here), where simple tensed verb phrases are coordinated absent a second pronounced subject, which was argued (section 3.2.0) to represent a coordination smaller than one involving tensed auxiliary verb phrases:
266
Chapter 3
(120) No rocker shimmied and shook (to that funky disco music). (121) No student of mine studies syntax and publishes semantics. Not any student of mine studies syntax and publishes semantics. In all of the above except at the top ((118)–(119), (323)–(326)), there is some structural asymmetry between the subjects—the result of an overt movement of the first subject out of the coordination—that affords the first a superior position from which to look down on the second. In contrast to (118)–(119) and (323)–(326), the first subject of (337) includes within its scope the pronounced second subject, which is the subject of an embedded participial phrase, as in (191) and (192): (337) No rocker has (ever) shimmied and (any of) his mama(s) shaken to that funky disco music. As in (191), (192), and (337), demoting in (333)–(336) the second subject to a participial phrase rescues the use of singular number agreement: (338) Not any student of mine has yesterday conjectured and any one of yours today proven (between them) p and not p. No student of mine has in one instance conjectured and any one of yours in the next proven (between them) p and not p. (339) Not any student of mine has yesterday conjectured p and any one of yours today proven not p. No student of mine has in one instance conjectured p and any one of yours in the next proven not p. Similar observations apply to smaller coordinations. In (340), there is no respect in which the subject of the second conjunct is in a position subordinate to the position in which the first is pronounced. They are the pronounced subjects of conjuncts of equal dimension, and correlatively their scope is confined to their conjuncts, also with “telescoping” dissembling in (341): (340) *Ed Sullivan gave no rocker a hand and (any of) his mama(s) a hoot. (341) Ed Sullivan gave every rocker a hand and his lead mama a hoot. Again, extreme Right-Node Raising allows the first quantifier to escape the coordination: (342) Ed Sullivan gave no rocker and (any of) his mama(s) a really big hand. The contrast between (340) and (342) is in truth the same one that contrasts the use of singular and plural number agreement in (326)–(327) and (332)–(333). The phrases a hand and a hoot in (340)–(342), Right-Node Raised or not, are like all phrases under supermonadicity tied to the sentence by an event pronoun, ‘[℩Ei:pro]’ in (343), referring to what the rocker did, or to what his mama did, or by one referring to what they both did.
PredP and PredP: Of Subjects and Ancient Grievance
267
(343) … [℩Ei:pro][℩Ej:pro]O[Ei,Ej][A x : Hx] … ∃Ej Theme[Ej,x] … That is, there is slight difference between number agreement so-called and any of the other event pronouns holding the sentence together, each of which occurs either in the singular to refer to what she or he did or in the plural to refer to what they did. In (340) and (341), absent Right-Node Raising, there is no copying between a hand and a hoot; the phrases are prefixed by event pronouns unrelated to one another. The pronouns that occur there refer to what is locally described within their respective conjuncts. The first refers to what the rocker does, since of course he and not his mama gets a hand, and the latter to what his mama does, since what she does and not what he does gets a hoot. That is, the event pronouns prefixing these phrases are singular, as it were, denoting different events in the two conjuncts. Thus the structure of the smaller coordination in (340) matches in this respect the structure of its more full-blooded counterparts (323)–(326) and (333)–(336). In contrast, (342) presents no indication of sense or form that its Right-Node Raised phrase is prefixed by a singular event pronoun rather than a collectivizing plural pronoun saying that what they, rocker and mama, get is a really big hand.42 If so, it falls in with (327) and (332), where overt plural agreement seems to provide license for the quantifier to escape the first conjunct. Supermonadicity plays a crucial role in assimilating the contrast in opacity in smaller coordinations to that in the larger ones. All of these coordinations begin by ascribing to their subjects subevents in which they alone participate: (344) … [No x: rocker(x)] … ∃E W[E,x] … and [any y : mama(y)] … ∃E W[E,x] … As in (332) in a large coordination, what is done alone may also be further qualified within the smaller coordination: (345) (Appointments with) Frege gave no colleague of mine yesterday and any student of hers today anything the two of them could collaborate on. (Appointments with) Frege gave no colleague of mine yesterday and any student of hers today more than a couple problems the two could collaborate on. Now, under supermonadicity, any continuation that is to describe an effect or an overlapping state or any other event at all, which many another morpheme is bound to do, requires the intervention of an event pronoun. Any such phrase that is tokened only within the first clause, such as a hand in (340), engages just what is described there through an event pronoun in the singular so-called, referring to what the rocker does. As a result, this small, secondary predicate comes to resemble the full-blooded clauses in (323)–(326), where singular number agreement in the first clause is in plain sight. The adverbs in (345) do not themselves induce opacity and
268
Chapter 3
all else there has been Right-Node Raised and presumably collectivized under a plural event pronoun. Opacity quickly follows, however, once the conjuncts are large enough to contain independent predicates to be related to what the singular individual does: (346) *(Appointments with) Frege gave no colleague of mine one problem yesterday and any student of hers another one today (which the two of them could collaborate on). *(Appointments with) Frege gave no colleague of mine yesterday one problem and any student of hers today another (one) (which the two of them could collaborate on). Granting the descriptive generalization, what remains to relate singular number agreement to opacity? Section 2.2.1 accounts for the contrast between singular and plural agreement in those cases where they have both been Right-Node Raised, as in (326) vs. (327), (332) vs. (333), and as it is exemplified in chapter 2, in (347) vs. (348): (326) *No rocker and (any of) his mama(s) has (ever) shimmied and shaken to that funky disco music. (327) No rocker and (any of) his mama(s) have (ever) shimmied and shaken to that funky disco music. (332) *Not any student of mine yesterday and any one of yours today has conjectured and proven (between them) p and not p. *No student of mine in one instance and any one of yours in the next has conjectured and proven (between them) p and not p. (333) Not any student of mine yesterday and any one of yours today have conjectured and proven (between them) p and not p. No student of mine in one instance and any one of yours in the next have conjectured and proven (between them) p and not p. (347) *No Columbia student and any Harvard student was ever on a strike committee in which the Columbia student led and the Harvard student followed. (348) No Columbia student and any Harvard student were ever on a strike committee in which the Columbia led and the Harvard student followed. Recall that the contrast between singular and plural number agreement rests on a difference of phrase structure that requires for singular agreement an extra layer of event quantification: (349) No Columbia student secretly and no Harvard student openly was on a strike committee in which the Columbia student led and the Harvard student followed.
PredP and PredP: Of Subjects and Ancient Grievance
269
(350) No Columbia student at any address in a frame of referenceF secretly—and no Harvard student at another address in that frame of referenceF openly— [℩f : proF][∃f′ : within(f′, f)] [℩E0 : sg.pro0] was on a strike committee … ‘No Columbia student secretly and no Harvard student openly somewhere there was on a strike committee …’ (351) No Columbia student secretly and no Harvard student openly were on a strike committee in which the Columbia student led and the Harvard student followed. (352) No Columbia student at any address in a frame of referenceF secretly—and no Harvard student at another address in that frame of referenceF openly— [℩f : proF] [℩E0 : pl.pro0] was on a strike committee … ‘No Columbia student secretly and no Harvard student openly there were on a strike committee …’ If the descriptive content of number agreement is so impoverished that the singular event pronoun denotes the participation there of one and the plural event pronoun denotes the participation there of more than one, number agreement must rely on what there is to vary the events denoted. Given that the initial frame of reference F in (349)–(352) frames one Columbia student and one Harvard so that the Columbia student and the Harvard student may refer successfully to the Columbia student in F and the Harvard student in F, the participation in F is the plural participation of more than one. For singular number agreement, an operator must intervene, as shown in (350), to narrow the frame of reference to an f′ so that the participation in f′ is indeed the participation of just one. As a result, the clause within the scope of the subject quantifiers is a trifle larger when number agreement is singular but large enough that the coordination of such clauses confines the subject quantifiers within the conjuncts. With no position outside the coordination for the first quantifier to move to, it renders unacceptable examples (326), (332), and (347) in their attempt to combine singular agreement and wide scope for the first quantifier. This reasoning and explanation extend directly to the small clauses of (340), (342), (345), (346), and their subatomic event pronouns: (353) Ed Sullivan gave no rocker and no roller a hand so loud that the rocker and the roller blushed. (354) Ed Sullivan gave no rocker a hand and no roller a hoot so loud that the rocker and the roller blushed. (355) *Ed Sullivan gave no rocker a hand and (any of) his mama(s) a hoot so loud that the rocker and his mama blushed. (356) Ed Sullivan gave no rocker and (any of) his mama(s) a hand so loud that the rocker and his mama blushed.
270
Chapter 3
These event pronouns do not contain number morphology, but if their content is otherwise equally impoverished and thus denoting the participation in f, the rocker’s participation and only the rocker’s participation gets a hand according to (354) and the roller’s participation and only hers a hoot only if they each have a frame of reference f of their own so that referring to the participation in f suffices to refer to the individual’s participation. Again, such a frame of reference is not the one F that frames both rocker and roller, affording the rocker and the roller reference to the rocker in F and the roller in F. Again, the intervention of an operator is necessary to narrow the frame of reference, the effect of which is visible in (355) in the failed attempt to combine reference to the individual’s participation with wide scope for the first quantifier. A further corollary is that the event pronoun prefixed to a bear hug in (357) and (358) cannot be interpreted “sloppily” in (358), with the result that the only interpretation of (358) implies a group bear hug, unlike (357), which permits Ed Sullivan two bear hugs, of rocker and roller in turn: (357) Ed Sullivan gave no rocker and no roller a bear hug. (358) Ed Sullivan gave no rocker and (any of) his mama(s) a bear hug. This account comes of course with a stipulation that the coordination in (353) and (356), with event pronouns referring to collective participation, holds in reserve a position outside the coordination for the quantifier in (356) to move to, a position swallowed inside the coordination of larger conjuncts in (355).43 Recall the dilemma now resolved and first posed in section 1.5.1 (see also section 2.0.1). Alongside (326) vs. (327), where number agreement discriminates when the first quantifier is intended to include the second within its scope, the independent quantifiers in (359) and (360) support both plural and singular number agreement: (359) No rocker and no roller has ever shimmied and shaken to a funky disco beat. Not any rocker and not any roller has ever shimmied and shaken to a funky disco beat. (360) No rocker and no roller have ever shimmied and shaken to a funky disco beat. Not any rocker and not any roller have ever shimmied and shaken to a funky disco beat. If one decides a priori that in all these sentences, (326), (327), (359), and (360), quantificational DPs have been combined under coordination to form complex quantificational DPs, how does one proceed with the rest of the account? How is that the complex DP that is internally asymmetric happens to then combine only with plural predicates, but the symmetric, complex DP combines with either singular or plural predicates?
PredP and PredP: Of Subjects and Ancient Grievance
3.3
271
Null coordinative pronouns
And, that which is always a sentential connective, cannot be flanked by unsaturated expressions.44 Any predicate that occurs within a coordination occurs there with its full complement of arguments, pronounced or unpronounced. (361) Every soul man was shimmying to that funky disco music and was shaking to it (too). (362) James Brown was shimmying to that funky disco music and was shaking to it (too). (363) *No soul man was shimmying to that funky disco music and was shaking to it (too). So it is that the coordination in (361)–(363) is not a coordination of predicates deriving a predicate, or else there would be no telling apart (361) and (363) or telling apart (363) and (364): (364) No soul man was shimmying to that funky disco music and shaking to it (too). Schwarz (1999; 2000, chap. 3) forcefully makes this point, observing that the pragmatic conditions that govern descriptive anaphora and discriminate between (365) and (366) do not apply when the coordination is within the scope of the antecedent quantifier as in (367)–(368): (365) #Either someone stole your hat or he robbed a bank. (Kamp and Reyle 1993) (366) Either someone stole your hat or he took it thinking it was his. (367) Someone either stole your hat or robbed a bank. (368) Someone either stole your hat or took it thinking it was his. Since, as Schwarz argues, the pragmatic conditions discriminate (369) from (370), it must be that someone is confined to the first conjunct and the subject of the second conjunct is a null descriptive pronoun: (369) #Either someone stole your hat or robbed a bank. (370) Either someone stole your hat or took it thinking it was his. A clausal coordination putting subjects in both conjuncts allows the beginning of some account. As an alternative to what has been said earlier (section 3.2.0) about (361)–(363), one might have imagined that universal quantifiers and names have a longer reach than decreasing quantifiers (see Szabolcsi 1997). It might be that their reach is long enough to derive a contrast in logical form where the universal quanti-
272
Chapter 3
fier and name manage to bind simple variables in both conjuncts, while only the decreasing quantifier remains confined within the first: (371) [Every x : soul man](x was shimmying to that funky disco music and x was shaking to it) (372) [℩x : James Brown](x was shimmying to that funky disco music and x was shaking to it) (373) *[No x : soul man](x was shimmying to that funky disco music and x was shaking to it) *([No x : soul man] x was shimmying to that funky disco music and x was shaking to it) I have instead held that the quantifiers in the subject positions of tensed auxiliary phrases are similar in their movements. Unless asymmetric semantic conditioning between shimmying and shaking is intended, the quantifiers in (361)–(363) are all confined to the first conjunct. As in Schwarz’s examples and argument, that the apparent wide scope is conditioned indicates that it is not true wide scope. Instead, a null descriptive pronoun in the second conjunct, interpreted via “telescoping” and thus subject to the conditioning noted, distinguishes (361) and (363). The conclusion to this section offers further argument for the null descriptive pronoun and against (371) and (372). But so far, the argument has been to display the “telescoping” adverb as the source of further variation, the contrast between (145) and (146) tracking (147) and (148), all of which would be effaced if the universal quantifier could as in (371) and (372) simply include the entire coordination within its scope without the adverbial intermediary. (145) Every race car was skidding and (then) (it) ate up track heading for the viewing stands. (146) ?*Every race car has skidded and (then) (it) ate up track heading for the viewing stands. (147) Every race car was skidding and when a race car was skidding (it) ate up track heading for the viewing stands. Every race car was skidding and (when) skidding (it) ate up track heading for the viewing stands. (148) ?*Every race car has skidded and when a race car has skidded (it) ate up track heading for the viewing stands. ?*Every race car has skidded and (when) having skidded (it) ate up track heading for the viewing stands. Now if in the end (371) and (372) are correctly rejected, it sets aside derivation of (361) and (362) by across-the-board movement out of a coordinate structure, as in (374) and (375), which could not escape translation as (371) and (372):
PredP and PredP: Of Subjects and Ancient Grievance
273
(374) Every soul man was shimmying to that funky disco music and every soul man was shaking to it ⇒ Every soul mani [[ti was shimmying …] and [ti was shaking …]] (375) James Brown was shimmying to that funky disco music and James Brown was shaking to it ⇒ James Browni [[ti was shimmying …] and [ti was shaking …]] Not that this loss is to be greatly lamented as across-the-board movement holds mysteries of its own, but it has always provided comfort when the need to license null subjects under coordination arises.45 Rejecting such comforts only makes the syntax of the null subject in the second conjunct more precarious, and I will soon be cornered into an empirical claim that Godard (1989) and Lasersohn (1995, 103f.) have already laid at the feet of anyone who holds views like mine. To dispel lingering doubt, I would like to cull some further arguments for embracing null subjects rather than wishing them away. Icelandic is among languages where nonnominative nominals occupy subject position without triggering number agreement, “quirky” subjects so-called, as in the second conjunct of (376): Icelandic (Rögnvaldsson 1990) (376) við vorum þreyttir og okkur fannst myndin leiðinleg. we.nom were.1pl tired and us.dat found.3sg the-movie boring (377) við vorum þreyttir og fannst myndin leiðinleg. we.nom were.1pl tired and found.3sg the-movie boring (378) *við vorum þreyttir og fundumst myndin leiðinleg. we.nom were.1pl tired and found.1pl the-movie boring Icelandic (Rögnvaldsson 1982) (379) Þeim líkar maturinn og borða mikið. them.dat likes.3sg the-food and eat.3pl much (380) *Þeim líkar maturinn og borðar mikið. them.dat likes.3sg the-food and eat.3sg much As in (377)–(380), predicates that select quirky subjects freely coordinate with those selecting nominative subjects. Rögnvaldsson (1982, 1990) observes that the single pronounced subject cannot satisfy the (inconsistent) case and number requirements of both conjuncts and argues that a null subject answers for case and number in the second conjunct, crucially in (379), where plural number on the verb could not be derived except by agreement with the subject.
274
Chapter 3
Enlarging on this point, Bresnan and Thràinnson (1990) draw out an argument from the combination of mixed coordinations with subject inversion. When a predicate selecting nominative subject complements an auxiliary verb as in (381), the subject may invert either with the auxiliary verb (382) or with the entire predicate phrase (383): (381) a. maður í hvítum buxum mun kaupa fiskinn. ‘The man.nom in the white pants will buy the fish.’ b. maður í hvítum buxum mun keyra hann heim. ‘The man.nom in the white pants will take it home.’ (382) Það mun maður í hvítum buxum kaupa fiskinn. it will man.nom in white pants buy fish-the (383) Það mun kaupa fiskinn maður í hvítum buxum. it will buy fish-the man.nom in white pants A coordination of such unquirky predicates allows, as before, inversion with the auxiliary verb as in (385), and both inversion with the predicate phrase within the first conjunct as in (386) and inversion with the entire coordination as in (387): (384) maður í hvítum buxum mun kaupa fiskinn og keyra hann heim. man.nom in white pants will buy fish-the and drive it home ‘The man in the white pants will buy the fish and take it home.’ (385) Það mun maður í hvítum buxum kaupa fiskinn og keyra hann heim. it will man.nom in white pants buy fish-the and drive it home (386) Það mun kaupa fiskinn maður í hvítum buxum og keyra hann heim. it will buy fish-the man.nom in white pants and drive it home (387) Það mun kaupa fiskinn og keyra hann heim maður í hvítum buxum. it will buy fish-the and drive it home man.nom in white pants Already in (386) (and again in (395) below) there is some argument for a null pronoun in the second conjunct. The position of the pronounced subject between the predicates precludes their composition into a complex predicate. If, instead, an across-the-board movement relates the subjects of conjoined clauses as in (388a), the final word order of (386) requires a movement (388b) that some may find too onerous, extraction of the VP kaupa fiskinn from within the first conjunct and its adjunction to the coordination: (388) a. [It [will [DPi [[ti [buy fish-the]] and [ti [drive it home]]]]]] ⇒ b. [It [will [[buy fish-the]j [DPi [[ti tj] and [ti [drive it home]]]]]]] Abandoning across-the-board movement of the subject in favor of a null pronoun accommodates a more pedestrian syntactic derivation and structure:46
PredP and PredP: Of Subjects and Ancient Grievance
275
(389) [It [will [[[buy fish-the] DPi ] and [proi [drive it home]]]]], or [It [will [[[buy fish-the] DPi ] and [[drive it home] proi]]]] Further argument for the null subject emerges from consideration of quirky predicates. In contrast (Gunnar Ólafur Hansson, p.c., 2002), quirky subjects invert with auxiliary verb (391) but not with the quirky predicate (392): (390) manni í hvítum buxum mun líka fiskurinn vel. man.dat in white pants will like fish-the well (391) Það mun manni í hvítum buxum líka fiskurinn vel. it will man.dat in white pants like fish-the well (392) *Það mun líka fiskurinn vel manni í hvítum buxum. it will like fish-the well man.dat in white pants When coordinating with a quirky predicate, inversion is ruled out, as Gunnar Ólafur Hansson (p.c.) points out, whenever the quirky subject precedes its pronounced antecedent. Thus, if an unquirky predicate precedes in coordination with a quirky predicate, inversion with the entire coordination (396) and only this inversion is excluded, as Bresnan and Thràinnson (1990) originally observed: (393) maður í hvítum buxum mun kaupa fiskinn og líka hann vel. man.nom in white pants will buy fish-the and pro.dat like it well (394) Það mun maður í hvítum buxum kaupa fiskinn og líka hann vel. it will man.nom in white pants buy fish-the and pro.dat like it well (395) Það mun kaupa fiskinn maður í hvítum buxum og líka hann vel. it will buy fish-the man.nom in white pants and pro.dat like it well (396) *Það mun kaupa fiskinn og líka hann vel maður í hvítum buxum. it will buy fish-the and pro.dat like it well man.nom in white pants *Það mun kaupa fiskinn og líka hann vel manni í hvítum buxum. it will buy fish-the and pro.dat like it well man.dat in white pants In addition, if a quirky predicate is first in the coordination, inversion within the first conjunct (400) is also then ruled out, as it is in (392), structures where the quirky subject precedes its antecedent:47 (397) manni í hvítum buxum mun líka fiskurinn vel og kaupa hann. man.dat in white pants will like fish.the well and buy it (398) Það mun manni í hvítum buxum líka fiskurinn vel og kaupa hann. it will man.dat in white pants like fish.the well and buy it
276
Chapter 3
(399) *Það mun líka fiskurinn vel maður í hvítum buxum og kaupa hann. it will like fish.the well man.nom in white pants and buy it *Það mun líka fiskurinn vel manni í hvítum buxum og kaupa hann. it will like fish.the well man.dat in white pants and buy it (400) *Það mun líka fiskurinn vel og kaupa hann maður í hvítum buxum. it will like fish.the well and buy it man.nom in white pants *Það mun líka fiskurinn vel og kaupa hann manni í hvítum buxum. it will like fish.the well and buy it man.dat in white pants There is a straightforward description of all this if null subjects are admitted: quirkyCase null pronouns do not participate in backward anaphora. Their antecedents must precede them. This property of theirs is invariant and preserved under coordination.48 As compelling a description eludes the other accounts of the coordination, presenting them with a further difficulty beyond the one met when pronouncing the subject between the conjuncts. If coordination composes a predicate from predicates, and it is decided in (384)–(385) and (393)–(394) that the result is applied to nominative subjects, which invert with their predicates (383), it is not easy to reconcile that the resulting predicate inverts in (387) but not in (396). The differentiating behavior of quirky predicates under coordination similarly challenges the false symmetry that across-the-board movement imposes on the conjuncts. If an acrossthe-board movement to the left results in grammatical output in (384)–(385) and (393)–(394), then why should such a movement to the right succeed in (387) yet fail in (396) (no matter the case marking)? If, as argued, the allegedly missing subject of a quirky predicate is indeed a null pronoun, we cannot look to its absence to explain a difference in logical form between the pronounced and the unpronounced (Bresnan and Thràinnson 1990, 364n9): (401) fáir menn fóru snemma og gekk vel. few men.nom left.3.pl early & pro.dat succeeded.3.sg well ‘Few men left early and had a successful trip.’ (402) fáir menn fóru snemma og þeim gekk vel. few men.nom left.3.pl early & them.dat succeeded.3.sg well ‘Few men left early, and they had a successful trip.’ Like English (113) and (115), only where the second subject is unpronounced, (401), does the decreasing quantifier include within its scope the entire coordination, reflecting what could only be contrast between a big clause and an even bigger one when the subject is pronounced. Null subjects are therefore not to be wished away, the evidence warranting their existence being too strong to ignore. Yet, claiming their existence, as Godard (1989)
PredP and PredP: Of Subjects and Ancient Grievance
277
and Lasersohn (1995, 103f.) object, posits null elements—call them coordinative pronouns—that appear in coordinative structures and not many other places if any (neither English nor Icelandic being pro-drop languages). Undaunted, Rögnvaldsson (1990) extends the claim to null objects in Icelandic, as in the second conjunct of (405). The null object does not appear outside coordination, (404), its antecedent must be the object of the preceding conjunct, and the second subject must be unpronounced (406): (403) ég dái þig. I admire you (404) *ég dái. I admire (you) (405) ég elska þig og dái. I love you and admire (you) (406) *ég elska þig og ég dái. I love you and I admire (you) Rögnvaldsson further points out that the null objects discharge the functions of their pronounced counterparts, serving as antecedents for PRO and for reflexives and as the subjects of small clauses or secondary predicates. The experimentalist discovers a new particle; the theorist proves that it could not have been otherwise. Rögnvaldsson dutifully sets out to revise the contemporary syntactic theory to accommodate his discovery, thus answering Godard and Lasersohn, in principle. I will not be so virtuous. Despite a preoccupation with whether they exist and what they mean, I will not pursue why coordinative pronouns never appear outside coordination.49 Rather, given my project (e.g., the account of the contrast between (94) and (95)50), it is more urgent to show that there are coordinative, null event pronouns too. In Japanese, a final tensed verb is coordinated with a preceding verb that is unmarked for tense, as in (407): (407) kinoo Chris-ga kaeri, asita Pat-ga kaer-u. yesterday Chris.nom return.∅ tomorrow Pat.nom return.impf ‘Chris returned yesterday, and Pat will return tomorrow.’ (Tomioka 1993) Verbs appear unmarked for tense only in such coordinations and so appear to reflect outwardly an across-the-board movement of Tense. Arguing, however, against the shared Tense implied by such an analysis, Tomioka (1993) points out that the clauses need not agree in temporal reference and may host contrary temporal adverbs as shown in (407). He concludes rather that the first clause contains an independent, null Tense morpheme, unrelated by movement to Tense in the second clause. He
278
Chapter 3
further remarks that the coordinated clauses are full-blooded and independent in all other respects as well, hosting double nominative foci, topicalizations, and epistemic modals (surely, perhaps), so that even if an extraction of Tense were attempted, it would find nowhere to go outside the coordination. On a priori grounds, a naive view (see section 4.2 and discussion above of (50) and (51)) according to which pronounced verbal morphemes are tokened in logical form as they are tokened overtly would also reject Tense movement and force a null Tense morpheme in the first clause. Despite leaving unexplained its curious confinement to coordinations, Tomioka’s arguments amount to the discovery of just such a morpheme, foreclosing as they do on an alternative across-the-board Tense movement.51 There is no cause for dismay unless one holds that no languages ever present morphology peculiar to coordinative constructions (such as switch-reference morphology in some languages; see Stirling 1993 and Camacho 2003). (Nor should it be forgotten that across-theboard movement, although as familiar and comfortable as an old shoe, is itself a construction-specific stipulation—the wrong one.) So let there be null coordinative pronouns referring to objects and to events. 3.4
Reference under the eaves
In the very large coordination shown in (408), the highest event variable occurring in the second conjunct fails to occur within the scope of the pronounced subjects or within the scope of the adverb its articulation introduces: (408) Kunstler is sitting and is standing. [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … be[E0] … sitting[Ei] …), and [℩E0 : (… Present[E0] … be[E0] … sitting[Ei] …)][∃E1 : N[E0,E1]] (… Present[E1] … be[E1] … standing[Ei] …) The null coordinative event pronoun that occurs in the second clause can be nothing else other than a definite description of the antecedent events, with the result that the standing is said to coincide with the sitting. The absurdity implied is avoided in (409)–(411): (409) Kunstler is sitting, and Kunstler is standing. [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … be[E0] … sitting[Ei] …), and [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] (… Present[E0] … be[E0] … standing[Ei] …)
PredP and PredP: Of Subjects and Ancient Grievance
279
(410) Kunstler is sitting and standing. [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]] … Present[E0] … be[E0] … (… sitting[Ei] … and standing[Ei] …) (411) Kunstler sits and stands. [℩x : ∃E Kunstler[E,x]][℩E : Kunstler[E,x]][∃E0 : N[E,E0]]((… Present[Ei] … sit[Ei] …) and (… Present[Ei] … stand[Ei] …)) In (409), the pronounced subject of the second clause cues articulation of its own adverb, and so no matter how much larger the conjuncts of this coordination are, temporal reference within the conjuncts is dependent on separate adverbs. In (410) for sure and in (411) under the hypothesis that simple tensed verbs project smaller clauses than tensed auxiliary verbs, the event variable in the second clause falls within the scope of the adverbialization of the subject of the first clause, again avoiding as shown, the implication that the standing coincided with the sitting. Sentence (408) illustrates one effect on meaning that a null coordinative pronoun has when it stands at the edge of its conjunct in a coordination that is not itself under any other operator. These effects are felt across the range of interpretations that come up in the discussion of coordination. 3.4.0
Disjunctive interpretation excluded
Recall the disjunctive interpretation so-called in chapter 1 that allows the twenty rockers to be divided between those who shimmy and those who shake in (412)–(414): (412) Twenty rockers shimmied and shook. (413) Twenty rockers were shimmying and shaking. (414) Twenty rockers have shimmied and shaken. The interpretation is derived when twenty rockers are said to be the participants (‘W[E,X]’) in that, referring to some shimmying (by some) and some shaking (by some), and there is room enough in the structure for a pronoun referring collectively to these events. In striking contrast to (412)–(414), (415)–(416) remove the disjunctive interpretation and entail that twenty rockers shimmied and they shook, too: (415) Twenty rockers were shimmying and were shaking. (416) Twenty rockers have shimmied and have shaken. But this is to be expected from a very large coordination where the second conjunct contains a null coordinative event pronoun referring to antecedent events of shimmying in which the twenty have already been said to participate, and the shimmying is then said to coincide with shaking (see note 11). In (412)–(414), event reference in the second conjunct is not cornered in this way; chapter 6 takes up the syntax
280
Chapter 3
and semantics of the collective reference to events that contrasts (412)–(414) and (415)–(416). 3.4.1
Reconstruction interrupted
As in section 3.3, none of the very large coordinations in (361)–(363) (repeated here) falls within the scope of the pronounced subject, the illusion to the contrary in (361) an effect of “telescoping”: (361) Every soul man was shimmying to that funky disco music and was shaking to it (too). (362) James Brown was shimmying to that funky disco music and was shaking to it (too). (363) *No soul man was shimmying to that funky disco music and was shaking to it (too). Recall from section 3.1 the reconstruction of a rocker into the subject of each conjunct to derive in (68) (repeated here) and (417)–(418) an interpretation where for every bump, there was a rocker who shimmied through it, and on every grind, a rocker who shook: (68)
All night long, a rocker has shimmied through every bump and shaken on every grind.
(417) All night long, a rocker was shimmying through every bump and shaking on every grind. (418) All night long, a rocker shimmied through every bump and shook on every grind. Now if reconstruction into a position implies prior movement from it (see Fox 1995, 2000) and the very large coordination in (419)–(420) is too large for the across-theboard movement as would be required, the sentences should fail to support the corresponding interpretation. Indeed, in contrast to (68) and (417)–(418), the only interpretation of these sentences entails that a single rocker was shimmier and shaker through all the bumps and grinds. (419) All night long, a rocker has shimmied through every bump and has shaken on every grind. (420) All night long, a rocker was shimmying through every bump and was shaking on every grind. Similarly Fox’s (2000) (79) contrasts with (79′), which puts the same guard in many places at once: (79) A guard is standing in front of every church and sitting at the side of every mosque.
PredP and PredP: Of Subjects and Ancient Grievance
281
(79′) A guard is standing in front of every church and is sitting at the side of every mosque. If movement is prerequisite to reconstruction, then reconstruction in (421) and (422) again indicates across-the-board extraction from smaller conjuncts as in (417)– (418), perhaps, as shown, in combination with an elided participle in (421) and an elided participle and preposition in (422): (421) A guard is standing in front of every church and at the side of every mosque. A guardi is [[ti standing in front of every church] and [ti standing at the side of every mosque]] (422) A guard is standing in front of every church and every mosque. A guardi is [[ti standing in front of every church] and [ti standing in front of every mosque]] 3.4.1.0 Reconstruction or “telescoping”?
Some celebrated examples, (423) and (424) (Partee and Rooth 1983, 369; also see note 7), look like reconstruction in their equivalence to counterparts (425) and (426): (423) An easy model theory textbook is badly needed and will surely be written within this decade. (424) A tropical storm was expected to form off the coast of Florida and did form there within a few days of the forecast. (425) An easy model theory textbook is badly needed, and an easy model theory textbook will surely be written within this decade. (426) A tropical storm was expected to form off the coast of Florida, and a tropical storm did form there within a few days of the forecast. But the very large coordination precludes its falling within the scope of the pronounced subject, which is wholly contained within the first conjunct. It is therefore more telling that (423) and (424) are also equivalent to (427) and (428), where the overt pronunciation of the second subject leaves no doubt about the structure of the coordination or about its own rendering as a definite description: (427) An easy model theory textbook is badly needed, and it will surely be written within this decade. (428) A tropical storm was expected to form off the coast of Florida, and it did form there within a few days of the forecast. Here is “modal subordination” of the pronoun to the context antecedently described (Roberts 1989; Ludlow 1994; Poesio and Zucchi 1992; also see (167)–(170) above),
282
Chapter 3
achieved via a “telescoping” adverb that restricts the modal shift in the second conjunct: (429) An easy model theory textbook is badly needed, and where an easy model theory textbook is badly needed, the model theory textbook badly needed there will near there in the future surely be written within this decade. Without the apparatus for modal shift in the second conjunct, the sentence becomes anomalous. That is, in contrast to (431), (430) seems to say that the very book will first be needed de re and then written or—equally unintended—that the need for a particular textbook coincides with the writing: (430) #An easy model theory textbook will be needed and written within this decade. (431) An easy model theory textbook will be needed and (it) will be written within this decade. Sentence (430) also thus differs from (432), which is felicitous when the second subject is an overt indefinite description:52, 53 (432) An easy model theory textbook will be needed and an easy model theory textbook written within this decade. The very large coordinations of (423) and (424) thus seem to exemplify a structure with null coordinative pronouns in the second conjuncts, which via “telescoping” come to approximate what could be mistaken for reconstruction. That is, (423) and (424) look closer to (427) and (428) than to (425) and (426). The question is, I think, settled with the construction of examples where pronoun or definite description and reconstructed indefinite description are no longer equivalent. The structure of (433) is at issue: (433) Many new tons of rice are needed and will surely be farmed and distributed to those who pay. An interpretation that reconstructs the subject in the second conjunct reports a sad truth if continued as in (434): (434) Many new tons of rice are needed and many new tons will surely be farmed and distributed to those who pay, but it’s only a cruel maybe that the rice that is needed will be delivered to those who need it. On the other hand, it is false and infelicitous for the speaker to entertain the possibility that the poor will get what she has said is certain to go to the rich, as happens when the second subject is a pronoun or definite description: (435) #Many new tons of rice are needed and they will surely be farmed and distributed to those who pay, but it’s only a cruel maybe that the rice that is needed will be delivered to those who need it.
PredP and PredP: Of Subjects and Ancient Grievance
283
#Many new tons of rice are needed and the tons of rice needed will surely be farmed and distributed to those who pay, but it’s only a cruel maybe that the rice that is needed will be delivered to those who need it. Where there is no equivalence between reconstruction and null definite description, the very large coordination falls in with the latter, as expected:54 (436) #Many new tons of rice are needed and will surely be farmed and distributed to those who pay, but it’s only a cruel maybe that the rice that is needed will be delivered to those who need it. 3.4.1.1 One or two null coordinative pronouns
If indeed movement is a prerequisite to reconstruction, it explains away the interpretation missing from (419)–(420) and (79′) that there is no across-the-board extraction from very large coordinations. But then, without across-the-board movement in (419)–(420) and (79′), some other account of the null second subject in these examples must be forthcoming, as the presence of some subject follows from and being a sentential connective. It might be enough for existential closure to dispose of it as in (437)–(439) if the coincidence of the shimmying and the shaking implied by the coordinative null event pronoun in (419)–(420), for example, is strong enough to entail that the shaker is the shimmier, as would be the case if only the same participants could participate in coincident events (see note 11). (437) All night long, [[a rocker has shimmied through every bump] and [someone has shaken on every grind]]. (438) All night long, [[a rocker was shimmying through every bump] and [someone was shaking on every grind]]. (439) [[A guard is standing in front of every church] and [someone is sitting at the side of every mosque]]. But if the events’ coincidence does not alone establish that the shakers are the shimmiers, a null coordinative pronoun in subject position will, referring to the participants in the antecedently described events. Both coordinative null pronouns, the one denoting events and the one in subject position, are definite descriptions by nature. On this account of (419)–(420) and (79′), it is still crucial that movement is a prerequisite to reconstruction in order to deny to (419)–(420) the interpretation where a rocker’s shimmying is said to coincide with a rocker’s—possibly another’s—shaking: (440) *All night long, [[a rocker has shimmied through every bump] and [a rocker has shaken on every grind]]. (441) *All night long, [[a rocker was shimmying through every bump] and [a rocker was shaking on every grind]].
284
Chapter 3
(442) *[[A guard is standing in front of every church] and [a guard is sitting at the side of every mosque]]. With reconstruction disabled without a prior across-the-board movement, it leaves null coordinative pronouns as the only license for unpronounced elements in the second conjunct, and these are all definite descriptions. 3.4.1.2 Scope inversion and the position of null definite descriptions
The account can be improved so that a contrast less stark divides what is reconstructed from null coordinative pronouns. It need not be that the null element in subject position is sometimes a definite description and sometimes an indefinite description as across-the-board movement allows. Recall from section 3.1.0 that even when syntactic conditions are favorable to reconstruction, the “sloppy” or distributed interpretation may sometimes be absent as from (67), in contrast to (68) (here repeated): (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. It does not indicate a failure of a rocker to reconstruct into the second conjunct. Rather, in (67) the event quantifiers are distributed so that both tokens of a rocker happen to apply to the same events. The same remark, that the participants in these events were a rocker, is thus asserted twice over. But in (68), the pronouns describing these events fall within the scope of different event quantifiers, those concurrent with the raised universal quantifiers, every bump and every grind, resulting in the “sloppy” or distributed interpretation. What is necessary for a “sloppy” interpretation whether of an indefinite or definite description is that its tokens fall within the scope of different event quantifiers. This dependence on event quantification can now be exploited to rescue the null element in subject position from an ambiguity conditioned by across-the-board movement. Appearances to the contrary in (68), suppose that a null element is always a definite description whether an underlying coordinative pronoun or derived by (across-the-board) movement (see Elbourne 2005). This description is restricted by reference to events so that what is “reconstructed” in the second conjunct of (67) or (68) is, in effect, “the rocker there.” As before, if in (67) the event pronouns refer to the same events, the rocker there is the same for both shimmying and shaking (443): (443) All night long, a rocker ∃E0 (has[E0] ([℩x: Rx W[E0,x]] shimmied through the bumps and [℩x: Rx W[E0,x]] shaken on the grinds)).
PredP and PredP: Of Subjects and Ancient Grievance
285
(444) All night long, a rocker ∃E0 (has[E0] ([Every y : bump(x)][℩E1: E1≤E0 W[E1,x]] … [℩x: Rx W[Ei,x]] shimmied through … and [Every y : grind(x)][℩E1: E1≤E0 W[E1,x]] … [℩x: Rx W[Ei,x]] shaken on …)). In (68), the pronouns referring to events fall within the scope of different event quantifiers, so that “the rocker there” refers in the second conjunct to the rocker in a shaking (444).55 In very large coordinations, (419)–(420) and (79′), the null definite descriptions referring to antecedent subject and event occupy syntactic positions outside the scope of inverted universal quantifiers, (445). (445) All night long, (a rocker ∃E0 has[E0] [Every y : bump(x)][℩E1: E1≤E0 W[E1,x]] … [℩x: Rx W[Ei,x]] shimmied through …) and ([℩E0 : pro][∃E1 : N[E0,E1]] [℩x: Rx W[E0,x]] has[E1] [Every y : grind(x)] [℩E2: E2≤E1 W[E2,x]] … shaken on …)). Recall from section 3.2.1.3 that scope inversion in a simple sentence such as (446) or (448) requires two movements (Johnson and Tomioka 1997; Johnson 2000b; Hornstein 1995): raising the quantifier from object position and, crucially, a lowering of the subject quantifier to a position low enough for negation to disrupt it, as in (447) and (449): (446) Some rocker has shimmied through every bump. (∃∀, ∀∃) (447) Some rocker hasn’t shimmied through every bump. Some rocker has not shimmied through every bump. (∃¬∀, ∃∀¬, *¬∃∀, *¬∀∃, *∀∃¬, *∀¬∃) (448) Some linguist speaks some language from every family. [some linguist][every family][some language] [every family][some linguist][some language] (Johnson and Tomioka 1997) (449) Some linguist doesn’t speak some language from every family. [some linguist][every family][some language] not *[every family][some linguist][some language] not (450) Ken Hale doesn’t speak some language from every family. (∀∃¬) The freedom of the object quantifier to raise outside negation in (447) (“∃∀¬”), (449), and (450) shows that negation must instead constrain the subject’s lowering. The position to which it lowers is of course well below the peripheral location of the coordinative null pronouns in (419)–(420) and (79′). Let me stipulate for the moment that coordinative null pronouns are not themselves subject to Quantifier Lowering. Thus eluding the scope of the universal
286
Chapter 3
quantifiers altogether, the coordinative pronouns in (419)–(420) can only refer to the shimmying as described by the first conjunct and its rocker, entailing as desired that the shimmier and the shaker must be the same rocker. The contrast between (419)–(420) and (417)–(418) amounts to the following. In (419)–(420), definite descriptions, the rocker there and the events there, find themselves at the edge of a conjunct so large that they cannot be captured by any quantifier that the conjunct contains. In contrast, in (417)–(418), these same definite descriptions occupy positions that fall within the scope of an inverted quantifier and their reference covaries appropriately. This account continues to rely on syntax to tell us that no logical form for (419)–(420) reconstructs an indefinite description into subject position. That is, an indefinite description would appear in logical form in this position only if one had been moved from it (and perhaps, as just proposed, not even then) and the coordinations in (419)–(420) are just too large for such an across-the-board movement. If not sanctioned by movement, there should be no other way for the indefinite description to be repeated. In particular, this account holds the line at the worst excesses of Conjunction Reduction transformations, which would let one think “a rocker” in the second conjunct after having already said it in the first, via an ellipsis unconcerned with conditions on movement. On the other hand, if Conjunction Reduction were let loose, an account of the contrast between (419)–(420) and (417)– (418) survives. Retreating into the semantics to explain away logical forms effectively equivalent to (437)–(439), it would have to be said that the coincidence imposed by the coordinative event pronoun entails that the participant in the one event, the shimmying, is the same as the participant in the coincident shaking. 3.5
Quantifier Lowering into collectivized Right-Node Raised constituents
In what follows, reconstruction looks poised to slip the bonds of movement, appearing where movement could not sanction it. This does not threaten to get meaning wrong—it does not threaten very large coordinations with “sloppy” interpretations they do not have, provided, as remarked above, the coordinative event pronoun remains at the left edge of its conjunct, referring once and for all to the antecedent events, and all the free reconstruction falls to its right. What is at stake is rather the logical syntax itself and the role that has been assumed for Quantifier Lowering in the syntax of reconstruction. Quantifier Lowering as deployed in Fox 1995, 2000; Hornstein 1995; Johnson and Tomioka 1997; Johnson 2000b; and many others is spared only with a commitment to chapter 2’s treatment of collectivized Right-Node Raising. This chance to hold hostage an idea so central to the syntax/semantics interface and provide myself with unexpected confirmation, I cannot miss. On the other hand, a reader satisfied to catalog the intended logical forms for predicative coordination could skip this section without harm.
PredP and PredP: Of Subjects and Ancient Grievance
287
Sentences (451) and (452) are familiar examples of Quantifier Lowering (May 1977): (451) A Democrat is likely to speak at every pro-choice rally. (452) A libertarian and a progressive are likely to speak at every pro-choice rally. Sentences (453)–(455) introduce a collective predicate to Quantifier Lowering: (453) Some policy wonks are likely to lock horns at every state caucus. (454) Some libertarians and some conservatives are likely to confront each other at every state Republican caucus. (455) A libertarian and a conservative were expected to debate at every state Republican caucus. One might think to assimilate (454) and (455) to (453) and so to whatever is the favored treatment of Quantifier Lowering in (451) and (452). But the coordination may be overtly clausal, containing adverbs that plainly modify a Right-Node Raised phrase:56 (456) Some libertarians to hisses & boos and some conservatives to wild applause were likely to clash at every state Republican caucus. (457) A Democratic incumbent to hisses & boos and a Republican challenger to wild applause are expected to debate at every Republican slugfest broadcast on Fox TV. (458) Some Republican eagerly and some Democrat reluctantly are likely to cosponsor every bipartisan effort to further immiserate Iraq. The intended interpretations of (456)–(458) do find alternative expression in (459)– (461), but how are these interpretations of (456)–(458) to be derived by Quantifier Lowering? (459) It is likely for some libertarians to hisses & boos and some conservatives to wild applause to clash at every state Republican caucus. (460) It is expected for a Democratic incumbent to hisses & boos and a Republican challenger to wild applause to debate at every Republican slugfest broadcast on Fox TV. One expects a Democratic incumbent to hisses & boos and a Republican challenger to wild applause to debate at every Republican slugfest broadcast on Fox TV. (461) It is likely for some Republican eagerly and some Democrat reluctantly to cosponsor every bipartisan effort to immiserate Iraq. Parsing out the conjoined remnants in (456) as a constituent, one could not then proceed to lower it into the antecedent for its own properly contained deletion sites: (462) *[j[Some libertarians to hisses & boos Δi] and [some conservatives to wild applause Δi]] [iwere likely tj to clash at every state Republican caucus].
288
Chapter 3
Despite the absence of a well-behaved derivation, the plain fact that (456) can mean (459) threatens a derivational relationship between them, as if one could reconstruct at will in the lower clause some libertarians to hisses & boos and some conservatives to wild applause, ignoring syntax and derivational history. The coordination of clausal remnants and collectivized Right-Node Raising similarly threatens the Quantifier Lowering that Johnson and Tomioka 1997 and Johnson 2000b count on in reckoning scope. In (463), as expected, the universal quantifier escapes from the scope of negation in an assertion that there are some Republicans and Democrats who oppose every effort, not cosponsoring any, but it cannot include the subject within its scope to assert that for every effort, there were some Republicans and Democrats or others who did not cosponsor it. (463) Some Republicans nervously and some Democrats gladly did not cosponsor every bipartisan effort to immiserate Iraq. By their account, when inversion with subject becomes possible in the absence of negation as in (464), Quantifier Lowering is implicated: (464) Some Republicans nervously and some Democrats gladly cosponsored every bipartisan effort to immiserate Iraq. Yet, as in (462), lowering the conjoined clausal remnants is incoherent: (465) *[j[Some Republicans nervously Δi] and [some Democrats gladly Δi]] [i cosponsored tj every bipartisan effort to immiserate Iraq]. As we have just seen, the introduction of adverbs and Right-Node Raising does not disturb the basic paradigm of facts implicating Quantifier Lowering. It rather enlarges the relevant cases just as it upsets that analysis. The same can be said about Quantifier Lowering across the board as alleged to play a role in deriving the “sloppy” interpretations in (79)–(417) and the contrast between these examples and (79′)–(420): (79)
A guard is standing in front of every church and sitting at the side of every mosque. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. (417) All night long, a rocker was shimmying through every bump and shaking on every grind. (79′) A guard is standing in front of every church and is sitting at the side of every mosque. (419) All night long, a rocker has shimmied through every bump and has shaken on every grind. (420) All night long, a rocker was shimmying through every bump and was shaking on every grind.
PredP and PredP: Of Subjects and Ancient Grievance
289
As before, the “sloppy” interpretation puts a different couple behind every kiosk and a different couple under every canopy in (466), (467), and (469), and it is thwarted when the predicative coordination is very large. In (468) and (470), the same couple is all over the park (“*” marks only the unacceptability of the “sloppy” interpretation): On a spring day in Central Park— (466) A lover cautiously and a beloved tentatively embraced behind every kiosk and snuggled under every canopy. (467) A lover cautiously and a beloved tentatively were embracing behind every kiosk and snuggling under every canopy. (468) *A lover cautiously and a beloved tentatively were embracing behind every kiosk and were snuggling under every canopy. (469) A lover cautiously and a beloved tentatively have embraced behind every kiosk and snuggled under every canopy. (470) *A lover cautiously and a beloved tentatively have embraced behind every kiosk and have snuggled under every canopy. Similarly, (471), (472), and (474) may convey merely that each tree needed one backhoe and one crane to be either uprooted or planted; the very large coordinations in (473) and (475) force it to be understood that a single backhoe and crane did all the landscaping: (471) One backhoe from below and one crane from above uprooted every elm and planted every beech. (472) One backhoe from below and one crane from above were uprooting every elm and planting every beech. (473) *One backhoe from below and one crane from above were uprooting every elm and were planting every beech. (474) One backhoe from below and one crane from above have uprooted every elm and planted every beech. (475) *One backhoe from below and one crane from above have uprooted every elm and have planted every beech.57, 58 (476) *[j[One backhoe from below Δi] and [one crane from above Δi]] [iwere [tj uprooting every elm and tj planting every beech]] Yet Quantifier Lowering is not yet at the end of its rope while chapter 2 holds out some hope. Recall that the whole point was to devise a means for collectivized Right-Node Raising that allowed tokens of the Right-Node Raised constituent to occur within each conjunct with its collective meaning, as if to resolve (456) as (477).
290
Chapter 3
(477) Some libertarians to hisses & boos (with conservatives) were likely to clash at every state Republican caucus, and some conservatives to wild applause (with libertarians) were likely to clash at every state Republican caucus. If, despite the collective meaning, chapter 2 succeeds in restoring the Right-Node Raised constituent to each conjunct, it cuts the knot for Quantifier Lowering, which may then apply within each conjunct of (477) to derive the target interpretation inverting the scope of the universal and indefinite. Similarly, for Johnson and Tomioka’s benefit, the logical form for (464) comes to resemble (478), where within each conjunct Quantifier Lowering and Quantifier Raising may accomplish the scope inversion as originally proposed: (478) Some Republicans nervously (with Democrats) cosponsored every bipartisan effort to immiserate Iraq, and some Democrats gladly (with Republicans) cosponsored every bipartisan effort to immiserate Iraq. For cases that call on Quantifier Lowering into coordinate structures, the RightNode Raised constituents in (474) are restored as in (479): (479) One backhoe from below (with crane) have uprooted every elm and planted every beech, and one crane from above (with backhoe) have uprooted every elm and planted every beech. Within each conjunct Quantifier Lowering applies across the board, with scope inversion to follow: (480) ___ have one backhoe from below (with crane) uprooted every elm and one backhoe from below (with crane) planted every beech, and ___ have one crane from above (with backhoe) uprooted every elm and one crane from above (with backhoe) planted every beech. If the same were attempted with the very large coordination (475), event pronouns at the left edge of every second tensed conjunct will remain outside the scope inversion with the same result seen in simpler cases that the same backhoe and crane do all the landscaping. (481) ___ have one backhoe from below (with crane) uprooted every elm and __ have one backhoe from below (with crane) planted every beech, and ___ have one crane from above (with backhoe) uprooted every elm and __ have one crane from above (with backhoe) planted every beech. These remarks point to what looks to be the only way to rescue from confusion the interaction among clausal remnant coordination, collectivized Right-Node Raising, and Quantifier Lowering. There needs to be a way as in chapter 2 to undo RightNode Raising while preserving its collective meaning.59
PredP and PredP: Of Subjects and Ancient Grievance
3.6
291
Appendix: Economy and reconstruction
The interpretation of (68) (repeated below) derived from inverting the object quantifiers with tokens of a rocker argues, following Fox (1995, 2000), for the constant presence of a position for subjects within the conjoined participial phrases, even in (67), since it hardly differs from (68): (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. (68) All night long, a rocker has shimmied through every bump and shaken on every grind. (70) All night long, a rocker has shimmied through the bumps and a rocker shaken on the grinds. The gross structure of what is coordinated, the type of conjunction, is not what tells (67) and (68) apart. Granted that (67) does not mean the same thing as (70), the force of Fox’s argument is that without a difference of type between these coordinations, we must look elsewhere to explain their difference of meaning. I have looked to the distribution of event quantifiers. The overt pronunciation of the subject in (70) signals a difference in event quantification sufficient to distinguish its meaning from its counterpart in (67) even if reconstruction should restore a token of rocker to the second conjunct. It turns out that the same remark, that the participants in such-and-such events were a rocker, is asserted of the same events twice over in (67) and asserted one time each of different events in (70). A virtue of this account is then to have assimilated the contrast between (67) and (68) to one previously encountered in chapter 2 between (74) and (76), between (74) and (75), and between (72) and (77): (74) šerbou alia w marwaan anninet pepsi. drank.pl Alia and Marwan bottle Pepsi (75) šerbou alia mbeereH w marwaan lyom anninet pepsi. drank.pl Alia yesterday and Marwan today bottle Pepsi (76) šerbet kell bent w kell Sabi anninet pepsi. drank.3fs every girl and every boy bottle Pepsi (72) keenou l-snoubraat w l-sendyeeneet xfeef bi mantʕa were.3mp the-pines and the-oaks sparse in region Hadd mantaʕa bašariyyeh. near region human ‘The pines and the oaks were sparse in a region near human settlement.’
292
Chapter 3
(77) keenou l-snoubraat men 5000seneh w l-sendyeeneet were.3mp the-pines from 5000year and the-oaks men 1000 seneh xfeef bi mantʕa Hadd mantaʕa bašariyyeh. from 1000 year sparse in region near region human ‘The pines 5000 years ago and the oaks 1000 years ago were sparse in a region near human settlement.’ The Right-Node Raised constituent in (72) and (77) is sparse in a region near human settlement. Recall it meaning that the pines are sparse and so are the oaks, without implying that the forest of pines and oaks is sparse. The Right-Node Raised constituent must be restored to both conjuncts in both (72) and (77) and yet there varies the interpretation of a phrase it properly contains, in a region near human settlement, according to whether the tokens of the Right-Node Raised constituent end up within the scope of different event quantifiers. Fox (1995, 2000) looks elsewhere to explain the difference between (67) and (70). Their phrase structure and the structure of coordination is the same, following the argument already given. Moreover, if a rocker were to lower into both conjuncts of (67), it would in fact mean the same as (70), but Fox proposes that a constraint on movement, Economy, prevents it from ever doing so in (67) while continuing to allow it in (68) where the lowered subjects end up within the scope of the object quantifiers. I do not see that Economy would also assimilate the variation in RightNode Raising, (74)–(77),60 and I hope I will not be faulted for appending some skeptical remarks about an alternative to the one presented here. Economy is a principle of “least effort” according to which some logical forms are ruled out if their derivations are more complex than others that are equivalent. The equivalence classes from which the most economical logical forms are chosen are circumscribed by what the logical vocabulary, the determiners and logical connectives, alone express without regard for inferences based on lexical or other knowledge. Thus, the sentences in (482) and (483) each have two logical forms, since universal and existential quantifiers do not in general permute preserving truth. It doesn’t matter that (483)’s two logical forms happen in this instance to be equivalent thanks to the peculiar meaning of one and only. In contrast, the sentences in (484) and (485) each have only one logical form, as it is a general fact about determiners that a name or definite description may permute with a universal quantifier without affecting truth. (482) Everybody serves a god. (483) Everybody serves a one and only true god. (484) Everybody serves God. (485) Everybody serves the one and only true god.
PredP and PredP: Of Subjects and Ancient Grievance
293
One can imagine evidence hard to come by that these sentences have one, the other, or both logical forms meaning the same thing, but Fox’s (1995, 2000) method for uncovering such evidence is ingenious. Ellipsis constructions require a formal identity between two sentences. Any restriction on the logical form of the one is therefore reflected in the other, where a difference in meaning can now be observed that could not be found in the sentence subject to the restriction. By Economy, Mary admires everyone has only one logical form (since all permutations of its DPs are equivalent), so the antecedent for the ellipsis in (487) must parallel this logical form, with the result that the first clause of (487) is unambiguous: (486) A boy admires every teacher, and a girl does too. (487) A student admires every teacher, and Mary does too. In contrast, ambiguity remains in (486). Within each clause, there is no equivalence between the candidate logical forms, none of which Economy will therefore rule out. Parallelism requires that the same permutation be chosen, leaving (486) exactly two-ways ambiguous. Parallelism and Economy apply to the sentences that preoccupy me as follows. First, Economy is set as a standard that applies within a clause to winnow out any uneconomical derivation affecting that clause. In (67), the reconstruction or lowering of a rocker within any conjunct is a “costly” one that does not enhance what can be expressed within that conjunct: (67) All night long, a rocker has shimmied through the bumps and shaken on the grinds. In particular, permuting a rocker with the bumps is equivalent to not doing so, and likewise for permuting it with the the grinds. As a result, Economy blocks the lowering altogether even though lowering would of course result in a nonequivalent interpretation of the coordination as a whole (cf. (70)). In (68), on the other hand, permutation with the universal quantifier does change the meaning of the individual conjunct: (68) All night long, a rocker has shimmied through every bump and shaken on every grind. Of course, (68) still cannot mean that the shimmiers were the same one rocker, and the shakers were also all the same but a different rocker. As far as Economy is concerned, a standard applied within conjunct, this meaning is no more economical than the one stricken for (67). From the point of view of the individual conjunct, it could have been achieved without lowering. Thus, if there is lowering at all it must be accompanied by scope inversion with the universal quantifier so as to have an immediate impact on the conjunct into which there has been lowering. A parallelism is noted so that scope inversion within one conjunct must be met with the same in
294
Chapter 3
the other, and inversion must satisfy Economy in both. The combination of these conditions serves to correctly keep Quantifier Lowering from applying in (488) and (489):61 (488) All night long, a rocker has shimmied through every bump and shaken to the beat. (489) All night long, a rocker has shimmied to the beat and shaken on every grind. According to the account on offer, whenever candidate logical forms break away from equivalence, ambiguity should remain. Thus, the claim is that (491) and (492) are ambiguous—the permutations of almost everyone in the film festival and every movie are not equivalent, nor are the permutations of most visitors to the film festival and every movie. In contrast, (490) is unambiguous, where permuting universal quantifiers are equivalent: (490) One of the film critics admires every movie. Everyone in the film festival does too. (491) One of the film critics admires every movie. Almost everyone in the film festival does too. (492) One of the film critics admires every movie. Most visitors to the film festival do too.62 (493) One of the film critics admires every movie. The person who produced it does too. (494) One of the film critics admires every movie. The person who produced the film festival does too. (495) One of the film critics admires every movie. Every person who was involved in its production does too. (496) One of the film critics admires every movie. Every person who was involved in this production does too. Because of Economy, the (b) sentences in (497)–(500) have unique logical forms, while the (a) sentences remain scopally ambiguous: (497) a. b. (498) a. b. (499) a. b. (500) a. b.
A girl admires every teacher. Mary admires every teacher. The person who produced it admires every film. The person who produced the film festival admires every film. Every person who was involved in its production admires every movie. Every person who was involved in this production admires every movie. Almost everyone in the film festival admires every movie. Everyone in the film festival admires every movie.
PredP and PredP: Of Subjects and Ancient Grievance
295
Given how the equivalence classes for Economy are defined (see the Revised Modularity Hypothesis (Fox 2000, 68)), a direct counterexample to the proposal would be any coordination and ellipsis where ambiguity survives a name, definite description (free of bound variables), or unmodified universal quantifier (free of bound variables) appearing as the subject of the second sentence. Also of interest are cases where Economy does not disambiguate the second conjunct ((a) examples), but the coordination and ellipsis is nevertheless unambiguous. Such cases are not in themselves inconsistent with Economy, but whatever steps in to explain the unexpected loss of ambiguity in such cases threatens to preempt Economy in those cases of disambiguation originally attributed to it. There are several degrees of separation between a girl and Mary in (497), besides the difference in logical determiner. First, A girl is a weak DP and the name is a strong one, and it should be kept in mind that the strong/weak distinction correlates better with partitivity (Milsark 1977; Herburger 1997), a property of the restriction, than with choice of determiner. Second, Mary is not just any definite description, but an incomplete one, one that is resolved with a peculiar indexical condition, ‘the Mary that we know’ as it were (Higginbotham 1988), and one that is used more referentially than descriptively. Sentences (501) and (502), in their choice of quantifiers many and most, hold a special significance in establishing the ambiguity expected under Economy: (501) A Canadian flag is in front of many buildings, and an American flag is too. (502) A Canadian flag is in front of most buildings, and an American flag is too. The sentences can be true although no more than a few buildings fly both flags. Therefore, separate tokens of many/most buildings must occur within both conjuncts and include the subject within their scope. The same can be said for the following, despite the violation of Economy prompted by the name and definite description: (503) A foreign flag stands on many distant shores, but Old Glory does too. A foreign flag stands on many a distant shore, but Old Glory does too. (504) A foreign flag stands on many distant shores, but the American flag does too. A foreign flag stands on many a distant shore, but the American flag does too. Similarly, inverse scope is favored in the following: (505) A tattered Confederate flag still flies over every Southern state capitol, and Old Glory does too. A tattered Confederate flag still flies over many a Southern state capitol, and Old Glory does too. A tattered Confederate flag still flies over many Southern state capitols, and Old Glory does too.
296
Chapter 3
(506) A tattered Confederate flag still flies over every Southern state capitol, and the American flag does too. A tattered Confederate flag still flies over many a Southern state capitol, and the American flag does too. A tattered Confederate flag still flies over many Southern state capitols, and the American flag does too. Apart from the violation of Economy and unexpected ambiguity that these examples of ellipsis present, the neighboring cases of Quantifier Lowering and scope without ellipsis inversion are still as unacceptable as Economy would predict. Like (506), the ellipsis in (507) allows lowering and inversion: (507) A tattered Confederate flag still flies over many a state flag in the South, and/but Old Glory/the American flag does too. A tattered Confederate flag still flies over many a state flag in the South, and/but Old Glory/the American flag does too. The politics of the South make it quite natural that the many state flags flying under tattered Confederate flags are not the many flying under Old Glory. The same politics fit (508) and (509), cases without ellipsis where Quantifier Lowering and scope inversion are consistent with Economy: (508) A tattered Confederate flag still waves above many a state flag in the South and ((waves) beside) many an American flag too. A tattered Confederate flag still waves above many a state flag in the South and ((flies) beside) many an American flag too. (509) A tattered Confederate flag has waved above many a state flag in the South and flown beside many an American flag too. A tattered Confederate flag is waving above many a state flag in the South and flying beside many an American flag too. This interpretation that for some reason (507) tolerates still fails in (510) and (511), attempting Quantifier Lowering and inversion without ellipsis: (510) A tattered Confederate flag still waves above many a state flag in the South and ((waves) beside) Old Glory/ the American flag (too). A tattered Confederate flag still waves above many a state flag in the South and ((flies) beside) Old Glory/the American flag (too). (511) A tattered Confederate flag has waved above many a state flag in the South and flown beside Old Glory/the American flag (too). A tattered Confederate flag is waving above many a state flag in the South and flying beside Old Glory/the American flag (too).
PredP and PredP: Of Subjects and Ancient Grievance
297
To be sure, (510) and (511) have an interpretation that locates a tattered Confederate flag within the scope of many a state flag. But it is an interpretation, the “telescoping” interpretation (512), that goes on to assert that these same Confederate flags are also beside American flags, resulting in a very confused Southern politics of Confederate-Union institutions and households. (512) [Many a y : state flag(y)][A x : Confederate flag(x)] x wave above y, and whenever a Confederate flag waves above a state flag, that Confederate flag flies beside Old Glory too. The contrast between (507) and (510)–(511) thus presents a further dilemma for Economy. As it is formulated, (507) stands as a counterexample to Economy. If, however, it were to be revised to accommodate (507), it is difficult to imagine how it would not thereby admit a mistaken interpretation of (510)–(511), since the economies of (507) and (510)–(511) are so very much alike at least according to the measure proposed. Returning to ellipsis under Economy, on the other side of the coin, we can find disambiguation despite indefinite subjects: (513) #A tattered Confederate flag still flies over every Southern state capitol, and one of the American flags in this photo does too. #A tattered Confederate flag still flies over many a Southern state capitol, and one of the American flags in this photo does too. #A tattered Confederate flag still flies over many Southern state capitols, and one of the American flags in this photo does too. Kyle Johnson (p.c., September 2002) wanted to know what was in the photo. It was explained that it was from the manufacturer’s catalogue with model numbers indicated, which gave the speaker reason to know that these are indeed the flags being flown, and yet no knowledge of which flag went to which capitol (hence, targeting a nonspecific indefinite). But more important is that it is not a photo of flags flying over state capitols, so there is no mistaking what is seen in the photo for what the sentence is describing. Johnson, with whom I concur, rejected inverted scope, finding it incongruous with the direct epistemic access imposed by the partitive construction. The disambiguating effects of partitives can be observed again in the following. First, note that (514) is ambiguous as expected—the same Les Pauls are heard on every track, or not—and (515) is acceptable under its only sensible reading, which requires scope inversion: (514) In my early recordings, a vintage Les Paul graces every track, and in my recordings after the Grammy, many do.
298
Chapter 3
(515) After one year, a scratch marred every LP63 in my collection, and after another year, several (more) / more (than one) did. After one year, a scratch could be found on every LP in my collection, and after another, several (more) / more (than one) could be. The intrusion of a partitive forces it to be the same Les Pauls on every track and makes nonsense of (517). That is, the partitive is disambiguating against inverse scope: (516) In my early recordings, a vintage Les Paul graces every track, and in my recordings after the Grammy, many of them do. (517) *After one year, a scratch marred every LP in my collection, and after another year, several (more) of them / more (than one) of them did. *After one year, a scratch could be found on every LP in my collection, and after another, several (more) of them / more (than one) of them could be. Combining these observations of names and (in)definite descriptions running contrary to expectation, it seems that the content of DP restrictions matters a great deal to the course of disambiguation. In general, if the name or definite description (or universal quantifier) is made more descriptive than directly referential, less dependent on context for its completion, and less dependent on indexical reference, it need not be disambiguating. And, correlatively, the intrusion of these elements in the restriction to an indefinite does appear to be disambiguating. Further examples follow, to be compared with (487), (490), (494), and (496), repeated here: (487) A student admires every teacher, and Mary does too. (494) One of the film critics admires every movie. The person who produced the film festival does too. (490) One of the film critics admires every movie. Everyone in the film festival does too. (496) One of the film critics admires every movie. Every person who was involved in this production does too. The logical properties Economy invokes are held constant, and the examples depart from (487), (490), (494), and (496) in the other respects mentioned above, suppressing, for instance, implicit indexical or demonstrative content as in Mary, the film festival, and this production. Expectations learned from (487), (490), (494), and (496) can be thwarted if the first sentence provides cues that the first subject is not to be taken as a discourse referent on which hangs the rest of the story. Reference to scratches, schmears, flaws, and wounds are such cues, appealing to the real-world knowledge that tells us to locate the flea on the dog rather than the dog among the
PredP and PredP: Of Subjects and Ancient Grievance
299
fleas, and cues can be made more explicit as when a dependent expression such as local occurs in the description. The following sentences are divided in two, which will matter shortly, between those where the second subject is a name or definite description and those where it is a universal quantifier. In contrast to (487), (490), (494), and (496), none of the following disambiguates its first sentence, all allowing scope inversion: With definite descriptions as the second subject— (518) A successor follows every natural number, and the least limit ordinal does too. (519) A state seal is affixed to every state document, and the Great Seal of the United States is too. A state seal seals every state document, and the Great Seal of the United States does too. (520) An assistant DA is responsible for every case, and (of course) the DA is too. (521) A special prosecutor had power of subpoena in every special investigation this quarter, and (naturally) the attorney general did as well. (522) A schmear of sunblock protected every nose from sunburn, and the big cabana did too. (523) An individual foil wrapper enveloped every chocolate morsel in gold, and the box that contained them all did too. (524) The seal of a local magistrate adorns every municipal ordinance, and the Queen’s seal does too. The seal of Her Majesty’s local magistrate adorns every municipal office building, and Her Majesty’s seal does too. With universal quantifiers as the second subject— (525) A successor follows every natural number, and every limit ordinal does too. (526) An open wound compromised every one of my patients’ chances for survival, and every airborne pathogen did too. (527) A second internal hard disk backs up every one of the house servers, and every networked mainframe does too. (528) A chip implanted inside tracks every robot, and every satellite in geosynchronous orbit does too. (529) A local national guard unit guards every nuclear facility, and every USAF fighter squadron does too. Given that scope inversion is observed in the first sentence, it can now be asked whether inversion corresponds to a “telescoping” interpretation or to the interpretation that Economy regulates, with independent, parallel scope inversion in both
300
Chapter 3
conjuncts. As remarked about (501) and (502), it requires examples with many or most to replace the universal quantifier in the first sentence. The examples have been modified, as best I could, so that the “telescoping” interpretations are either false or as unlikely as a Southerner fighting both sides of the Civil War ((507)– (511)).64 Economy’s position now discriminates between the two classes of example. Permutation of a definite description with many or most is truth-preserving, hence Economy proscribes independent, parallel scope. Thus, with a definite description as second subject, scope inversion in the first clause is predicted to yield only the “telescoping” interpretation. Examples (530)–(537), which can be heard as true, join (506) and (507) as counterexamples to Economy. Permutation of many or most with a universal quantifier gives a nonequivalent interpretation, and here Economy expects parallel scope inversion. It could very well be that these sentences are ambiguous between the “telescoping” interpretation and the parallel one. But parallel scope inversion becomes unexpectedly rather remote in (538)–(544). The universal quantifiers as second subject seem to compel the (false) “telescoping” interpretation. The nonequivalence of permuting the universal quantifier with many surely does not improve the prospects for parallel scope inversion.65 With definite descriptions as the second subject— (530) A limit ordinal beyond ω is the least limit ordinal for many ordinal numbers, and ω is too. (531) A state seal is affixed to many court documents, and the Great Seal of the United States is too. A state seal seals many a court document, and the Great Seal of the United States does too. (532) Last year, a state district attorney prosecuted many a trespass onto government property, and the federal district attorneys did too. (533) A special prosecutor had power of subpoena in many special investigations this quarter, and (naturally) the attorney general did as well. (534) An open wound killed many of my patients, and congestive heart failure did too. (535) A schmear of sunblock alone between naked skin and direct sunlight protected many noses from a serious burn, and the big cabana did too.66 (536) (In the refrigerator,) a portion-sized dairy container hid/stored many a kosher treat, and the meat locker did too. (537) The seal of a local rabbi adorns many marriage contracts, and the Pope’s seal does too. The seal of a local rabbi adorns many a marriage contract, and the Pope’s seal does too.
PredP and PredP: Of Subjects and Ancient Grievance
301
With universal quantifiers as the second subject— (538) A limit ordinal above ω2 is the least limit ordinal for many ordinal numbers, and every limit ordinal below ω2 is too. (539) (In the refrigerator,) a portion-sized dairy container hid/stored many a kosher treat, and every meat container did too. (540) The seal of a local rabbi adorns many marriage contracts, and the seal of every local priest does too. The seal of a local rabbi adorns many a marriage contract, and the seal of every local priest does too. (541) An open wound killed many of my patients, and every airborne pathogen did too. (542) A second internal hard disk backs up (with a second set of secure files) many of the house servers, and every networked mainframe does too.67 (543) A single chip implanted inside (alone) tracks many robots, and every satellite in geosynchronous orbit does too. (544) A local U.S. unit guards many a nuclear facility, and every (local) Soviet unit does too. If one holds that “telescoping” is outside the purview of Economy and parallelism and then does not yield to examples like (501) and (502) relying on many or most to discriminate the target from “telescoping,” the descriptive alternative posed by “telescoping” threatens to undermine contrasts taken earlier to confirm Economy, such as between (493) and (494) and between (495) and (496): (493) One of the film critics admires every movie. The person who produced it does too. (494) One of the film critics admires every movie. The person who produced the film festival does too. (495) One of the film critics admires every movie. Every person who was involved in its production does too. (496) One of the film critics admires every movie. Every person who was involved in this production does too. Fox puzzles over the absence of a weak crossover effect that parallel scope inversion predicts for the second sentences of (493) and (495). Were it rather “telescoping,” no weak crossover would be expected since there would be no crossing over. I would not risk anything valuable on the judgment, but the parallel inversion found earlier in (506) does seem absent under threat of weak crossover in (545), which has only the “telescoping” interpretation. Economy predicts to the contrary that (545) should be more rather than less favorable to parallel inversion.
302
Chapter 3
(506) A tattered Confederate flag still flies over many a Southern state capitol, and the American flag does too. (545) A tattered Confederate flag still flies over many a Southern state capitol, and the American flag that graces it does too. 3.7
Summary
And, that which is always a sentential connective, cannot be flanked by unsaturated expressions, and therefore conjuncts that contain predicates also contain their subjects and all their other arguments too. When a contrast is observed, syntactic or semantic, between a conjunct with an unpronounced argument and one with argument pronounced, it is a difference between a clause of a certain size and one even bigger. The pronunciation of an argument cues an articulation (section 3.2) interposing an adverb the semantic force of which distinguishes (94) and (95) (repeated here), where nothing else could: (94) #Kunstler is sitting and is standing. (95) Kunstler is sitting, and Kunstler is standing. Kunstler is sitting, and he is standing. The earlier sections began with the pronunciation of subjects in tensed auxiliary phrases like (94) and (95). Section 3.2.1 went on to the effects of pronunciation on smaller clauses, where again pronunciation of an argument enlarges the conjunct that contains it (see also (83)–(89) above). The size of a coordination is tailored according to both verbal morphology and argument pronunciation. The very largest coordinations of tensed auxiliary phrases with subjects pronounced in both conjuncts confine everything within so that nothing escapes to include the entire coordination within its scope: (119) *No childi is watching Sesame Street, and heri teddy bear is falling asleep. A coordination of tensed auxiliary phrases with second subject unpronounced shrinks to allow the pronounced subject to escape outside the coordination as in (177) while its adverbialization remains confined, (176): (150) ?No child is watching Sesame Street and (watching Sesame Street) is looking unhappy. (176) * [No x : NPi] ([℩E : NPi[E,x]][∃E0 : N[E,E0]](… Tense[E0] …) and [℩E : NPi[E,x]][∃E0 : N[E,E0]](… Tense[E0] …)) (177) [No x : NPi]
([℩E : NPi[E,x]] [∃E0 : N[E,E0]](Φi … Tense[E0] …) and [℩E0 : Φi] [∃E1 : N[E0,E1]](… Tense[E1] …))
PredP and PredP: Of Subjects and Ancient Grievance
303
Without a pronounced second subject and an adverbialization to shift event reference, one continues with those events under discussion that have been antecedently described. Silence is resumption. The only way to close off the event variable exposed in the second conjunct, E1 in (177), is with an event pronoun denoting the antecedent events. It imposes an asymmetric conditioning relation between the conjuncts, which fits (150) but fails the intended meaning elsewhere: (124) #No soul man was shimmying and was shaking (to that funky disco music). If the thought is not for the second conjunct to comment on the events described by the first, smaller conjuncts are required so that both fall within the scope of the pronounced subject’s adverbialization and both can make independent remarks on the subject’s participation in whatever: (120) No soul man shimmied and shook (to that funky disco music). [No x : NPi][℩E : NPi[E,x]] ([∃E0 : N[E,E0]](… Tense[E0] …) and [∃E0 : N[E,E0]](… Tense[E0] …)) (123) No soul man was shimmying and shaking (to that funky disco music). [No x : NPi][℩E : NPi[E,x]][∃E0 : N[E,E0]]… Tense[E0] … be[Ej] … ((… shimmy[Ek] …) and (… shake[Ek] …)) A coordination of simple tensed verb phrases manages this in (120), as does the coordination of participial phrases complementing the tensed verb in (123). It is a plain fact that subjects go missing in (361) and (362), and the unacceptability of (363) is a gatekeeper for what may pass as its analysis: (361) Every soul man was shimmying to that funky disco music and was shaking to it (too). (362) James Brown was shimmying to that funky disco music and was shaking to it (too). (363) #No soul man was shimmying to that funky disco music and was shaking to it (too). Despite the missing subjects, (361) and (362) do not contain a predicate derived by predicative coordination. For, if this means anything at all, it means that the quantificational closure of such a predicate by any other quantifier should be just as good, which (363) halts. Conceding clausal coordination and the presence of null subjects, one must still guard against an across-the-board movement that would allow (363) in the course of deriving (361). As just remarked, (150) attests that the quantifier can sometimes include the coordination within its scope. The persistent unacceptability of (124) and (363) has rather to do with restrictions on reference to events in the second conjunct, which are met felicitously in (150) but not in (124) and (363). The thought in (150) is to deny that Sesame Street induces unhappiness. In (124) and
304
Chapter 3
(363), the disjointed herky-jerky of shimmying and shaking puts us in no mind to see a connection between them, and thus there is no analysis of these sentences that affords the pronounced subject the widest scope.68 The same considerations and the observation that neither (361) nor (362) entrains some new twist between shimmying and shaking imply that the widest scope for the subject is not the source for the felicitous use of these sentences. Rather, the pronounced subjects remain within the first conjuncts, and the null subjects in the second conjunct are null descriptive pronouns, interpreted via “telescoping” to derive covariation in (361). “Telescoping” itself discriminates among quantifiers, and introducing it in (361) is not expected to provide (363) with an unwanted, alternative derivation. Recall also that the “telescoping” adverb can be the source of further variation, between (145) and (146), which would be lost if the subject quantifier could simply include the entire coordination within its scope without the adverbial intermediary: (145) Every race car was skidding and (then) (it) ate up track heading for the viewing stands. (146) ?#Every race car has skidded and (then) (it) ate up track heading for the viewing stands. The ancient grievance against Conjunction Reduction worries that (63) should not mean the same thing as (64), here repeated: (63) All night long, a rocker has shimmied and (has) shaken. (64) All night long, a rocker has shimmied and a rocker (has) shaken. Without worry, Conjunction Reduction may apply helter-skelter reconstructing indefinite or definite description, and no “sloppy” interpretation of (63), equivalent to (64), will threaten if the reconstruction ends up describing the participants in the same events described by the antecedent clause. A “sloppy” interpretation is arrived at only when the (in)definite description’s pronounced token and its reconstructed, unpronounced token end up in the scope of different event quantifiers, and even then, neither event quantifier should be a definite description referring to the events described by the other clause. The type and distribution of event quantifiers regulate “sloppy” interpretations and distinguish the logical syntax of (63) and (64). When a null coordinative pronoun referring to the events of an antecedent clause is forced to the edge of its own conjunct and beyond the reach of any inverting quantifier, as when tensed auxiliary phrases are coordinated, the effects of fixing that non-“sloppy” reference are felt across a range of interpretations (section 3.4). Reconstruction, in particular, appears to fail in that “sloppy” interpretations disappear altogether. The questions treated in the preceding sections are just those that attend the singular thesis that and is a sentential connective. Their treatment owes something
PredP and PredP: Of Subjects and Ancient Grievance
305
to Eventish—certainly it is a Davidsonian conceit that reconstruction will innocently fail to generate unwanted interpretations because the token of ‘[A x : rocker(x)] W[E,x]’ in the second conjunct happens to apply to the same events as its token in the first. Event quantification and adverbialization from chapter 10 have also figured prominently, and to a lesser degree to the extent necessary to wriggle out of a syntactic puzzle (section 3.5), chapter 2’s treatment of collectivized Right-Node Raising joins the account reconciling the nature of and to the facts. I turn now to engage coordination and supermonadicity, Eventish substructure, to coordinations of smaller phrases.
4
PredP and PredP: Coordination vs. Subordination
To begin with the conjunction itself, and means ‘&’. Hence, DP and DP is a covert coordination of clauses. It is not however the case that and may take the place of ‘&’ wherever the latter has appeared in earlier logical forms. One cannot after all say *“Jones and buttered and the toast and in the bathroom and with a knife and at midnight.” The notation ‘&’, to be sure, is merely expedient. The formal claim should be that concatenation may express logical conjunction as in (1) in composing the meaning of rotate slowly, as well as quantification in Every planet rotates, rather than, say, function application throughout (see Pietroski 2005, 2006). Or, alternatively, all concatenation could be quantificational as in (5), the method favored in Plurals and Events. That is, nominal quantification derived by QR, [y″ [D NP]Y″], translates as [D : NP]Y″ and the composition of thematic relations, other functional projections such as Tense, and event concepts, all instances of X0 in [x″ Q [x′X0 Y″]], translate as [Q : X0]Y″. If correct, this amounts to the following eccentricity: in natural language, some restricted quantification is phrased, ‘(Q A)B’, so that the restriction is in construction with the quantifier—this is the case of nominal quantification (and in fact any phrase that has undergone movement)—and, some restricted quantification is phrased ‘Q(A,B)’, with the restriction in construction with the matrix predicate, the case of in situ event quantification.1 (1) (2) (3) (4) (5) (6) (7)
σ satisfies ΦΨ if and only if σ satisfies Φ & σ satisfies Ψ Thaetetus flies. ∃e(Ce([℩x : Tx](W(e,x)(Present(e,u)(Agent(e,x) fly(e)))))) [c″∃e [c′Ce[AgrS″ [ιx : Tx][AgrS′W(e,x) [t″Present(e,u) [θ″Agent(e,x) [V″ fly(e)]]]]]]] [∃e : Ce][℩x : Tx][℩E : W(e,x)][℩E : Present(e,u)][℩E : Agent(e,x)] fly(e) Venus rotates. Mars rotates. All the other planets rotate too. Venus rotates, and Mars rotates, and all the other planets rotate too.
Yet the question about and and ‘&’ remains in a slightly different guise. There may be little difference between the meaning of and and concatenation to tell apart (6)
308
Chapter 4
and (7), yet it remains that and cannot freely take the place of concatenation, slowly (*and) rotates. A difference in grammar between coordination and subordination survives the Davidsonian view of the latter. In section 2.4, it was suggested that as a point of grammar rather than logic, some morphemes demand complements, and that coordination, unlike subordination, whether represented by ‘&’ or by concatenation, is opaque to complementation. Subordination was taken to indicate scope relations among the sentence’s event quantifiers. The subordinated phrase falls entirely within the scope of all the quantifiers, event quantifiers in particular, introduced with the superior phrase. Thus the only way to parse the concatenation of Γ in (8) is as shown (cf. Kayne 1994): (8) [Qe1 : Φ] … [Qen : Φ]Ψ Γ ⇒ [Qe1 : Φ] … [Qen : Φ](Ψ Γ), that is—[Qe1 : Φ] … [Qen : Φ](Ψ & Γ) Recall also that across any sequence of event quantifiers, an outer quantifier restricts the events in the domain of the next one in its scope, so that in some sense subordinate phrases are always describing subordinate events. It was observed that the Slavic comitative construction (9) is in effect to be analyzed as in (10): (9) Boris s Natashej vyigrali $100. Boris.nom with Natasha.inst won.pl $100. (10) Boris participated (‘W’); in some of this, Natasha was an accomplice (with); and all of that was winning $100. Coordination, in contrast, escapes the subordination of event quantifiers and makes possible the parse in (11): (11) ([Qe1 : Φ] … [Qen : Φ]Ψ) and Γ But when the Slavic coordination in (12) is so parsed, it strands the thematic relation W in the first conjunct and compels its complements to be reconstructed at Δ: (12) Boris i Natasha vyigrali $100. Boris.nom and Natasha.nom won.pl $100 (13) a. (Boris W Δ) and Natasha W won $100. Alternatively, b. Boris ((W Δ) and Natasha W won $100) Given that and is univocal, sentential, and also opaque, the complementation of any phrase within a conjunct must be complete within that conjunct. Thus a coordination of DPs, which are themselves sentential operators, is a coordination of clauses as before, but every coordination of DPs must also instantiate Right-Node Raising as in (13) so that the first conjunct may be complete. Finding Right-Node Raising in every DP coordination should not incite further rebellion if one has already acquiesced to and as univocal and sentential. Surely if the first conjunct is allowed some sentential content, such as the thematic relation W, no one would expect to find its
PredP and PredP: Coordination vs. Subordination
309
complements in the next clause over. If there is W or any other thematic relation, then, uncontroversially I think, there is also Δ to provide local complements: (14) DP W Δ and DP Φ With a Δ given any W, one still has room to quarrel over the location of Δ’s antecedent, over whether the Right-Node Raised constituent is properly contained within the second conjunct or outside the coordination entirely. (As in chapter 2, I have only to remark that the semantics proposed there is consistent with either view.) 4.0 PredP and PredP: Coordinating supermonadic PredPs. Phrasing—complementation—quantification
Still to be determined is the structure of coordination that the above considerations do not fix. The above observes only that the apparent coordination of DPs indicates a coordination of clauses, large enough in (14) that one cannot mistake the second conjunct for a grammatical complement to the preceding conjunct or to any phrase it properly contains. Moreover, any such phrase, ‘W’, in need of grammatical complement Δ must find it within the first conjunct, given the parse in (15). (15) (DP W Δ) and (DP Φ) But absent an overt subject DP in the second clause, it is less certain how a coordination of predicative phrases should go. Suppose both Δ1 and Δ2 to be complements for ‘W,’ then either parse in (16) or (17) satisfies the constraints considered so far to pertain to scope, subordination, and coordination. (16) DP W (Δ1 and Δ2) DP ∃e(W(e) (Δ1(e) and Δ2(e))) (17) DP (W Δ1 and W Δ2) DP ∃e(W(e) Δ1(e) and W(e) Δ2(e)) DP (∃eW(e) Δ1(e) and ∃eW(e) Δ2(e)) With supermonadic PredPs, the same question is revisited several times over. In (18), it could be that there is only a small coordination of event concepts while thematic relations and Tense remain outside as in (19): (18) The rockers shimmied and shook. (19) ∃e[the X : … Rx …](W(e,x)(Past(e,u)(Agent(e,x) (shimmy(e) and shake(e))))) Perhaps each conjunct instead contains a token of the thematic relation Agent in addition to the verb while only Tense and ‘W’ remain outside, or perhaps tokens of Tense are included in each conjunct too, or perhaps the conjuncts are even larger to the point where each contains tokens of all the thematic roles, Tense morphemes,
310
Chapter 4
verbs, and so on. As far as I can see, any of these structures would be coherent, and for every verbal projection, we might wonder about its disposition when it tangles with predicative coordination. Thus the question may also be asked of the freestanding auxiliary verb and its participial complements in (20). It could be that there is only a small coordination of participles while the auxiliary verb and Tense remain outside as in (21), or with both conjuncts containing tokens of the auxiliary verb and Tense, it could be that those in the second conjunct are merely unpronounced as in (22): (20) a. The rockers have shimmied and shaken. b. The rockers have shimmied and the rollers shaken. (21) a. The rockers have (shimmied and shaken), i.e., The rockers Present have (shimmied and shaken) b. The rockers Present have (shimmied and the rollers shaken) (22) a. The rockers (have shimmied and have shaken), i.e., The rockers (Present have shimmied and Present have shaken) b. The rockers (have shimmied and the rollers have shaken), i.e., The rockers (Present have shimmied and the rollers Present have shaken) As all of these elements express binary relations between events or between events and objects, there is little to expect from their logical type to tell them apart. If they differ in their interaction with predicative coordination, some further point of grammar is at issue. 4.1
(Tense+) Aux sharing
In the case of a freestanding auxiliary verb, whatever grammatical relation holds between it and its participial complement, this one appears to tolerate occasional interruptions—The rockers have sinuously shimmied. If coordination is also among the tolerated interruptions, the auxiliary verb in (20) manages to find its participial complements despite their containment within the coordination proper, as in (21). If, on the other hand, coordination gets in the way, (20) should be parsed as in (22), with tokens of the auxiliary verb within the conjuncts and nearer their participles. Movement (23) and ellipsis (24) tell that the auxiliary verb in (20) stands outside the coordination as in (21), precluding the structure in (22) as a source for (23) and (24). (23) Though shimmied and shaken the rockers have for sure, not an ankle was twisted nor an elbow sprained. (24) The rockers have shimmied and shaken, and the rollers have too.
PredP and PredP: Coordination vs. Subordination
311
Chapter 5 provides further argument that (20) is at least sometimes parsed as in (21), with all the semantic force of the auxiliary verb coming from a single token outside the coordination. If such a parse be possible, then whatever binds auxiliary and participle is clearly not taut enough to compel the alternative in (22). Absent such compulsion, one might then conclude from further consideration of economy conditions that (20) is never parsed as in (22) since ellipsis would be gratuitous. Some evidence in chapter 5 also supports this stronger conclusion that (21) is the only way to parse (20).2 The argument of chapter 5, like the plain syntax of (23) and (24), is robust in that it neither relies on any of my contrivances nor provides any particular support for them. Next, chapter 6, calling on an important difference in meaning between (20) and (25) already encountered in section 3.4.0, corroborates that (20) has an analysis unavailable to (25). (20) The rockers have shimmied and shaken. (25) The rockers have shimmied and have shaken. Again, the meaning of (20) will endorse (21), where Tense and the auxiliary verb are tokened exactly as they appear overtly. The interest of this section is not, however, to second a well-established parse. Rather, while skating along with a naive and secure parse, the section manages to explain an unexpected variation in the meaning of predicative coordinations. To properly render that variation takes the full power of supermonadicity, which, if true, defeats alternative approaches to the meaning of coordination—what’s in it for me. 4.2
Bound morphemes and affixation under coordination
Tense morphemes, the abstract causative and topological relations of supermonadic PredPs, and the thematic relations are all taken to be bound morphemes, affixes, no different in this respect from the verbal modifiers re- and co-. I can no more say “Sig’d” to mean that Sig was an agent than I can say “Sig re’d” to mean that he did it again. In contrast to an auxiliary verb with participle, the grammatical relation between a bound morpheme such as Tense and its host is easily frayed by interruptions—*The rocker shimmy-sinuously-ed, *The rocker re-sinuously shook—including coordination. Note that affixes re- and co- do not modify coordinated verbs:3 (26) *Sig will re-(varnish and string) the Amati before mounting it over the mantelpiece. (27) *The surgeons will co-(laborate and operate) in the OR. (28) Through the miracle of science, we can now rehydrate and *(re-)constitute whole potatoes. (Cf. The mashed potatoes have been de- and re- hydrated for your dining pleasure. (Wilder 1994))
312
Chapter 4
As Lin (2000) remarks, the same is true of Tense: the conjuncts of (29) cannot share the past-tense morpheme, expressing the equivalent of (30) (or (31)): (29) *The rocker shimmied and shake. The rocker Past+(shimmy and shake). (30) The rocker did shimmy and shake. (31) The rocker shimmied and shook. The rocker ((Past+shimmy) and (Past+shake)) Coordination cannot separate a bound morpheme from its host.4 Unlike the freestanding auxiliary verbs, tense morphemes appear within both conjuncts, affixed to the verb. The story of re-, co-, and Past mercifully decides much of the structure of (31) ((18) above) if we continue with the thought that the elements of a Davidsonian decomposition are bound morphemes too. (31) The rockers shimmied and shook. (19) ∃e[the X : … Rx …](W(e,x)(Past(e,u)(Agent(e,x) (shimmy(e) and shake(e))))) In (19), none of the verbal, bound morphemes will reach their host verbs. If these are indeed bound morphemes, they must be tokened within both conjuncts, and the constituents of coordination must be fairly large (Wilder’s (1994, 1995, 1997) “largeconjuncts-with-deletion”). It isn’t that V and V or VP and VP might not make sense. Rather, it happens for reasons poorly understood that in sentences simple or complex, one doesn’t say a VP without including some functional relations expressed by bound morphemes. It is their presence that compels the coordination of larger constituents. Some may choose not to see the moral in the story of re-, co-, and Past, preferring to consign it to something called word formation, but I cannot in good faith join them, having just written the decomposition of predicates across the face of syntax. It would appear then that thematic relations and most other elements of the supermonadic PredP are all tokened within each conjunct. Morphology and syntax together seem to dictate a structure for (31), imposing the largest conjuncts on its coordination. If the morphosyntax is fixed on the largest conjuncts, any further logical form for (31) placing Tense outside the coordination, as in (19), comes with a claim that the overt morphology inside is anaphoric or merely grammatical agreement with what is interpreted outside: (19) ∃e[the X : … Rx …](W(e,x)(Past(e,u)(Agent(e,x) (shimmy(e) and shake(e)))))
PredP and PredP: Coordination vs. Subordination
313
Rather, let’s pursue the idea that the morphosyntax decides logical form. That is, tokens of bound morphemes are semantically effective in situ. Sentence (31) contains two tokens of the past-tense morpheme and therefore two temporal quantifiers, and thus its logical form remains distinct from that of (30) with only the one.5 In contrasting auxiliary verbs and bound verbal morphology, elements of similar logical type end up with a rather different distribution. Nevertheless, the treatment to be offered settles into a comfortably naive view of pairs like (30) and (31). Roughing out their phrasing, one hears in (30) a single auxiliary verb pronounced, and without cause to suppose there is another, another is not heard unpronounced. In (31), one hears two past-tense morphemes pronounced, and without cause to imagine otherwise, one takes what has been heard at face value. For all morphemes, both the bound morphemes such as Tense and the freestanding auxiliaries, it turns out that overt tokens of these morphemes are tokened in logical form where they overtly appear to be tokened. A nice arrangement it is, to make a pronouncement and be taken to have arranged the things pronounced as they are intended to be understood. Even sweeter is to convey a thought and have it represented in the mind subject to the same grammar as if one had pronounced it. Assuming as much will lead throughout what follows to certain patterns of reasoning. So, for example, discovering ambiguity in (33) between a collective interpretation (33b) and a distributive one (33a) equivalent to what (32) expresses, suppose one hastens to characterize the difference with an operator, say a quantifier ‘[∀x : Xx]’ that applies for the sake of one but not for the other. (32) The rockers each lifted the piano. [The X: rocker[X]][each x : Xx]∃e lift-the-piano[e,x] (33) The rockers lifted the piano. a. [The X: rocker[X]][∀x : Xx]∃e lift-the-piano[e,x] b. [The X: rocker[X]] ∃e lift-the-piano[e,X] Then one has as good as said that a larger phrase expresses the distributive interpretation of (33) than expresses its collective interpretation. For to be a semantic operator is to be a morpheme or an assembly of morphemes, and so a phrase of morphosyntax (“Generative Semantics Meets Minimalism”). As such, there may not be room enough for a semantic operator, pronounced or unpronounced, in some kinds of phrases while it finds a place in others (see section 6.1). Yet further consequences for logical syntax await what else may be discovered about morphosyntax. So, if every semantic operator is one morpheme or several, and morphemes are known to come bound or freestanding, and if there is a tendency for zero—that is, unpronounced—morphemes to be bound (Pesetsky 1995), and for the unpronounced distributive operator to be a bound verbal morpheme, then like the tense morpheme
314
Chapter 4
discussed above, the distributive interpretation of (34) must only be parsed as in (36) locating the distributive operator within the conjuncts. A bound morpheme lying outside the coordination, as in (35), will never reach its hosts. (34) The rockers lifted the piano and smashed a guitar. (35) *[The X: rocker[X]][∀x : Xx](∃e lift-the-piano[e,x] and ∃e smashed-a-guitar[e,x]) (36) [The X: rocker[X]]([∀x : Xx]∃e lift-the-piano[e,x] and [∀x : Xx]∃e smashed-a-guitar[e,x]) It would then follow on morphosyntactic grounds that there is no distributive over a conjunctive (35) although it is as sensible as, and equivalent to, the conjunctive over distributive (36) (cf. section 4.2.0). Highlighting again the traffic between logical form and morphosyntax, suppose now that after having decided on the semantic analysis in (33) and the conclusion that the phrase representing the distributive interpretation is the larger, one stumbles across (syntactic) evidence that sentence (32), containing a pronounced each, is larger still than any parse of (33). Whether it is the one quantifier pronounced ‘[each x : Xx]’ or another unpronounced ‘[∀x : Xx]’, there is, on the face of it, no basis to expect a larger phrase in (32). Rather, the evidence of a larger phrase must betray the presence of some as yet undetected morpheme, as in (37). And correlatively, unless there is a morpheme without meaning, it should be cause to doubt the superficial equivalence of (32) and the distributive interpretation of (33), prompting a search for some telltale, discriminating context. (37) The rockers each lifted the piano. [The X: rocker[X]][each x : Xx][? [∃e lift-the-piano[e,x]]] 4.2.0
Why not the semidistributive of the disjunctive?
The morphosyntax of bound morphemes, even as it applies to unspoken morphemes, does some real semantic work, thus shaping logical syntax as well and providing further argument that The rocker shimmied and shook is not an alternative disguise for a logical form shared with The rocker did shimmy and shake. In a semidistributive interpretation, a distributive quantifier occurring with a collective predicate appears, in effect, to quantify over groups. Thus, (38) reports that few Columbia students were among the groups (of Columbia students) that gathered, as in (39). (38) Few Columbia students gathered. (39) [Few x: Sx][∃X: Xx & ∀x(Xx → Sx)]gather[X]
PredP and PredP: Coordination vs. Subordination
315
Indefinite or definite descriptions, when understood distributively, also support the semidistributive interpretation. Thus (40) asserts that any Columbia student is to be found among those gathered at some meeting point or another. (40) The Columbia students (all) gathered (at various meeting points). In Schein 1993, chapter 8 (see also section 13.0 below), the semidistributive interpretation derives from an operator that modifies thematic relations so that Co-θ[e,x] is true of an object and an event just in case that object along with perhaps some other objects bears the thematic relation θ(e,α) to that event. The logical forms for (38) and (40) turn out along the lines of (41) and (42): (41) [Few x: Sx]∃e(Co-θ[e,x] & gather(e)) (42) [the X: … Sx …][∀x: Xx] ∃e(Co-θ[e,x] & gather(e)) The details of the analysis are immaterial here.6 What proves crucial for present purposes is just that the operator charged with semidistributivity, modifying either thematic relations or verbs themselves, is a bound morpheme. Not all collective predicates can participate in semidistributive interpretations, as Dowty (1987) and Taub (1989) have discussed. It requires the predicate to have some property.7 Whatever it is, both huddle in the center and walk in single file have it, since these are among those predicates that obviously allow a semidistributive interpretation, as (43) and (44) attest. (43) a. Few combatants walked on the outside in single file. b. The combatants all walked on the outside in single file. (44) a. Few combatants huddled in the center. b. The combatants (all) huddled in the center. Thus (43a) has an interpretation according to which few combatants are said to be walking on any of the lines forming on the outside. Similarly, (43b) can be understood to allow all the combatants to be distributed among several groups walking in single file. In (44), the combatants may be distributed among several huddles. Now observe that these predicates may be conjoined in a collectivized, disjunctive interpretation. Suppose, for example, that the combatants belong to opposing sides that have chosen different tactics, one side preferring a more defensive huddle and the other choosing to encircle offensively. Sentences (45) and (46) can allow that some of the combatants huddled and others of them walked in single file. (45) The combatants huddled in the center and walked on the outside in single file. (46) Five thousand combatants huddled in the center and walked on the outside in single file.
316
Chapter 4
The predicate derived by such a coordination denotes some combatants just in case they consist of some huddlers in the center and some walkers in single file. The derived predicate is itself a collective predicate, and one may now inquire whether it too supports semidistributivity. To this end, consider (47)–(49): (47) Few combatants huddled in the center and walked on the outside in single file. (48) All the combatants huddled in the center and walked on the outside in single file. (49) The combatants all huddled in the center and walked on the outside in single file. These sentences do have the acceptable semidistributive interpretations in (50) and (51). These, however, impose a conjunctive condition on each combatant, requiring that he participated (with others) in both actions. (50) Few combatants are such that he huddled in the center with other combatants, and he walked with other combatants on the outside in single file. (51) All the combatants are each such that he huddled in the center with other combatants, and he walked with other combatants on the outside in single file. What is absent is an interpretation that, in effect, quantifies over groups that can be divided between huddlers and walkers. Such an interpretation of (47) would assert that few combatants are among combatants that huddled in the center and walked in single file. It would yield a disjunctive condition on the individual combatant: (52) *Few combatants are such that he huddled in the center with other combatants, or he walked with other combatants on the outside in single file. (53) *All the combatants are each such that he huddled in the center with other combatants, or he walked with other combatants on the outside in single file. The absence of this interpretation is striking and unexpected since all it does is combine the mechanism for semidistributivity, whatever that is, with the collective predicate derived by coordination: (54) [Few x: Cx][∃X: Xx & ∀x(Xx → Cx)] [VP[VP[… huddle …] and [VP … walk …]] (X) It is easy enough to dig oneself into a hole and not emerge: suppose that in deriving the disjunctive interpretation, and is an operator composing phrases of a given logical type to derive a third phrase of the very same type, expressing the derived, collective predicate, and suppose further that the semidistributive operator—what-
PredP and PredP: Coordination vs. Subordination
317
ever it is—applies, without a morphological footprint, “in the semantics,” where phrases of the same type, coordinated or not, will look much alike. The semidistributive of the collective predicate derived via disjunctive coordination in (47)–(49) should be no worse than the semidistributive of the simple collective predicates in (43) and (44), contrary to fact. Let Co- be instead an unpronounced bound verbal morpheme. The facts of (47)– (49) follow immediately, given that coordination is opaque to bound morphemes. First, suppose, for the sake of argument, that the subjects of huddle and walk involve different thematic relations. Any attempt at a semidistributive interpretation of the coordination expressing the disjunctive, collective predicate must then start from (55), which has no grammatical outcome. Co- has been located beyond the reach of its potential hosts, whether they be thematic relations or verbs. (55) … Co-[[θ1[e,α] huddle(e) …] and [θ2[e,α] walk(e) …]] In contrast, Co- may of course occur within both conjuncts as in (56), but that derives just those interpretations where a relevant combatant is one who meets the conjunctive condition (see (50)–(51)) of having both huddled with others and walked with others in single file.8 (56) [Few x: Cx] [[Co-θ[e,x] & huddle(e) …] and [Co-θ[e,x]walk(e) …]] Second, suppose that the subjects of the constituent predicate phrases share the same thematic relation as is surely the case for (57)–(59): (57) Few combatants huddled on the grassy knoll and huddled near the stone wall. (58) All the combatants huddled on the grassy knoll and huddled near the stone wall. (59) The combatants all huddled on the grassy knoll and huddled near the stone wall. Then (55) might not be the only structure that underlies the conjunction in these sentences. Perhaps the syntactic structure factors out the common thematic relation allowing smaller verbal projections to be conjoined as in (60): (60) Co- Agent[e,α] [[huddle(e) …] and [huddle(e) …]] The agents’ action according to (60) results in huddling on the grassy knoll and huddling near the stone wall. Applying Co- to the unique thematic relation derives a predicate that is true of a combatant just in case he with some others were the agents in some action that resulted in huddling on the grassy knoll and huddling near the stone wall. He himself may have participated in the one or the other, escaping from the conjunctive condition. But (60) is also ungrammatical, as now two
318
Chapter 4
bound morphemes Co- and Agent- (not to mention Tense in higher position) are beyond the reach of their hosts. The curious anomaly that (47)–(49) and (57)–(59) are missing the semidistributive of the disjunctive collective is but evidence that the semidistributive operator is a bound, verbal morpheme. Not much happens in logic without its reflection in morphosyntax, pronounced or unpronounced. 4.2.1
Appendix: Bound morphology vs. phrasal complementation
Wary of innovations in syntax, I have let my morphemes, C-, O-, Agent-, and so on—with the exception of W—fall in with the better-known bound morphemes and thus be subject to the opacity of coordination (section 4.2). Yet, prior to consideration of the morphosyntax, it has been a convenience in paraphrase and in logical form to treat the topological/causative relations as if they were freestanding verbs, as is the practice when explaining the decomposition of transitives as causative constructions. Thus in (61) what the students did is said to coincide with surrounding the Pentagon and being crowded into the Mall. The placement of O- in (62) outside the coordination suggests a higher verb with the coordination as its clausal complement. (61) Twenty thousand students have surrounded the Pentagon and crowded into the Mall. (62) [∃X: 20,000(X) & …][℩E : NPi[E,X]][∃E0 : N[E,E0]]Wi[E0,X] [℩E0: proi]have[E0] [℩E0: proi][ιE1:proj]O[E0,E1] (j(∃E1∃XW[E1,X] … surround …) and (∃E1∃XW[E1,X] … crowd …)) But the force of the interpretation (see chapter 6) rests on what is denoted by the event pronouns, the actual arguments to the Overlap relation. The first proi denotes whatever the students did (having) and the second proj refers collectively to the events of surrounding the Pentagon and being crowded into the Mall. The pronoun proj in ‘[℩E0: proi][℩E1:proj]O[E0,E1] (jΦ and Ψ)’ follows the example of its overt counterpart in A kind word triggering it, Tiff shouted obscenities and Biff swore revenge. Provided that the content of the event pronouns remains fixed, it matters little where ‘O[E0,E1]’ appears. So, to humor opacity, even if O- is a bound morpheme and tokened within the conjuncts, the intended interpretation is represented just as well by (64), where event pronouns alone are exported outside the coordination, as in schema (63).
PredP and PredP: Coordination vs. Subordination
319
(63) … [℩E0: proi][℩E1: Φ Ψ]((O[E0,E1] Φ) and O[E0,E1] Ψ). (64) [∃X: 20,000(X) & …][℩E : NPi[E,X]][∃E0 : N[E,E0]]Wi[E0,X] [℩E0: proi]have[E0] [℩E0: proi][℩E1:proj] (j(O[E0,E1]∃E1∃XW[E1,X] … surround …) and (O[E0,E1]∃E1∃XW[E1,X] … crowd …)) Notice in (63) and (64) that the event pronoun outside the coordination reaches down into the conjuncts to pull out Φ and Ψ, ignoring the presence of ‘O’ within each conjunct. This eccentricity is overtly attested in (65) and (66) (Fiengo and May 1994, 194–200). (65) Connors can win the U.S. Open, and McEnroe can win Wimbledon. I predict that one of them will, but the other won’t. (66) It will beat the odds if it all happens, but Connors boasts he will win the U.S. Open, McEnroe wants to win Wimbledon, Lendl is predicted to take the French Open, and Sea Biscuit promises to place at Preakness. The VP Ellipsis in (65) denotes events of winning the U.S. Open or Wimbledon, suppressing the modal can in the antecedent clauses. In (66), the pronoun manages to collectively denote Connors winning the U.S. Open, McEnroe winning Wimbledon, Lendl taking the French Open, and Sea Biscuit placing at Preakness, which it does while ignoring quite a bit more material in the antecedent clauses. In (63) and (64), eccentricity becomes the rule. In search of a more attractive solution, let us stipulate that the event pronoun takes up as much of the content of the coordination as does not contain free occurrences of its own variable. In this way the pronoun’s content is fixed by the complement clause so-called. Although somewhat obscured now, the basic claim remains that Φ and Ψ describe some events, the event pronoun simply denotes the events antecedently described, and what the students did is said to overlap these events. The stipulated rule is tribute paid in service to the idea that O- must be a bound morpheme tokened in both conjuncts. This exercise illustrates how the position of event pronouns and the determination of their content by their position could emulate complementation (see chapter 6 for further discussion)—that, for example, the “complement” of O- (or C-) is an entire coordination and its subject the W-ing—despite migration of all the bound morphology onto the verbs. So noted, I will nevertheless let the relations O- (and C-) remain outside coordinations and leave the more transparent (62) as is, to guide translation and rough paraphrase.
5
PredP and PredP: (Tense+) Aux Sharing
This chapter assembles the evidence for the conclusion in section 4.1 that auxiliary verbs are indeed shared in logical form when their overt syntax suggests as much. To start, some observations about scope and meaning urge (2) as an analysis of (1). (1) a. The rockers have shimmied and shaken. b. The rockers have shimmied and the rollers shaken. (2) a. The rockers have (shimmied and shaken), i.e., The rockers Present have (shimmied and shaken) b. The rockers Present have (shimmied and the rollers shaken) Observe that the decreasing quantifier in (3) and (4) may include the second conjunct within its scope, binding a pronominal variable and licensing negative polarity, provided the second conjunct omits pronouncing an auxiliary verb. Its pronunciation in the second conjunct otherwise confines the decreasing quantifier to the first clause in (6) and (7). (3) a. b. c. d.
No rocker was shimmying and shaking (too). No rocker was ever shimmying and shaking. ?No rocker was ever shimmying and ever shaking. No rocker was shimmying and shaking, ever.
(4) a. No rocker has shimmied and his mama shaken to a throbbing disco beat. b. No rocker has shimmied and any of his mamas shaken to a throbbing disco beat.1 (5) Not any student of mine has conjectured a single theorem and any student of yours proven anything different. (6) a. *No rocker was shimmying and was shaking (too). b. *No rocker was ever shimmying and was (ever) shaking. c. *No rocker was shimmying and was shaking, ever.
322
Chapter 5
(7) a. *No rocker has shimmied and his mama has shaken to a throbbing disco beat. b. *No rocker has shimmied and any of his mamas has/have shaken to a throbbing disco beat. (8) a. *No rocker has shimmied and his mama has shaken to a throbbing disco beat. b. *No rocker has shimmied and any of his mamas has/have shaken to a throbbing disco beat. (9) *Not any student of mine has conjectured a single theorem and any student of yours has proven anything different. The contrast tells of some difference of logical form between (3)–(5) and (6)–(9) and so discourages the thought that they are the same but for the speaker’s negligence in (3)–(5) in pronouncing second tokens of the tensed auxiliary verbs. Rather, if in (3)–(5) a single token of the auxiliary verb includes the coordination within its scope, it follows without further comment that the decreasing quantifier includes both conjuncts within its scope as well, while in (6)–(9) (and, for that matter, in (3)–(5) too) the decreasing quantifier simply never escapes from a tensed clause. The contrast is prima facie argument that a freestanding auxiliary verb may be tokened in a single position outside the coordination, without resumption in the second. Yet, as the difference in meaning between (3)–(5) and (6)–(9) might amount to nothing more than a difference of scope for a nominal quantifier, perhaps it could be said that the process relating an unpronounced verb to its antecedent itself suffices to scramble the decreasing quantifier into a higher position commanding wide scope. As much must already be conceded to derive the contrast in (10) (see Johnson, [1996] 2003, 2009, 2002, 1996; Lin 2000, 2001, 2002):2 (10) a. No rocker shimmied slow and his mama fast. b. *No rocker shimmied slow and his mama shimmied fast. *No rocker shimmied and his mama shook. Better then to look for differences of meaning that go beyond what can be couched as variation in DP scope.3 More to the point is Johnson’s ([1996] 2003, 2002, 2009) and Lin’s (2000, 2001, 2002) observation, deriving from Siegel 1984, 1987 and Oehrle 1987, that a modal auxiliary verb or negation is compelled by a coordination of its complement phrases to be interpreted outside the coordination: (11) Kim didn’t play bingo and Sandy sit at home all evening. (12) Ward can’t bathe and his guests watch. (13) Ward can’t slurp caviar and his guests feed on beans.
PredP and PredP: (Tense+) Aux Sharing
323
That is, (11) can only be taken to deny an evening where Kim plays bingo and Sandy sits at home, and (12) and (13) only to report on Ward’s observance of etiquette— although he may bathe and slurp caviar often, he refrains from the former when guests are watching and from the latter when there is nothing for them but beans.4 Given the familiar interaction of modals, negation, and conjunction, the plain meaning of (11)–(13) is certain argument that the second conjuncts leave no room for unspoken tokens of modal verbs or negation. The following sections extend this result to the auxiliary verbs progressive be and perfect have, although the differences of meaning are more subtle distinguishing (Tense+) Aux sharing from the coordination of (Tense+) Aux phrases. 5.0
Progressive be
To elicit one such difference, recall (96) from section 3.2 repeated here as (14), where during trial, counsel sometimes stands before the court and sometimes remains seated. These postures cannot be simultaneous, but that does not preclude (14) from asserting that Kunstler’s action in progress will ceteris paribus amount to some sitting and some standing. Perhaps his peculiar courtroom antics, jumping up and down and thus alternating the postures, is what has warranted the assertion: (14) Kunstler is sitting and standing. In contrast, (15) describes a scene that is impossible but for contortions imagined to be simultaneous sitting and standing: (15) #Kunstler is sitting and is standing. What (15) demands is that what is now in progress be a sitting, and it also be a standing. When both actions, as in (17), can be true of the same moment, there is no harm in repeating the auxiliary verb: (16) Kunstler is glaring at the judge and formulating his rebuttal. (17) Kunstler is glaring at the judge and is formulating his rebuttal. In fact, if the speaker’s point is that Kunstler’s glaring at the judge is his formulating his rebuttal, she is better off with (17), as (16) leaves open that Kunstler may switch between studying his papers to prepare the rebuttal and glaring up at the judge in an unrelated expression of contempt. These observations suggest a difference of logical form between (14) and (15): the coordination of participial phrases in (14) is not the coordination of auxiliary phrases in (15). That is, (14) omits an unpronounced second token of the tensed auxiliary.
324
Chapter 5
Filling out the paradigm, these observations join those of section 3.2 showing a difference of logical form between (15) and (18). They both token two tensed auxiliary verbs but differ apparently only in pronouncing the second subject: (15) Kunstler is sitting and is standing. (18) Kunstler is sitting, and Kunstler is standing. (19) Kunstler is sitting. Kunstler is standing. While at first glance (18) describes the same contradictory scene, recall that a coherent one is reached if one consciously allows that the coordination in (18), like the narrative in (19), advances the moment of evaluation, the now, from one sentence to the next. In contrast, (15) is impervious to such manipulation and cannot thereby escape contradictory description of the very same moment. Recall that pronunciation of the second subject in (18) and (19) cues the articulation of an adverb restricted by the subject’s content, which provides an occasion, in contrast to (15), to recalibrate the temporal anchor in the second clause. Otherwise, repeating the tensed auxiliary verb in the second clause merely invokes reference to the time of the first, an effect of the null coordinative event pronoun standing at the edge of its conjunct (section 3.4). Of course (14) manages to avoid the contradictory scene without advancing the time referred to from one conjunct to the next. As we will see in chapter 6, the logical syntax of coordinating participial phrases proper allows collective reference to sitting and standing so that (14), as remarked above, asserts that action at the present time, the single moment or interval referred to, will amount to some sitting and some standing. A three-way contrast thus emerges among (14), (15), and (18),5 beginning with the absence from (14) of a second tensed auxiliary verb in the second conjunct with any semantic force: (14) Kunstler is sitting and standing. (15) Kunstler is sitting and is standing. (18) Kunstler is sitting, and Kunstler is standing. It further supports contrast in logical form that the observed effects on temporal reference in (14)–(18) are magnified by sequence of tense and reflected in nominal reference: (20) Sigmund is chasing and fleeing the woman he is thinking about. (21) Sigmund is chasing the woman he is thinking about and fleeing her. (22) Sigmund is chasing and is fleeing the woman he is thinking about. (23) Sigmund is chasing the woman he is thinking about, and is fleeing her. (24) Sigmund is chasing and Sigmund is fleeing, the woman he is thinking about. (25) Sigmund is chasing the woman he is thinking about, and Sigmund is fleeing her.
PredP and PredP: (Tense+) Aux Sharing
325
The thinking coincides in (20) and (21) with the chasing and fleeing, the same woman is both hunter and quarry, and the sentences (see (14)) are best taken to report an alternation cat-and-mouse between Sigmund on the chase and Sigmund in flight. In (22) and (23), again the same woman is hunter and quarry, and the thinking coincides with the chasing and with the fleeing, but as the fleeing now coincides with the hunting (see (15)), Sigmund suffers from a paralysis of action. To my ear, (24) and (25) may narrate some chasing and then some fleeing with the further variation now that the thinking that coincides with the chase may be of one woman and the thinking that coincides with flight may be of another. This would follow if the two clauses within (24) and (25), but not those within (20)–(23), offer distinct temporal anchors with respect to which the definite descriptions (overt, tacit, or concealed as pronoun of laziness) are evaluated separately. In sum, (14) and (20)–(21) present a single token of Tense (and so a single token of the auxiliary verb), (15) and (22)–(23) present two tokens of Tense (along with supporting auxiliary verb) with coreference between them, and (18) and (24)–(25), two tokens of Tense without coreference. The earlier scope facts, (3)–(7), taken to suggest that auxiliary verbs are not tokened unpronounced, include cases such as (4) where the second conjunct omitting the tensed auxiliary verb nevertheless pronounces an overt subject. Let us now survey what temporal effects carry over to such cases, distinguishing, say, between (26) and (27): (26) Sigmund conflicted, his heart is chasing and his mind fleeing. (27) Sigmund conflicted, his heart is chasing and his mind is fleeing. But since the heart will chase while the mind flees, I turn to an example where a conflict in the action has sharper consequences. In some controlled but treacherous maneuvers, a race car may skid tires, in which they suffer undue compression and motion lateral to its course, and it may also slip tires, in which their contact with the road and frictional forces are diminished below the norm for straight-and-level traffic. As it attempts to navigate a slalom course, the car when banking to the left skids the left tires and slips the right tires, and then banking to the right, it skids the right tires and slips the left tires. With higher speed and tighter turns, the banks, skids, and slips all become more acute, but if the car remains under control, a regular rocking motion persists, alternating between skidding the left tires while slipping the right tires and skidding the right tires while slipping the left tires. But if at any time the front tires, for example, skid while the rear tires slip (imagine even that the rear tires have lost contact with the road), lateral motion overtakes longitudinal motion, and the race car loses control and careens off course. The prior circumstances are repeated in (28)–(30) in the new setting. Sentence (28) describes the behavior of any tire on the slalom course as it alternates between skid and slip. Sentence (29) describes a physical impossibility, a tire that skids
326
Chapter 5
(greater-than-normal road contact) and slips (less-than-normal road contact) at the same time. Sentence (30), as before, allows the narrative to sample moments in sequence during the ongoing slalom: (28) The tire is skidding and slipping. (29) The tire is skidding and is slipping. (30) The tire is skidding, and the tire is slipping. With a plural subject, note that the alternation may be in unison as in (31), the left tires rock in a controlled slalom between skidding together and slipping together, or counterbalanced as in (32), where it alternates between the left front tire skidding with the right front one slipping and the left front tire slipping with the right front one skidding. (31) The left tires are skidding and slipping (and the right tires are, too). (32) The front tires are skidding and slipping (and the rear tires are, too). Like the overt pronunciation of a second tensed auxiliary verb, it appears that the pronunciation of a second subject, even without a pronounced auxiliary verb, keeps the conjuncts from describing different moments: (33) The left front tire is skidding and the right front tire slipping. (34) A left tire is skidding and a right tire slipping. (35) The left front tire is skidding and the left rear tire slipping. (36) A left tire is skidding, and a left tire slipping. One left tire is skidding, and the other slipping. Sentences (33) and (34) describe what is true of a controlled slalom at alternating moments, at any one of which the left tires are in a skid and the right tires are in a slip. In contrast, sentences (35) and (36) only describe catastrophe, a moment at which front tires are in a skid and rear tires in a slip, and the car is out of control. Note that the contrast between (35) and (28) is especially striking if one is inclined to think that the second conjunct of (28) itself contains a subject DP, either a variable or unpronounced token of the first subject as in (37), (37) The tire is skidding and the tire slipping, pressing harder the question why (37) but not (35) may describe the circumstances where skidding and slipping alternate.6 In light of these further temporal effects, the three-way contrast of (14), (15), and (18) has become four-way: (38) (39) (40) (41)
The The The The
left left left left
front front front front
tire tire tire tire
is is is is
skidding and slipping. skidding and is slipping. skidding and the left rear tire slipping. skidding, and the left rear tire is slipping.
PredP and PredP: (Tense+) Aux Sharing
327
Each of (38)–(41) has a logical form distinct from the others’, and thus none of them differ merely on pronunciation. If one holds that all the second conjuncts contain a subject, then some further difference of logical structure must accompany the decision to pronounce it in (40) if one is to gain any purchase on the difference of meaning between (40) and (38) and between (41) and (39). For the latter, section 3.2 discovered that pronunciation cues the articulation of an adverb, which including a Tense Phrase in its scope in (41) allows a shift forward in narrative time denied to (39). For the contrast between (40) and (38), adverbialization, prompted whenever a DP is pronounced, influences the interpretation of the smallest phrases containing thematic relations, as further elaborated in discussion of the lifetime effect in section 10.0. Section 3.2.1 took up pronunciation’s effects on the logical syntax of smaller phrases and provided for the contrast between (40) and (38). For present worries about (Tense−) Aux sharing, it is enough that the four-way contrast in (38)–(41) demonstrates differences of logical form between clauses with and without a pronounced, tensed auxiliary verb in both contexts—with and without a pronounced subject. With what is already understood about tense and the progressive, the obvious suggestion is that the unpronounced tensed auxiliary verb is altogether absent here. When it is plain, as in (39), that both conjuncts contain tensed auxiliary verbs, the sentence says that the tire at a given moment is in a condition that if persistent coincides with some skidding and at that very same moment, it is in a condition that if persistent coincides with some slipping. If the second tensed auxiliary verb is omitted as suggested for the logical form (38), the same moment is not said to coincide with one behavior and to coincide with the contrary behavior. Rather, (38) says that the tire at a given moment is in a condition that if persistent coincides with some skidding and slipping. As befits two tensed clauses in (41), each conjunct again says of some moment that a condition that holds then, if persistent, coincides with something. But what has been observed is that pronouncing the subject in the second clause further affords the choice of a different moment in the second clause, escaping the description of a contradictory scene.7 5.1
Perfect have
Evidence that coordinated participial phrases share a single token of a tensed auxiliary verb is also found in the perfect construction with have.8 5.1.0
Consequent states of the perfect
Eurydice is a boa constrictor.9 Since digestion is slow and meals rather large, the present perfect (42) reports a valid inference from her behavior last week to her current satiated condition, the expected consequent state of eating: (42) Eurydice ate last week and has therefore eaten herself full. (43) ??Eurydice ate last week and therefore ate herself full.
328
Chapter 5
The simple past (43), in contrast, makes no reference to present conditions, and the resultative herself full is thus left to measure the event of last week and suggest surprisingly that because she ate at all, she must have become full. Sentence (45) says more plausibly that a cow is a full meal, and (44), that a cow last week is enough to be sated now: (44) Eurydice ate a cow last week and has therefore eaten herself full. (45) Eurydice ate a cow last week and therefore ate herself full. The inference in (46) is logical and uninformative, but the inference in (47) to a present consequent state relies on a theory of eating and satiation that subsumes causal laws about how things should turn out ceteris paribus—all things remaining equal.10 (46) Eurydice ate herself full at dinner. So, Eurydice ate herself full. (47) Eurydice ate herself full at dinner. So, Eurydice has eaten herself full. The inference to the good consequences of eating well is unjustified if not all things have remained equal according to the implicit theory: (48) #Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia has eaten herself full. At best, the inference holds of a past prior to the events that fall outside what the causal law covers: (49) Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia had eaten herself full (but will soon be back for dessert). All of (47)–(49) assert that one ate herself full, but the reasoning to present consequences follows a nonmonotonic pattern—What is a consequence of eating oneself full? Being full. What is a consequence of eating oneself full and purging oneself? Not being full.11 Now if the complement participial phrase in a perfect construction is just eaten herself full, then the consequences at issue are just those of eating oneself full. As observed, the inference to a present consequent state is unwarranted under deviant circumstances, and further assertions about the current condition do not improve it. That is, (50) has all the flaws of (48), which can as before be addressed with a shift into the past tense, (51): (50) #Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia has eaten herself full and (she) has still put herself at risk from malnutrition. (51) Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia had eaten herself full and (she) has still put herself at risk from malnutrition.
PredP and PredP: (Tense+) Aux Sharing
329
In contrast, if the complement phrase is a coordination, eaten herself full and still put herself at risk, then the consequences at issue are those of eating oneself full and putting oneself at risk, and these are exactly those that can be inferred if one eats and purges: (52) Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia has eaten herself full and still put herself at risk from malnutrition. The second sentence of (52) manages to avoid attributing to Bulimia’s present condition contrary consequences, those of having eaten herself full, and so being full, and those of being put at risk from malnutrition, that is, of not being full. Crucially, as a result of sharing the tensed auxiliary verb, only those consequences are invoked that follow from the joined condition of having eaten oneself full and put oneself at risk. The effects just observed and the argument for sharing tensed auxiliary verbs carry over to coordinations with pronounced second subjects. Thus, shifting to the past tense (54) or sharing the tensed auxiliary verb (55) is necessary to avoid the anomaly in (53):12 (53) #Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia has eaten herself full, and an eating disorder has put her still at risk from malnutrition. (54) Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia had eaten herself full and an eating disorder has put her still at risk from malnutrition. (55) Bulimia ate herself full at dinner and is purging herself in the bathroom. So, Bulimia has eaten herself full and an eating disorder put her still at risk from malnutrition. 5.1.1 have [[be+en…] and [V+en…]]
Between conjuncts that share a tensed auxiliary verb, there are semantic conditions from which the coordination of Tensed phrases is exempt. In (56)–(61), an active participial phrase is conjoined to an antecedent, passive participial phrase describing a durative process. Sharing the same auxiliary verb have, the second conjunct must describe a coincident process, so that (56) and (57), for example, report constant proof of the fabric’s indestructibility being given throughout repeated insult during the 20-minute test. Similarly, (58)–(61) imply a continuous erosion of the material’s integrity, more and more of it failing as the test proceeds. As the examples also illustrate, the effect holds whether a second subject is pronounced or not. (56) The fabric had been tested (for 20 minutes) and proven itself indestructible. (57) The fabric had been tested (for 20 minutes) and new fibers proven it indestructible.
330
Chapter 5
(58) The fabric had been tested (for 20 minutes) and worn its finish away. (59) The fabric had been tested (for 20 minutes) and friction worn it away. (60) The fabric had been tested (for 20 minutes) and torn its stitching in shreds. (61) The fabric had been tested (for 20 minutes) and its stitching torn it in shreds. Any departure from this relation between the conjuncts renders the shared tensed auxiliary verb anomalous. Thus, (62) and (63) are odd if proof of the fabric’s mettle comes only from some 20-minute test, with no measure intended of its performance during redesign and remanufacture: (62) #The fabric had been redesigned, remanufactured, and tested over 20 years and proven itself indestructible. (63) #The fabric had been redesigned, remanufactured, and tested over 20 years and new fibers proven it indestructible. The intended report requires rather a second token of Tense and auxiliary verb: (64) The fabric had been redesigned, remanufactured, and tested over 20 years and had proven itself indestructible. (65) The fabric had been redesigned, remanufactured, and tested over 20 years, and new fibers had proven it indestructible. Similarly, sharing a tensed auxiliary is awkward if the second conjunct is understood to describe a punctuated event that occurs within a fraction of the test, (66), (69), and (72); and, as above, a tensed second conjunct rescues the intended meaning, (67)–(68), (70)–(71), and (73)–(74): (66) #The fabric had been tested (for 20 minutes) and come apart (in 20 seconds). #The fabric had been tested (for 20 minutes) and come apart a bit. (67) The fabric had been tested (for 20 minutes) and came apart (in 20 seconds). The fabric had been tested (for 20 minutes) and came apart a bit. (68) The fabric had been tested (for 20 minutes) and had come apart (in 20 seconds). The fabric had been tested (for 20 minutes) and had come apart a bit. (69) #The fabric had been tested (for 20 minutes) and torn a stitch. (70) The fabric had been tested (for 20 minutes) and tore a stitch. (71) The fabric had been tested (for 20 minutes) and had torn a stitch. (72) #The fabric had been tested (for 20 minutes) and a bullet torn through it. (73) ?The fabric had been tested (for 20 minutes) and a bullet tore through it. (74) The fabric had been tested (for 20 minutes), and a bullet had torn through it.
PredP and PredP: (Tense+) Aux Sharing
331
This semantic condition is in evidence just where the coordination, conjoining passive and active participial phrases, is thereby compelled to be the complement of perfect have. A coordination of passive participial phrases, taken as complement to be, is exempt, providing in (75) and (76) minimal pairs with (69) and (72):13 (75) The fabric had been tested (for 20 minutes) and torn (through) (a bit) (in 20 seconds). (76) The fabric had been tested (for 20 minutes) and a hole torn through (it(s surface)) (in 20 seconds). The fabric had been tested for 20 minutes and its surface torn (through) (in 20 seconds). The facts argue a strong conclusion: not only must (20) (repeated below) and its analogs above ((62), (63), (66), (69), (72)) have a logical form (21) where the coordination shares a solitary token of the tensed auxiliary verb, but they must also be denied logical forms (22) that repeat the tensed auxiliary verb in the second conjunct. (20) a. The rockers have shimmied and shaken. b. The rockers have shimmied and the rollers shaken. (21) a. The rockers have (shimmied and shaken), i.e., The rockers Present have (shimmied and shaken) b. The rockers Present have (shimmied and the rollers shaken) (22) a. The rockers (have shimmied and have shaken), i.e., The rockers (Present have shimmied and Present have shaken) b. The rockers (have shimmied and the rollers have shaken), i.e., The rockers (Present have shimmied and the rollers Present have shaken) Otherwise, the semantic condition could not be observed, should these sentences share logical forms with their felicitous counterparts ((64), (65), (68), (71), (74)).14 5.1.2 have [[V+en…] and […]]
The semantic condition observed when an active participial phrase is conjoined to an antecedent, passive participial phrase takes in second conjuncts with both pronounced and unpronounced subjects. A semantic condition affecting coordination with an antecedent, active participial phrase provides further argument for tensed auxiliary sharing. But, in this case, it shows up only when the second subject is pronounced and thus joins evidence from sections 3.2 and 5.0 suggesting that more than pronunciation tells apart sentences with pronounced and unpronounced subjects. Tensed auxiliary sharing in (77) and (78) implies some causal or circumstantial connection between the events that the conjuncts describe.15
332
Chapter 5
(77) The wind has battered the window and the rain gotten in. (78) The clouds have parted and the sun come out. The fog has lifted and the sun come out. The darkness has retreated and the sun come out. Failing to grasp what that connection could be makes the like structures in (79) and (80) infelicitous: (79) # The wind has battered the window and the rain gotten in through the chimney. # The wind has battered the window and the doorbell rung. (80) #The waters have parted and the sun come out. #The flood has retreated and the sun come out.16 But neither the coordination of Tensed Phrases in (81) and (82) nor a coordination with unpronounced subjects as in (83) requires such a causal connection—no matter what else may be implicated about the temporal relations between the events described: (81) The wind has battered the window, and the rain has gotten in through the chimney. The wind has battered the window, and the rain got in through the chimney. The wind has battered the window, and the doorbell has rung. The wind has battered the window, and the doorbell rang. (82) The waters have parted, and the sun has come out. The waters have parted, and the sun came out. The flood has retreated, and the sun has come out. The flood has retreated, and the sun came out. (83) The wind has battered the window and rung the doorbell. Pace the effect’s absence when subjects are unpronounced, the contrast between (79)–(80) and (81)–(82) again argues that tensed auxiliary verb sharing has no alternative analysis as a coordination of tensed phrases—and this despite the pronounced nominative17 subject in the second conjunct, which to some will suggest reduction from a tensed clause.18 5.1.3
The present perfect puzzle
McCoard 1978 attributes to Diver 1963 the observation that participial coordination appears to sanction adverbs normally prohibited within a simple present perfect (the
PredP and PredP: (Tense+) Aux Sharing
333
present perfect puzzle19), (84) vs. (86), which also serves to distinguish participial coordination from the coordination of Tensed Phrases, (85) vs. (86). (84) #Hillary has driven a race car on Monday. #Hillary has gone sailing on Wednesday. #Hillary has flown a turboprop on Friday. (85) #Hillary has driven a race car on Monday, has gone sailing on Wednesday, and has flown a turboprop on Friday. (86) Hillary has driven a race car on Monday, gone sailing on Wednesday, and flown a turboprop on Friday. That a coordination of simple present perfects, (85), contrasts in this respect with the present perfect of a coordination of participials, (86), shows that the logical forms of (85) and (86) should not be confounded. Without recourse to a scope effect, the likely difference is that (86) has an analysis that does not conceal repeated tokens of the tensed auxiliary verb within its conjuncts. The present perfect puzzle and the relief from it under participial coordination can be reproduced even where second subjects are pronounced, provided care is taken to respect the semantic condition from section 5.1.2 eliciting a causal or circumstantial connection between the conjuncts: (87) #Plimpton has driven a race car on Monday morning, and has written about it on Monday evening. (88) Plimpton has driven a race car on Monday morning and written about it on Monday evening. (89) #Plimpton has driven a race car on Monday morning, and the sports columns have written about it on Monday evening. (90) Plimpton has driven a race car on Monday morning and the sports columns written about it on Monday evening. (91) #Captain Tyne has piloted the Andrea Gail on Monday, and has sunk (her) on Monday night. (92) Captain Tyne has piloted the Andrea Gail on Monday and sunk (her) on Monday night. (93) #Captain Tyne has piloted the Andrea Gail on Monday, and the perfect storm has sunk her Monday night. (94) Captain Tyne has piloted the Andrea Gail on Monday and the perfect storm sunk her Monday night. In all these cases, the logical form of (Tense-) Aux sharing must be distinct from that of the coordination of Tense Phrases.
6
PredP and PredP: Complementation as a Condition on Subatomic Event Anaphora
Chapter 4 decides the phrasing for sentences such as (1)–(3), how tokens of morphemes, pronounced and unpronounced, are phrased within and without coordinations. One hears in (1) a single auxiliary verb pronounced, and without cause to suppose there is another, another is not heard unpronounced, and thus (1) is not a counterpart to (2), with unpronounced elements. In (3), one hears two past-tense morphemes pronounced, and without cause to imagine otherwise, one takes what has been heard at face value. For all morphemes, both the bound morphemes such as Tense and the freestanding auxiliaries, it turns out that overt tokens of these morphemes are tokened in logical form where they overtly appear to be tokened. (1) a. The rockers did shimmy and shake. b. The rockers have shimmied and shaken. c. The rockers are shimmying and shaking. (2) a. The rockers did shimmy and did shake. b. The rockers have shimmied and have shaken. c. The rockers are shimmying and are shaking. (3) The rockers shimmied and shook. When translating these sentences into Eventish, chapter 4 decides how to coordinate braids, supermonadic structures like (5) for the simple sentence (4), in which event descriptions dovetail in a causal/topological chain.1 (4) The rockers have shimmied. (5) [The X : … Rx …]W[E0,X] Present[E0] Have[E0] O[E0,E1] ∃XW[E1,X] O[E1,E2] ∃XAgent[E2,X] C[E2,E3] shimmied[E3] It has been decided, for example, that (6) should go over into something like (7), and in contrast (8) into (9):2
336
Chapter 6
(6) The rockers will have shimmied and shaken. (7) [The X : … Rx …] W[E0,X] Present[E0] Will[E0] O[E0,E1] ∃XW[E1,X] Have[E1] O[E1,E2] ∃X W[E2,X] O[E2,E3] ∃X Agent[E3,X] C[E3,E4] shimmied[E4] and ∃X W[E2,X] O[E2,E3] ∃X Agent[E3,X] C[E3,E4] shaken[E4] (8) The rockers will have shimmied and will have shaken. (9) [The X : … Rx …] W[E0,X] Present[E0] Will[E0] O[E0,E1] ∃XW[E1,X] Have[E1] O[E1,E2] ∃X W[E2,X] O[E2,E3] ∃X Agent[E3,X] C[E3,E4] shimmied[E4] W[E0,X] Present[E0] Will[E0] O[E0,E1] ∃XW[E1,X] Have[E1] O[E1,E2] ∃X W[E2,X] O[E2,E3] ∃X Agent[E3,X] C[E3,E4] shaken[E4]
and
What (5)–(9) display in earnest are just the sentence’s lexical content, spoken and unspoken, its argument positions and their phrasing. A proper logical form next demands an overlay of event quantification and descriptive pronouns cross-referring to events and holding the sentence together. The argument from chapter 1 (based in part on examples like The Columbia students and the Harvard students surrounded the Pentagon and were crowded into the Mall) concludes that pronouns must be deployed to refer to causes E1 and effects E2, ‘[℩E1 : pro1][℩E2 : pro2]C[E1,E2]’, and likewise to refer to the events related by overlap, ‘[℩E0 : pro0][℩E1 : pro1]O[E0,E1]’. Generalizing this conclusion, the pronoun is taken to be the basic instrument for referring to events described by other phrases. A logical form for (4) thus comes to look like (10): (10) [The X : … Rx …]∃E0W[E0,X] [℩E0 : pro0]Present[E0] [℩E0 : pro0]Have[E0] [℩E0 : pro0][℩E1 : pro1]O[E0,E1] ∃E1∃XW[E1,X] [℩E1 : pro1][℩E2 : pro2]O[E1,E2] ∃E2∃XAgent[E2,X] [℩E2 : pro2][℩E3 : pro3]C[E2,E3] ∃E3shimmied[E3] ‘The rockers participate in the present in having, and there is action, and that participation (in having) exactly overlaps that action, and there is shimmying, and that action causes that shimmying.’ Recall that many an event pronoun is a backward anaphor, following the example of an overt counterpart in A kind word triggering it, Tiff shouted obscenities and Biff swore revenge. Given the structure of a braid, ‘[℩Ej : proj]’ can be counted on to be backward in every instance of ‘[℩Ei : proi][℩Ej : proj]C[Ei,Ej]’ or ‘[℩Ei : proi][℩Ej : proj] O[Ei,Ej]’, since the reference of this pronoun anticipates the complement to the causal/topological relation.
PredP and PredP: Complementation as a Condition
337
In deriving the disjunctive interpretation of (6), where there is a coordination of participial phrases, recall that the interpretation allowing rockers to be divided between shimmiers and shakers requires collective reference to some shimmying and shaking. The logical form for the disjunctive interpretation of (6) parsed as in (11) with a backward event pronoun is shown in (12), where the backward pronoun ‘[℩E2:ΦΨ]’, its content fixed by the conjoined phrases, describes shimmying and shaking (according to (13)), the collective effect of what the rockers did:3 (11) The rockers will have [℩E2:pro2](2shimmied and shaken). The rockers will have[℩E : proΦ,Ψ] (Φshimmied) and (Ψshaken). (12) [The X : … Rx …] ∃E0W[E0,X] [ιE0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] ∃E1∃XW[E1,X] [℩E1:pro1]Have[E1] [℩E1:pro1] [℩E2:ΦΨ]O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4]) and ∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4]) (13) [℩E2:ΦΨ]E2e ↔ [℩E : [∃E2 : Φ]∀e(E2e → Ee) & [∃E2 : Ψ]∀e(E2e → Ee)]Ee
(Appendix 2)
In contrast to (6), where participles conjoin, it was remarked in section 3.4.0 that disjunctive interpretation fails when the coordination is large, as in (8), where tensed auxiliary phrases conjoin: (6) The rockers will have shimmied and shaken. (8) The rockers will have shimmied and will have shaken. The explanation offered was that a second conjunct of a large coordination contains at its left edge a pronoun that refers in the case of (8) to what the rockers did as described by the first conjunct. Then, if the rockers have already been said to shimmy, that ends there a disjunctive interpretation that could divide them between shimmiers and shakers. This explanation presumes without comment that the first conjunct says indeed that they will have shimmied. Yet the presumption and the explanation that rests on it are undermined unless one further constrains the (backward) event pronoun referring to the effects of what the rockers did, the pronoun that in (11) and (12) manages to refer to some shimmying and some shaking. More generally, any attempt to correlate a difference of meaning with the size of coordination will founder on an event pronoun left unconstrained, despite the very real differences of phrasing and lexical content that already tell apart (6) and (8). So suppose one had set out
338
Chapter 6
as in (14)–(15) to overlay event quantifiers on the structure (9) that section 4.0 gives for (8): (14) *The rockers [℩E : proΦ,Ψ] will have (Φshimmied) and will have (Ψshaken). (15) *[The X : … Rx …][℩E2:ΦΨ] ∃E0W[E0,X] [ιE0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] ∃E1∃XW[E1,X] [℩E1:pro1]Have[E1] [℩E1:pro1] O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4]) and ∃E0W[E0,X] [℩E0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] ∃E1∃XW[E1,X] [℩E1:pro1]Have[E1] [℩E1:pro1] O[E1,E2]∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4]) Peering through all the clutter, it remains that the first conjunct asserts that there will be some shimmying and otherwise does not assert the existence of shaking, and likewise, the second conjunct asserts the existence of shaking without asserting shimmying. But what engages the rockers are exactly the events E2 that within each conjunct are said to coincide with what they did. A backward pronoun refers to these events, and, as in (11) and (12), with its antecedents within the coordination, it refers to some shimmying and shaking. Then, according to (15), the first conjunct of (8) says that what the rockers did was some shimmying and shaking, and the second conjunct says the same. That is, (8) is rendered, in effect, as (16), which, far from excluding the disjunctive interpretation, rather says it twice over. (16) The rockers will have shimmied and shaken and will have shimmied and shaken. In (14)–(15), a single token of the offending pronoun occurs outside the coordination. The same anomaly and false reading of (8) will occur if tokens occur within the conjuncts, as in (17): (17) *The rockers will have [℩E : proΦ,Ψ](Φshimmied) and will have [℩E : proΦ,Ψ] (Ψshaken). An event pronoun that manages to take for its antecedents the participles shimmied and shaken threatens to make synonyms of (6) and (8), and (18) too, despite any other differences in their structure: (6) The rockers will have shimmied and shaken. (11) The rockers will have[℩E : proΦ,Ψ] (Φshimmied) and (Ψshaken). (18) The rockers will have shimmied and have shaken.
PredP and PredP: Complementation as a Condition
339
(8) The rockers will have shimmied and will have shaken. (14) *The rockers [℩E : proΦ,Ψ] will have (Φshimmied) and will have (Ψshaken). (17) *The rockers will have [℩E: proΦ,Ψ](Φshimmied) and will have [℩E : proΦ,Ψ] (Ψshaken). Something must be wrong with the use of this pronoun in (14) and (17) if synonymy is to be avoided. To be sure, a tacit event pronoun in its meaning and logical form closely resembles the overt kind, both being definite descriptions, but there is no reason to presume that it will be as free or far-reaching in seeking out its antecedents. After all, these pronouns are subatomic, effective over a limited range, and subject certainly to some grammatical constraints. In (11) the event pronoun sits beside its coordinated antecedents, and it is as if a coordination of participles shimmied and shaken fixes collective reference once and for all on some shimmying and shaking. Intending the same reference in (14) and (17) involves the pronoun in seeking out antecedents deeply embedded within the conjuncts either below (14) or next door (17), a syntax that will turn out to be ungrammatical. If, instead, event pronouns in (6), (18), and (8) are held to the same syntax, collective reference in (6) is indeed to some shimmying and shaking, but we should rather expect in (18) reference to some having shimmied and having shaken—a state distinct from any events of shimmying and shaking—and in (8), to some will-ing have shimmied and will-ing have shaken, whatever that is if anything. The sections that follow spell out such conditions on the syntax of event pronoun anaphora as will correlate the distribution of the disjunctive interpretation with the size of coordination. The causative analysis of transitivity, which a braid such as (10) imposes on the logical form for (4), says that the small clause headed by ‘shimmied’ is a complement so-called to the causative morpheme ‘C’ (and again that the small clause headed by the causative morpheme is a complement to the morpheme expressing the Overlap relation, ‘O’). A descriptive grammar might casually add that the small-clause complement is an “argument” of the causative, which is an unhelpful remark to the extent that it obscures the fact that the causative morpheme’s arguments, ‘E1’ and ‘E2’ in ‘C[E1,E2]’, denote events. A complement clause just happens to provide some further description of the caused events. The traditional notion of complementation has no formal realization as such in a logical syntax that is all conjunction, quantification and anaphora. Yet the traditional notion correctly draws attention to some intimate connection between phrases, as between the causative ‘C’ and its smallclause complements so-called. As all relations between phrases denoting events amount here to pronominal cross-reference, complementation of the causative ‘C’ corresponds to a grammatical condition fixing the content of the pronoun ‘[℩E2 : pro2]’ denoting the caused events. The observed locality conditions on subatomic event pronouns are merely the formal realization—so it can be pleaded—of an already well-entrenched notion, complementation. Despite a logical syntax of conjunction,
340
Chapter 6
some feeling is thus left intact for the notion that the conjoined phrase in (6) and (11)–(12) is a grammatical complement to something higher, in that this phrase and only this phrase fixes the reference to events for the next predicate up. With complementation now realized as a condition on anaphora, a modification of those conditions in section 6.3 will make room for an account of the eccentric distribution of disjunctive interpretations. 6.0
The distribution of the disjunctive interpretation
The disjunctive interpretation, so-called in chapter 1, allows that the 20,000 students of (19) be divided between those who surrounded the Pentagon and those crowded into the Mall: (19) Twenty thousand students surrounded the Pentagon and were crowded into the Mall. So too do coordinated participial phrases, (20)–(22), allow twenty rockers to divide between shimmiers and shakers: (20) Twenty rockers (will) have shimmied (all night long) and shaken (all night long). (21) Twenty rockers were shimmying (all night long) and shaking (all night long). (22) Twenty rockers (will) have been shimmying (all night long) and shaking (all night long). In contrast, the coordination of some larger predicative phrases hinders the disjunctive interpretation.4 Sentences (23)–(25) coordinating tensed auxiliary phrases (section 3.4.0) all entail that twenty rockers shimmy and twenty rockers shake: (23) a. Twenty rockers have shimmied (all night long) and have shaken (all night long). b. Twenty rockers will have shimmied (all night long) and will have shaken (all night long). (24) Twenty rockers were shimmying (all night long) and were shaking (all night long). (25) a. Twenty rockers have been shimmying (all night long) and have been shaking (all night long). b. Twenty rockers will have been shimmying (all night long) and will have been shaking (all night long). To my ear, the intermediate coordinations in (26)–(28) tolerate the disjunctive interpretation, although not as well for the reasons urged below.
PredP and PredP: Complementation as a Condition
341
(26) Twenty rockers will have shimmied and have shaken. (27) Twenty rockers have been shimmying and been shaking. (28) a. Twenty rockers will have been shimmying and been shaking. b. ??Twenty rockers will have been shimmying and have been shaking. The contrasts between (20)–(22) and (23)–(25), between (20)–(22) and (26)–(28), and between (23)–(25) and (26)–(28) tell of differences in logical form. It cannot be, for example, that (20) and (23) are the same except for what is pronounced. It has already been argued in chapter 5 that auxiliary verbs absent from pronunciation are absent from logical form as well. It remains to be seen how exactly that matters for the disjunctive interpretation throughout the above paradigm. Also, whatever is to be said about this variation, it exists alongside the coordination of simple tensed verb phrases, (29), where the disjunctive interpretation is just as good as when participial phrases are coordinated: (29) Twenty rockers shimmied and shook. Something further (section 6.3) will have to be said to explain this unexpected contrast between simple tensed verb phrases and tensed auxiliary phrases. 6.1
The logical syntax of the disjunctive interpretation
Recall that the disjunctive interpretation of (30) rests on three features of its logical form (31): (30) Twenty thousand students surrounded the Pentagon and were crowded into the Mall. (31) ∃Ei[∃X : 20,000(X) & …]W[Ei,X] O(proi,proj) ∃E′j(∃XAgent[E′j,X] C(pro′j,pro″k) ∃E″k (surround[E″k] Theme[E″k,p])) and ∃E′j(∃XTheme[E′j,X] C(pro′j,pro″k) ∃E″k (crowd[E″k] Into[E″k,m])) First, the subject positions within the conjoined phrases must be existentially closed off from twenty thousand students, as the logical form for this interpretation should not entail either that 20,000 students were Agents surrounding the Pentagon or that 20,000 were Themes being crowded into the Mall:5 (32) ∃Ei[∃X : 20,000(X) & …]W[Ei,X] O(proi,proj) ∃E′j(∃XAgent[E′j,X] … surround[E″k] …) and ∃E′j(∃XTheme[E′j,X] … crowd[E″k] …) If within each conjunct several positions were to be associated with the subject, as one might imagine for (33), they must all be existentially closed off from twenty thousand students to derive the disjunctive interpretation, as in (34). Direct anaphora
342
Chapter 6
to the subject from any one of these positions would spoil the vagueness as to who was where and consign all 20,000 students to at least one of the venues.6 (33) Twenty thousand students will have surrounded the Pentagon and have been crowded into the Mall. (34) ∃Ei[∃X : 20,000(X) & …]W[Ei,X] will[Ei] O(proi,proj) ∃E′j(∃XW[E′j,X] … have[E′j] … ∃E″(∃XAgent[E″,X] … surround[E″] …)) and ∃E′j(∃XW[E′j,X] … have[E′j] … ∃E″(∃XW[E″,X] … been[E″] … ∃E‴(∃XTheme[E‴,X] … crowd[E‴] …))) With all this existential closure, it must however then be guaranteed that what was some surrounding the Pentagon and being crowded into the Mall is exactly what 20,000 students did indeed do. Thus, event pronouns, the second crucial feature, are recruited in (31), proj to denote some surrounding and crowding and proi to denote what the students did (W-ing). The content of this last pronoun referring to just what they, the students, and only they did does not itself say anything about surrounding or crowding: (35) ∃Ei[∃X : 20,000(X) & …]W[Ei,X] O(proi,proj) ∃E′j(… surround …) and ∃E′j(… crowd …) Now of course it is not enough to simply token pronouns, one referring to what the students did and another referring to some surrounding and crowding. Something has to be said to relate the two. There must be room enough in the structure to express that relation, represented here as ‘O’ (overlap), the third crucial feature for the disjunctive interpretation: (36) ∃Ei[∃X : 20,000(X) & …]W[Ei,X] O(proi,proj) ∃E′j(… surround …) and ∃E′j(… crowd …) In sum, we may expect that the distribution of the disjunctive interpretation is governed by three considerations: i. The distribution of the existential closure of subject positions ii. The distribution of event pronouns, their content, and any locality conditions that hold between them and their antecedents iii. The projection in phrase structure of the causal/topological relations that host event pronouns 6.2
Deriving the distribution of the disjunctive interpretation
As observed in (8) and (23)–(25), the coordination of tensed auxiliary phrases disrupts the disjunctive interpretation, and it also degrades somewhat when any aux-
PredP and PredP: Complementation as a Condition
343
iliary phrases are conjoined, (26)–(28). Which of the three features of logical form are in play? It is unlikely to be existential closure of an inferior subject position that tells apart (20) and (23) and fails the disjunctive interpretation in the latter. (20) The rockers have shimmied and shaken. (23) The rockers have shimmied and have shaken. Why, after all, should the larger phrasing in (38) disrupt the existential closure of the embedded participles, in contrast to (37)? (37) The rockers have ∃X shimmied and ∃X shaken. (38) *The rockers have ∃X shimmied and have ∃X shaken. Perhaps it could be that existential closure in Tensed Phrases discriminates: (39) [The X : rockers] X have ∃X shimmied and ∃X shaken. (40) *[The X : rockers] ∃X have shimmied and ∃X have shaken. (41) [The X : rockers] X have shimmied and X have shaken. Failing existential closure (40), the subjects of the conjoined Tensed Phrases in (23) must be bound by the rockers as in (41), deriving an interpretation that does not allow the rockers to divide between shimmiers and shakers. But such a broad prohibition against existential closure overextends to block the disjunctive interpretation of the conjoined Tensed Phrases in (42) and (43), which plainly allow some to shimmy while others shake: (42) Twenty rockers shimmied and shook. (43) The rockers shimmied and shook. Consider instead the event pronouns crucial to the disjunctive interpretation, and in particular the phrasal positions of proj referring to some shimmying and shaking and prok referring to what the rockers did: (44) The rockers prok have proj(jshimmied and shaken), or The rockers have prok proj(jshimmied and shaken) In (44), with participles coordinated, the pronoun proj can be adjacent to the phrase that fixes its content, an antecedent describing some shimmying and shaking. There remains elsewhere in the structure, higher up, room enough for another pronoun prok referring to what the rockers do (‘W[E,X]’) and for a relation, e.g., ‘O[Ek,Ej]’from one to the other. Larger coordinations disturb this arrangement one way or another.7 What goes wrong with the event pronouns will rule out the disjunctive interpreta-
344
Chapter 6
tion in the case of a coordination of tensed auxiliary phrases, and get a grip on why the interpretation degrades elsewhere: (45) a. *The rockers have proi,j(ishimmied) and have (jshaken). b. *The rockers have proi(ishimmied) and have proj(jshaken). (46) a. *The rockers proj(jhave shimmied and have shaken). b. *The rockers proi,j have (ishimmied) and have (jshaken). Suppressing in (45) and (46) the pronoun prok referring to what the rockers did, notice that the pronoun meant to refer to some shimmying and shaking if it remains adjacent to one participle phrase as in (45) is now far removed from the other, and it may be stipulated as above that the distant participle embedded within the neighboring conjunct is too remote an antecedent. Failing collective reference, event pronouns in these positions denote either some shimmying or some shaking, falling short of what is required for the disjunctive interpretation. The interpretation will simply have no outlet in a larger coordination unless the event pronoun seeks higher ground outside as in (46). Yet, even from higher ground, the disjunctive interpretation is elusive. In (46b), the antecedents denoting shimmying and shaking are again too distant for the pronoun to refer to some shimmying and shaking. Rather, if the pronoun takes up proximate antecedents, as in (46a), it refers instead to some having shimmied and some having shaken, which, according to supermonadicity, are different events or states from the shimmying and shaking to which they are related at bottom. If this is correct, none of (47) and (26)–(28) share logical forms in representing their disjunctive interpretations. (47) The rockers will have shimmied and shaken. The rockers will have been shimmying and shaking. (26) Twenty rockers will have shimmied and have shaken. (27) Twenty rockers have been shimmying and been shaking. (28) a. Twenty rockers will have been shimmying and been shaking. b. ??Twenty rockers will have been shimmying and have been shaking. Only (47) involves reference to some shimmying and shaking; (26), to some having shimmied and having shaken; (27) and (28a), to some been shimmying and been shaking; and (28b), to some having been shimmying and having been shaking. These may not all be equally successful. To whatever extent reference to such events or states is felicitous, completing a disjunctive interpretation still demands room enough outside the coordination for the pronoun referring to what the rockers did and for the relation bridging what they did to whatever events the coordination denotes. In (47) and (26)–(28), there is at least a tensed auxiliary verb outside and whatever higher structure comes with it. Some details are provided below, but the structural implications and the argumentative point should be plain. The disjunctive interpretation rests on a canonical configuration relating a collective event pronoun
PredP and PredP: Complementation as a Condition
345
to its phrasal antecedents as in (44). That configuration proves to discriminate between participial coordination and the coordination of larger predicative phrases, correlating with the failure of coordinated tensed auxiliary phrases to support a disjunctive interpretation and its gradient acceptability among somewhat smaller phrases. The logical form for the disjunctive interpretation of (6) parsed as in (11) is shown in (12), where pro2 in (11) corresponds to ‘[℩E2 : ΦΨ]’, the descriptive content being fixed by the conjoined clauses and denoting what shimmying and shaking there was (see (13)) . (6) The rockers will have shimmied and shaken. (11) The rockers will have [℩E2:pro2] (2shimmied and shaken). The rockers will have[℩E : proΦ,Ψ] (Φshimmied) and (Ψshaken). (12) [The X : … Rx …] ∃E0W[E0,X] [℩E0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] ∃E1∃XW[E1,X] [℩E1:pro1]Have[E1] [℩E1:pro1] [℩E2:ΦΨ]O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][ιE4:pro4]C[E3,E4] ∃E4shimmied[E4]) and ∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4]) The logical form (12) must be distinguished from candidates for representing an equivalent disjunctive interpretation of the somewhat larger coordination in (18). (18) The rockers will have shimmied and have shaken. In none of (48)–(55) does a grammatical use of the event pronoun refer to some shimmying and shaking. In (48), it should be that if both conjuncts are to serve as antecedents, the pronoun cannot reach from a position properly contained within one conjunct to pluck an antecedent from within the other— *… (… [℩Ej : Φ Ψ]O(Ei, Ej) Φ) and (…[℩Ej : Φ Ψ]O(Ei, Ej) Ψ). (48) *The rockers will have [℩E : proΦ,Ψ](Φshimmied) and have [℩E : proΦ,Ψ](Ψshaken). (49) *[The X : … Rx …] ∃E0W[E0,X] [℩E0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] (∃E1∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1][℩E2:ΦΨ]O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (∃E1∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1][℩E2:ΦΨ]O[E1,E2] ∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4]))
346
Chapter 6
In contrast, both tokens of the event pronoun in (50) find their antecedents within their own clauses, conforming to this constraint. But with the first token referring to some shimmying and the second to some shaking, no event pronoun in (50) refers to some shimmying and shaking, as the disjunctive interpretation requires: (50) *The rockers will have[℩E : proΦ](Φshimmied) and have [℩E : proΨ](Ψshaken). (51) *[The X : … Rx …] ∃E0W[E0,X] [℩E0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] (∃E1∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1][℩E2: Φ]O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (∃E1∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1][℩E2: Ψ]O[E1,E2] ∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4])) Even where the pronoun is outside the coordination and includes its antecedents within its scope, further constraint appears to keep it from seeking antecedents that are embedded within the conjuncts. Thus, the coordination of the larger auxiliary phrases in (18) deprives the pronoun of smaller participial antecedents and closes off yet another route to the disjunctive interpretation: (52) *The rockers will [℩E : proΦ,Ψ] have (Φshimmied) and have (Ψshaken). (53) *[The X : … Rx …] ∃E0W[E0,X] [℩E0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] [℩E2:ΦΨ] (∃E1∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1] O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (∃E1∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1] O[E1,E2] ∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4])) In (54), the event pronoun conforms to all the locality conditions that in section 6.1 are taken to characterize complementation, and it ends up referring to some having shimmied and having shaken: (54) The rockers will [℩E : proΦ′,Ψ′] (Φ′ have shimmied) and (Ψ′ have shaken).
PredP and PredP: Complementation as a Condition
347
(55) [The X : … Rx …] ∃E0W[E0,X][℩E0:pro0]Present[E0][℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:Φ′Ψ′]O[E0,E1] (∃E1(Φ’ ∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1] [℩E2:pro2]O[E1,E2] ∃E2∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (∃E1(Ψ’ ∃XW[E1,X] [℩E1:pro1]Have[E1][℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2 ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4])) These are not the same events as the shimmying and shaking referred to in (6): (6) The rockers will have shimmied and shaken. (18) The rockers will have shimmied and have shaken. If imagining how the participation of some rockers is divided among states of having shimmied and having shaken is somewhat harder than seeing them arranged across discrete actions of shimmying and shaking, it is just as well and suffices to underwrite the awkwardness of (18) and (26)–(28) noted earlier. What is important for distinguishing (18) from (6) is that complementation refuse it an event pronoun that refers to some shimmying and shaking, lest it masquerade as (6) and be mistaken for saying the very same thing. With an obligatory difference of interpretation, there can be room for a difference in judgment. Although degraded and more awkward than in (6), the disjunctive interpretation of (18) shows that the fault in (25), with an even larger coordination of tensed auxiliary phrases, cannot merely be reference to some having shimmied and having shaken:8 (25) The rockers have shimmied and have shaken. Of course, successful reference to some having shimmied and having shaken is not sufficient to derive the disjunctive interpretation. In addition, recall that the phrase structure must project a causal/topological relation that relates the having shimmied and having shaken to what the rockers do (‘W’). As shown in (55), the logical form for (18), the having shimmied and having shaken, is related by ‘O[E0,E1]’ to events E0 described by phrasing outside the coordination, as must be the case if the coordination itself is entirely taken up describing the events E1. The logical form for (18) differs from (57), the one under consideration for (25), only in the location of the tensed auxiliary verb. Surely (57) expresses a disjunctive interpretation as well as (55) does. It must rather be that (57) is ungrammatical, which deprives (25) of disjunctive interpretation altogether.
348
Chapter 6
(56) *The rockers [℩E : proΦ′,Ψ′] (Φ′ have shimmied) and (Ψ′ have shaken). (57) *[The X : … Rx …] ∃E0W[E0,X][℩E0:pro0] [℩E1:Φ′Ψ′]O[E0, E1] (∃E1(Φ′ ∃XW[E1,X][℩E1:pro1]Present[E1][℩E1:pro1]Have[E1] [℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (∃E1(Ψ′ ∃XW[E1,X][℩E1:pro1]Present[E1] [℩E1:pro1]Have[E1] [℩E1:pro1][℩E2:pro2]O[E1,E2] ∃E2 ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4])) The confinement of the tensed auxiliary verb within the conjuncts appears to matter. There is after all no reason to expect that the projection of causal/topological relations, such as O, expressed by bound verbal morphology iterates unbounded (or that the projections of W are unbounded). If the entire tensed auxiliary phrases are taken up into the description of the pronoun, as they are in (57), referring to events of present having shimmied and present having shaken, the sentence structure has been entirely consumed, without room remaining to relate these events to anything else. The second line of (57) oversteps what can be projected.9 In sum, the event pronoun under consideration refers, as complementation (section 6.1) requires, to some shimmying and shaking in (20) and (47) (repeated below). The structure outside the coordination is rich enough to express a relation between these events and what the rockers did. (20) Twenty rockers have shimmied and shaken. Twenty rockers have [℩E : pro](shimmied and shaken). (47) Twenty rockers will have shimmied and shaken. Twenty rockers will have [℩E : pro](shimmied and shaken). (26) Twenty rockers will have shimmied and have shaken. Twenty rockers will [℩E : pro](have shimmied and have shaken). (25) Twenty rockers have shimmied and have shaken. *Twenty rockers [℩E : pro](have shimmied and have shaken). (23) Twenty rockers will have shimmied and will have shaken. *Twenty rockers [℩E : pro](will have shimmied and will have shaken). In (26), complementation requires instead reference to some having shimmied and having shaken, but there is yet room enough outside the coordination to relate these events to the rockers. In (25) and (23), in contrast, the disjunctive interpretation fails altogether. To satisfy complementation, the event pronoun in (25) refers to some having shimmied and having shaken and in (23) to some going to have shimmied
PredP and PredP: Complementation as a Condition
349
and going to have shaken, and the structures have been exhausted describing these events. Nothing is left to relate these events to the rockers, all the candidate relations having been caught well inside the conjuncts. 6.3
Coordinating simple tensed verbs
A coordination of simple tensed verb phrases like (18), which readily entertains a disjunctive interpretation, undermines the above account to the extent that (18) better resembles the coordination of tensed auxiliary phrases in (25) than (20)’s coordination of participial phrases. (18) Twenty rockers shimmied and shook. (25) Twenty rockers have shimmied and have shaken. (20) Twenty rockers have shimmied and shaken. A tempting response declares that (18) disguises a logical form shared with (58), which falls in with (20) and derives the disjunctive interpretation as expected: (58) Twenty rockers did shimmy and shake. A similar logical form—so the response continues—would not inadvertently be available to rescue a disjunctive interpretation for (25), since the auxiliary verb displaces the overt expression of the target structure (59) and presumably its tacit counterpart as well. (59) *Twenty rockers do have shimmied and have shaken. (*Twenty rockers do have shimmied.) The contrast between (58) and (59) appeals to the difference between auxiliary verbs and main verbs, which must be acknowledged even if no analysis yet does much more than stipulate it. I have no principled objection to this response, and the alternative to be offered does not escape from stipulating some difference between auxiliary and main verbs, having already stipulated (sections 3.2–3.5) that a difference in size between tensed auxiliary phrases and tensed main verbs is the basis for their differences under coordination. But some empirical considerations introduced below argue against the particular misstep of confounding (18) and (58), observing that (18) and the like contain two independent tenses, contrary to what is in (58). There are also the general considerations from section 4.2 that coordination is opaque to bound morphology, precluding the assimilation of (18) to (58). Besides the empirical and analytic concerns, I would like to show that the logical syntax recommended here thrives even under the harsh conditions of a naive view where pronounced tokens of verbal morphemes, both freestanding verbs and bound morphemes such as Tense, are
350
Chapter 6
tokened in logical form (with their full semantic force) exactly where they appear to be tokened overtly. Such a view compels the disjunctive interpretation of (18) to be derived despite the presence of coordinated Tense Phrases. On the other hand, if one abandons the naive view and puts to rest the empirical concerns, an easy way out always beckons treating (18) as (58). Recall prior differences between tensed auxiliary phrases and simple tensed verb phrases. A quantifier may itself escape from a coordination of the former ((150), (152), (154)), but its adverbialization remains confined within the conjunct, (177) vs. (60).10 (150) ?No child is watching Sesame Street and is looking unhappy. (152) ?No child has watched Sesame Street and has gone to bed unhappy. (154) ?No student is reading the text and is taking notes on what she reads. (177) [No x : NPi]
([℩E : NPi[E,x]] [∃E0 : N[E,E0]](Φ i … Tense[E0] …) and [℩E0 : Φi] [∃E1 : N[E0,E1]](… Tense[E1] …))
(60) *[No x : NPi] [℩E : NPi[E,x]][∃E0 : N[E,E0]] ((… Tense[E0] …) and (… Tense[E0] …)) Absent the adverbialization of a pronounced subject in the second conjunct ((176)), the event variable exposed can only be covered by descriptive anaphora to the events antecedently described by the first conjunct as in (177). (176) *[No x : NPi] ([℩E : NPi[E,x]][∃E0 : N[E,E0]](… Tense[E0] …) and [℩E : NPi[E,x]][∃E0 : N[E,E0]](… Tense[E0] …)) (177) [No x : NPi]
([℩E : NPi[E,x]] [∃E0 : N[E,E0]](Φ i … Tense[E0] …) and [℩E0 : Φi] [∃E1 : N[E0,E1]](… Tense[E1] …))
Some (causal) connection is thus understood to hold between the events described by the first conjunct and those to which the second conjunct applies. Failing to see a connection is infelicitous, with the fragile conditions making (150)–(154) possible being broken in (124)–(128) (see also chapter 3, note 68 and (145)–(146)): (124) *No rocker was shimmying and was shaking (to that funky disco music). (126) *No rocker has shimmied and has shaken (to that funky disco music). (128) *No student of mine is studying syntax and is publishing semantics. *Not any student of mine is studying syntax and is publishing semantics. Characteristic of descriptive anaphora, recall also that an error of modal subordination, contrasting (159)–(163) with (150)–(154), may thwart the event pronoun, an observation confirming its presence and interpretation:
PredP and PredP: Complementation as a Condition
351
(159) *No child is ever watching Sesame Street and is looking unhappy. (161) *No child has ever watched Sesame Street and has gone to bed unhappy. (163) *No student is ever reading the text and is taking notes on what she reads. Simple tensed verb phrases differ, it was argued, in allowing adverbialization to follow its quantifier outside a coordination: (181) [No x : NPi][ ℩E : NPi[E,x]][∃E0 : N[E,E0]] ((W[E0,x] … Tense[Ei] …) and (W[E0,x] … Tense[Ei] …)) Event reference in the second conjunct thus needs to be no more dependent on the first conjunct than the first conjunct’s event reference needs to be on the second. Both conjuncts may amount to independent remarks on what the subject is up to and so escape the conditions that hamstring coordination when reference to events in the second conjunct must be to those described by the first: (120) No rocker shimmied and shook (to that funky disco music). (121) No student of mine studies syntax and publishes semantics. Not any student of mine studies syntax and publishes semantics. (61) (62) (63) (64)
No child ever watched Sesame Street and went to bed unhappy. No child ever watches Sesame Street and goes to bed unhappy. No student ever read the text and took notes on what she read. No student ever reads the text and takes notes on what she reads.
Juxtaposing their logical forms below resumes the difference. The very large coordination of tensed auxiliary verb phrases in (177) allows only the pronounced subject DP to be raised outside the coordination. The lesser coordination of simple tensed verb phrases in (181) also allows the escape of the articulating adverbial phrase, itself involving a definite description of events: (177) [No x : NPi]
([℩E : NPi[E,x]][∃E0 : N[E,E0]](Φ i … Tense[E0] …) and [℩E0 : Φi] [∃E1 : N[E0,E1]](… Tense[E1] …)) E.g.,*No rocker was shimmying and was shaking (to that funky disco music).
(181) [No x : NPi][℩E : NPi[E,x]][∃E0 : N[E,E0]] ((W[E0,x] … Tense[Ei] …) and (W[E0,x] … Tense[Ei] …)) E.g., No rocker shimmied and shook (to that funky disco music). It remains now to discover how such a difference of logical form should matter enough for a coordination of simple tensed verb phrases to rescue the disjunctive interpretation, which sections 3.4.0 and 6.2 have put beyond the reach of tensed auxiliary verb phrases.
352
Chapter 6
The logical form for (18)—a coordination of simple tensed verb phrases—is (65), displaying both its adverbialization and internal structure: (65) [The X : ∃E rocker[E,X]][℩E : rocker[E,X]] ([∃E0 : N[E,E0]]W[E0,X] Past[E0][℩E0 : pro0] [℩E1 : pro1]O[E0,E1] ∃E1∃YAgent[E1,Y][℩E1 : pro1][℩E2 : pro2]C[E1,E2]∃E2shimmy[E2]) and ([∃E0 : N[E,E0]]W[E0,X] Past[E0] [℩E0 : pro0] [℩E1 : pro1]O[E0,E1] ∃E1∃YAgent[E1,Y][℩E1 : pro1][℩E2 : pro2]C[E1,E2]∃E2shake[E2]) As with the coordinated tensed auxiliary phrases in (25) and (23), a pronoun based outside the coordination in (65) denoting the events E1 the coordination describes would fail to find room to express a relation between these events and what the rockers did. Rather, what a coordination of simple tensed verb phrases licenses is, in effect, clitic climbing, the movement of a lower event pronoun to a position outside the coordination. So much has already been admitted for the sake of the adverbialized term ‘[℩E : rocker[E,X]],’ which moves outside along with the subject when simple tensed verb phrases are coordinated. The difference between (18) and the tensed auxiliary verb phrase in (25) is that the latter is opaque to such clitic climbing in a pattern attested among overt clitics in Piedmontese such as the locative clitic y/ye ‘there’ (Burzio 1986; Kayne 1993; Manzini and Savoia 2005; Tortora 2010):11 Piedmontese (Burzio 1986, 122–126) (66) i client a y rivu. the clients subj.cl there arrive ‘The clients arrive there.’ (67) i client a sun rivaye. the clients subj.cl are arrived.there ‘The clients have arrived there.’ (68) *i client a y sun riva. the clients subj.cl there are arrived The clitic climbing that derives the disjunctive interpretation for (18) is both across the board and collectivizing (see the discussion of collectivized Right-Node Raising in section 2.1): (69) [The X : ∃E rocker[E,X]][℩E : rocker[E,X]][∃E0 : N[E,E0]][℩E1 : ΦΨ] ((∃E0W[E0,X] Past[E0][℩E0 : pro0][℩E 1 : pro1]O[E0,E1] (Φ∃E1∃YAgent[E1,Y][℩E1 : pro1][℩E2 : pro2]C[E1,E2] ∃E2shimmy[E2])) and (∃E0W[E0,X] Past[E0] [℩E0 : pro0] [℩E 1 : pro1]O[E0,E1] (Ψ∃E1∃YAgent[E1,Y][℩E1 : pro1][℩E2 : pro2]C[E1,E2] ∃E2shake[E2])))
PredP and PredP: Complementation as a Condition
353
The opacity of the auxiliary verb have rules out parallel efforts to clitic-climb in (25): (70) *[The X : ∃E rocker[E,X]][℩E : rocker[E,X]][∃E0 : N[E,E0]][℩E2 : ΦΨ] ((∃E0W[E0,X] Present[E0][℩E0 : pro0] [℩E1 : pro1]O[E0,E1] ∃E1Have[E1][℩E1 : pro1][℩E 2 : pro2]O[E1,E2] (Φ∃E2∃YAgent[E2,Y][℩E2 : pro2][℩E3 : pro3]C[E2,E3] ∃E3shimmy[E3])) and (∃E0W[E0,X] Present[E0][℩E0 : pro0][℩E1 : pro1] O[E0,E1] ∃E1Have[E1][℩E1 : pro1][℩E2 : pro2]O[E1,E2] (Ψ∃E2∃YAgent[E2,Y][℩E2 : pro2][℩E3 : pro3]C[E2,E3] ∃E3shake[E3]))) (71) *[The X : ∃E rocker[E,X]][℩E : rocker[E,X]][∃E0 : N[E,E0]][℩E1 : ΦΨ] ((∃E0W[E0,X] Present[E0][℩E0 : pro0] [℩E1 : pro1]O[E0,E1] (Φ∃E1Have[E1][℩E1 : pro1][℩E2 : pro2]O[E1,E2] ∃E2∃YAgent[E2,Y][℩E2 : pro2][℩E 3 : pro3]C[E2,E3] ∃E3shimmy[E3])) and (∃E0W[E0,X] Present[E0][℩E0 : pro0] [℩E1 : pro1]O[E0,E1] (Ψ∃E1Have[E1][℩E1 : pro1][℩E2 : pro2]O[E1,E2] ∃E2∃YAgent[E2,Y][℩E2 : pro2][℩E3 : pro3]C[E2,E3] ∃E3shake[E3]))) Turning now to cases that mix coordination of a simple tensed verb with a tensed auxiliary verb, recall that the disjunctive interpretation was introduced by such an example, (19), and (72) also seems to support it: (19) Twenty thousand students surrounded the Pentagon and were crowded into the Mall. (72) Twenty thousand students surrounded the Pentagon and were crowding into the Mall. The copula is no different from have in that (73) and (74) show the expected contrast, with a disjunctive interpretation available to (73) and not to (74): (73) a. Twenty thousand students were surrounded at the Pentagon and crowded into the Mall. b. Twenty thousand students were surrounding the Pentagon and crowding into the Mall. (74) a. Twenty thousand students were surrounded at the Pentagon and were crowded into the Mall. b. Twenty thousand students were surrounding the Pentagon and were crowding into the Mall.
354
Chapter 6
It appears that a clitic-climbing context in the first conjunct of (19) is sufficient license for extraction from the coordination, an effect that is asymmetric insofar as (75) does not, to my ear, tolerate a disjunctive interpretation: (75) Twenty thousand students were crowded into the Mall and surrounded the Pentagon. Asymmetric extraction via both A- and A′-movements is not without precedent (see Johnson ([1996] 2003, 2002, and the references cited there), to which I return below.12 The Piedmontese gambit, my appeal to the syntax of clitic climbing, is urged only on the failure of (18) to assimilate to (58), in the event that (18) turns out to contain two semantically effective expressions of Tense: (18) Twenty rockers shimmied and shook. (58) Twenty rockers did shimmy and shake. Yet we have already seen in section 3.3, following Tomioka’s (1993) discussion of (407), that even where there is an overt invitation to share a single tense morpheme, the only one pronounced, there are rather two, each semantically effective in its own clause. (407) Kinoo Chris-ga kaeri, asita Pat-ga kaer-u. Yesterday Chris.nom return.∅ tomorrow Pat.nom return.impf ‘Chris returned yesterday, and Pat will return tomorrow.’ (Tomioka 1993) Given the opacity of coordination (section 4.2) to bound morphemes, Japanese speakers know a priori to dismiss the possibility that conjuncts could be sharing tense. In English, the overt morphosyntax tells against sharing from the start, and there is some evidence in the next section that logical form is no departure, or at least, that nothing is to be gained on the meaning of (18) and its kind by pretending that the tense morphemes are any fewer than they appear to be. 6.3.0
Mixing tenses in coordinated PredPs
The allure of reducing (18) to (58) would quickly fade if (18)’s conjuncts were not in the same tense, the simple past.13 Yet examples of the disjunctive interpretation of conjoined phrases in different tenses, sure evidence of the tenses’ independence, are not easy to come by, as witnessed by (76) and (77), which, if acceptable, mean only that twenty rockers shimmy and they also shake, disallowing their division into shimmiers and shakers: (18) Twenty rockers shimmied and shook. (58) Twenty rockers did shimmy and shake.
PredP and PredP: Complementation as a Condition
355
(76) Twenty rockers shimmy (now) and shook (5 minutes ago). (77) Twenty rockers shimmied (5 minutes ago) and shake (now). Such examples can, however, be composed, along with a suggestion for why they are rare. If so, it secures a crucial formal point: not all (if any) structures like (18) reduce to (58) by across-the-board movement of Tense. Thus a direct argument joins the more oblique evidence from Japanese for the same conclusion. The examples mixing tenses will also at the same time instantiate the disjunctive interpretation. Now without other recourse, the disjunctive interpretation can only be obtained from clitic climbing according to the Piedmontese gambit. A felicitous example mixing tenses is (78), where the lifetimes (see section 10.0) of the 2500 businesses themselves compile reference to both past and present: (78) 2500 diverse businesses spanning a century broke ground for the Chrysler Building in 1928, kept it in high style for seven decades, and today manage an icon of the city welcoming millions of visitors every year. None of the conjuncts alone can truthfully be attributed to the 2500 businesses, as the 2500 did not all exist at the relevant moment, and many of them did not in any case provide the services described. Nowadays cement mixers do not welcome visitors to the Chrysler Building, and so it cannot be said of the construction companies among the 2500 businesses that they participate in its current management. The coordinated predicates are thus interpreted disjunctively, describing a collective action. Similarly, tenses can be felicitously mixed in (79), where the point of the discourse is to show the history of a family and its rise to power through the lifetimes of four of its scions: (79) It has been a busy century for the family. Four Rockefellers (between them) founded Standard Oil, governed New York State, and chair the Council on Foreign Relations.14 Apparently, the collection of disparate lifetimes the subject refers to centers and scales up a historical perspective wide enough and coherent enough to arrange both past and present events. The logical forms for (78) and (79) provide only a single token of the perspectival relation ‘W[E0,X]’ to relate the events described by the coordinated predicates:15 (80) [∃X : 2500(X) ∃E diverse business[E,X]][ ℩E : diverse business[E,X]] [∃E0 : N[E,E0]]W[E0,X] … (81) [∃X : 4(X) ∃E Rockefeller[E,X]][℩E : Rockefeller[E,X]] [∃E0 : N[E,E0]]W[E0,X] …
356
Chapter 6
Recalling section 2.5 and anticipating sections 10.0.0, 10.0.8, and 14.2, it is no surprise that ‘W[E0,X]’ goes beyond the expression of mere participation in some arbitrary events. Rather it also imposes a perspective on that participation. Given any perspective, some events will conform to its topography and others will not, it being an empirical question to discover the exact contours of perspective. The speaker will not, however, entertain the conditions for placing these events under an appropriate perspective unless there is an event pronoun referring to them all, which requires clitic climbing outside the coordination. As with conjuncts of like tense, an intrusion of auxiliary verbs disrupts the disjunctive interpretation in the mixed cases too: (82) 200 diverse businesses spanning seven decades kept the Chrysler Building in high style for half a century and today manage an icon of the city welcoming millions of visitors every year. (83) #200 diverse businesses spanning seven decades have kept the Chrysler Building in high style for half a century and today manage an icon of the city welcoming millions of visitors every year. (84) #200 diverse businesses spanning seven decades were keeping the Chrysler Building in high style for half a century and today manage an icon of the city welcoming millions of visitors every year. (85) #200 diverse businesses spanning seven decades have been keeping the Chrysler Building in high style for half a century and today manage an icon of the city welcoming millions of visitors every year. (86) #200 diverse businesses spanning seven decades kept the Chrysler Building in high style for half a century and today are managing an icon of the city welcoming millions of visitors every year. (87) #200 diverse businesses spanning seven decades kept the Chrysler Building in high style for half a century and today have been managing an icon of the city welcoming millions of visitors every year. (88) #200 diverse businesses spanning seven decades have kept the Chrysler Building in high style for half a century and today are managing an icon of the city welcoming millions of visitors every year. Sentences (83)–(88) compel the 200 businesses to both maintain the Chrysler Building for a half century and participate in its current management. Similarly, (89)–(93) imply falsely that four Rockefellers founded Standard Oil, these same governed New York State, and they also currently chair the Council on Foreign Relations: (89) #It has been a busy century for the family. Four Rockefellers (between them) have founded Standard Oil, (have) governed New York State, and chair the Council on Foreign Relations.
PredP and PredP: Complementation as a Condition
357
(90) #It has been a busy century for the family. Four Rockefellers (between them) (have) founded Standard Oil, were governing New York State for three terms, and (now) chair the Council on Foreign Relations. (91) #It has been a busy century for the family. Four Rockefellers (between them) founded Standard Oil, governed New York State, and are (now) chairing the Council on Foreign Relations. (92) #It has been a busy century for the family. Four Rockefellers (between them) have founded Standard Oil, (have) governed New York State, and are (now) chairing the Council on Foreign Relations. (93) #It has been a busy century for the family. Four Rockefellers (between them) have founded Standard Oil, were governing New York State for three terms, and are (now) chairing the Council on Foreign Relations. Not only does the explicitly historical perspective of several lifetimes offer a coherent presentation of past and present events, but so may a spatial presentation of the subject itself if the current state under observation directly reflects the effects of the past events described. If, for example, the police stumble on a cache of assorted artwork, (94) can report that the 250 pieces uncovered were forgeries copied from Old Masters or reproducing modern photographs: (94) 250 forgeries copied Old Masters and reproduce modern photographs. In contrast, (95) does not felicitously divide the talents of the forgers. It commits the 250 to both having copied Old Masters and now reproducing modern photographs (see (76)–(77)). Of course, there is no obvious presentation of the forgers that shows in their faces the effects of their forgeries: (95) #250 forgers copied Old Masters and (now) reproduce modern photographs. On similar grounds, (96) and (97) contrast. The first sentence need not imply that there were 613 asteroids. Presumably there were few impacts, and the 613 geological formations are divided among impact sites and the fossilized remains of the asteroids’ long-term and far-flung effects on climate. In contrast, (97) puts 613 geologists at the impact sites and these same surveying the sedimentary strata: (96) 613 geological formations recorded Cretaceous asteroid impacts and trace out their effects on climatic conditions in later sedimentary strata. (97) #613 geologists recorded Cretaceous asteroid impacts and trace out their effects on climatic conditions in later sedimentary strata.
358
Chapter 6
Again, the intrusion of auxiliary verbs deprives (94) and (96) of their disjunctive interpretations: (98) #250 forgeries have copied Old Masters and reproduce modern photographs. (99) #613 geological formations have recorded Cretaceous asteroid impacts and trace out their effects on climatic conditions in later sedimentary strata. (100) #613 geological formations recorded Cretaceous asteroid impacts and are tracing out their effects on climatic conditions in later sedimentary strata. (101) #613 geological formations have recorded Cretaceous asteroid impacts and are tracing out their effects on climatic conditions in later sedimentary strata. To summarize these observations, the reason the coordination of mixed tenses under a disjunctive interpretation fails in (76), (77), (95), and (97) and is elsewhere rare, even without the intrusion of auxiliary verbs, is from the start that ‘W[E,X]’ is not a relation without substantive content, a bleached expression of participation. In the logical form for the disjunctive interpretation, a single term, an event pronoun gathers up and refers to all the events described by the conjoined predicates and a single token of ‘W[E,X]’ views them under the same perspective. It is not often that we come across past and present events on the same plane or perspectival field, as it were. The more imminent perspectives of everyday experience are those that are themselves fixed in a particular moment and place surveying proximate events in a local spatiotemporal neighborhood defined with respect to the standard metric. So, for (76) and (77), imagine presentations of twenty rockers under ordinary conditions of observation. Such a presentation could well be of some present shimmying, or translating into the past this perspective (along with the imaginary observer and her conditions of observation), it could be of some past shaking. It is, however, hard to imagine a presentation of the rockers as we might ordinarily find them that is also a scene of both present-shimmying and past-shaking. When mixed tenses succeed in conveying the disjunctive interpretation, as in (78), (79), (94), and (96), we manage to solve for ‘W[E,X]’, locating a single, unrealistic perspective that escapes the ordinary to take in distant events. Sections 2.5, 10.0.0, 10.0.8, and 14.2 show ‘W[E,X]’ to express more than mere participation, importing perspective to distinguish (sub)events from the same temporal neighborhood. It can hardly be surprising that it also has something to say that constrains events that are temporally distant. Meeting those constraints is what makes mixing tenses a hardship. If so, those cases that thrive nevertheless are robust instances where the tenses are indeed independent. Without recourse to an across-the-board movement of Tense, only clitic climbing and the Piedmontese gambit are left to distinguish simple tensed verbs from auxiliaries in representing the disjunctive interpretation.
7
PredP and PredP: Conclusion
Eventish is the logical language with supermonadicity, descriptive event anaphora, and adverbialization among its defining characteristics. Let me first cull from chapters 3–6 some of the evidence for the logical language itself, prior to some remarks on the grammar for translation into it. With and always the univocal sentential connective, any illusions to the contrary are an effect of plural event pronouns cloaked in silence (chapter 2). Deriving either the collective interpretation of (1)–(2) or the disjunctive interpretation of (4) engages two crucial event pronouns: (1) On the first day of spring in Central Park, a lover and a belovèd embraced behind Cleopatra’s Needle. (2) On the first day of spring in Central Park, a lover cautiously and a belovèd tentatively embraced behind Cleopatra’s Needle. (3) [[A lover] Wi (cautiously) Δ∂] and [[a belovèd] Wj (tentatively) [∂[℩Eij: proij][℩Ek: prok]O[Eij, Ek] [embracedk …]]] (4) On the first day of spring in Central Park, the many lovers scattered in that earthly paradise embraced under its canopies and entwined in its bowers. (5) [[The … lovers …] Wi … [℩Ei: proi][℩Ejk: projk]O[Ej, Ejk] [[embracedj …] and [entwinedk …]]] For (1)–(2), one event pronoun, [℩Eij: proij] in (3), refers to what the lover and the belovèd do and the other pronoun [℩Ek: prok], to the embrace. These are brought together only with the intervention of some relation, ‘O[Eij, Ek]’ in (3). Similarly for the disjunctive interpretation of (4), one event pronoun in (5) refers to what the lovers do, another, ‘[℩Ejk: projk]’, to some embraces and entwinements, and the former are scattered among the latter when a relation says they coincide. To assert that and is always ‘&’ and to choose Eventish for the logical language, without commitment to any further particulars, comes with two corollaries important for this chapter. First, in contrast to a lexical ambiguity shoehorned into and, the disjunctive interpretation here needs room outside the coordination so interpreted to store all its apparatus. There needs to be room for the event pronoun the content of which the
360
Chapter 7
coordination fixes and room to express a relation between the events the pronoun refers to and events described elsewhere in the sentence. And why not? It should be heavy lifting to take and and turn it around to mean its dual. So it is that the disjunctive interpretation for which there is room enough in (6) does not fit into (7): (6) 613 geologists will be recording Cretaceous asteroid impacts and tracing out their effects. (7) 613 geologists will be recording Cretaceous asteroid impacts and will be tracing out their effects. When the disjunctive interpretation does fit, as in (6), it is achieved asserting that what the geologists do W-ing (cf. (5)) coincides with some recording and tracing. If ‘W’ itself proves to have a content that demands what they do to fit under the same perspective, then as the recording and the tracing coincide with what they do, they too must fit under the same perspective. As seen in the last section, that further content to ‘W’ emerges as a semantic condition distinguishing (8) from (9) that holds only of the disjunctive interpretation: (8) 613 geological formations recorded Cretaceous asteroid impacts and trace out their effects on climatic conditions in later sedimentary strata. (9) #613 geologists recorded Cretaceous asteroid impacts and trace out their effects on climatic conditions in later sedimentary strata. The Eventish syntax for this interpretation compels a fit under the same perspective. The conjunctive interpretation, on the other hand, requiring no apparatus outside the coordination and twice-tokening ‘W’ within, affords to each conjunct its own perspective, localized to the time under description—“613 geologists W recorded Cretaceous asteroid impacts and pro W trace out their effect on climatic conditions.” Thus, (9) is felicitous if its meaning is that 613 geologists recorded earlier Cretaceous asteroid impacts and the 613 now trace out their effects. The disjunctive meaning is not such a dual to the conjunctive after all, as it relies on this ‘W’ sprinkled all about that means more than simple participation or membership in a participating party. If, again, the disjunctive interpretation merely reflected an ambiguity in and, any deviation in the distribution of the two interpretations is unexpected. Unexpected it is that the semidistributive of the conjunctive but not the semidistributive of the disjunctive is a meaning for (10) (section 4.2.0): (10) Few players huddled at one end of the field and crowded around their captain at the other. Rare is the player, so the sentence says, who has huddled with other players at one end of the field and crowded with others around the captain at the other end. It cannot say that rare is the player who has participated in the one or the other. Rare
PredP and PredP: Conclusion
361
is the player who has done both, and not rare is the player who has done either— despite their division in (11), thanks to and, between those who did the one and those who did the other: (11) The players huddled at one end of the field and crowded around their captain at the other. Yet if huddled at one end of the field and crowded around their captain at the other, as it occurs in (11), is no different in its logical type from the conjuncts separately or their coordination with the conjunctive meaning, the disjunctive meaning should not then have failed in (10). The second corollary for Eventish comes with its commitment to plural event pronouns as the bearers of collective reference, even for the general case of collectivized Right-Node Raising as in (2). As these pronouns are descriptive— ‘[℩Eij: proij]’ in particular in (3)—they and their copies may describe what they describe wherever in the sentence they are tokened—either in the second conjunct or in the first conjunct at the deletion site Δ∂ of Right-Node Raising (see section 2.1.3). There is no implication in Eventish that the coordinated phrases in (1)–(2) must first compose a collective referent to which the collectivized predicate is then applied in situ and only in situ. There is no implication that collectivized Right-Node Raising is inconsistent with, well, Right-Node Raising. Eventish is striking in this respect, in its definitive answer to the spurious argument that and must not be a sentential connective in (1) (or, even (2)) since the reduction from (12) to (1) ((2)) would deprive it of collective meaning: (12) *On the first day of spring in Central Park, a lover (cautiously) embraced behind Cleopatra’s Needle, and a belovèd (tentatively) embraced behind Cleopatra’s Needle. These chapters have found empirical evidence (section 3.5) where this corollary to Eventish resolves the syntactic puzzle presented by examples such as (13) when under reconstruction it is interpreted to mean what (14) does: (13) On the first day of spring in Central Park, a lover cautiously and a belovèd tentatively are certain to be entwined under every canopy, bower, or trellis in that earthly paradise. (14) On the first day of spring in Central Park, it is certain for under every canopy, bower, or trellis in that earthly paradise to be a lover cautiously and a belovèd tentatively entwined. The indefinite descriptions a lover and a belovèd are to be lowered within the scope of certain and every canopy. … If the collective predicate entwined implied that a lover cautiously and a belovèd tentatively must remain a phrase composing a collective referent for it, the phrase to be lowered would not be a constituent for
362
Chapter 7
movement. Worse, as this is an instance of Right-Node Raising, of be certain to be entwined …, the imagined lowering would lower the host into the phrase extracted from it, to a position deep within it. In Eventish, the collective predicate entwined comes with no such implication. The Right-Node Raising is undone, restoring be certain to be entwined … to both conjuncts without loss to its collective meaning. The indefinite descriptions then lower each within its own conjunct, each of which is about what a lover did with the belovèd or a belovèd with the lover. That plural event pronouns are the tiny bearers of collective reference has allowed Eventish to escape from the syntactic puzzle by dissociating the Right-Node Raised phrase as a whole from collective reference. And if, to explain the difference for the disjunctive interpretation between simple tensed verb phrases and tensed auxiliary phrases, the plural event pronoun displays the clitic climbing of Piedmontese (section 6.3), it is a pronoun with a vengeance. In any case, the Eventish use of descriptive event pronouns presupposes supermonadicity and thus the syntax of (13) joins the asymmetries noted above in the distribution of the alleged disjunctive and conjunctive meanings for and as evidence for these two essential aspects of the logical language, supermonadicity and descriptive event anaphora, and their corollaries. Adverbialization, which joins supermonadicity and descriptive event anaphora in the formal characterization of Eventish, finds in these chapters mainly constraint rather than direct argument. But if it exists according to argument elsewhere (see chapters 8–15 and passim), it is to be represented in logical form so as to be subject to grammatical constraint. There is no daylight between any two of (15)–(17) if all there is to them is the predication of Kunstler of the same pair of properties. A difference of meaning is achieved only modulating the reference to events (or times) (section 3.2): (15) Kunstler is sitting and standing. (16) Kunstler is sitting and is standing. (17) Kunstler is sitting and Kunstler is standing. So it is proposed that the contrast between (16) and (17) in particular reflects as in (18) and (19) a difference in the distribution of adverbialization: (18) Kunstler while Kunstler is sitting and still there is standing. (19) Kunstler while Kunstler is sitting and Kunstler while Kunstler is standing. Absent adverbialization and the requantification over events it introduces, the second clause in (16) must refer back to the antecedent events, as in (18). If this is to be the explanation, then adverbialization if it exists must indeed not occur in the second conjunct. But note that this explanation is available just as well to any other grammar that would link existential event quantification to spoken subjects and
PredP and PredP: Conclusion
363
pronominal event reference to unspoken subjects. So too goes the contrast between (21) and (23) (section 3.2.0): (20) No child is watching Sesame Street and looking unhappy. (21) No child is ever watching Sesame Street and looking unhappy. (22) ?No child is watching Sesame Street and is looking unhappy. (23) *No child is ever watching Sesame Street and is looking unhappy. The conjoined participial phrases in (21) (and (20)) are squarely within the scope of any existential event quantification the rest of the sentence may provide, but the second clause of (23) (and (22)) is not within the scope of any event quantification introduced by the first clause. The descriptive pronoun that closes off the second conjunct is subject to semantic conditions, including modal subordination, which rule it out (23) while tolerating it in (22). If that is to be the explanation and no child adverbializes, its adverbialization must be confined to the first clause of (23) (and (22)), as in (25), in contrast to the logical form in (24) for (21) (and (20)): (24) [D NP][℩E: NP][∃E: N[E, E′]] [ Φ and Ψ] (25) [D NP] [[℩E: NP][∃E: N[E, E′]] Φ and Ψ] Granted that the logical language is Eventish, translation into it from the natural language is open at first: (26) A soul man will have been shimmying and shaking. Sentence (26) has over a dozen ways it might be parsed. Since it is small clauses all the way down, there is no worry on logical grounds that and need conjoin anything larger than two verbal roots within the scope of (and agreeing with) a participial operator, [-ing[shimmy and shake]]. With every higher verbal phrase, there is a decision whether to include it or not within the coordination. Since Eventish is committed to the higher verbal phrases and has introduced a few of its own, some decision is due. As for the nominal phrase, the subject, Eventish is indifferent and Conjunction Reduction is promiscuous. So alongside the dozen or so parses induced by the presence spoken and unspoken of verbal morphology, one parse for (26) might just turn it into (27): (27) A soul man will have been shimmying, and a soul man will have been shaking. That is Conjunction Reduction, the horror movie. Let it in and it turns sentences into what they are not, except in formal semanticists’ nightmares—(26) into (27), or more simply but with as much menace, (28) into (29): (28) A soul man shimmied and shook. (29) A soul man shimmied, and a soul man shook.
364
Chapter 7
Yet, to the Davidsonian, prior to any thought of Eventish per se, Conjunction Reduction is necessary but hardly sufficient to reduce (28) to (29). It is so reduced only if the resulting logical forms are identical in their event quantification too, with both conjuncts introducing their own existential event quantifiers. Only then, with its unspoken subject the agent of an event different from the one described in the first conjunct, could (28) come to mean what (29) does. A rule for the Davidsonian is that an ellipsis unspoken goes unnoticed unless it and its antecedent fall within the scope of different existential event quantifiers. Only then will the elided phrase have a “sloppy” (i.e., nonrigid) interpretation and the likes of (28) come to resemble (29). It does exactly that in (30) (see (13) above; cf. Fox 1995, 2000), where elided copy and antecedent fall within the scope of the different event quantifiers introduced along with the distributive quantifiers: (30) A soul man shimmied on every upbeat and shook on every downbeat. Without the intervening event quantifiers, an extra token of a soul man is harmless in (28). With supermonadicity, the distribution of “sloppy” interpretations reduces to and probes the distribution of event quantifiers, as seen in the preceding chapters. The fact of (30) and every other “sloppy” interpretation encountered, an empirical finding, is the warrant for Conjunction Reduction, since without its promiscuity (30) and the like would not have been possible. Note that if (31) is a coordination of participial phrases and nothing larger, as argued in chapter 5, the “sloppy” interpretation of the subject within the scope of the distributive quantifiers indicates that the participial phrases themselves contain subject positions, and again it is Conjunction Reduction, that is, ellipsis, that delivers to the second participial phrase the token of a soul man that occurs in its subject position: (31) A soul man will have been shimmying on every up beat and shaking on every downbeat. Eventish is of course congenial to it and Conjunction Reduction proves necessary in (30) and (31), but nothing a priori anticipates that Conjunction Reduction or any manner of ellipsis is unconstrained. The “sloppy” interpretation is absent from the coordination of tensed auxiliary phrases (section 3.4.1): (32) A soul man will have been shimmying on every up beat and will have been shaking on every downbeat. Partly it reflects what has been observed in (23) (and (22)), namely, the second clause leads with a descriptive pronoun referring to the events described in the first. But this is no assurance that every downbeat will not raise to capture within its scope any event description the second clause may contain. If it may, and if a soul man has been elided in the second clause, it threatens a “sloppy” interpretation for (32).
PredP and PredP: Conclusion
365
The easy way out has been to follow Fox (1995, 2000) in assuming that the elision under discussion is a reconstruction undoing a movement. But no such movement obtains in (32) since the clauses conjoined are too large for a soul man to find a landing site outside the coordination, and therefore the subject of the second clause is rather a null descriptive pronoun referring to the soul man mentioned in the first clause. Whatever is the disposition of the second subject, it is certain that the first subject does not occupy a phrasal position that includes the second within its scope when tensed auxiliary phrases are conjoined. For if it did, all quantifiers should be allowed if any are, contrary to the facts of (33): (33) a. James Brown was shimmying to that funky disco music and was shaking to it (too). b. Every soul man was shimmying to that funky disco music and was shaking to it (too). c. *No soul man was shimmying to that funky disco music and was shaking to it (too). The objection applies a fortiori to any thought that an unspoken subject implies the Generalized Conjunction of unsaturated predicates to derive an unsaturated predicate, which by definition would be helpless telling apart (33c) from (33a) or (33b). There is instead something here to recommend a null pronoun assimilating (33) to (34): (34) a. James Brown was shimmying to that funky disco music and he was shaking to it (too). b. Every soul man was shimmying to that funky disco music and he was shaking to it (too). c. *No soul man was shimmying to that funky disco music and he was shaking to it (too). Both predicative coordination and reconstruction of the same phrase may be rejected alike if ever the subject of one conjunct needs to differ in meaning from the subject of the other, as happens in (35) (see section 3.4.1.0), which means what is meant when, as in (36), the first subject is an indefinite description or quantifier but the second subject is a definite description. The sentence has no meaning equivalent to (37) or (38): (35) Many easy model theory textbooks are badly needed and will surely be written within this decade. (36) Many easy model theory textbooks are badly needed, and they will surely be written within this decade.
366
Chapter 7
(37) Many easy model theory textbooks are badly needed, and many easy model theory textbooks will surely be written within this decade. (38) #Many easy model theory textbooks will be needed and written within this decade. On the other hand, a null pronoun in (35) correctly assimilates its meaning to that of (36). This chapter, expanding the Eventish vocabulary on empirical grounds, corroborates earlier studies discovering null coordinative pronouns in English and German (Schwarz 1999; 2000, chap. 3), in Icelandic (Rögnvaldsson 1982, 1990; Bresnan and Thràinnson 1990), and in Japanese (Tomioka 1993). Unspoken subjects are subject to early objections to Conjunction Reduction whenever an alleged unspoken subject and the spoken one it most resembles diverge (Godard 1989; Lasersohn 1995, 103f.): (39) ?No child is watching Sesame Street and is looking unhappy. (40) *No child is watching Sesame Street and she is looking unhappy. A substantive claim about translation into Eventish has been that (41) Pronunciation cues adverbialization; silence—resumption. The adverbialization of the pronounced subject pronoun in (40) results in a conjunct so large that its coordination with like conjunct traps the subject of the first conjunct within. The coordination in (39) is just small enough not to be opaque to Quantifier Raising.1 The connection just invoked between the addition of a logical operator, the adverbialized adverb, and larger clause size is an instance of a more general thesis also at work in the discussion above of the apparatus for the disjunctive interpretation and in discussions throughout (“Generative Semantics Meets Minimalism”): there is no meaning without representation and no representation without morphological weight. Morphological weight can matter even with something as small as number agreement. What can be mistaken for a DP, no rocker and no roller in (42)–(43), appears to quantify over what can answer to singular and plural predicates: (42) No rocker and no roller has ever grooved to a funky disco beat. (43) No rocker and no roller have ever grooved to a funky disco beat. The dilemma for such a DP (section 1.5.1) is that singular number agreement in (42) and (44) appears to restrict the allowed scope relations internal to the alleged DP it agrees with—not to mention the mystery of what no rocker and any roller could mean if it were a DP (see section 3.2.2): (44) *No rocker and any roller has ever grooved to a funky disco beat. (45) No rocker and any roller have ever grooved to a funky disco beat.
PredP and PredP: Conclusion
367
Instead, these are to be recognized as extreme Right-Node Raising. The logical form for singular number agreement requires one more operator here, a tacit each, than does plural agreement. As a result, the clauses conjoined in (44) are just large enough to block no rocker from raising outside the coordination to include within its scope the other conjunct and its subject any roller, as it must to license the negative-polarity item. This is an example of the “squeeze plays” that throughout relate clausal opacity and clause size. Finally, it is a point of morphology that resolves how to parse (26) and fixes translation into Eventish. First, it is a universal of complementation that no morpheme μ finds its complement XP in an adjacent conjunct (c-command condition): *[… μ …] and … XP … . This condition, as it applies to the thematic relations concealed in an apparent DP-coordination, is what compels full Right-Node Raising to provide the required complements within both conjuncts. Second, morphemes are sorted into the bound and the freestanding. The latter suffer their complements to be coordinated, μ[XP1 and XP2], but bound morphemes do not: *-μ[XP1 and XP2], ok [-μXP1 and -μXP2]. A further condition on freestanding morphemes stipulates the minimal parse tolerated, so that indeed (26) tokens have and be exactly once and the logical form for (26) is far removed from that of (27). This morphology joins the claim in (41) in fully determining the translation into Eventish, generating the variations in clause structure and their effect on meaning that have been seen to distinguish sentences such as (46)–(50) and others throughout chapters 3–6. (46) (47) (48) (49) (50)
Kunstler Kunstler Kunstler Kunstler Kunstler
is is is is is
sitting sitting sitting sitting sitting
and standing. and Weinglass standing. and is standing. and Kunstler is standing. and Weinglass is standing.
8
8.0
Introducing Adverbialization and Cinerama
Previously …
The previous chapters have absolved the sentential connectives in (1) from collective and divided reference: (1) The Columbia students fearlessly and the Harvard students fearfully surrounded the Pentagon and were crowded into the Mall. Instead, unspoken plural event pronouns intercede: one referring to what the students participate in fearlessly and fearfully (‘[℩E1,2: pro1,2]’), and another pronoun referring to events at the Pentagon and on the Mall (‘[℩E3,4: pro3,4]’): (2) [The X: ∃E C[E,X]] ∃E1 (W[E1,X] fearlessly[E1]) and [The X: ∃E H[E,X]] ∃E2 (W[E2,X] fearfully[E1]) [℩E1,2: pro1,2][℩E3,4: pro3,4]O[E1,2, E3,4] ∃E3 ∃X Agent[E3,X] [℩E3: pro3][℩E5: pro5] Cause[E3, E5] ∃E5 surround[E5] … and ∃E4 ∃X Agent[E4,X] [℩E4: pro4][℩E6: pro6] Cause[E4, E6] ∃E6 be crowded[E6] … Recall that the vagueness of (1) with respect to who was at what demonstration requires the neutral participation relation ‘W’, vague about participation as an Agent at the Pentagon or as a Theme crowded into the Mall. Collective reference to the Columbia and Harvard students per se is an illusion of the plural definite description referring to events in which they participate. The same apparatus is at work within the DP itself in the apparent coordination of NPs in, for example, nine boys and girls. The sentential connective and begins with a coordination of clauses that the nouns provide: (3) [∃X : nine[X] [℩E1,2: pro1,2]W[E1,2, X] ∃E1∃X boys[E1,X] and ∃E2∃X girls[E2,X]]
370
Chapter 8
Since the DP is vague about which of the nine children are boys and which girls, the variables are existentially closed off, and the nine can only be described as the subjects or participants in some events or states of boyhood and girlhood. Even if the manner of their participation in such were exactly the same, it could not be described drawing on the verbal vocabulary of thematic relations such as Agent, Theme, Patient, or Goal. So leave it at ‘W’, understanding that any event or state of boyhood has a boy as its unique participant (and similarly for girlhood and girls): (4) i. (W(e,x) & ∃x boy(e,x)) ↔ boy(e,x) ii. (boy(e,x) & boy(e,y)) → x = y Since the NPs are clauses conjoined by and, the route in (3) to collective reference is by now well worn. The plural descriptive event pronoun, ‘[℩E1,2: pro1,2]’, refers backwards to events of boyhood and girlhood. The indefinite description comes to denote the participants, boys and girls, only with the intervention of some relation such as W-ing to the events referred to. All this apparatus sits outside the coordination proper, and thus there must be room enough within the DP to host at least the plural event pronoun,‘W’ and the accompanying variables and quantification. Beyond this minimal logical syntax, if there is a further thought that DPs have a canonical logical form (5) to which nine boys and girls should conform, it does so only if there is existential quantification over events in which they all participate— nine such that there are some events in which they participate and these are events of boyhood and girlhood—as in (6) or (7): (5) ⌜[D ξ : ∃Ei Φ]⌝ (6) [∃X : ∃E(nine[X] ([℩E1,2: pro1,2] E =E1,2 [℩E1,2: pro1,2]W[E1,2, X] ∃E1∃X boys[E1,X] and ∃E2∃X girls[E2,X]] (7) [∃X : ∃E(nine[X] ∃E0 W[E0, X] [℩E0: pro0][℩E: pro] O[E0, E] Events[E] [℩E: pro][℩E1,2: pro1,2] O[E0, E1,2] ∃E1∃X boys[E1,X] and ∃E2∃X girls[E2,X])] The logical form in (7) says little more than (3) or (6) except to make it look like a full-blooded clause projected from very light verbs. There could be more to it if the DP manages to conform to (5) in virtue of a vocabulary of perspective and spatiotemporal reference special to nominal phrases, as will be suggested—nine such that there are scenes of them and these are scenes of boyhood and girlhood: (8) [∃X : ∃E(nine[X] ∃E0 W[E0, X] [℩E0: pro0][℩E: pro] O[E0, E] Scenes[E] [℩E: pro][℩E1,2: pro1,2] Scenes-of[E, E1,2] ∃E1∃X boys[E1,X] and ∃E2∃X girls[E2,X])] In any case, whether the logical syntax is the minimal (3) or enlarges as in (6)–(8) to answer further considerations, the apparatus for collective reference remains as
Introducing Adverbialization and Cinerama
371
ever outside the phrases and conjoins. Joining previous chapters, the remaining ones will provide further witness (sections 12.2 and 14.2) to the morphological weight, syntactic position, and meaning of the peripheral material, which I will take at its minimum in (3) for the following illustration of that argument. Once upon a time, it begins, a diva found twelve pralines on her divan. Sentences (10), in contrast to (9), say they were in one box, as an counts one boxed-ness: (9) a. Twelve boxed pralines were on her divan. b. Twelve boxed pralines in two boxes were on her divan, one of noix and one of noisettes. (10) a. A boxed twelve pralines were on her divan. b. #A boxed twelve pralines in two boxes were on her divan, one of noix and one of noisettes.1 Likewise, in another time and place, twelve passersby are near the same site in a one nearby-hood, according to (12), which cannot be the scattered accidents themselves: (11) Twelve nearby passersby were accident victims yesterday (in scattered accidents). (12) #A nearby twelve passersby were accident victims yesterday in scattered accidents. With this in mind, note that the reports in (13) and (14) allow, for all we know, the dumb luck that every accident victim was twice the victim, once in the morning as a pedestrian or motorist and again later in the day as the other: (13) Some pedestrians and motorists were accident victims yesterday (in scattered accidents). (14) A pedestrian and a motorist were accident victims yesterday (in scattered accidents). Now, in contrast, (15) demands two bodies, without apparent explanation: (15) A pedestrian and motorist were accident victims yesterday (in scattered accidents). It cannot be that plural number agreement needs the bodies, since the same poor bastard may be the victims of (14). If indeed in (15) the bodies are two, pedestrian and motorist, then a what is there one of? Not a boxed-ness or a nearby-hood, but a W-ing, for a syntax as in (3): (16) A W-ing pedestrian and motorist were accident victims yesterday (in scattered accidents).
372
Chapter 8
In contrast to (15), (13) and (14), parsed respectively as “Some.pl W-ing.pl pedestrians and motorists” and as “a (W-ing) pedestrian and a (W-ing) motorist” (with or without ‘W’), do not imply that any pedestrian and motorist participate in the same one W-ing. As it will turn out, to participate in a W-ing is more than to participate in the same event, no matter the event. To be a W-ing is to be a scene tracking with fixed, uninterrupted gaze the things it is a scene of. A scene of a pedestrian and motorist is a scene of two bodies, since one body cannot walk and drive in the same scene. It can of course walk and look in the same scene, as a pedestrian and gawker, and it can reappear, as (13) and (14) allow, as a pedestrian or a motorist in different scenes. This observation about (15)’s meaning, separating it from (13) and (14), is evidence that joins the argument of preceding chapters in showing that there is morphology with syntactic position outside the coordination, weighty enough to be modified by the indefinite article, and denoting events that can in fact be individuated and counted. Coordination is not of itself an expression for collective reference to pedestrians and motorists. That is achieved only with the intercession of a relation, as minimal as ‘W’ or more complex, that puts what is referred to into scenes, a relation with specific, perspectival content. If so, so much the worse for sterile translation of and as an operator in the language of mereology or classes, damning again the common view of collective reference in (1). 8.1
The several faces of NP1 and NP2: Puzzles of extensional substitutivity
The novelty of the argument notwithstanding, there is nothing yet new in the logical syntax of (3), (7), or (8). The innovations in the following chapters are adverbialization and an apparatus for spatiotemporal orientation. The latter expands Eventish and the descriptive content of NPs with a vocabulary already gestured at in (8). Adverbialization augments the combination of a DP with its scope. The plain meaning of (17)–(20), among which there is no contradiction even if all motorists at some times are pedestrians at other times, is represented only if unspoken adverbial phrases derived from the NPs intervene as paraphrased:2 (17) Few motorists were injured in collisions under 10 mph. Few motorists while motorists were injured in collisions under 10 mph. (18) Many pedestrians were injured in collisions under 10 mph. Many pedestrians while pedestrians were injured in collisions under 10 mph. (19) Few motorists were injured in collisions under 25 mph with pedestrians. Few motorists while motorists were injured in collisions under 25 mph with pedestrians while pedestrians. (20) Many pedestrians were injured in collisions under 25 mph with motorists. Many pedestrians while pedestrians were injured in collisions under 25 mph with motorists while motorists.
Introducing Adverbialization and Cinerama
373
Adverbializing every DP, adverbialization itself implies supermonadicity, defended on other grounds in earlier chapters: (21) A careless camper killed the unconscious drunk passed out in bed. The action while a careless camper, a cigarette flicked into the brush, is not the effect endured while unconscious at home several miles away, even as he is victim to the forest fire caused by his own negligence. Thus, the derived adverbial phrases must modify different event descriptions supermonadicity provides. Adverbialization is the solution to anything only if supermonadicity comes with it. What in this book is the occasion for all this—universal adverbialization and a vocabulary for spatiotemporal orientation—are some puzzles of substitution under identity that coordinate NPs are prey to, empirical problems peculiar to the likes of boys and girls or vocalists and instrumentalists that earlier chapters have not encountered. In one class of substitution puzzles, opposing simple NP0 and coordinate NP1-andNP2 that are presumed coextensive—musicians and vocalists and instrumentalists (or children and boys and girls)—DPs such as many musicians and many vocalists and instrumentalists do not substitute salva veritate. Another class of puzzles opposes an (in)definite description containing a coordinate NP, some/an/the NP1-and-NP2, and the coordinate description derived from its coordinated NPs, some/an/the NP1 and some/an/the NP2. In such puzzles, neither a motorist and pedestrian and a motorist and a pedestrian (see (14)–(15)) nor the motorists and pedestrians and the motorists and the pedestrians substitute salva veritate. (22) ⌜Ft⌝ (23) ⌜[D ξ : NP[ξ]] Φ[ξ]⌝ Under the spell that (22)–(23) is the structure of predication and quantification in natural language, logic will insist after a failed substitution of many children and many boys and girls that the simple NP and the coordinate NP do not denote the same things after all, implicating and in some sleight of hand to distinguish their reference. If, in turn, the motorists and the pedestrians is taken to be a referring expression, then also whatever it refers to is not what the motorists and pedestrians refers to, failing to substitute salva veritate, and it looks again like and is up to no good coordinating NPs. The substitution puzzles are thus a clear and present danger for the thesis that and is always the sentential connective—hence, the present chapters. Illustrating the first substitution puzzle, to be a musician is to be an instrumentalist or to be a vocalist (24), and so the musicians are the instrumentalists and vocalists (25): (24) musician(e,x) ↔ (instrumentalist(e,x) ∨ vocalist(e,x))3 (25) ∃E musicians[E,X] ↔ ∃E instrumentalists and vocalists[E,X]
374
Chapter 8
Yet simple NP and coordinate NP do not substitute salva veritate, alternating between the banal truths of (31)–(35) and the incoherent (26)–(30): (26) #Many musicians are a dynamic duo. #Many musicians are dynamic duos. (27) #Many musicians are a fearsome foursome. #Many musicians are fearsome foursomes. (28) #Many musicians are a chamber group. #Many musicians are chamber groups. (29) #Many musicians are two dueling dynamos. (30) #Many musicians are four mop tops impersonating the Beatles. (31) Many instrumentalists and vocalists are a dynamic duo. Many instrumentalists and vocalists are dynamic duos. (32) Many instrumentalists and vocalists are a fearsome foursome. Many instrumentalists and vocalists are fearsome foursomes. (33) Many instrumentalists and vocalists are a chamber group. Many instrumentalists and vocalists are chamber groups. (34) Many instrumentalists and vocalists are two dueling dynamos. (35) Many instrumentalists and vocalists are four mop tops impersonating the Beatles. For a syntax conforming to (36) or (37), all that should have mattered to preserve truth is that the things X that simple NP and coordinate NP denote are the same, as in (25) (whatever passes in the events E): (36) [Many X : ∃E NP[E,X]] are-a-dynamic-duo[X]. (37) [Many x : ∃X(Xx & ∃E NP[E,X])] are-a-dynamic-duo[x]. The puzzle is aggravated, where the baseball players are the fielders, who are also the batters;4 but they are not the fielders and batters, so it appears from a failed substitution: (38) #Many baseball players are a powerhouse squad. (39) Many fielders and batters are a powerhouse squad. (40) #Many baseball players are nine powerhouses. (41) Many fielders and batters are nine powerhouses. If the baseball players, the fielders, and the batters are all the same, then except for what and might manufacture from them, how else could simple NP and coordinate NP come to denote a difference? Something is indeed to be manufactured if there is a further thought that be a dynamic duo and be two dueling dynamos in (26), (29), and (42) (as well as all the
Introducing Adverbialization and Cinerama
375
other predicates of (26)–(41)) are collective predicates that require more wholesome wholes than the collective predicates in (44)–(45) do: (42) Many ensembles (of instrumentalist and vocalist) are a dynamic duo. Many ensembles (of instrumentalist and vocalist) are two dueling dynamos. (43) Many musicians are in a dynamic duo. (44) Many musicians are paired with dynamic effect. (45) Many musicians are dueling, dynamic partners. If so, musicians in (26)–(29) and (43)–(45) does not denote the same things as instrumentalists and vocalists in (31), which denotes instead the wholier ensembles that can be dynamic duos, two dueling dynamos, fearsome foursomes, chamber groups, powerhouse squads, and nine powerhouses. If the logical form for these sentences conforms to (36) or (37), there is little to conclude other than that and manufactures reference to such ensembles. But if musicians does not denote what instrumentalists and vocalists does, how is it to be explained that no interpretation of (46) is true (cf. (47))? (46) F The musicians are not the instrumentalists and vocalists. (47) The musicians are not the ensembles (of instrumentalist and vocalist). The bling ontology and alternative meaning for and founder on the false assertion in (46) of the supposed ontological distinction. A solution to the substitution puzzle here must derive the contrast between many musicians and many instrumentalists and vocalists in combining with certain predicates such as be a dynamic duo and be two dueling dynamos without denying the identity of the musicians and the instrumentalists and vocalists. The logical form to be proffered (chapter 13) manages this and maintains and as the sentential connective. Whatever and is up to in (31) and (39), it looks to be up to something else again in certain examples where identity statements themselves host the substitutions: (48) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned. (49) The lover and belovèd in either love note are the lovers exchanging them. The substitution in (48) under the identity (49) is false: (50) F The lover and belovèd in the first love note are not the lovers exchanging them.5 When a simple NP lovers and a coordinate NP lover and belovèd refer to the same things and yet their substitution does not occur salva veritate, there is an illusion that and is to blame and temptation for and under new guise to fashion from lovers their ordered pairs and for the coordinate NP to denote. Simple lovers and coordinate lover and belovèd end up referring to
376
Chapter 8
different things, sparing substitutivity on impulse but ending all the same in contradiction and confusion in (49) and (50). Again, identity statements belie the bloated semantics for and. Alongside substitutions between simple and coordinate NPs, the other class of puzzles relates a DP containing a coordination of NPs to a coordination of DPs containing the coordinated NPs, comparing a/the sunrise and sunset and a/the sunrise and a/the sunset: (51) A sunrise and a sunset have preceded two twilights’ bloody battle. The sunrise and the sunset have preceded two twilights’ bloody battle. (52) A sunrise and sunset have preceded two twilights’ bloody battle. The sunrise and sunset have preceded two twilights’ bloody battle. Despite plural reference to two sun transits, as reflected in the plural number agreement in (51) and (52), only the coordination of DPs (51) may report events that interleave sun transits and battles, sunrise—battle—sunset—battle. With coordinate NPs, (52) demands the sun transits to both precede both battles, sunrise—sunset— battle—battle. The single temporal interval demanded in (52) finds an analog in a single scene or presentation that (54) demands (see a pedestrian and motorist (15)). Both the hapless cub reporter Jimmy Olsen, always missing the action, and senior reporter Clark Kent, Superman’s alter ego, never manage to catch Superman at a crime scene: (53) A reporter and a superhero were at the crime scene interviewing witnesses. The reporter and the superhero were at the crime scene interviewing witnesses. (54) A reporter and superhero were at the crime scene interviewing witnesses. The reporter and superhero were at the crime scene interviewing witnesses. Both (53) and (54) may indifferently narrate the adventures of Jimmy Olsen and Superman. But only (53) may do the same for Clark Kent and Superman, as if the single DP in (54) containing coordinate NPs demanded reporter and superhero to sit for the same photograph, which of course Clark Kent and Superman cannot do. If it is held that the reporter and the superhero and the reporter and superhero are both referring expressions, and one has not already gagged on the metaphysics of a twosome of Clark Kent and Superman, it is just more poison to swallow that these two, unlike Jimmy Olsen and Superman, are not two enough for the reporter and superhero to refer to. On the other hand, it appears that the sunrise and sunset refers to less a twosome—again a more wholesome whole—than the sunrise and the sunset as far as what there is that can be said to be separated in time. A contrast between coordinate NPs and their counterpart coordinate DPs extends to the interpretation of relative clauses restricting them:
Introducing Adverbialization and Cinerama
377
(55) Some Apostles and some Saints who were (both) twelve in number were martyred in the All-Star game last Friday with many injuries. (56) Some Apostles and Saints who were (*both) twelve in number were martyred in the All-Star game last Friday with many injuries. (57) The elms and the beeches that are (both) dense in the middle of the forest darken the forest floor most of the day. (58) The elms and beeches that are (*both) dense in the middle of the forest darken the forest floor most of the day. In (56), twelve is unambiguously the number of all those Apostle or Saint martyred on the playing field. In contrast to (55), it cannot be understood that such Apostles were twelve and such Saints were too.6 Similarly, (58), in contrast to (57), cannot be taken to describe the elms as dense and the beeches too, describing as dense only the whole grove. Given the received syntax for predication and quantification in natural language, the substitution puzzles are an argument that the musicians are not the instrumentalists and vocalists that they are; the baseball players, not the fielders and batters; and the two lovers, not the lover and belovèd. The sunset and sunrise are not the same as the sunset and the sunrise, nor are the reporter and superhero the same as the reporter and the superhero, and they are two-ish to different standards. These substitution puzzles look to implicate and in metaphysical mischief and multiple meanings. 8.2
Spatiotemporal orientation and adverbialization
Metaphysical mischief and multiple meanings for and is an illusion dissolved in two steps by the next chapters’ two innovations, adverbialization and an apparatus for spatiotemporal orientation, Cinerama. The first step is to recognize a more general puzzle of extensional substitutivity, which, occurring among simple NPs such as musicians, instrumentalists, and vocalists, must find its solution without recourse to and or to the novel objects it might refer to. Consider then a world in which all musicians are like John, Paul, George, and Ringo in all being instrumentalists and all being vocalists. Suppose further that all of them maintain their acquired instruments in perfect condition, and yet all of them overuse their vocal folds to the point of scarring. Sentence (59) is certain truth; (60), certain falsehood; and (61), certain confusion:7 (59) Few instrumentalists abuse their instrument. (60) Few vocalists abuse their instrument. (61) Few musicians abuse their instrument.
378
Chapter 8
If the logical form for these sentences conforms to (23), from which adverbialization is absent, the failure of substitution salva veritate among them indicates that no two among instrumentalists, vocalists, and musicians are coextensive, and yet, by hypothesis, they are. In the general case, extensional substitution appears to fail, despite the identity of the instrumentalists and the vocalists, when the events they are said to participate in are not identical—playing an instrument is not doing vocals. In a logical form with adverbialization, there is no failure of extensional substitutivity per se here, since the derived adverbials, while playing an instrument and while doing vocals, fail from the start to be coextensive descriptions of events. The sentences they modify may indeed diverge in truth unremarkably—what happens while playing an instrument need not be what happens while doing vocals.8 All the puzzles of extensional substitutivity similarly rely on adverbialization for their resolution. Yet adverbialization alone does not suffice in all instances, in particular, not when the substituted NPs appear to describe the same events in addition to describing the same participants. The events of performing music are the collected events of playing instruments and doing vocals. What happens while playing instruments and doing vocals is what happens while performing music, and thus even under adverbialization, vocalists and instrumentalists and musicians might have been expected to substitute salva veritate: (62) musician(e,x) ↔ (instrumentalist(e,x) ∨ vocalist(e,x)) fielder(e,x) ↔ (infielder(e,x) ∨ outfielder(e,x)) child(e,x) ↔ (boy(e,x) ∨ girl(e,x)) (63) F musicians[E,X] ↔ instrumentalists and vocalists[E,X] F fielders[E,X] ↔ infielders and outfielders[E,X] F children[E,X] ↔ boys and girls[E,X] That they still do not owes its explanation to the second step introducing an apparatus for spatiotemporal orientation. The naked logic of substitution is not to be denied. Despite the underlying concepts (62), the NPs of natural language that are ultimately pronounced must invalidate the alleged equivalences (63) between simple and coordinate NPs. The plural NP [npmusicians] or the plural, coordinate NP [npinstrumentalists and vocalists] says something about events E or things X that the other does not. If all there is are lexical nouns voicing the concepts in (62), the plural morpheme, and and, it falls to and to invalidate (63),9 and there is no exit from the blind alleys of the preceding section. We have seen the likes of this before in the conjunction of tensed sentences: (64) They got married, and she got pregnant by him. (65) She got pregnant by him, and they got married.
Introducing Adverbialization and Cinerama
379
If the implied temporal sequencing that distinguishes (64) and (65) is semantic, then the meaning of and is at risk if the sentences conform to ⌜p and q⌝ and ⌜q and p⌝. The difference of meaning can be located elsewhere only with a difference elsewhere in the sentences ⌜p and q⌝ and ⌜q′ and p′⌝. What is spoken is an illusion masking that no two of the four clauses are identical. To that end, it is quickly observed that none of the clauses are identical if Tense as temporal anaphora (Partee 1973, 1984) explicitly represents that the time of the event described by the second clause follows that of the event described by the first. It then corroborates the dissociation of temporal meaning from and that the difference between (64) and (65) survives and’s absence: (66) They got married. She got pregnant by him. (67) She got pregnant by him. They got married. What varies in (64)–(67) has nothing to do with and and everything to do with Tense. The formal point carries over to DPs if it is assumed that NPs do not occur in DPs without morphology, addresses that describe spatiotemporal location: (68) [DP the X : ∃E (here0[E] musicians[E,X])] [DP D [AdrP Adr NP]] the musicians here0
(Cf. [CP C [TP Tense VP]] )
(69) [DP D AdrP and AdrP] (70) [DP the X : [℩E1,2: pro1.2]W[E1,2, X] ∃E1∃X here1[E1] instrumentalists[E1,X] and ∃E2∃X here2[E2] vocalists[E2,X]] [DP D [AdrP Adr NP] and [AdrP Adr NP]] the instrumentalists here1 and vocalists here2 Suppose that every NP locates the events and the participants it describes within a scene: “musicians here0” and “vocalists here1 and instrumentalists here2.” The simple NP and coordinate NP may very well describe the same performances but the latter presents them in scenes that segregate vocal and instrumental, the one here1 and the other here2,10 which the simple NP musicians does not, commingling them all here0. So enlarged to describe scenes of events, the adverbials derived from the simple and coordinate NPs are no longer coextensive and so are not substituted for one another salva veritate, as desired. The nonequivalence of simple and coordinate NPs is necessary for explaining the substitution failure that (71) and (72) exemplify, but again, it is hardly sufficient without adverbialization: (71) #Many musicians are a dynamic duo. (72) Many instrumentalists and vocalists are a dynamic duo.
380
Chapter 8
After all, the musicians are the instrumentalists and vocalists, no matter the mode of their presentation. If the logical forms of (71) and (72) were as in (73) or (74), it should be enough license for substitution salva veritate that simple and coordinate NPs are related by the condition (75), which is weaker than relational coextension (cf. (63)). (73) [Many X : ∃E NP[E,X]] are-a-dynamic-duo[X]. (74) [Many x : ∃X(Xx & ∃E NP[E,X])] are-a-dynamic-duo[x]. (75)
∃E NP0[E,X] ↔ ∃E NP1and2[E,X]
The divergence of simple and coordinate NPs does, however, matter for the logical forms obtained by adverbialization of the NPs. For then, to substitute the such-andsuch for the so-and-so in a sentence of the natural language is always a double substitution in logical form—the such-and-such while such-and-such-ing for the soand-so while so-and-so-ing. To be valid, it is not enough that the such-and-such are the so-and-so. It must also be that while such-and-such-ing is while so-and-so-ing— that is, that the derived adverbs describe the same events too: (76) [Many X : ∃E NP[E,X]][℩E: NP[E,X]] [∃E0: while[E,E0]] are-a-dynamic-duo[E0,X] (77) [Many x : ∃X(Xx & ∃E NP[E,X])][℩E: ∃X(Xx & ∃E NP[E,X])] [∃E0: while[E,E0]] are-a-dynamic-duo[E0,x] Now, with adverbialization in place, even where the events might coincide, musical events being instrumental or vocal, if NPs always occur with addresses, scenes and the location of the events within may restrict the events described and thus how the matrix events are framed: (78) #Many musicians at addresses in scenes while musicians at their addresses in those scenes are a dynamic duo. (79) Many instrumentalists at addresses in scenes and vocalists at other addresses in those scenes while instrumentalists at their addresses in those scenes and vocalists at their addresses in their scenes are a dynamic duo. As described in (79), the scenes of a dynamic duo are scenes presupposed to be wide enough for both instrumentalists and vocalists to find addresses within. In contrast, the scenes of musicianship in (78) are not so structured, affording then some difference of syntax and meaning to be exploited for substitution non salva veritate without the conclusion that the musicians are not the instrumentalists and vocalists (cf. (46)). At least, this is the prospect for a general account of the first class of substitution puzzles. Before the present example, chosen with deliberate malice, yields to it, there must be some prior account of why only a certain class of collec-
Introducing Adverbialization and Cinerama
381
tive predicates, those denoting wholier wholes, (80)–(83) vs. (84)–(88), reject a certain class of DPs, (83) vs. (90): (80) (81) (82) (83)
#Many musicians are a dynamic duo. #The many musicians are a dynamic duo. #The four musicians are a dynamic duo. #Many musicians are a chamber group.
(84) (85) (86) (87) (88)
Many musicians pair up. The many musicians pair up. The four musicians pair up. Many musicians team up. Many musicians congregate in chamber groups.
(89) The four musicians are a dynamic quartet. (90) The many musicians are a chamber group. Only with some purchase on what goes wrong in (71) and (80)–(83) can it be understood what makes it right in (72) to substitute a coordinate NP describing scenes different from that of the simple NP replaced. Some remarks in sections 8.2 and 9.4.2.2 will anticipate the account of predication and quantification in chapter 13 of these sentences. Adverbialization and the apparatus for spatiotemporal orientation also step in to resolve the second class of substitution puzzles mentioned earlier. In this class (see (14)–(15), (51)–(52), (53)–(54)), an (in)definite description containing coordinate NPs—a motorist and pedestrian, a sunrise and sunset, a reporter and superhero—does not substitute salva veritate with the corresponding coordination of (in)definite descriptions—a motorist and a pedestrian, a sunset and a sunrise, a reporter and a superhero. What could tell apart the single DP the motorists and pedestrians from the alleged coordination of DPs the motorists and the pedestrians? Both the solitary DP and the coordinated DPs segregate motorists and pedestrians in their scenes, and what happens while motorists and pedestrians is what happens while motorists and while pedestrians. It is that the solitary DP puts the motorists and pedestrians at different locations in the same scene, and the coordinated DPs further afford reference to different scenes. What happens while motorists and pedestrians are in the same scene is not the same as what happens to them in possibly scattered scenes, as already noticed in (14) and (15) above. It is to be added as a point of grammar that DP is the level for quantifying over scenes so that NPs that share a DP, a motorist and pedestrian, ascribe locations within the same scene, whereas different DPs, a motorist and a pedestrian, may be about different scenes. Applied to (91) and (93), adverbialization in the former derives an adverbial denoting events while sunrise
382
Chapter 8
and sunset in the same scene but the latter, an adverbial denoting events that occur variously while sunrise in some scenes and while sunset in others: (91) A sunrise and sunset have preceded two twilights’ bloody battle. (92) A sunrise in a scene and sunset in that scene while a sunrise and sunset in that scene have preceded two twilights’ bloody battle. ‘A sunrise and sunset have preceded two twilights’ bloody battle.’ (93) A sunrise and a sunset have preceded two twilights’ bloody battle. (94) A sunrise in a scene while a sunrise in that scene and a sunset in a scene while a sunset in that scene have preceded two twilights’ bloody battle. If DPs—but not NPs—may introduce new scenes, the substitution in natural language of ⌜[DPa/the NP1 and NP2]⌝ for ⌜[DPa/the NP1] and [DPa/the NP2]⌝ is also not a substitution of DPs into an identical context in logical form. The logical forms now invite surmise that the substitution failure has something to do with it being easier to interleave two scenes among the battles than just one. Likewise, it may just be harder on Clark Kent and Superman to appear in the same scene than in different ones: (95) A reporter in a scene and superhero in that scene while a reporter and superhero in that scene were at the crime scene interviewing witnesses. ‘A reporter and superhero were at the crime scene interviewing witnesses.’ (96) A reporter in a scene while a reporter in that scene and a superhero in a scene while a superhero in that scene were at the crime scene interviewing witnesses. ‘A reporter and a superhero were at the crime scene interviewing witnesses.’ Lastly, adverbialization and the apparatus of spatiotemporal orientation help resolve the remaining substitution puzzle from section 8.1, which creates the illusion in (97) that and derives reference to ordered pairs and then undoes it with identity statements such as (98): (97) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned. (98) The lover and belovèd in either love note are the lovers exchanging them. If, as just supposed, DPs introduce new scenes, perhaps speakers are mindful enough of continuity from one scene to the next to comment on their orientation:
Introducing Adverbialization and Cinerama
383
(99) In a mutual declaration of requited love, the lover here in a scene1 and belovèd there in that scene1 in the first love note while lover here and belovèd there in that scene1 are not the lover here in a scene2 and belovèd there in that scene2 in the note returned while lover here and belovèd there in that scene2, that scene1 oriented the same way as that scene2. ‘In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned oriented the same way.’ According to (99), the speaker relies on a tacit adverb to deny that they are them arranged a certain way. In using a coordination of addressed NPs, the speaker intends scenes that segregate lover and belovèd, and she denies in (99) that the scenes in which they appear share the same orientation. The speaker never entertains that the two tokens of the lover and belovèd might differ in their reference or differ from that of the lovers exchanging love notes. Any conclusion otherwise rests on an error of syntax, a parse of the puzzle sentences that omits adverbialization and the morphology of spatiotemporal reference. For NP-coordination, a minimal logical form is just that of the schema in (100), earlier exemplified in (3) and now with AdrPs to host the morphology of spatiotemporal reference: (100) [DX Φ [℩E1,2: pro1,2]W[E1,2, X] ∃E1∃X1 [AdrPAdr NP1] and ∃E2∃X2 [AdrPAdr NP2]] The X are AdrP1-ers and AdrP2-ers who are the participants in one or more scenes E1,2, W-ings. That is, the X are exactly those referred to by Eventish means for collective reference. [D AdrP1-and-AdrP2] is then subject to adverbialization, the formal scheme in (101) subsuming all DPs:11 (101) Adverbialization [D ξ : ∃Ei Φ] Ψ ⇒ [D ξ : ∃Ei Φ] [℩Ei : Φ] [∃Ej : N[Ei, Ej]] Ψ12 Any deviation from the expected meaning that the substitution puzzles might invite reflects descriptive content missing from the DP and its effect under adverbialization on the entire sentence. The remaining exercise is to discover what exactly that content should be. In the corrected syntax, all DPs will retain their customary and ordinary reference, and when all is said and done, and remains throughout the univocal sentential connective. The preamble for the coming chapters has been these empirical problems, puzzles of substitution under identity, that look to corrupt and with adventitious meanings when conjoining NPs. Their resolution supposes that there is so much else going on in the grammar, it is hardly necessary to look at and at all.
384
8.3
Chapter 8
Coming attractions
The argument is to entrench elsewhere—all without consideration of coordination—adverbialization, the supermonadicity it entails, and the apparatus for spatiotemporal orientation. All DPs adverbialize, and all DPs contain AdrPs, in particular those DPs containing simple, uncoordinated NPs. As remarked earlier, argument for the new grammar begins with the recognition that what has been at issue with the substitution of coordinate NPs is just a special case of classic substitution puzzles that include substitutions between coextensive NPs that are both simple, where a new meaning for and can be no help. Adverbialization proves necessary for the resolution of all the puzzles of substitution under identity. The resolution is not, however, ad hoc for the puzzles. The first sections of this chapter set out to establish that adverbialization is basic grammar. Chapter 10 shows that adverbialization and supermonadicity are essential for the syntax and semantics of Tense and temporal adverbs in order to represent their interaction with DPs. Section 10.4 then goes on to survey sundry aspects of meaning and grammar indicative of adverbialization, including, for example, the eccentric behavior of the adjective respective. Embedding it within a DP seems from the contrast between (102) and (104) to be contingent on the choice of preposition in the matrix: (102) The two newlyweds stood beside their respective fathers. The two newlyweds stood behind their respective fathers. (103) The two newlyweds stood between their fathers. (104) *The two newlyweds stood between their respective fathers. On independent grounds, every DP turns out to adverbialize, and so it can be no surprise that coextensive NP1 and NP2 substitute salva veritate only if while NP1-ing is while NP2-ing also—that is, only when the derived adverbial phrases are coextensive too. If all DPs adverbialize, it must be that all sentences in which they occur have a clause structure fit to host adverbialization, thereby entailing supermonadicity in all sentences, as remarked above for (21): (21) A careless camper killed the unconscious drunk passed out in bed. Every sentence is about at least as many events as there are DPs in it. This mandate touches not only an action sentence such as (21) with obvious distance between cause and effect but also sentences as unforthcoming in multiple events as identity statements (Hesperus is Phosphorus), those figuring in the substitution puzzles ((46), (48)–(50), and (109)), and those involving attributions of number (The Apostles are twelve), which are among the attributions of wholier wholes, facetiously dubbed so earlier because they do not appear to present any events an individual can easily participate in, let alone several. Even so, there are problems in the syntax
Introducing Adverbialization and Cinerama
385
and semantics of these sentences that require for their resolution adverbialization, spatiotemporal reference, and their conformity to a canonical, supermonadic clause structure (sections 10.3, 11.1, 13.1, and 13.2). This is just as well, since to parse identity statements as closures of ⌜x = y⌝ is a garden path into the substitution puzzles without exit, as the sentences can only chase after different values for the variables. Adverbialization and supermonadicity aside, any analysis of identity statements or attributions of number should make room for Tense and for the overt modification in (106). The conditioned identity statement in (106) differs from identity simpliciter (105) only in the presence of a secondary predicate and, as imagined, in truth. (105) The oranges in the first still life are the oranges in the second. They are them. (106) F The oranges in the first still life are the oranges in the second arranged the same way. F They are them that way. If, on the other hand, identity statements are supermonadic, there is some purchase on how a temporal adverbial might felicitously frame an event or state that holds while Hesperus or while the evening star but not while Venus: (107) For some evenings in 1892, Venus was Hesperus (aligned with a crescent moon). For some evenings in 1892, Venus was the evening star (aligned with a crescent moon). (108) #For some evenings in 1892, Hesperus was Venus. #For some evenings in 1892, the evening star was Hesperus. In conforming to the supermonadic logical form of other sentences, identity statements are then also ready to host adverbialization and to appeal to it in (99) to resolve their own substitution puzzle (48)–(49) /(97)–(98) (see section 10.3 and chapter 15). With adverbialization in place after chapter 9, chapter 10 turns to number and counting, for which the substitution puzzles such as the one in (109)–(111) are not resolved without reference to the epistemic conditions for counting, including a scene and protocol for the measurement events. (109) The passengers who crowded National Airlines’ routes last year were the frequent fliers loyal to it. (110) Three million passengers crowded National Airlines routes last year. (111) ⊭Three million frequent fliers crowded National Airlines routes last year. (After Gupta 1980, 23; Moore 1994) (112) Three million passengers had three million opinions about the food on National Airlines.
386
Chapter 8
(113) #Three million passengers have three million opinions about the food on National Airlines. The identity statement (109) fails to license the substitution and inference from (110) to (111), where frequent fliers recounted under (110)’s protocol keep it from implying (111). Moreover, although the reference to more passengers than there are persons continues in (112) as a continuation to (110) recounting the passengers’ experience on board, it cannot do so in (113) in the present tense, despite the three million opinions still held about flights remembered all too well. Here the meaning of the DP is contingent on Tense, subject to a condition that Doetjes and Honcoop (1997) have called sequencing of events. Instead of the bare, atemporal “3,000,000(X),” imagine that three million translates as “(now) counted (this way) to three million,” a description of events of measurement (e.g., counter clicks) under an explicit protocol. What is counted and what the indefinite description three million passengers refers to is naively what there is, persons, who are counted and under some protocols recounted. With the counting now entered into the description, adverbialization derives (114) and (115): (114) Three million passengers while counted to three million had three million opinions about the food on National Airlines. (115) #Three million passengers while counted to three million have three million opinions about the food on National Airlines. Clicking the counter as passengers go by frames or locates opinion recorded contemporaneously as in (114), but there is no counting to three million that frames current conditions, which, according to (115), is what defeats (113). The speaker is secure in her constant reference to naive, familiar objects by a sleight of hand that manipulates her epistemic conditions, in this case, conditions of measurement. A fundamental relationship in turn exists between the conditions of measurement and the scenes of what is counted, which under adverbialization frame the events reported. In the narrow scene in (116), there are exactly two green regions (or uncountably many). (Note that the green regions are shown as gray in (116).) There is only one in the wide scene. This is not a metaphysical claim, for surely the regions counted in the narrow scene still exist as parts of the shaded ring in the wider scene. They are just not to be counted there, under conditions inappropriate for their measurement. (116)
Introducing Adverbialization and Cinerama
387
Think of the design of an optical counter, blind to what is counted—passengers or planets if you say so. To utter two is to take on a perspective from which what is reported projects a scene the optical counter measures as 2. Counting two commits to the narrow scene in (116) and to a report of what the events counted two may frame within it, whether counting passengers or planets, as reflected in number agreement in the following: (117) The morning star and the evening are playing hide-and-seek in the twilight, season after season. (118) *From the vantage point of Jupiter, (you can see that) the morning star and the evening star are circling the sun. From the vantage point of Jupiter, (you can see that) the morning star and the evening star is circling the sun. Narrow scenes while the morning star and while the evening star exist only from Earth. From Jupiter, if anything, what is while the morning star and while the evening star converges on the same large scene of Venus in orbit, where there is exactly one Venusian region, undermining the plural number agreement. To be n is to be counted n in a scene, and the scene while so counted frames under adverbialization the participation of the n in the events reported, which forges the fundamental relationship between measurement and scenes. The events counted under one protocol may differ in their temporal resolution from those counted under another, counting passengers three million vs. counting frequent fliers one million, or in their spatial resolution as in (117) vs. (118). The relationship of counting to scenes is what requires a vocabulary for spatiotemporal reference inside DPs, the AdrPs subsequently adverbialized. Uttering in the natural language three million invokes a protocol and a scene and frame of reference for the events counted that satisfy the spatiotemporal conditions for counting. Expanding three million to include an explicit parameter for such a scene or frame of reference, “now counted so to three million” dissolves further problems of nonequivalence among DPs that at first glance appear otherwise (section 12.3). It turns out that no two of many children, many a child, and many a one or more children are synonymous. In appreciation of the problem, note that to begin with the thought that children denotes pluralities of one or more children, “children(X),” and many quantifies over them, “[many X : NP(X)],” is to forgo daylight between many children and many a one or more children. The latter manages however to break away from the former if it quantifies over scenes or frames of reference, exploiting the parameter f made available,“[many f : ∃E∃X … a … count[E,X,f, ≥1] …]” or, as eventually adopted, “[∃F: many(F) & ∃E∃X … a … count[E,X,F,≥1] …].”
388
Chapter 8
A contrast between singular and plural scenes or frames of reference is behind a contrast between perfective measurement (120) when the article is present and an imperfective, scanning measurement (119) when the article is null (section 13.1): (119) Baseball legends by the hundreds were Jews. (120) #Some baseball legends by the hundreds were Jews. This contrast between articles in referring to singular or plural frames of reference is also reflected in the failure to apply a predicate of wholier wholes (see (26)–(31) and (42) above) across multiple scenes or frames of reference in (121) and (125): (121) (122) (123) (124)
#Unsolved murders were a cluster near the Green River in Washington. Unsolved murders were in a cluster near the Green River in Washington. Unsolved murders were clustered near the Green River in Washington. Some unsolved murders were a cluster near the Green River in Washington.
(125) (126) (127) (128)
#Unsolved murders that were a cluster had a common modus operandi. Unsolved murders that were in a cluster had a common modus operandi. Unsolved murders that were clustered had a common modus operandi. Some unsolved murders that were a cluster had a common modus operandi.
Distributive quantification and singular and plural reference to scenes or frames of reference will thus warrant their explicit representation in the object language as supposed in the solution to the substitution puzzles. All NPs (three million passengers, many fielders, baseball legends, some legends, and so on) address scenes or frames of reference that under adverbialization then frame the events reported. What will prove necessary in the logical form of sentences with only simple NPs (fielders, three million passengers) carries over just the same to coordinate NPs (infielders and outfielders). Rather than a special and, the only thing special about the coordination, nine infielders and outfielders or the infielders and outfielders, is that there are two NPs and therefore twice the addresses of a simple NP, nine fielders or the fielders. For the most part, these chapters are therefore detached from and and investigate the empirical phenomena, the semantic puzzles and other aspects of meaning, that reflect the effects of adverbialization and spatiotemporal orientation in scenes and frames of reference. The good it all does and, emphasized in the earlier introductory remarks, becomes a corollary to the larger study. As in the preceding chapters, to assert that and doesn’t mean much at all is to study at length what and is not.
9
Cinerama Semantics
The events that supermonadicity and adverbialization introduce decompose the goings on of the speaker’s conscious report in (1) even as she might be unaware of the resolution into finer-grained events of, say, cause and effect: (1) The Columbia students fearlessly and the Harvard students fearfully surrounded the Pentagon and were crowded into the Mall. Further annotations to logical form will indicate perspective, scene, or frame of reference, representing the epistemic conditions for the observation of the events reported. Cardinality predicates are to be relativized to the contingent, fragile circumstances of measurement and in expressing a relation to protocol and frame of reference presuppose some fix or perspective on the goings-on being reported. Only by such means are there truths such as (2) without metaphysical dereliction: (2) The three million passengers who crowded National Airlines’ routes last year were one million frequent fliers loyal to it. It is not, however, to be imagined that perspectives, scenes, and frames of reference spring to life only where a measurement is at issue or just to rescue and and a metaphysical thesis that the three million passengers really are the one million frequent fliers. Other vocabulary and thoughts presuppose it too. Reference to scenes is required independently for the resolution of the problem of singular reference to indiscernibles (Schein 1993, 219–237; section 1.5, appendix 1, and section 9.0 of the present book). Three examples of reference to frames of reference follow. First, consider the meaning of left. If the frame of reference for (3) is taken to be that of the ballerina’s proximate visual experience, the sentence puts her partner in synchronous orbit around her (possibly leading from an outstretched arm): (3) The ice ballerina pirouetted with her partner on the left across the rink.
390
Chapter 9
Alternatively, the frame of reference could be fixed at a moment by the line of sight pointing across the rink and tangent to the ballerina’s curved trajectory, tracking the partner in a parallel course, on one side or the other depending on whether the speaker faces the ballerina and the frame of reference is centered on the speaker’s point of view or on the ballerina’s. The sentence is thus three-ways ambiguous, and it communicates no determinate thought at all unless the speaker and hearer agree on the frame of reference intended. A lexical ambiguity—left1, left2, and left3—to sort them out would insult the concept’s integrity, which must rather be left in f, for different choices for frame of reference f. In contexts that are not demonstrative for f, as in third-party transmission, to have understood (3) is to have grasped some descriptive intention like in (4), absent which, again, no determinate thought is expressed: (4) In the orbital frame of reference f centered on the agent’s proximate visual experience, the ice ballerina pirouetted with her partner on the left in f across the rink. Sentence (5) is also three-ways ambiguous—mercifully, not the nine ways it would be if the frame of reference for the second token of left were independent and possibly discontinuous from the first’s: (5) The ballerina will pirouette with her partner on the left to the middle of the ice, do a triple somersault, and pirouette with her partner on the left the rest of the way. The instruction and knowledge not to switch frames of reference within sentence is itself a point of grammar. There is no thought here without description of a frame of reference, and its representation is subject to grammatical constraint. For the second example,1 note that the orbital frame of reference just mentioned, locating the partner in fixed position within a pirouetting frame of reference, departs from the static perspective of stationary observer to introduce motion to frames of reference, hence to be thought of as subspaces on which a coordinate system of some kind is imposed. Frames of reference in motion also enter into the meaning of the spatial prepositions above, below, ahead, behind, and so on. Zwarts (1997) remarks that once the speaker’s static perspective fixes the direction of, say, behind, she may ambiguously intend a location either directly behind an object (6i) or behind a horizon fixed by the object (6ii):
Cinerama Semantics
(6)
(i)
391
(ii)
(fig. from Svenonius 2007) i. The jeep is behind the aircraft parked on the airfield. ii. Anything behind the aircraft parked on the airfield is in a military restricted area. What goes unremarked is that the alleged ambiguity disappears if the aircraft is airborne and tracked from the ground. When the observer says (7), she locates the helicopter directly behind: (7) The helicopter is behind the aircraft. It is unexpected that what counts as behind an object should narrow to a single heading (6i) just because it moves. Consider instead that spatial prepositions do not reference arbitrary objects but specifically landmarks—that is, objects at fixed location for their frame of reference—and to be behind a landmark means unambiguously to be behind the horizon it fixes for the understood frame of reference. Parked, the aircraft is a landmark both for the entire airfield and for an envelope that conforms to it. Airborne and observed from the ground, the aircraft can be a landmark only for a frame of reference that moves with it, an envelope and not an airfield. All that is to be determined in context, without any other ambiguity, is the value of the parameter f, the intended frame of reference, in which respect left and behind are alike. For both, values of f must admit frames of reference in motion. Corroborating these observations, note that from the pilot’s seat the frame of reference for which the aircraft is at fixed position stretches to the horizon, and in high-speed maneuvers, the pilot may indeed not have much more of a fix on what chases him than to say that the target is behind—somewhere within a wide 180°. The third example is simply the observation that implicit reference to mobile frames of reference reaches even the most austere expressions of direct reference such as here and there. Addressing his subjects from the prow of his flagship in harbor, King Zero says of his sons, Primo, Secondo, and Tertio, as they step in succession to the exact same spot on the podium that Zero points to with thrice-spoken heres, (8) Here Primo shall remain as regent in my absence. [Secondo replacing Primo on stage] Here Secondo shall stand as admiral of the fleet throughout the campaign, and [Tertio replacing Secondo] here too Tertio shall serve by my side as trusted advisor wherever fate takes me.
392
Chapter 9
For the princes to stay in place here throughout the long expedition to follow, it has to be that here is where three frames of reference intersect, one stationary and two mobile—the ship of state, the flagship of the admiralty, and the king’s retinue. With each token of here, the king demonstrates the same point of intersection but has in mind to refer to its address in a different frame of reference, at which address it is said the prince will remain steadfast throughout. As these three frames of reference do indeed intersect at the same point on the occasion of King Zero’s address, it could be that their salience in context is sufficient for the king and his subjects to refer to and grasp the intended values of f in which the point demonstrated is to be located and thereby to express and convey a determinate thought. Yet these frames of reference are not present for third-party report as in (9): (9) On the eve of a war long forgotten, an unknown king addressed his subjects and proclaimed that his first son shall remain there as regent in his absence, and his second shall stand there as admiral of the fleet throughout the campaign, and his third shall serve there too by his side as trusted advisor wherever fate takes him. Like (4), speaker (9) must rather have in mind some description of the values of f answering where there is that allows all the princes to remain in situ.2 These remarks about the spatiotemporal here and there parallel what is plain for demonstrative reference, this and that, to objects and events. It must be grasped what object or event that lies at the point demonstrated is intended—nose, face, head, upper torso, body, or person; kiss, embrace, tussle, or love life. It is only to be remarked that demonstration is to an intersection as crowded with locations in different frames of reference as it is with the parts of different objects or events, as spatiotemporal location is not an absolute but an address in a frame of reference.3 Vocabulary in the object language—overt spatiotemporal terms such as left, behind, here, cardinal predicates such as three million—and the interpretation of singular reference in the object language—singular definite descriptions referring to indiscernibles, and even singular number agreement (see chapter 2 and section 9.0)—all imply an object language rich in unpronounced expressions of demonstrative and descriptive reference to scenes and frames of reference. Although such parameters are neither asserted nor consciously attended to, there is no thought expressed or understood without grasping the values intended. This last remark holds of course of anything referred to, overtly or not, and in no way privileges frames of reference or terms referring to them, even if it is discovered that natural language conceals a lot more talk about the spatiotemporal. Perspective, scenes, and frames of reference acquire privilege when it is recognized that navigational communication is a benchmark for language design and language use.
Cinerama Semantics
393
Among the communicative intentions the design of language supports, I include the intention to convey to a companion witness accurate enough to guide and coordinate our navigation. It is then an empirical question what needs to be communicated from among those parameters already in place for spatial orientation and navigation. Analysis of the visual scene and the rest of the somatosensory scene includes all those parameters, even if subdoxastic, justified by the agent’s accurate navigation. I will idealize the task as twofold. The first, scene analysis, is the concurrent, inverse projection of the streaming, incoming visual and somatosensory scene onto an egocentric frame of reference for the ambient environment. The second, path integration (Aloimonos 1997; Gallistel 1990; Golledge 1999a, 1999b; Redish 1999), is the subsequent, often partial translation of the egocentric frame of reference and the features perceived therein to allocentric frames of reference, as a result of which transient perception ends in a map for an environment that is invariant to the conditions of observation. I see new construction several blocks ahead of me (scene analysis) and recognize that Manhattan has a new tower at 620 Eighth Avenue (path integration). As Gallistel (1990, 193) reports, “The reader may have had the experience of emerging from a subway station or movie theater in a grid city like Manhattan 180° misoriented. One walks with this unwitting misorientation until one fails to find some expected building or street at the spot one takes oneself to have arrived at. There follows a hard-to-describe sense of something rotating inside one’s head to produce the proper alignment between the perceived city and one’s cognitive map.” Scene analysis affords one the orientation to navigate the ambient space of a city block or so. In that sense, a scene will be said to be a scene of the ambient space it projects. Only after path integration, the experience of which may be delayed at exit from subway or theater, is the same scene, without alteration to it as such, a scene of the larger space Manhattan, with which it now comes into proper alignment. The scene is at first orienting or navigational for a city block, and then the same scene is so for a city island.4 Consider then a language of communication between agents engaged in coordinated navigation and exploration such as the drones, NCC-1700 and NCC-1701, on a 5-year mission to survey Manhattan and construct a definitive 3-D map. Their language could be such that their signal transmissions are images that successively update the 3-D map as they conduct their survey, with positions indicated for NCC1700 and NCC-1701 at the time of transmission. They thereby communicate only the results of their own path integration, triangulating any discrepancies where they disagree. Perhaps each marks in red the details of the other’s map where it dissents, assenting otherwise. In any case, their transmissions are assertions about and representations of How-the-World(-of Manhattan)-Is, and at every turn, they are accepted as true (“Yes, true”) or rejected as false (“No, false”). Alternatively,
394
Chapter 9
one may imagine a different, richer language designed for these drones, where their transmissions include the scenes of their scene analysis prior to path integration—that is, a scene within Manhattan and of it from the perspective of a drone’s current position and orientation. The richer language, to be concrete, will at the very least include an indexical expression like now-en-scène. One can well imagine how such a language would contribute to the overall robustness of the mission. If the drones relied on slightly different knowledge bases for their path integration, it would accelerate their survey if what they each saw was also subjected to the other’s path integration. Accepting frequent error, it would further hasten its correction if not only the conclusions of their path integration are communicated but also the apparent evidence. The richer language does not deter them from their mission or primary objective—that is, to represent How-the-World(-of Manhattan)-Is, a description of it in all its detail in the earlier language without now-en-scène or like expressions. As before, there is language “Yes, true” and “No, false” dedicated to the acceptance and rejection of such representations. When “Yes, true” (or “No, false”) is applied to a scene, it means that the result of that scene’s path integration is an update to the 3-D map of Manhattan that is true (or false) of Manhattan. Other responses to the scenes qua scenes, if there are any, will have to resort to other language, unsurprising in a language that starts out as an enrichment of the earlier one.5 As these two languages are meant to illustrate, the drones’ singleminded dedication to their mission and the special language of assent and dissent reserved for its product are in themselves uninformative as to which language is their language of communication, the one without or the one with now-en-scène. If it is the latter language, in which scenes are the imagery transmitted, it is a tendentious mistake to insist that the transmission’s “content” is what is implied for How-the-World(-of Manhattan)-Is or a direct representation of it like the 3-D map. Probing further, suppose that the communications between NCC-1700 and NC-1701 belong to the second language (with now-en-scène), transmitting scenes and not only updates to the 3-D model of How-the-World(-of Manhattan)-Is, and therefore the semantics for any signal transmission at t0 may include reference to scenes, perspectives, projections, and so on. Allow however that at t0 + Δt after path integration of an incoming scene, drone memory archives only the entailed update to the 3-D model so that indeed any later query as to the events at t0 can only elicit a response couched in the first language (i.e., the second language restricted to the vocabulary it shares with the first). To inquire after the drones’ “commitments” is to query what they remember and then to soon find that one commitment differs from another only if they differ in How-the-World(-of Manhattan)-Is. The study of drone design per se (read: cognitive science) takes only a glancing interest in cataloging drone commitments.6 Rather, the signal analyst (read: linguist) sets out to
Cinerama Semantics
395
discover the syntax of the language of transmission at t0 and to discover what invariances hold at t0 among any signal of like syntax, the mental state it induces at t0, and the conditions at t0 under which it is transmitted—its semantics. It may be that memory consolidation subsequently results in translation from the language of transmission into a more impoverished archival language, which then proves to be, fault of memory, the language of deliberation, the language in which the drones themselves and their outside observers traffic in commitment. Entertaining that there may be some distance between cognition, the thought, as prompted by the linguistic moment t0, and the aftermath in memory, one might refrain from an abbreviation of the process that identifies the meaning or content of the signal with its eventual decay into the commitment taken to have been asserted. It is not fair speculation to assume that the archival language is the language of transmission, presuming that t0 = t0 + Δt. It was remarked above that the logical forms for sentences (3), (5), (7), (8), and (9) demand at the very least reference to a frame of reference f if they are to be taken to have expressed a determinate thought with truth conditions that can be probed. Frame of reference is just one of several parameters that will annotate logical form by chapter’s end without the speaker’s conscious awareness of the annotations or their reference. They are surely an embarrassment if logical form is mistaken for a direct representation of the speaker’s commitment to How-the-World(-of Manhattan)-Is or something that strives toward it. Yet, since the signal transmissions between NCC-1700 and NC-1701 are (often) not the 3-D representations of How-the-World(-of Manhattan)-Is, neither are the effects of their communication merely assertions or denials of How-the-World(-of Manhattan)-Is. Communication is more telepathic—to induce in the other the same thought, so that the other is seeing what one is seeing at the moment of transmission. The logical form of such a transmission should be tricked out to choreograph and calibrate the speaker’s and hearer’s thought and perception determinate enough for cooperative, concurrent spatial orientation and navigation.7 In particular, the scene in progress under an assumed frame of reference, now-en-scène, is as much an accessible feature of context as the index that provides the reference for the indexicals in an utterance of You and I are here now. Such a model of communication and its implication for logical form need not deflect from a primary, parallel mission to settle once and for all How-the-World(-of Manhattan)-Is, to which end every transmission, the thought it induces, and the motive for transmission of this thought at this time are judged for what has been asserted obliquely about How-the-World(-of Manhattan)-Is.8 If narrative comprises thoughts for which the logical forms include annotations to indicate perspective, scene, frame of reference, and so on, representing the epistemic conditions for the observation of the events under narration, then the context
396
Chapter 9
for narrative must be rich enough to support the communicative intentions of its narrator, real or imagined (and always imagined if not real). It is shortsighted to suppose that the only parameter for narrative continuity is temporal—a timeline for world history and a succession of points or intervals along it as the narrative advances its ‘now’. Even a ‘here and now’, the indices for a context of utterance, narrowly conceived, will not fix the thoughts expressed in a spatial vocabulary including left. Here we now are indeed at the landing site on schedule, and yet we are disoriented and disabled for visual navigation if the faulty periscope from our windowless rover swivels about randomly, no matter how detailed the image transmitted or acute my report of it. A sequence of scenes or their narration is orienting and navigational for observer, speaker, or hearer only if their lines of sight are known and they constitute a survey conforming to some natural conditions on their continuity—conditions that regulate over the course of survey and narrative such parameters as needed to constitute a natural (stereoscopic, i.e., 3-D) panorama and its narration, that is, a cinerama. Continuity conditions are several in kind. There are inertial conditions that relate scenes in progress, the space they subtend (i.e., their perspective), and the position or motion of the observer relative to that space. A natural panorama amounts to a survey: (i) a dynamic scene analysis that inversely projects the visual stream into an egocentric frame of reference for the ambient environment, and (ii) concurrent local, path integration of the space the panorama’s scenes are subtending into a single space and allocentric frame of reference that the entire panorama will have subtended by its end. One surveys the city block and one’s location within it, still disoriented perhaps with respect to any landmark beyond. So much acknowledges that there is an imminent perception of a uniform, integrated space despite the diverse scenes from which it is constructed, much as a seamless scene emerges from unconsciously many saccades. Egocentric inverse projection and survey per se are dissociated under extreme conditions of high-speed hot pursuit. Imagine several superheroes so engaged in aerial combat. Their skills at targeting, chasing, and eluding each other imply the rapid inverse projection of the incoming visual stream into their egocentric frames of reference for the mobile spaces centered on them. That they are lost in space at the end of their skirmish is evidence that none of their visual experience constituted a survey (except in the defective sense of surveying their own egocentric frames of reference), despite vivid recall of theirs and their opponents’ trajectories. Without alteration to their visual experience, suppose that their combat had been a gladiatorial contest confined to an arena enforced by a force field, itself driven randomly among the stars like the motion of a plastic ball full of hamsters. A superhero’s visual experience constitutes a survey of that arena just in case the space it had subtended during the contest happens to coincide with the arena’s space, and at any moment and for a frame of reference fixed for that
Cinerama Semantics
397
arena, the superhero was aware of the coordinates of whatever her scene is a scene of, from which she is also given to know her own coordinates within the arena. Of course even if the superheroes have all so surveyed the arena, they remain just as lost in outer space. Limited by its visual content, their experience could not have been a survey—as I have defined it—of anything larger than the arena. Yet invested with further powers, say a constant awareness of their coordinates for a larger frame of reference fixed on the center of the universe at the Empire State Building, their visual experience can be said to be orienting and oriented for that larger space too. Thus, scenes with the same visual content may be further discriminated by the degree of spatial orientation they afford. At a minimum, absent pathology, a scene may be thought of (like the image through a cyborg’s eyes) as overlaid with annotations for points of interest indicating their coordinates of altitude, azimuth, and radial distance for an egocentric frame of reference at which the observer is always at , even in motion. For a survey, the scenes are further overlaid with annotations translating the egocentric coordinates to coordinates for an allocentric frame of reference fixed for the space surveyed. Yet another overlay may provide translation to coordinates for a frame of reference within which is tracked the arena’s position from the Empire State Building, a space too large for any survey. These scenes may be variously oriented and orienting for only the egocentric space, or for that space and the allocentric space of the arena, or for those two spaces and also the space of the civilized universe. In the following sections, narration is understood sometimes to include intentions to refer into scenes (now-en-scène) and to refer to spaces and frames of reference for which the scenes are intended to be oriented and orienting. Given that path integration is automatic and relentless (even if defeated in high-speed hot pursuit), narrative reports of scenes are often taken to be reports of scenes that are oriented and orienting for some allocentric frame of reference it is also the speaker’s intention to refer to. There is always, as it were, a survey in progress. Whether orientation for a larger frame of reference is also intended is rather a matter to be finessed with every discourse. For narration and any history that is its subject, continuity also has segmentation conditions that relate scenes continuous or en montage to the history they are intended to be a cinematic presentation of. For the accompanying narration, the discrete sentences of which can do no better than a series of “snapshots,” there presumably must also be segmentation conditions to anchor what is worth a sentence to salient transitions in what might otherwise be a long, continuous scene.9 The 3-D cinerama is thus a depiction of history and an experience of scene, frame of reference, and space observed, with voice-over narration, which all can be realized within a darkened theater or at least in a holodeck aboard the starship Enterprise. It idealizes what it means for an agent merely to be ambulatory, visually guided, and reporting—survey in progress—independent of and perhaps without any concurrent
398
Chapter 9
global positioning for what is under report. It will be a further discovery in the sections that follow that reference to the scenes and frames of reference drawn from cinerama also answers to conditions that relate cinerama to global frames of reference and spaces through which the cinerama is intended to afford orientation and navigation—again, city block to city island. The communications between navigating agents conducting a joint visual survey are a benchmark for language design. The characterization of that linguistic moment coordinating report and spatiotemporal orientation is criterial for a syntax and semantics adequate for the language then in use. But of course not all discourse is exigent and navigational or tethered to somatosensory experience at the time it is voiced. Even where narration had been joined to concurrent navigation, the visual experience recorded and its accompanying narrative survive as artifacts in memory often revisited and retold and also survive sometimes in representations and recitations to a public reclined around a festive meal or slouched in some other seating. Lest the primary mission be forgotten to settle once and for all How-the-World(-of Manhattan)-Is, many discourses with a view toward codifying and archiving just that strive to construct a stable guide to what is persistent and invariant across multiple perspectives and thus take up a pageant of eyewitness reports from multiple observers, like the scene changes for a history comprising scattered events. It is no surprise then that an arbitrary discourse should contain cues, stage directions as it were, indicating such shifts in perspective, and equally unsurprising is that there should be continuity conditions on such shifts in perspective (see sections 9.2–9.3). The logical form of such discourse will quantify over the scenes, perspectives, or frames of reference of others’ eyewitness narration. As with any contextual parameter, thoughts may be more or less dependent on reference to scene or frame of reference, and some narratives may wash out that dependency altogether in adopting an omniscient narrator or an arbitrary, unknown, or neutral perspective: (10) Hershey chocolate is made in Hershey, PA. (11) No three positive integers a, b, and c can satisfy the equation an + bn = cn for any integer n greater than 2. Not every sentence includes now-en-scène or any other indexical. Nevertheless the parameters in support of such are always there for the asking. Sections 9.0–9.4 argue for the elements of cinerama. Section 9.0 reviews the evidence of earlier sections for reference to scenes and to addresses within a frame of reference for the space that a scene subtends (i.e., reference into the space in view). Such reference occurs within the logical form of
Cinerama Semantics
399
nominal expressions, in which DPs are expanded to include AdrPs properly containing NPs: (12) [DPsome X : ∃E (here[E,f] NP[E,X])] [DP D [AdrP Adr NP]] Some chessmen (13) [DP D AdrP, AdrP and AdrP] a bishop, rook, and pawn Whether or not language is always cinerama, frames of reference must be uncontroversial for any language when it uses overt spatial vocabulary (e.g., left, right, etc.). As remarked above, sentences in such a vocabulary express no determinate thought without definite reference to a frame of reference. Section 9.1 comments on the robustness of such reference in two respects. First, section 9.1.0 shows that the parameter f in (12) is not merely a faint index for a frame of reference salient in context, as might appear from (3)–(8). It may also be bound by a definite description of a remote frame of reference, so that this description too occurs in the logical form of the thought expressed. The logical object language includes tacit, fullblooded definite descriptions referring to frames of reference. Second, section 9.1.1 demonstrates that the frame of reference deployed in understanding the left ice skater is not only an instrument to locate the locatum in the location the spatial vocabulary describes. It also frames the action the sentence reports, as expected if the AdrP in (12) adverbializes (chapter 10): Adverbialization (14) … [Dα : ∃E AdrP[E,α]] (… θ[Ej,β] Ψ) ⇒ (15) … [Dα : ∃E AdrP[E,α]][℩E : AdrP[E,α]][∃Ei : N[E,Ei]](… θ[Ej,β] Ψ) What happens to the left skater happens where and while on the left. The reference to frame of reference is thus twice-tokened, in the AdrP inside the host DP and again in the AdrP copied into the adverbial phrase, reflecting that it is fully present in the object language. Spaces, frames of reference, projection from the 3-D onto 2.5-D manifolds to create scenes and inverse projection from the 2.5-D onto the 3-D are all abstract geometry, and so reference to scenes, frames of reference, and spaces per se does not necessarily engage an (imaginary) agent’s visual experience and perception, let alone her spatial orientation and navigation. It suffices in sections 9.0–9.1 that ‘here[E,f]’ in (12) is an address in a frame of reference f for a space or region of space in view at the moment of utterance. Narration nevertheless implicates reference to those scenes and frames of reference constitutive of a narrator’s concurrent visual experience, to which her narration is anchored. Given the logical syntax of (12) and (13) laid down in sections 9.0–9.1,
400
Chapter 9
sections 9.2–9.4 animate narrator and navigator. Frames of reference and the projection of the 3-D onto a 2.5-D scene are geometric abstractions, without much life to them on their own. Even to say that the projective geometry of a scene has been fixed in an actual event of projection is only to hire a camera, without yet engaging the perception, spatial orientation, and navigation of an (imaginary) agent who holds it. Further annotations that might etch a scene with coordinates under this or that frame of reference are themselves just more geometric information. But the scenes referred to and quantified over in natural language are just those in which their annotation reflects the path integration of these scenes into those frames of reference for which they are orienting for some navigator. In section 9.2, the semantics of the adverbs still and again imports the relation that holds for a given visual navigation, between a scene and any space or frame of reference to which it is path-integrated. Many of us visually navigate byzantine interiors, likes rats in a maze, and from our visual survey construct accurate 3-D maps of the interior, while remaining clueless about our orientation to the compass or to any features of the landscape outside. The scenes of that survey in the course of such a navigation stand in the intended relation to the interior space surveyed and a frame of reference for it, and not to a space or frame of reference outside it. This is an essentially intentional relation. It is not a fact of projective geometry that a scene is orienting for one space larger than it subtends but not for another. It is knowledge about visual navigation that informs us that the scenes from within an interior that a narrator refers to are likely to be orienting for her for that interior, which is the space we infer it is her intention to talk about, and not for an exterior one that contains it. Correlatively, a narrator referring to scenes while navigating is taken to refer to scenes oriented and orienting for the frame of reference that is the target of her path integration. The communications of a narrator qua navigator are a benchmark for language design. That is, the language must at least be equipped for when it is so used, with an expressive power equal to the demands then placed on it. So it may be conceded that for the sake of such communications, there is warrant for all the vocabulary of spatiotemporal orientation and navigation and for all the technical apparatus for reference to scenes and frames of reference deployed in the course of such communications. It may be conceded that the logical language is the one detailed in this chapter, a syntax and semantics with the resources for effective navigational communication. But not all narration reports somatosensory experience in progress at the time it is voiced, nor do the semantic puzzles for nominal coordination ⌜NP and NP⌝ (chapter 8) now at issue attach only to navigational communication. If cinerama semantics is to be foundational, it remains to be seen where the scenery is when language is in use with none in plain sight. The thought (section 9.3) is that all narrative or discourse, including relics buried in the sand, relates to original
Cinerama Semantics
401
navigational communication via displacement: (i) as compilation from one or more proximate or distal original narrators, (ii) as recitation that is original communication or its repetition, or (iii) as narration for scenery guiding concurrent navigation or for scenery that is such scenery in reprise. Understanding always imagines how what is to be understood is compilation, recitation, or reprise displaced from its original narrators qua navigators. As just remarked, some narrative and cinema is frozen documentary assembled en montage from untold witnesses and borrowing from these authors their concurrent reportage. A sentence narrating a scene of another’s eyewitness may also be taken to have her as its author, subjecting the sentence to the epistemic conditions of reportage at the scene. In section 9.3, such is found in narrative in the present tense that is narration to faux documentaries If You Were There, which transport viewers to important moments in history, such as the eve of the Battle of Britain. As its outcome was unknown then, nothing that would betray otherwise can be said even now with hindsight in a narrative If You Were There that deploys the historical present tense despite its Churchillian tone. Hindsight is strictly in the past tense. The historical present in restricting what can be said is thus more than a vivid way for a scriptwriter to refer to the past and more than an alternative to the past tense, to which it might have been equivalent. Its semantics must relate the scene narrated to an (imaginary) author for the sentence in which it is tokened, whose authorship is displaced from the present present. To allow for a displaced narrator, the semantics for a token of the present tense must not refer to the absolute time of its utterance. In section 9.3, it is further shown that temporal-frame adverbials modifying a token of the historical present tense are rejected if they imply that the sentence’s displaced author is in two different places at once. The narrator qua navigator is a constant presence in the background to any discourse, and the semantics allows for more than one to be in play, shifting among them. Absent pathology, distinct scenes in the course of visual navigation—that is, distinct events of projection from the perspective of a mobile navigator—are of distinct spatiotemporal regions. Various records of such navigation, whether atlases or surveys, conform to this condition on representation: what looks different is different. There are not multiple perspectives on the very same spatiotemporal region. In generalizing to cinerama, this anticonvergence condition so-called (section 9.4.1) must allow for the shifts in perspective mentioned above. It cannot be that the scenes narrated all cohere as projection from the perspective of a single navigator. Rather, if twelve spies are sent to scout out the land of Canaan (Numbers 13.1–14.9), they are organized so as not to be in the same place at the same time if the mission is a topographical survey, or so as not to track the same actors if their mission is a chronicle of local history and custom. It can then be said of the documentary constituted from the scenes of their witness that distinct scenes oriented to the same
402
Chapter 9
frame of reference, Canaan, are of distinct subject matter, where the distinction in subject matter can be represented in the archival language mentioned above that is free of perspectival-dependent language such as now-en-scène. Certain semantic and pragmatic effects encountered in later chapters find no explanation unless the imaginary or real narrator is taken to be reporting scenes of visual navigation (section 11.0.0). The continuity conditions mediating scenery and spatiotemporal orientation prove to explain conditions on counting and measurement. Recall that counting one or two greens (shown in the figure as gray) is dependent on the scene being narrow or wide and the resulting presentation of what is green: (16)
(17) Two greens flank white. (18) *Two greens surround white. (19) One green surrounds white. The scene chosen determines the count of what is counted. But when a speaker intends a certain scene for counting what there is, the scene so intended must also conform to continuity conditions. It happens that sentences are judged defective if in the scenes reported, things counted n in a given scene are tracked to a scene where they would be counted fewer, n−k for k > 0. There is little hope for principle to directly relate counting (as prompted by the cardinal predicates within NP) to conditions on how the scenes the rest of the sentence describes unfold. It will rather be that the anticonvergence condition on the scenery of visual navigation happens to be breached when it tracks from one scene where what is counted are counted n to another where they are counted fewer. The condition on counting will follow from the presupposition that the scenes quantified over are just those that conform to the condition on cineramic narration mentioned above. It is sometimes remarked (see, e.g., Barker 1999; Carlson 1982; Doetjes and Honcoop 1997; Moore 1994; Zimmermann 2005) that the felicity of event counting discrepant with object counting as in (20) (cf. (21)) is contingent on the speaker representing herself as ignorant of or unable to recognize the identities that undermine the inflated tally (see also chapter 10, note 52), which in turn finds the best examples to involve large numbers:10
Cinerama Semantics
403
(20) Three thousand clients that attended speed-dating events last month exchanged phone numbers at the end of the date. (21) The 3000 clients that attended speed-dating events last month were 1000 lonelyhearts alone in a city of eight million. But anyone who knows any baseball knows that players make multiple appearances as batters and that an offensive roster of twenty-seven would be illegal in the game: (22) On September 9, 1965, in the greatest game ever pitched, Sandy Koufax retired twenty-seven consecutive batters for the Chicago Cubs with fourteen strikeouts, the most ever recorded in a perfect game. It is a law of baseball that twenty-seven batters are fewer players. Koufax’s game is a perfect one in that no batter managed to get on base, and twenty-seven is their smallest number possible in a regulation nine-inning game. Here it would be infelicitous and obscure to count otherwise, even though it is true that the Cubs fielded eleven offensive players that glorious day (nine starters and two pinch hitters): (23) #On September 9, 1965, in the greatest game ever pitched, Sandy Koufax retired eleven Chicago Cubs with fourteen strikeouts, the most ever recorded in a perfect game. Nor could it be said that the speaker of (22) represents himself as ignorant of these facts in a sports bar, where he would forfeit his manhood doing so. Notice rather that the scenes evoked of this ballgame are not orienting or oriented for the speaker for any frame of reference or space outside the game itself. What modulates the felicity of discrepant event counting is the extent to which the speaker can plausibly represent herself to be disoriented in any and all (larger) frames of reference within which the anticonvergence condition would be violated. Plausible deniability is itself gradient: (24) Three thousand clients that attended speed-dating events last month exchanged phone numbers at the end of the date. (25) ?(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers in a little black book. (26) ??(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers in mind. (27) ???(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers on the fridge.
404
Chapter 9
How can the speaker, it must be asked, have known either by visual tracking or by inference from the known facts that the 3000 clients do as reported without seeing that some clients are the same lonelyheart? One could imagine a warrant for (27) in which the speaker receives transmissions from personal cameras that have been implanted at the speed-dating events, with a field of vision just wide enough for client and refrigerator. If, on the other hand, the scenes in witness of (27) are visits to the clients’ homes, the speaker in uttering (27) represents herself as somehow subject to a lapse in the path integration that would put some of these homes at the same address containing the same fridge and the same client. Thus, it will be seen that in the conditions on counting and measurement are all the parameters of a cineramic semantics centered on the communications of navigating agents. Section 9.4 is a handbook for cinerama semantics, almost all11 of it in one place, with definitions and explanations that the reader can expect to revisit as their linguistic applications are encountered in the chapters ahead. Section 9.4.0 introduces a vocabulary for the geometry that relates scenes, frames of reference, and visual projection from space onto surface, a vocabulary for both “snapshot” scenes and cinematic scenes. Section 9.4.1 is about the intentional relations of spatial orientation to a frame of reference, and the path integration of visual experience to frames of reference. This section provides the formal definition of now-en-scène, fundamental for the language of spatial orientation and navigation, and here too is the anticonvergence condition that governs any piece of cinerama, obtained by compilation, recitation, or reprise from any number of original sources. Section 9.4.2 defines visual counting, explaining its dependence on scene as illustrated in (16)–(19) above, while leaving room for (28)–(29) although the Catskills are a single, unbroken land mass and the single forest fire nothing but many scattered fires: (28) There are many mountains in the Catskills. (29) There is a fire across the Catskills. Some counting is a summation of several events of measurement (section 9.4.2.0), and if they occur under different frames of reference, it imposes severe conditions (section 9.4.2.1) on what can be counted if anything at all (section 9.4.2.2). 9.0
Reference mise-en-scène
Besides overtly spatial vocabulary, such as left, and cardinality predicates such as two in (17) treated here as relating to the one scene of measurement and not the other, there are precedents in earlier chapters for tacit reference to scenes and to
Cinerama Semantics
405
the spaces and frames of reference they subtend—to that which is in view from a given perspective:12 (30) Two men came to the office today. They tried to sell encyclopedias. (31) Leif has a chair. It is a Regency or Biedermeier. (After Kadmon 1990) In contrast to (30), which escapes the implication that all the men visiting the office today tried to sell encyclopedias, (31) seems to imply that the unique chair that Leif owns is Regency or Biedermeier. As remarked in section 1.5 and appendix 1, descriptive anaphora suspend maximal reference contingent on the accessibility to speaker and hearer of a perspective on the events reported that is itself selective in its purview (Schein 1993, 219–237), governing both overt anaphora and the tacit event anaphora introduced with supermonadicity. On the one hand, witness to events at the office is rarely witness to them all. The encyclopedia salesmen may very well have been all the men who came to the office today under that perspective without being all the men who came to the office. On the other hand, it is not obvious what neutral stance affords a perspective on the state of chair possession that isolates Leif’s having one chair from his having others. Under what conditions for the speaker is the chair Leif has here not all the chairs he has? In extremis, singular reference to participants otherwise indistinguishable (“the problem of indistinguishable participants” (Heim 1990)) succeeds only under a succession of scenes to distinguish them, and thus (32) and (33) are fitting narration to some films about identical princesses and not others: (32) Once upon a time no one knows when, there was a twin. She had an identical twin. She was really, really identical to her. She loved her as she loved herself. But she did not love her nearly as well. She envied her her beauty, which she would have all to herself. (Appendix 1) (33) Once upon a time no one knows when, there was a tuplet. (At some time then,) An idiot prince visiting the castle mistook her for an identical tuplet. She was really, really identical to her, and so she forgave that he loved her and not herself, because he was an idiot, after all. In (32)–(33) the scene need change only with a new sentence. But if the problem of indistinguishable participants as it occurs within a single clause is to be subsumed under the same solution, scenes change subatomically too: (34) Once upon a time no one knows when, a twin had an identical twin. She was really, really identical to her … It hardly matters to the narration or to the reference of the pronouns in the last sentence, whether the story begins as it did in (32) or now as it does in (34). Given Conjunction Reduction and supermonadicity, it hardly matters for the pronouns
406
Chapter 9
whether the story begins as in (35) or as in (36), a spoken counterpart.13 Whatever is said for (36) will apply the same mutatis mutandis to (35): (35) A triplet, an identical triplet, and another identical triplet were courted by the same idiot prince and despised what he gave her, what he gave her, and what he gave her—all for one and one for all! (36) A triplet was courted. An identical triplet was also courted, and another identical triplet was courted. They despised what was given her. They despised what was given her, and they despised what was given her. It has been a commonplace in semantics that interpretation varies with the world and time at which what is described is located, representing two parameters of discourse more or less explicitly in the object language. This is fine as far as it goes, which is not far enough to model the cinematic structure of narrative intentions. It means that canonical logical form will have to be expanded to include relations to scenes to reflect the dependence of singular reference on the narrator’s actual visual experience when the only thing unique about the referent is its mise-en-scène. Moreover, any such singular or nonmaximal reference answers to conditions on the observability of the referent within the scene presented: (37) Eighteen barrels filled some bottles with Rhône varietals. They were 80% Grenache. Consider the scene from appendix 1 of a simultaneous filling up of 5400 bottles. Nonmaximal reference to merely 60 bottles (they in (37)), a fraction of them, is contingent on a spatial arrangement that affords recognition of that fraction’s participation in a spatially discrete event.14 It might have sufficed, for all that has been observed above, that the cinematic in grammar be realized simply by replacing Tense with Spatiotense, as it were—just a richer vocabulary for a similar verbal morphology. Note that the singular reference to indistinguishable participants instanced in (35) does not provide much guidance where in logical form to locate the perspectival relations, whether within the DP, in its scope, or both: (38) a. [DP∃X : an[X] ∃E … ∏[E,X] … triplet[E,X])] ∃E W[E,X] b. [DP∃X : an[X] ∃E triplet[E,X])] ∃E W[E,X] … ∏[E,X] … c. [DP∃X : an[X] ∃E … ∏[E,X] … triplet[E,X])] ∃E W[E,X] … ∏[E,X] … The descriptive content for the pronoun her may reach indifferently for the ∏ triplet that W-s, the triplet that W-s & ∏s, or the ∏ triplet that W-s & ∏s. The conclusion that the DP itself contains perspectival relations for reasons beyond those pertaining to measurement and counting has required different examples, such as the NP-
Cinerama Semantics
407
coordination in (39). Eavesdropping on the chess tutor teaching my son the endgame, I overhear: (39) A bishop, rook, and pawn will always checkmate in ten moves, but a rook, pawn, and bishop only stalemate. The well-paid tutor does not intend contradictory falsehoods, namely, that the same three chessmen will always checkmate and only stalemate. He rather quantifies over chessboards, frames of reference within which the chessmen are assigned fixed addresses. For them to be at different positions on the chessboard, each NP must occur with its own address as (40) paraphrases (along with adverbialization): (40) A bishop here1, rook here2, and pawn here3 while a bishop here1, rook here2, and pawn here3 will always checkmate in ten moves, but a rook here1, pawn here2, and bishop here3 while a rook here1, pawn here2, and bishop here3 only stalemate. The frames of reference quantified over frame the events reported—games—none of which is in progress during the lesson. Examples such as (39) suggest that NPs occur in DPs with a morphology that describes location in a frame of reference as in (41) and applied to every NP when they are coordinated as in (42): (41) [DPsome X : ∃E (here[E,f] NP[E,X])] [DP D [AdrP Adr NP]] (Cf. Some chessmen
[CP C [TP Tense VP]])
(42) [DP D AdrP, AdrP and AdrP] a bishop, rook, and pawn That NPs specifically are addressed is also in evidence in the ambiguity of (43), with both a true and a false interpretation: (43) Not many a proton and not many an antiproton annihilate each other. It is true in the sense that most protons have not and will not annihilate and be annihilated by an antiproton. Protons rarely meet antiprotons. The sentence is, however, false if proton and antiproton are understood to be confined in a frame of reference with a high probability of collision: (44) Not many a proton near and not many an antiproton near annihilate each other. To represent this meaning, there is no alternative to restriction on the NP. The sentence that removes the spatial vocabulary to the matrix is again true: (45) Not many a proton and not many an antiproton annihilate each other near each other.
408
Chapter 9
The NP-coordination in (46) exerts a subtle preference for the false interpretation in seeming to locate proton and antiproton in the same scene and thus compelling the speaker to have some grounds for so regarding them: (46) Not many a proton and antiproton annihilate each other. Whatever the explanation for the contrast between (43) and (46), the false interpretation of either can only be represented addressing the NPs to a shared scene and frame of reference. Otherwise, it is true that not many a scattered, arbitrary pair of proton and antiproton annihilate each other. Perhaps then all NPs are addressed to some scene in some frame of reference, and NPs within the same DP must be addressed to the same scene and same frame of reference, inviting then some contrast between (43) and (46) (section 8.2). Again, in section 2.0.0, as (47) and (48) limit the membership only of committees with a single Columbia student, it proves necessary to address frames of reference with only one Columbia student in them and for singular reference thus to refer to the Columbia student there: (47) Not many a Columbia student (secretly) and no more than thirty-five Harvard students (openly) were a strike committee in which the Columbia student led and the Harvard students followed. (48) Not many a Columbia student launched and no more than thirty-five Harvard students joined a strike committee in which the Columbia student led and the Harvard students followed. Recall from section 2.0.0 that the reference to frame of reference and addresses within it can clearly be descriptive and nondemonstrative, as shown when the NPs are within nonincreasing quantifiers (also the case for (43) and (46) and (47) and (48)): (49) Not many a bishop, rook, and pawn ever checkmated in ten moves without the sacrifice of the bishop, the rook, or the pawn. (50) Not many a bishop from any position on a chessboard ever retreated, not many a rook from a second position ever advanced, and not many a pawn at a third position ever remained in place to checkmate in ten moves without the sacrifice of the bishop, the rook or the pawn. The force of (50) is that not many a bishop, not many a rook, and not many a pawn have been in endgames of three, with one each, without the sacrifice of one of them. Making no claims about any chessboards other than those so laid out, (50) is understood as in (51), containing tacit forward- and backward-looking anaphora analogous to the overt herein, such, the same in (52). Sentence (52) is plainly not falsified by the many Arthurian legends with no resemblance to The Faerie Queene that are not about king, knight, succubus, and true lady:
Cinerama Semantics
409
(51) Not many a bishop from any position on a chessboard with a rook in a second position and a pawn in third position ever retreated, not many a rook from a second position on a chessboard with a bishop in first position and a pawn in a third position ever advanced, and not many a pawn at a third position on a chessboard with a bishop in first position and a rook in second position ever remained in place to checkmate in ten moves without the sacrifice of the bishop, the rook, or the pawn. (52) Not many a king in legends with the characters herein, not many a knight in such, not many a succubus in the same, and not many a true lady joining them ever tell a story very different from that of Arthur, the Redcrosse Knight, Duessa and Una in Book I of The Faerie Queene. Whether the addresses are an ordinal series, first, second, and third, or each described simply as other than those previous, relative position among them makes little sense except under the same frame of reference. What may formally present itself as a proper sequence of spatiotemporal addresses, first, second, third, and so on as in (50), is nevertheless understood as in (51). A certain kind of chessboard is at issue, with three positions on it occupied (by one side), but its description only unfolds as the sequence of NPs reveals its details, just as the description of the legends in (52) known to be at issue from the start unfolds and is completed only with the last DP. If narration narrates the narrator’s visual (i.e., cinematic) experience of the events under report, it is no surprise that the narration should contain cues, stage directions as it were, indicating the locations of actors in the scenes of the narrator’s experience. Many discourses, recalling witness from memory or the reports of other witnesses, may in turn quantify over the scenes, perspectives, or frames of reference of such witnesses’ narration, as reviewed in part above. The above also suggests NP as the locus for such reference, hence the introduction of AdrP. In the sections to follow, the quantification over perspectival objects and the perspectival relations that fill out AdrPs become explicit in pursuit of the problems outlined in chapter 8 that attach to the syntax and semantics of NP-coordination. 9.1 9.1.0
Referring to frames of reference By definite description in the object language
With any utterance, speaker and hearer reach common ground on the space to be navigated and on its frames of reference. These can be traded as when speakers face each other in a discussion of the space between them, and those on the table can be abstracted away from proximate perceptual experience as illustrated by the ambiguity of (53): (53) The ice ballerina pirouetted with her partner on her left across the rink.
410
Chapter 9
If the frames of reference are taken to be those of proximate (visual) perception, the sentence describes a partner in synchronous orbit around the ballerina. Alternatively, they could be those each of which is fixed at a moment by the line of sight pointing across the rink and tangent to the ballerina’s curved trajectory. The partner on those lefts tracks a parallel course, perhaps without any pirouettes at all. Sentences drawing on the vocabulary of spatial orientation and navigation express determinate thoughts only with respect to intended frames of reference, and speakers in interpreting such sentences are at the mercy of a rich set of assumptions about which frames of reference could be intended. Eavesdropping on another’s thoughts about a scene, I am confident of her intentions concerning left and right, despite the many frames of reference crowding her, until I observe her in conversation with another. As hinted at above, what is taken to be the intended frames of reference is informed by both the psychological states of observers, their intentions and purpose, and the physical conditions of observation. Let ‘∏[f,T,o]’ abbreviate the inertial conditions that lock a frame of reference f to the motions, postures, and physical condition of an object o taken at all times T to sit at the origin of frame of reference f related by ‘∏[f,T,o]’.15 For (53), the physical circumstances of the ice ballerina pirouetting are not alone sufficient to fix the frames of reference for a determinate thought. In any true utterance of (53), one must intend either the orbital frames of reference, ‘Orbital[f,T,o] & ∏[f,T,o]’, or the tangential ones, ‘Tangential[f,T,o] & ∏[f,T,o]’. Presumably, no speaker of (53) ever intends to refer both to the orbital and tangential frames of reference, perhaps knowing that it would be self-defeating to describe anything as on the left for both. Grammar too weighs in on reference to frames of reference, as already remarked (see (5)). Suppose you and I are at the show sitting across the ice directly opposite where the ballerina and her partner will make an entrance skating toward us, and I, having seen it before, say (54) to you: (54) The ballerina will pirouette with her partner on the left to the middle of the ice, do a triple somersault, and pirouette with her partner on the left the rest of the way. The sentence is three-ways ambiguous. I may have meant that the partner will be in synchronous orbit on the ballerina’s left all across the ice except for the somersault’s interruption, or the partner will maintain a parallel track on the same one side across the ice, meaning either that it will be on our left or that it will be left of the ballerina’s course, the opposite side. The sentence is mercifully not nine-ways ambiguous, as it would be if the frames of reference assumed in interpreting the second token of left were independent and possibly discontinuous from those assumed for the first:16
Cinerama Semantics
411
(55) [℩f: ∏orbital[f,T,b]] The ballerina will pirouette with her partner on the leftf to the middle of the ice, do a triple somersault, and [℩f: ∏orbital[f,T,b]] pirouette with her partner on the leftf the rest of the way. [℩f: ∏tangential[f,T,b]] The ballerina will pirouette with her partner on the leftf to the middle of the ice, do a triple somersault, and [℩f: ∏tangential[f,T,b]] pirouette with her partner on the leftf the rest of the way. [℩f: ∏u[f,T, us]] The ballerina will pirouette with her partner on the leftf to the middle of the ice, do a triple somersault, and [℩f: ∏u[f,T,us]] pirouette with her partner on the leftf the rest of the way. (56) *[℩f: ∏orbital[f,T,b]] The ballerina will pirouette with her partner on the leftf to the middle of the ice, do a triple somersault, and [℩f: ∏tangential[f,T,b]] pirouette with her partner on the leftf the rest of the way. *[℩f: ∏orbital[f,T,b]] The ballerina will pirouette with her partner on the leftf to the middle of the ice, do a triple somersault, and [℩f: ∏u[f,T, us]] pirouette with her partner on the leftf the rest of the way. *Etc. The instruction and knowledge not to switch frames of reference within sentence is itself a point of grammar. There is no thought here without definite reference to a frame of reference, and its representation is subject to grammatical constraint. In (55)–(56), I assume, without argument for the moment, that this definite reference is accomplished by definite description in the object language. Yet it could be with this example in which the frames of reference, orbital and tangential, are all in the scene before her that the saliency of one or the other combined with a demonstrative gesture or nod from the speaker suffices for definite reference, and the point of grammar that the same frame of reference is referred to may just be that the same free variable f must be tokened in both clauses, … leftf … leftf … and not *… leftf … leftg. … The warrant for definite description in the object language involves definite reference to frames of reference remote from the speaker, and, in the example to be constructed, it is only the definite description itself that is the same in both clauses, as it is interpreted nonrigidly in each to refer to a different frame of reference. Consider a circular library occupied only by facing club chairs aligned on a north-south axis, bisected by an east-west axis at opposite ends of which are the library’s only entrances. According to custom, the club chairs rotate position every equinox so that the white one is always aligned with hemispheric summer and the black one with hemispheric winter. It is also according to custom that morning visitors to the library arrive at the eastern entrance and evening visitors at its western entrance:17
412
Chapter 9
(57) The left chair and (the) right chair on my second visit to the library were not the left chair and (the) right chair on my first. (So, my visits were not at the same time of day in the same season, but that is all I can remember.) (58) The left chair and (the) right chair on Jorge’s visit to the library were not the left chair and (the) right chair on Luis’s. (So, their visits were not at the same time of day in the same season, and that is all that can be inferred from their reports.) In (58), the accessible frames of reference are those inertially related to Jorge’s and Luis’s experience and conditions of observation: (59) … [℩f: ∏u[f,T,j]] … left[E,f,X] … and … [℩f: ∏u[f,T,j]] … right[E,f,Y] … are not … [℩f: ∏u[f,T,l]] … left[E,f,X] … and … [℩f: ∏u[f,T,l]] … right[E,f,Y] … In contrast to the circumstances imagined to surround (57) and (58), a speaker of (60) without knowledge of their entrances expresses no determinate thought about their ongoing visit if she herself is not a visitor and they have been pacing throughout through unaligned frames of reference: (60) The left chair and (the) right chair on Jorge’s visit to the library are not the left chair and (the) right chair on Luis’s. Within a simple sentence multiple tokens of the same (spatial) vocabulary must in some sense be understood the same way (unless, and perhaps not even then, contrastively focused). In (54), even with only the ballerina’s point of view in mind, it violates this condition to first understand the directional to be defined with respect to the frames of reference of her proximate visual experience and to then understand it in terms of those fixed by the tangents to her trajectory across the ice. In that sense, notional left and right are indeed the same for Jorge and Luis in (58). What is to be explained for them is how the speaker’s understanding of the directionals has remained the same despite the shifted frames of reference. It is that the speaker has in mind throughout the same nonrigid definite description for it—for example, the egocentric frame of reference centered on the visitor to the library at e′ with the standard orientation: (61) … ∃e′ … ( [℩f: [ιx: visitor[x,e′]]∏u[f,T,x]] … left[E,f,X] … and … [℩f: [ιx: visitor[x,e′]]∏u[f,T,x]] … right[E,f,Y] …) are not … ∃e′ … ( [℩f: [ιx: visitor[x,e′]]∏u[f,T,x]] … left[E,f,X] … and … [℩f: [ιx: visitor[x,e′]]∏u[f,T,x]] … right[E,f,Y]) … The use of spatial vocabulary, left and right and the like, requires definite reference to some intended frames of reference. At the very least, speaker and hearer are salient enough for definite reference to frames of reference from their points of
Cinerama Semantics
413
view. Definite reference to frames of reference can also be grasped through their relation to objects such as ice ballerinas or visitors to the library that are themselves available for definite reference. Within a sentence such as (60), the different frames of reference must be fixed and oriented to their object in the same way, a sameness enforced only by saying that there are definite descriptions of frames of reference occurring in the object language, and those in a sentence such as (60) must be syntactically identical, perhaps the effect of a null-anaphoric relation among them.18 Definite description in the object language disappoints the wish to suppress reference to frames of reference except as a contextual parameter f with its interpretation concealed “in the semantics” and supplied by pragmatic context. 9.1.1
Frames of reference frame events
The frames of reference deployed, by demonstration or description, in understanding the left ice skater are not only an instrument for locating the locatum in the location the spatial vocabulary describes. They also frame the action the sentence reports, as expected if AdrPs adverbialize (chapter 10): Adverbialization (62) … [Dα : ∃E AdrP[E,α]] (… θ[Ej,β] Ψ) ⇒ (63) … [Dα : ∃E AdrP[E,α]][℩E : AdrP[E,α]][∃Ei : N[E,Ei]](… θ[Ej,β] Ψ) Adverbialized logical form copies the content of DPs into an adverbial description of events modifying a local thematic relation. Anything said about events in fixing nominal reference is said again to qualify or frame (via a neighborhood relation) the events described by the matrix predicate, as if to say the so-and-so while so-andso-ing …: (64) [The x : AdrP][while e: AdrP] VP(x,e) The AdrPs in (64) (NPs in common usage) are copies, the denominal adverbial incorporating all of the descriptive content of its antecedent nominal. The neighborhood relation ‘N[E,Ei]’, while, nearby, in the neighborhood of, holds in a lexical item a complex judgment dependent on context for a scale and precision to fix what counts as near. Still it can be presumed about ‘N[E,Ei]’ and AdrP that to chose a finer-grained description of yet smaller events is to imply a yet smaller neighborhood framing the matrix events VP. So much is inherent in any concept of nearness. Any variation in the events described by nominal description ripples through the neighborhood for the events satisfying the sentential assertion. If the spatial vocabulary left and right comes with parameters for scenes and frames of reference and so on, then such parameters are also copied into the derived adverbial descriptions. If the events so described can only be grasped or addressed relative to such
414
Chapter 9
parameters, their neighborhoods are likewise dependent on these parameters and the frames of reference for the events of the adverbial description turn out also to frame the events the matrix describes. The coordination in (65) allows three separate remarks to locate the chessmen in three different positions and thus engages three distinct states or events: (65) A bishop here, a rook here, and a pawn here will always checkmate in ten moves. The three positions must belong to the same frame of reference, the same (abstract) chessboard, and any subsequent (collective) reference to being in those positions and to the checkmate nearby refers to the same chessboard. Like everything else, the spatial vocabulary left and right in its lexical analysis is eventive, that is to say spatiotemporal, and when adverbialized along with the rest of the nominal description says, in effect, while left and while right. Adverbialized, its effect is illustrated in the following. The figure in (66) represents two skaters on sinusoidal and inverse sinusoidal paths of progressively increasing amplitude and wavelength. The skaters always oppose each other—the same time at the same y-coordinate.19 (66)
Describing their progress across the ice in (66), any of (67)–(69) with a plural definite description reports the increasing amplitude of their paths: (67) The skaters are skating farther apart. The skaters are skating away from each other. The skaters are diverging. (68) The left and right skaters are skating farther apart. The left and right skaters are skating away from each other. The left and right skaters are diverging.
Cinerama Semantics
415
(69) The sometimes left and sometimes right skaters are skating farther apart. The sometimes left and sometimes right skaters are skating away from each other. The sometimes left and sometimes right skaters are diverging. As there is however no unique left skater and no unique right one, a coordination of singular descriptions fails in this report: (70) #The left (skater) and (the) right skater are skating farther apart. #The left (skater) and (the) right skater are skating away from each other. #The left (skater) and (the) right skater are diverging. (71) #The sometimes left (skater) and (the) sometimes right skater are skating farther apart. #The sometimes left (skater) and (the) sometimes right skater are skating away from each other. #The sometimes left (skater) and (the) sometimes right skater are diverging. Singular description becomes possible only for smaller intervals for which there is a unique skater on the left and a unique one on the right in reports of their action while left and while right: (72) After crossing the axis, the left (skater) and (the) right skater are skating farther apart. After crossing the axis, the left (skater) and (the) right skater are skating away from each other. After crossing the axis, the left (skater) and (the) right skater are diverging. Obstructing the restriction of left and right to that interval again ends in a failure of singular reference: (73) #After crossing the axis, the sometimes left (skater) and (the) sometimes right skater are skating farther apart. #After crossing the axis, the sometimes left (skater) and (the) sometimes right skater are skating away from each other. #After crossing the axis, the sometimes left (skater) and (the) sometimes right skater are diverging. The interval quantification can itself be distributed and the singular descriptions nonrigid with respect to it, provided that the intervals sampled each contain a unique left and a unique right skater and frame the action reported: (74) The left (skater) and (the) right skater are crossing/cross at decreasing frequency. The left (skater) and (the) right skater are crossing/cross at greater intervals/ wavelengths.
416
Chapter 9
That is, the events in each of which the left skater in it and the right skater in it cross are at decreasing frequency. The nominal description describes only events with a unique left and a unique right skater, events that happen to be large enough to frame the events at crossings subsequently described in (72) and (74), as adverbialized logical form requires in deriving the sequence of event effects. In contrast, the larger event that is all of (66) is too large for an interval while there has been only one on the left and only one on the right, resulting in (70)’s failure to sequence events. This ice show allows construction of a minimal pair. Note that figure (66) (repeated here) shows three complete sinusoidal cycles, or six semicycles: (66)
A semicycle, the skaters’ path from one crossing to the next, is a zigzag. That is, first they zig away from the axis and then they zag back to it, completing the semicycle. Throughout a semicycle, a skater remains on the same side, and (75) manages to report that the semicycles are taking longer: (75) The left (skater) and (the) right skater are zigzagging/zigzag at decreasing frequency. A full cycle, on the other hand, needs a zig, a zag, and another zig, requiring the skaters to switch sides. Although it is equally true that the full cycles are taking longer, no cycle is an event that falls within an interval during which the skaters uniquely occupy their sides, and thus (76) in contrast to (75) fails the sequence of events required: (76) #The left (skater) and (the) right skater are zig-zag-zigging/zig-zag-zig at decreasing frequency. Sentences (75) and (76) are identical in every detail of logical form, except that the event concepts expressed by the verbs differ, with the first describing a zig-zag, and the second, a zig-zag-zig. This is enough to parse the same skating (66) into a different number of events of different duration. That it ends badly for (76) can be explained only if the sentence’s logical form implies that a zig-zag-zig occurs while
Cinerama Semantics
417
the left skater is on the left and the right skater on the right.20 The logical form bears that implication only if its AdrPs adverbialize, so that the frame of reference within which the skaters remain on their respective sides is also the frame of reference framing all the action described. 9.2
Scenes for spatial orientation and navigation
Emerging from the subway I see new construction several blocks ahead of me (scene analysis) and only later recognize that Manhattan has a new tower at 620 Eighth Avenue (path integration), experiencing a hiatus: as Gallistel (1990, 193) was quoted earlier (p. 393), “The reader may have had the experience of emerging from a subway station or movie theater in a grid city like Manhattan 180o misoriented. One walks with this unwitting misorientation until one fails to find some expected building or street at the spot one takes oneself to have arrived at. There follows a hardto-describe sense of something rotating inside one’s head to produce the proper alignment between the perceived city and one’s cognitive map.” That hiatus would impede none of the narration for a current scene canvassed in sections 9.0 and 9.1, which engages either reference into the current scene and to objects, spaces or frames of reference it subtends—to that which is in view from the current perspective—or reference by definite description to remote objects, spaces, or frames of reference of unknown current position. The traveler disoriented can still deliver a detailed report on the cityscape ahead. Her hiatus implies neither loss of any detail of light or sound nor lapse in the inverse projection from perceived scene into the ambient 3-D environment. Rather, the scenes with the same sensory content before and after the hiatus are distinguished by the degree of spatial orientation they afford. As it were, the initial scene analysis overlays the scene (as through a cyborg’s eyes) with annotations for points of interest indicating their coordinates of altitude, azimuth, and radial distance for an egocentric frame of reference at which the observer is always at , even in motion. The sighted construction recognized as several blocks ahead is annotated . During hiatus, [ego, NYC], the blank Manhattan coordinates flash expectantly until posthiatus the mental rotation fixes 620 Eighth Avenue.21 The scenes before and after hiatus differ only in that the former is oriented and orienting for only the egocentric frame of reference, and the latter, for both the egocentric frame of reference and Manhattan. If some communication is navigational and not merely narrative, it matters what frames of reference the scenes transmitted are intended orientation for. In this section, two adverbs, still and again as in “still on the left” and “again on the left” are found to differ in the continuity conditions they impose on the scenes that witness that being on the left still obtains or that it obtains again. For still, all such scenes must be orienting and oriented for
418
Chapter 9
all the same frames of reference. For again, it suffices that they all share orientation for one, the egocentric frame of reference of initial scene analysis. That two lexical items differ so is itself reason to import into the semantics orientation, this relation between a scene and the frames of reference to which it is path-integrated, which may be outside, properly containing the space the scene itself subtends. Communication is both cinematic and navigational, a cinerama, the language of which has a cinerama semantics. As with Jorge’s and Luis’s visits to the Library and its arrangement of club chairs (section 9.1.0), (77) may truthfully deny that the same earring was on the left and the same one on the right for both outings to Tiffany’s, even though the earrings were the same pair of diamond solitaires set in platinum, one brilliant cut (ʘ) and the other princess cut (▣): (77) The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were not the left (earring) and (the) right earring that Cathy Lane saw at Tiffany’s. These earrings sit inside Tiffany’s inside a glass display case inside a jewelry box lined in Tiffany Blue satin. With varying degrees of plausibility and depending on the context of utterance, any of these structures may provide the frame of reference necessary for understanding (77). Perhaps the speaker is stationed overlooking the entire store facing east from its Fifth Avenue entrance and intends (77) to deny that the same earring, the brilliant cut or the princess cut, has remained to the north of the other one for Patty and Cathy Lane’s viewings. Or perhaps left and right are anchored to the glass displays without regard to their orientation within the building or to the orientations of sales personnel and customers around them. With such a frame of reference in mind, the speaker is interested in whether the earring on the left side of whatever display case Patty Lane saw it in was the same as the one on the left side of whatever display case Cathy Lane saw it in. The same can be said for the jewelry boxes imagining for them an intrinsic left and right according to which way the imprinted company logo reads. A fix on these frames of reference can be understood without regard for the boxes’ orientation with respect to the display cases or to the building or whether the boxes were presented to Patty and Cathy Lane right-side up or upside down. In addition to the above frames of reference fixed on location, there are also those more plausible, especially for a speaker who is not taken to have been there and is otherwise unacquainted with Tiffany’s, the personal, egocentric frames of reference centered on Patty and Cathy Lane and oriented with their lines of sight. With these frames of reference, (77) is a comment on the Lanes’ visual experience without implications for the entrances they passed through, their approaches to the display cases or to the jewelry boxes. A tacit adverb
Cinerama Semantics
419
in (77) similarly oriented (see sections 8.2 and 15.0.1) thus manages to compare scenes of the earrings from rather diverse frames of reference. Introducing still in the sentence is more selective and in fact rules out the personal frames of reference salient for (77): (78) The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were (not) still the left (earring) and (the) right earring that Cathy Lane saw at Tiffany’s. Whether (78) is denying or asserting similar orientation, the use of still presupposes a continuity that is absent when cutting between the Lanes’ personal frames of reference. How could anything in Patty’s personal experience still be in Cathy’s? Joining the personal frames of reference with those above centered on location, it seems that still presupposes that the scenes compared are projected from frames of reference centered on the same object or point of view. There is only one Fifth Avenue entrance to Tiffany’s, and so (78) is felicitous with that frame of reference in mind, but those frames of reference centered on display cases or jewelry boxes are felicitous for (78) only if the Lanes visit the same display case or view the same jewelry box. Thus, (78) would not be a felicitous report of the events at t1 and t2 in (79) given the Lanes’ frames of reference: (79)
Tiffany & Co.
E 57th St t1 PL
+
5th Ave
CL
t2
From our stationary point of view floating above the shop floor facing uptown, (78) would be no better in that the left earring and the right earring fail to refer in that frame of reference at t1. On the other hand, (78) again becomes felicitous if the speaker manages to invoke the point of view of a salesperson, say, stationed at ‘+’ for the duration of the Lanes’ visits. While such stationary points of view are sufficient for still, they are not necessary. It cannot be said that the mere displacement in point of view, from Cathy’s to Patty’s
420
Chapter 9
or from one display case to another, undermines still. Observers navigate through many points of view, preserving their orientation (or not) to relevant landmarks: (80) a. The left (earring) and (the) right earring that Patty Lane fluttered over above the glass case at Tiffany’s were still the left (earring) and (the) right earring when she pressed her nose up against its side. The view was still to die for. b. The left (earring) and (the) right earring that Patty Lane fluttered over above the glass case at Tiffany’s were not still the left (earring) and (the) right earring when she pressed her nose up against its side. The view was still to die for. Sentences (80a–b) are vague about the time elapsed between Patty’s appreciation from above the display case and the one beside it, and it does not imply her constant attention. Other baubles or boys may have distracted her long enough for a prank22 switching the earrings. The sentences merely comment that the earring that was on her left from her point of view above the display case was (not) still on her left from a later point of view after she had, say, dropped to her knees. Of special interest is a context where there has been no prank, but in her exuberance Patty both drops to her knees and twists around to the opposite side of this small showcase for very special earrings. Sampling them from another angle, the earring on her left and the earring on her right are not still on her left and right respectively, but the view is just as entrancing, as (80b) may truthfully report. As the earrings have not moved, (80b) would however be a false report from an external, stationary frame of reference, which holds the earrings at the same addresses despite Patty’s gymnastics. In this context, (80b) is true only because still’s continuity conditions are met by a frame of reference that moves. Thus still does not presuppose a stationary frame of reference. Patty Lane’s snapshot, as it were, from above the display and her later snapshot alongside it belonged to the same frame of reference in motion. A presupposition that the frame of reference for what still obtains is the same throughout rules out cutting between Patty’s and Cathy’s personal frames of reference in understanding (78). But it accepts all the above stationary frames of reference and also allows Patty Lane to move hers. This continuity condition is necessary for still but still insufficient: the narrator may accompany Patty Lane shopping, crossing the intersection of Fifth Avenue and 57th Street from Tiffany’s to Van Cleef & Arpels with her, and yet the report (81) that the earrings at Van Cleef & Arpels were or were not still as they were at Tiffany’s is again infelicitous (even imagining the earrings’ transport and installation from one location to the other before her entrance at the latter):
Cinerama Semantics
421
(81) #The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were (not) still the left (earring) and (the) right earring that she saw at Van Cleef & Arpels. Whether twirling around the display case inside Tiffany’s or crossing the street, the only sense in which the earrings could still be on the left or right holds fast to Patty Lane’s frame of reference for left and right. Yet in addition, inside Tiffany’s, all her scenes of the earrings while twirling are also orienting and oriented for Tiffany’s interior or at least for that department in which the earrings are on display, as much as scene analysis is always (absent pathology) inversely projective into the ambient environment. During these scenes, coordinates or regions such as left and right within her egocentric frame of reference have immediate, concurrent translation into coordinates for Tiffany’s. In an obvious sense, these scenes all survey the same space, Tiffany’s interior. Inside Tiffany’s, even with a wide angle, the frame of reference is of and for an interior space that does not include Van Cleef & Arpels. Similarly, within Van Cleef & Arpels, Patty and the narrator sweep through its aisles expertly oriented for all that this space contains, which is other than Tiffany’s. Besides holding fast to the same frame of reference for left and right, still further presupposes that the scenes are all of and for the same space, thereby excluding (81). Departing from the naturalistic circumstances imagined for the narrative, one can, in turn, rescue (81), satisfying this further condition on spatial orientation. Imagine that the narrator floats above the intersection of Fifth Avenue and 57th Street, from which perch the goddess sees into the interiors of Tiffany’s and Van Cleef & Arpels. For a frame of reference for the space she commands, she may say (81) to mean, in effect, that the earring to the west of the other while the pair was at Tiffany’s was still to its west at Van Cleef & Arpels. Correlatively, one can hold to naturalistic assumptions if the space narrows to fit: (82) The left (earring) and (the) right earring that Patty Lane wore at Tiffany’s were (not) still the left (earring) and (the) right earring that she wore at Van Cleef & Arpels. If Patty does not adjust her earrings, they literally never move, remaining throughout the day in the same position for that space, one on the left and the other on the right, although the space itself is as mobile as a teenager on the avenue. Here the space navigated, the space to which the scenes are path-integrated, just is the egocentric one. Like still, again presupposes that the scenes compared are projected from a frame of reference centered on the same object or point of view, but unlike still, again omits presupposing that they are of the same space.
422
Chapter 9
(83) The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were (not) again the left (earring) and (the) right earring that Cathy Lane saw at Tiffany’s. The relevant interpretations of (83) concern whether or not the earrings that are left and right somewhere are elsewhere again left and right, with again focused only on being on the left and being on the right, without implying that again Cathy Lane sees them. As with still in (78), the frames of reference associated with a stationary point of view are all felicitous provided it is the same point of view for both the Lanes’ visits, and again the personal frames of reference centered on the Lanes’ themselves are rejected—in what sense do the earrings find again the same left and right? In contrast to still in (81), again does not however reject the change of venue: (84) a. The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were again the left (earring) and (the) right earring that she saw at Van Cleef & Arpels. b. The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were not again the left (earring) and (the) right earring that she saw at Van Cleef & Arpels. To her surprise (or not), the very same earrings from Tiffany’s are again in her experience of left and right gazing down at a display case at Van Cleef & Arpels. Again, unlike still, implies no continuity in the space under survey and fits a report of a punctuated experience whose interpretation as a cognitive lapse or as evidence of the secret movement of the things on display is at issue. This discussion of still and again is distilled in the definitions in (85)–(87), where a more formal presentation of addresses and projections awaits section 9.4.0 (see (173)): (85) en-scène[E, Σ,A,F] ↔df For frames of reference F and addresses A, events E are at A in F & scenes Σ are (projections) of A in F (86) still[E,Σ,F] ↔df ∃e1∃e2(e1≠e2 & Ee1 & Ee2) & ∃A en-scène[E,Σ,A,F] & ∀σ∀f((Σσ & Ff) → orienting(σ,f)) (87) again[E,Σ,F] ↔df ∃e1∃e2(e1≠e2 & Ee1 & Ee2) & ∃A en-scène[E,Σ,A,F] & ∃f(Ff & ∀σ(Σσ → orienting(σ,f)) As above, a scene σ is oriented and orienting for a frame of reference f just in case its addressing always includes coordinates for f. The traveler’s scenes emerging from the subway are not orienting for Manhattan until she has recovered from her hiatus. What persists to be either still such-and-such or again such-and-such is of course a plurality of events or states (if not a continuum)—hence the first clause in (86) and (87). Crosscutting between Patty Lane’s and Cathy Lane’s personal, egocentric
Cinerama Semantics
423
frames of reference, there is not any one frame of reference to which are addressed all the elements the scenes are scenes of. Thus, what obtains neither still obtains nor obtains again. If there is at least one frame of reference for all the addresses, such as Patty Lane’s egocentric frame of reference, then what obtains may obtain again. Under naturalistic conditions for both observation and perception, the scenes of visual experience orienting for Tiffany’s interior are not also orienting for the Van Cleef & Arpels interior, even imagining an extension of the frame of reference anchored inside Van Cleef & Arpels to include Midtown. Events inside Tiffany’s indeed have addresses abstractly in the Van Cleef & Arpels extension but none that the observer inside Tiffany’s is aware of. For this reason, what obtains inside Tiffany’s does not still obtain inside Van Cleef & Arpels. Section 9.1 began to measure out the extent to which we are up to our necks in utterances that do not express determinate thoughts let alone true ones without definite reference to frames of reference. The adverbs still and again enlarge the point in imposing continuity conditions on the scenes and frames of reference referred to. Since all these sentences (77)–(84) have been about the same two diamond solitaires, a brilliant cut and a princess cut, the only true denials (sentences with not) deny something about their orientation, conveyed, I have assumed, by a tacit adverb similarly oriented, the semantics of which is deferred until section 15.0.1. It remains to be asked now whether this tacit adverb, like still and again, also presupposes anything about the frames of reference from which are drawn the scenes said or denied to be similarly oriented. Already, the truth of (84b) (as well as any of the sentences from earlier sections comparing scenes from scattered or unknown venues) shows that tacit similarly oriented does not presuppose that the frames of reference are of and for the same space. Moreover, since (77) may felicitously report the Lanes’ discontinuous experience of jewelry boxes, crosscutting between them, it shows that not much is presupposed about the frames of reference from which the scenes are projected. The discontinuity does not however escape unnoticed. Keep in mind that the sentences can be judged true only denying that the earrings hold their relative position left and right since they are themselves the same ones throughout. For the distinct spaces already given as the context for (84b), (88) simply introduces another observer to disrupt the continuity from one point of view to another, resulting in a contrast between (84b)’s easy truth in the earlier context and (88)’s infelicity in (89): (88) #The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were not the left (earring) and (the) right earring that Cathy Lane saw at Van Cleef & Arpels.
424
(89)
Chapter 9
Van Cleef & Arpels F i f t h
t2
CL
W
57 A v e n u e
E t1 PL
Tiffany & Co.
The events of (88) have been contrived so that (88) is not true for a stationary frame of reference of an omniscient narrator floating above the intersection. It could be true if she rotates from east to west either above the intersection or at street level in it, or if she herself happens to be shopping, starting beside Patty Lane at t1 and then crossing the intersection to find herself beside Cathy Lane at t2. But I assume these are all too remote for (88) uttered out of the blue without richer context animating such a narrator. To grasp at a truth in (88), we are asked instead to jump from Patty Lane’s skin into Cathy Lane’s, which is uncomfortable, unlike accompanying Patty Lane as she alone moves from Tiffany’s to Van Cleef & Arpels in the context for (84b). Because one crosscuts between points of view at different venues and at different times, one cannot conclude from their dissimilar orientation anything about the earrings’ movements or position in absolute space absent a common frame of reference, and one also cannot take the asserted dissimilar orientation to mean anything for either Patty’s or Cathy’s frame of mind, since neither experiences it (cf. (84b)). If not, then what could have been the speaker’s purpose in denying that the two scenes are similarly oriented? Crosscutting between the Lanes may not violate any inertial conditions on the observer that similarly oriented presupposes, but it prompts a question about the relevance of orientation in the circumstances envisaged. The discomfort from (88) may itself be gradient, an effect in accord with the conversational pragmatic effect entertained. Thus, (90) presents the same arrest-
Cinerama Semantics
425
ing, crosscutting between the Lanes, but as a report of (79) from within the smaller space of Tiffany’s, it seems to get better with pleading (cf. (78), (83)). (90) #The left (earring) and (the) right earring that Patty Lane saw at Tiffany’s were not the left (earring) and (the) right earring that Cathy Lane saw at Tiffany’s. (79)
Tiffany & Co.
E 57th St t1 PL
+
5th Ave
CL
t2
Imagine a narrator from a natural, stationary point of view overlooking the shop floor and thus in command of a frame of reference that includes both the Lanes’ frames of reference. Knowing the Lanes’ positions, such a narrator could infer from the dissimilar orientation of their scenes that the earrings have in fact been moved from one place to another within Tiffany’s. Such a narrator would have grounds for (90) and so the sentence in the context of (79) seems better motivated. Crosscutting is equally disruptive in personal space, resulting in at least a passing contrast between (91) and (92), where again it must be remembered that these sentences can be true only in virtue of the earrings’ relative position on the bearers’ heads: (91) The left (earring) and (the) right earring that Patty Lane wore at Tiffany’s were not the left (earring) and (the) right earring that she wore at Van Cleef & Arpels. (92) #The left (earring) and (the) right earring that Patty Lane wore at Tiffany’s were not the left (earring) and (the) right earring that Cathy Lane wore at Van Cleef & Arpels. As before, left and right on Patty Lane seem incommensurate with left and right on Cathy Lane, and thus (92) comes with no basis for the comparison unless one is provided. Suppose it known that a girl who plays with the rich and marries for love wears a princess solitaire in her left ear and a brilliant solitaire in her right ear, but
426
Chapter 9
a girl who plays for love and marries rich wears the reverse; then (92) pointedly remarks that the Lanes are different kinds of girls. Alternatively, (92) can be restored to the same continuity as (91) satisfying the same inertial conditions, once it is realized that Patty Lane and Cathy Lane are the same person, Patty Duke, who changes her character along with her earrings between Tiffany’s and Van Cleef & Arpels.23 What matters is that the sentences of this section, given their spatial vocabulary and the tacit adverb similarly oriented, can express no determinate true thought without an essential reference to frames of reference. These must then variously satisfy the presuppositions of again and still, when these adverbs occur, and the speaker and hearer must attend to why the frames of reference referred to and the orientation of their associated scenes are at all relevant.24 This section has explored substantive constraints on the abstracta of scene and frames of reference quantified over. It is not merely reference to scenes in the narrator’s visual experience and to the space that these scenes are scenes of. The narrator is also a navigator for whom path integration is a dynamic relation among observer, scenes, and an allocentric frame of reference for the space navigated. The narrator may intend to refer to the scenes of her visual navigation, oriented and orienting for the frame of reference that is the target of her path integration. Certain semantic and pragmatic conditions on counting and measurement encountered in later chapters find no explanation unless the imaginary or real narrator is indeed taken to be reporting scenes of visual navigation subject to continuity conditions mediating scenery and spatiotemporal orientation. 9.3
The narration of visual experience, and narrative as artifact
Navigational scenes have been taken to be annotated for orientation to frames of reference, to at minimum an egocentric frame of reference for their projection, which is a frame of reference for a mobile space with one fixed landmark, the observatory platform at . Further annotations enrich the information states of these scenes with orientation for other frames of reference for other spaces, said in the previous section to be under survey. This characterization, the mention of information states in particular, is a reminder that navigational scenes and their orientation to frames of reference are in the first instance abstracta. To say, in turn, that the projective geometry is presumed to have been fixed in an event of projection further implicates an observer, a camera or eye, for that event. But absent some assumption that anchors discourse to narrator and observer, reference to scenes and the events of their projection does not preclude that a discourse’s sentences be desultory observations from scattered posts. If a discourse is a narration or communication for navigation, it ought to be that the scenes referred to cohere as the visual experience of some navigating agent. The scenes referred to are events of projection for
Cinerama Semantics
427
which she is the observer, and her path integration annotates them with coordinates for the frames of reference and spaces for which they orient her. In discourse, narrators are also navigators, in which narration is utterance, an action coordinated with scenes projected during navigation. At least, this is the benchmark epistemic condition underlying language use and understanding. Many a narrative, however, is storytelling in which the teller is removed from the events told, scenes of their witness, and any navigation among them. Even if such narrative is conceived of as a relic from live communication of navigation in progress, the interesting story now told is likely to have been compiled from an untold number of navigating agents. As much as the narrative now retold may be an artifact displaced from an earlier, original narration, any visualization that now accompanies the voice-over, whether recalled from memory or projected in a cinerama, may also be an artifact displaced from its original projection by an untold number of witnesses. For any event of narration Δ comprising subevents φ of phrases or sentences uttered, there are the events of their original authorship Δ© and φ©, which may be scattered insofar as the narrative draws on the testimony of many. Likewise, any visualization Π now in progress comprising uninterrupted and contiguous scenes π may reprise scattered, disjointed original events of projection π©. As it may be that a narrative is original commentary, with the benefit of hindsight, for a visualization in reprise, note that it has to be allowed that the coincidence of φ as commentary on π does not imply the coincidence of φ© and π©. Recitation and reprise are of course meant to preserve information and tradition, and thus events Δ and Π are transparent in their grasp and understood to relate directly to the conditions surrounding Δ© and Π©, that which they are reports and scenes of. Restricting ourselves for the moment to the navigation and concurrent communications of a single agent, note that any cineramic scene Π comes with an egocentric clock, ltΠ (local time), running from the scene’s opening to The End. It may be thought of as a running timestamp annotating the scene with nonnegative real number increments δt from 0. Let ltΠ be the stopwatch for Π that starts at δt0 when Π begins and ends at δtFINIS when it ends. For any fragment π of Π, let ltΠ(π) = be the interval of time that ltΠ clocks as this fragment’s start and finish. The same is to be said for the discourse Δ and its phrases φ narrating the scene and a clock/stopwatch ltΔ for it. As it may be useful to talk about times within fragments of scenes and of discourses, let (93) ⌜ltΠ(π)(δt)⌝ abbreviates ⌜ltΠ(π) = & δt0π ≤ δt ≤ δtfinisπ)⌝. (94) ⌜ltΔ(φ)(δt)⌝ abbreviates ⌜ltΔ(φ) = & δt0φ ≤ δt ≤ δtfinisφ)⌝. Discourse Δ narrating scene Π synchronizes their clocks. That is, for any bijective homomorphism from narrative fragments φ to narrated scene fragments π that
428
Chapter 9
preserves temporal order, the narrative fragment and the scene fragment it narrates clock the same: (95) ∀φ0∀φ1((Within(φ0, Δ) & Within(φ1, Δ) & φ0 < φ1) → (φ0) < (φ1)) → (Within(φ, Δ) → (LTΔ(φ) = ↔ LTΠ((φ)) = )) Narrations Δ and the visual experiences Π narrated are dated, historical events, and their local time coordinates map onto universal, absolute time, UTC:25 (96) Within(φ, Δ) → ∃t(UTCΔ(ltΔ(φ)) = t) (97) Within(π, Π) → ∃t(UTCΠ(ltΠ(π)) = t) With any discourse Δ, there is a function nowΔ dependent on its clock defining a window of absolute time centered on the local time to which it is applied: (98) ∀δt ((ltΔ(Δ)(δt) & UTCΔ(δt) = t) → overlaps(t, nowΔ(δt))) The function nowΔ underwrites the semantics for now, a predicate of events, in the object language. For simplicity, present tense will be presumed a synonym, also translated as now: (99) For discourse Δ and token u of ⌜now(vi)⌝ in Δ such that ltΔ(u) = δt, Σ satisfy u ↔ Within(nowΔ(δt), Σ(vi)) In appendix 1, quantification over scenes σ0 … σ6 in succession is introduced in (101) in support of singular reference to participants, identical tuplets, qualitatively indistinguishable (Heim 1990): (100) Once upon a time no one knows when, there was a tuplet. (At some time then,) An idiot prince visiting the castle mistook her for an identical tuplet. She was really, really identical to her; and so she forgave that he loved her and not herself, because he was an idiot, after all. (101) ∃Π0∃σ0∃σ1∃σ2∃σ3∃σ4∃σ5∃σ6 ||: [∃t : t = time(σ0) & Once upon a time(t) no one knows when] there was a tuplet[t,σ0,Π0]. [Some t′: t′ = time(σ1) & [℩t: then[t, time(σ0)]]Within(t′, t)] An idiot prince mistook [℩x: tuplet[x, Π0]] for an identical tuplet[t′, σ1]. [∃t′: t′ = time(σ2) & [℩t: then[t, time(σ1)]]Within(t′, t)] [℩x: tuplet[x,Π0]] was identical to [℩x: identical tuplet … [x, σ1]] [t′, σ2]. [∃t′: t′ = time(σ3) & Later(t′, time(σ2))] he mistook [℩x: tuplet[x, Π0]] for an identical tuplet[t′, σ3]. [∃t′: t′ = time(σ4) & [℩t: then[t, time(σ3)]]Within(t′, t)] [℩x: tuplet[x,Π0]] was identical to [℩x: identical tuplet … [x, σ3]] [t′, σ4].
Cinerama Semantics
429
[∃t′: t′ = time(σ5) & Later(t′, time(σ4))] he mistook [℩x: tuplet[x, Π0]] for an identical tuplet[t′, σ5]. [∃t′: t′ = time(σ6) & [℩t: then[t, time(σ5)]]Within(t′, t)] [℩x: tuplet[x,Π0]] was identical to [℩x: identical tuplet … [x, σ5]] [t′, σ6]. :|| It may be truer to (100) to replace the overt quantification over scenes with function calls to the internal clock—the identical tuplet now en-scène … the identical tuplet now en-scène … the identical tuplet now en-scène (see (85)):26 (102) Δ||: …[℩x: ∃e∃σ∃a∃f (now(σ) & Π0(σ) & en-scène[e,σ,a,f] & tuplet(e,x)] …:||Δ Navigation and concurrent narration may be a benchmark for language design, but as remarked above, natural language generalizes on the language in use then in two dimensions. The first is displacement: the visual experience recorded and its accompanying narrative survive as artifacts in memory revisited and retold and also survive sometimes in representations and recitations to the public. It is indeed the same narrative and cinema that resurfaces in enumerable events of recitation and representation. Second, any such artifact may in fact be a composite of displacements from multiple sources. A montage shifts the locations and frames of reference for the original narration and concurrent navigation it reprises. Let Δ stand indifferently for the narrative discourse qua artifact or for an event of its recitation leaving it to context to disambiguate, and similarly φ stands for either any phrase qua artifact or for an utterance of it. Let φ© stand for the event of its original authorship—that is, for its utterance within the original event of narration and concurrent navigation from which it is excerpted. Such an event, recall, coheres as the behavior of a communicating and navigating agent. The authorship function λφ.φ© is thus undefined if the phrase φ answers to multiple authors. For the discourse Δ deriving from multiple authors, Δ©(φ©) ↔ Δ(φ). A revised semantics for now and for present tense privileges the original events of narration. Any token μ of now,27 or of present tense, uttered in recitation Δ refers to the time coincident with when this token was uttered during its original narration Δμ©. Accordingly, (99) is revised:28 (103) For discourse Δ and token μ of ⌜now(vi)⌝ in Δ such that ltΔμ©(μ) = δt, Σ satisfy u ↔ Within(nowΔμ©(δt), Σ(vi)) The semantics for now and the present tense do not alone warrant that their tokens in my speech be taken to refer to the present time at which this speech occurs. Rather, to be so understood will always require the further pragmatic understanding that this particular recitation is also the original one, Δ = Δ©, for which I am the narrator.
430
Chapter 9
For a recitation Δ that concatenates Δ1, …, Δm from multiple authors and a visualization Π that is a montage Π1, …, Πn of the eyewitness of many, it is an editorial duty of course to calibrate their clocks to a uniform time for Δ and Π. Local time in Δi calibrated to Δ is offset to reflect the temporal location of Δi in Δ. That is, (104) For 1 ≤ i ≤ m, ltΔΔi = ltΔi + k, where Δi begins at k in ltΔ. And, similarly, (105) For 1 ≤ i ≤ n, ltΠΠi = ltΠi + l, where Πi begins at l in ltΠ. In turn, UTCΔ is the function that maps the δtth moment that ltΔ clocks from the start of Δ to the corresponding historical moment in this particular recitation. UTCΠ does the same relating the running timestamp index ltΠ to the actual moment in which this reprised visualization is experienced. These functions are of no particular semantic interest, as they will differ for the matinee and evening performances of Ten Days That Shook the World, UTCamTen Days That Shook the World ≠ Ten UTCpm Days That Shook the World. What will be of interest relates the δtth cineramic moment to the original authorial moment UTCΔ©(δt) and the epistemic conditions that held then for the original assertion, and it relates the scene at the δtth cineramic moment to the moment UTCΠ©(δt) of its original projection of the events it is a scene of. Evening or matinee, UTCamTen Days That Shook the World© = UTCpmTen Days That Shook the World©, these are identical functions into the same 10 days of October 1917. Given the shifts in spatiotemporal location and frame of reference among the multiple authors and eyewitnesses from which Δ and Π derive, UTCΔ© and UTCΠ© are composites of the many origination functions: (106) For 1 ≤ i ≤ m, UTCΔ©Δi = λδt. ltΔΔi(Δi)(δt): UTCΔi©(δ−k). m UTC ∆ = ∪ i = 1 UTC ∆ ∆i . (107) For 1 ≤ i ≤ n, UTCΠ©Πi = λδt. ltΠΠi(Πi)(δt): UTCΠi©(δt−l). n UTCΠ = ∪ i = 1 UTCΠ Πi . For any narration Δ or visualization Π, elapsed local time and the passage of historical time while that narration or visualization is in progress are identical and the temporal distances covered measure the same: (108) |UTCΔ(δti) - UTCΔ(δtj)| = |LTΔ(δti) - LTΔ(δtj)| (109) |UTCΠ(δti) - UTCΠ(δtj)| = |LTΠ(δti) - LTΠ(δtj)| But, of course, the 10 minutes that elapse from one scene to the next may span days in the events that they are scenes of, as must be the case for a 95-minute movie about 10 days that shook the world. Apart from scene changes of venue, note that for any (eccentric) visualization Π that represents its subject in variable accelerated
Cinerama Semantics
431
motion (i.e., in slow motion or fast at varying rates), the effect of temporal dilation or compression means a variable coefficient of difference between temporal distance in the visualization and temporal distance in real time, the time of original projection: (110) |UTCΠ©(δti) − UTCΠ©(δtj)| = c()(|ltΠ(δti) − ltΠ(δtj)|) Given the cineramic setting, understanding a communication attends to the coordination among current recitation, what current scene it narrates, and original narration and projection. It does so to an extent that becomes plain when recitation and original narration are dissociated in counterfactual narration as If You Were There. If You Were There29 deploys the historical present tense for a historical narrator remote from the recitation in progress, as if narration and scenes narrated were relived memory. Note first that in using the present present tense, if the Dow Jones index is up, I may assert (111) with confidence given what is known about market fluctuation: (111) The Dow Jones is now up and soon down. (112) The Dow Jones is now up 248 points and soon down 365 points. On the other hand, even if the Dow Jones index is in fact up 248, the second clause of (112) implicates me in price-fixing or fortune-telling about events beyond my horizon. The adverbs now and then and similar expressions may order past events from within the speaker’s perspective on a moment in history from two decades ago, as in (113):30 (113) On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average was now / (at) first / at this time up a modest gain and then/later down a record-breaking 508 points by the closing bell. On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average was now / (at) first / at this time up a modest gain and soon down a recordbreaking 508 points that ended trading. Being in the past tense, there is nothing clairvoyant about (113)’s assertions. Yet in the historical present, the narrator becomes a contemporary of the events observed. If in that moment of observation the stock market is up, it is not down and it cannot be seen that even in the near future it will fall 508 points:
432
Chapter 9
(114) #On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and then/later down (a recordbreaking) 508 points (by the closing bell). #On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and soon down (a record-breaking) 508 points (that ends/ended trading). A subsequent token of the historical present advances to a subsequent moment of observation, also contemporary with the events observed, at which the stock market is no longer up and the crash is in plain sight: (115) On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and is then/later down (a recordbreaking) 508 points (by the closing bell). On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and is soon down (a recordbreaking) 508 points (that ends/ended trading). For a speaker reciting (114) today, the historical present tense is not merely an alternative tense for referring to past events, one with a special effect that makes more vivid for speaker and audience a history that is otherwise well known and well rehearsed. Rather, the historical present also imposes the epistemic condition of the time referred to—opening bell on October 19, 1987. Recitations today of (113) and (114) assert the same proposition, and yet (114) is infelicitous in that its narrator then navigating the floor of the NYSE has no grounds for her prediction. In understanding any given recitation, unless the displacement of its narrator is entertained, it is mysterious that epistemic conditions other than those contemporary with the recitation itself should ever condition the assertibility of what is recited. To allow for a displaced narrator, the semantics for a token of the present tense must not refer to the absolute time of its utterance. Its semantics refers instead to its relative position in the narrative discourse—that is, to when it occurs in discourse local time, the temporal frame of reference, 0 to The End, that the discourse qua artifact carries with it invariant throughout all its recitations. For the scene under narration, it suffices for their local time to be synchronized for the present tense in the narrative to indicate present goings-on in the concurrent scene. For the absolute historical frame of reference, within which the time of this recitation occurs in the twenty-first century, the present tense is no better than the past tense in fixing historical time,
Cinerama Semantics
433
as is made plain when “October 19, 1987” is omitted from both (113) and (114). The recitations of both (113) and (114) and (113)’s original authorship are today, but the date of Black Monday and the date of (114)’s authorship are unknown. The present tense pronounced in a recitation defers to the time of its original authorship, which is perhaps remote from the recitation now in progress and perhaps otherwise unknown. Temporal-frame adverbials modifying the (historical) present tense may nevertheless narrow the range for the original authorship. They are however rejected if they imply that the author is in two different places at once, even if no such thing is implied about the events being reported. Consider the audience at the 3-hour pageant History of the Universe from the Big Bang to the Mission to Mars:31 (116) OMG, it’s the Big Bang. In 2½ hours, it will be the Dawn of Man. (117) OMG, it’s the Big Bang. In 14 billion years, it will be the Dawn of Man. (118) OMG, it’s the Big Bang. In 2½ hours and 14 billion years later, it will be the Dawn of Man. (119) OMG, it’s the Big Bang. In 2½ hours, it’s the Dawn of Man. (120) OMG, it’s the Big Bang. In 14 billion years, it’s the Dawn of Man. (121) #OMG, it’s the Big Bang. In 2½ hours and 14 billion years later, it’s the Dawn of Man. Sentences (116)–(121) contain one adverb, in 2½ hours, measuring elapsed time in the frame of reference of visual experience and another adverb, in 14 billion years, for time elapsed in an (absolute) allocentric frame of reference for the events the scenes are scenes of. Like spatial frames of reference, what knowledge there is if any of an allocentric frame of reference affords the use of either egocentric (i.e., visualized) experiential time, in 2½ hours, or allocentric, actual time, in 14 billion years. Recall from the discussion of (8) and (9) that demonstrative reference to spatiotemporal location is not to an absolute but to the demonstratum’s address in the frame of reference intended from among the plenitude that intersect at the point demonstrated.32 This ambiguity is plain in this context in the reference of the demonstrative or indexical pronouns in such phrases as just now, then, at this moment, etc., which also refer into either the egocentric, experiential frame of reference or the allocentric one for the actual time of the events depicted. Rather than countenance ambiguity in the adverbs themselves, it is taken here to reside in an unspoken demonstrative pronoun, in 2½ hours / 14 billion years from then, translating it either as ‘[℩t : t = UTCΠ©(ltΠ(π0))]’ referring to the actual time 14 billion years ago to which the local time of scene π0 is mapped, or to the local time [℩t : t = (ltΠ(π0))] itself (or alternatively, to the actual time, [℩t : t = UTCΠ(ltΠ(π0))], of this reprise performance of π1):
434
Chapter 9
(122) [℩t : t = (ltΠ(π0))] in[t, π1, 2½ hours] (123) [℩t : t = UTCΠ©(ltΠ(π0))] in[t, π1, 14 billion years] An amnesiac otherwise intact is accurate in her sense of the time elapsed even as she does not know the date, and what she actually witnesses she knows to be coincident with her witnessing it. The latter cannot be said of recalled memory or documentary footage in reprise, in which the visualization may be at unknown remove from that which it is of. Moreover, temporal compression or dilation in memory or in cinema (as in (116)–(121)) means that the translation of elapsed time (Δt) in the visualization to elapsed historical time is a function other than identity, which may also be unknown (see (110)). Altogether, the (re-)viewer of a visual experience may have no knowledge toward its integration into any temporal frame of reference other than that of the visual experience itself, in which respect temporal integration and spatial, path integration are alike. In the future tense, sentences (116)–(118) assert that the Dawn of Man is 2½ hours in experiential time from what is now en-scène, from which it is also 14 billion years in cosmological time:33 (124) OMG, [∃σ0: now(σ0)][℩e0: Big Bang(e0)] (∃α∃f en-scène[e0, σ0, α, f]. [∃σ1: [℩t : t = (ltΠ(σ0))] in[t, σ1, 2½ hours]] will[σ0, σ1][℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]) OMG, it’s the Big Bang. In 2½ hours, it will be the Dawn of Man. (125) OMG, [∃σ0: now(σ0)][℩e0: Big Bang(e0)] (∃α∃f en-scène[e0, σ0, α, f]. [∃σ1: [℩t : t = UTCΠ©(ltΠ(σ0))] in[t, σ1, 14 billion years]] will[σ0, σ1] [℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]) OMG, it’s the Big Bang. In 14 billion years, it will be the Dawn of Man. The future according to (124) is a future according to local time in progress. The future according to (125) is, on the other hand, a future with respect to what this scene is a scene of. Despite the cosmological expanse of what that visual experience is a scene of, the viewer, her egocentric frame of reference, and the visual experience are presumed stationary. There is no frame of reference in play within which they have moved. Looking out from the very same fixed position, the same future is felicitously located as at 2½ experiential hours away and as 14 billion cosmological years away: (126) OMG, [∃σ0: now(σ0)][℩e0: Big Bang(e0)] (∃α∃f en-scène[e0, σ0, α, f]. [∃σ1: [∃σ1: [℩t : t = (ltΠ(σ0))] in[t, σ1, 2½ hours]] will[σ0, σ1][℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]), and [∃σ1: [℩t : t = UTCΠ©(ltΠ(σ0))] in[t, σ1, 14 billion years]] will[σ0, σ1] [℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]) OMG, it’s the Big Bang. In 2½ hours Δi, and 14 billion years later, [Δi it will be the Dawn of Man. Δi].34
Cinerama Semantics
435
In the present tense, (119)–(121) would appear to make similar assertions about the Big Bang and the Dawn of Man. The infelicity when both adverbs combine with the present tense thus wants explanation. It seems that the adverbs interact with present tense to impose contrary requirements on the narrator that end up inconsistent when they occur together. In (120), with in 14 billion years, the second sentence tokens the historical present and relocates the original narrator, as if you were there (see above), as a direct witness to the Dawn of Man (recall from (103) that now translates both adverb and present tense): (127) OMG, [∃σ0: now(σ0)][℩e0: Big Bang(e0)] (∃α∃f en-scène[e0, σ0, α, f]. [∃σ1: [℩t : t = UTCΠ©(ltΠ(σ0))] in[t, σ1, 14 billion years]] now(σ1) [℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]) OMG, it’s the Big Bang. In 14 billion years, it’s the Dawn of Man. That is, according to (127), in 14 billion cosmological years from the events now en-scène (i.e., the Big Bang), the original narrative resumes with the Dawn of Man (which we now experience rebroadcast as this recitation continues). The two scenes now unfolding 2½ hours apart splice together scenes from 14 billion years apart. Analogously, for (119), the two tokens of now, now a few seconds apart, are lifted from the two scenes’ narration in which they occur 2½ hours apart: (128) OMG, [∃σ0: now(σ0)][℩e0: Big Bang(e0)] (∃α∃f En-scène[e0, σ0, α, f]. [∃σ1: [∃σ1: [℩t : t = (ltΠ(σ0))] in[t, σ1, 2½ hours]] now(σ1) [℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]) OMG, it’s the Big Bang. In 2½ hours, it’s the Dawn of Man. But when both these adverbs come to modify the present tense of the second sentence in (121), they come to require the narration to both keep pace with this recitation and visual experience and with cosmological time: (129) #OMG, [∃σ0: now(σ0)][℩e0: Big Bang(e0)] (∃α∃f en-scène[e0, σ0, α, f]. [∃σ1: [∃σ1: [℩t : t = (ltΠ(σ0))] in[t, σ1, 2½ hours]] now(σ1) [℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]), and [∃σ1: [℩t : t = UTCΠ©(ltΠ(σ0))] in[t, σ1, 14 billion years]] now(σ1) [℩e2:Dawn of Man(e2)] ∃α∃f en-scène[e2, σ1, α, f]) OMG, it’s the Big Bang. In 2½ hours Δi, and 14 billion years later, [Δi it’s the Dawn of Man. Δi].35 The logical form in (129) for the infelicitous (121) entails the logical form in (126) for (118). The latter says in effect that in 2½ hours from this scene of the Big Bang, there is a subsequent scene that is of the Dawn of Man at 14 billion years after the Big Bang. The second token of now in (129) says further that the time of this subsequent scene is itself also 14 billion years later, coinciding with the Dawn of Man.
436
Chapter 9
Therein is (121)’s flaw. One adverb would shift the location of the narrator, or shift to another narrator 14 billion years away, and the other would keep her in situ for 2½ hours. As (118) and (121) share a context, today’s showing of History of the Universe from the Big Bang to the Mission to Mars and the facts about the Big Bang and the Dawn of Man, the only account for the contrast between future tense and present tense is in the conditions that the latter imposes on a displaced narrator absent from today’s showing. As was to be shown, understanding cinerama attends to the coordination among current recitation, what current scene it narrates, and original narration and projection. 9.4
Cinema verité
Visual navigation is twofold: first is scene analysis, the concurrent, inverse projection of the streaming, incoming visual scene onto an egocentric frame of reference for the ambient environment. Second, path integration (Aloimonos 1997; Gallistel 1990; Golledge 1999a, 1999b; Redish 1999) is the translation of the egocentric frame of reference and the features perceived therein to allocentric frames of reference, as a result of which transient perception ends in a map for an environment that is invariant to the conditions of observation. I see new construction several blocks ahead of me (scene analysis) and recognize that Manhattan has a new tower at 620 Eighth Avenue (path integration). Scene analysis affords one the orientation to navigate the ambient space of a city block or so. In that sense, a scene will be said to be a scene of the ambient space it projects. Only after path integration, the experience of which may be delayed,36 is the same scene, without alteration to it as such, a scene of the larger space Manhattan, with which it now comes into proper alignment. The scene is at first orienting or navigational for a city block, and then the same scene is so for a city island, a change in its intentional relation to the beholder. In visual navigation solo, distinct scenes—that is, distinct events of projection from a mobile perspective—are of distinct spatiotemporal regions. Any redaction of such projections into, say, a continuous atlas of the celestial hemisphere from a fixed point of view or a survey of some landscape in transit across it conforms to this condition on representation: distinct fragments of what purports to be an atlas or survey are of distinct spatiotemporal regions. Narrative cinerama may contain multiple narrators, and thus any spatiotemporal condition relating scene and what it is a scene of must anticipate shifts in perspective. It cannot be that the scenes narrated will all cohere as projection from the perspective of a single navigator. There is, however, a cooperative principle when multiple perspectives are engaged with the intention that their paths are to be integrated to the same frame of reference. If twelve spies are sent to scout out the land of Canaan, they are organized so as to survey different territories or track different tribes. Then, of any cinerama that is a montage of scenes
Cinerama Semantics
437
they have witnessed, it can be said that distinct scenes oriented to the same frame of reference are of distinct subjects with distinct coordinates in that allocentric frame of reference. Their multiple perspectives would be disorienting and their efforts partly in vain if they should end up converging on the same subjects. This anticonvergence condition so-called (section 9.4.1) will eventually (section 11.0.2) be invoked to explain a constraint on counting (section 11.0.0) where sentences are judged defective if in the scenes reported, anything counted n in a given scene is tracked to a scene where it would be counted n−k, for k > 0. That is, what is first counted n should never be seen to converge on fewer than n. It happens that the anticonvergence condition on the scenery of visual navigation is breached when it tracks from one scene where the things counted are n to another where they are counted fewer. The condition on counting will follow from the presupposition that the scenes quantified over from multiple perspectives are just those that conform to this cooperative principle. Anticonvergence, as remarked above, is relative to a frame of reference. It is imposed only on scenery path-integrated and intended to be orienting for the same frame of reference. To count n when there exists a frame of reference counting fewer can be bad. How bad it is varies with the extent to which the speaker can plausibly represent herself as ignorant and without intention to refer to the offending frame of reference. Prior to the statement of the anticonvergence condition in section 9.4.1, section 9.4.0 lays out under a single heading most of the apparatus for spatial orientation and navigation, holding some of it in reserve until the phenomena it is designed for are introduced, and section 9.4.2 is about the scenery for counting. 9.4.0
Scenes, projections, and frames of reference for spatial orientation and navigation
There must be a fact of the matter concerning the coordinate systems and frames of reference in use in human navigation,37 and it may be that human cognition adapts coordinate system and frame of reference to the topography navigated, all of which I abstract away from in amateurishly borrowing from amateur astronomy the horizontal coordinate system of altitude, azimuth, and radial distance . An observatory o is persistent through time and may be transported from one point of view in absolute space or a given frame of reference to another. An observatory is an arbitrarily small sphere of radius r and center π, with a celestial horizon and celestial meridian that intersect at and an intrinsic handedness for positive and negative declination from , thus defining a coordinate system for egocentric frame of reference fo for o. For an observer at the observatory’s center π, positive altitude is degrees from the celestial horizon toward the zenith; negative altitude is degrees from the celestial horizon away from it (i.e., toward the nadir). Azimuth increases eastward along the horizon and decreases westward, as, for example, in the coordinate system centered on the center of the earth and oriented
438
Chapter 9
to align the zenith with polar north, the celestial meridian with the prime meridian, and the celestial horizon with the equator. The coordinate system for any observatory o projects to all of absolute space, but given its mobility, it is that at any time t, it assigns every point of space an ordered triple as its address at t. A frame of reference is any (sub)space and coordinate system for it containing at least one landmark. A landmark for a frame of reference is an object with spatial extension every point of which remains at a fixed address and thus locks the orientation of the space and coordinate system. An egocentric frame of reference has exactly one landmark, the centroid of which coincides with the center π of the frame of reference. Manhattan and the interior of a taxicab are frames of reference with many landmarks. An artifact of this definition is that there are infinitely many distinct frames of reference that share the same space Manhattan and all their landmarks, varying only with the location of π within Manhattan and the orientation of the coordinate system with respect to it.38 There is little reason to refer to coordinate systems apart from the spaces they are etched into; hence, frames of reference are spaces. There will, however, be occasion to refer to the (sub)spaces σ that are invariant despite the coordinate systems imposed on them. Two frames of reference are of the same (sub)space just in case they coincide when they are path-integrated into a frame of reference large enough to include them both. There is a spatial vocabulary for the location of objects and events in frames of reference. I will take it as primitive that objects and events with spatial extension overlap at any given time some regions of space and not others, conceding that vagueness in what constitutes an object or event carries over into vagueness about the space it occupies. An object or event is said to be ‘At’ a region of space addressed in a frame of reference just in case the space it occupies is properly contained within that region: (130) At(αf,ζ,t,f) iff object or event ζ is within region αf of f for the duration of t, a moment or interval of time. (131) At[AF,Z,T,F] iff ∀αf(AFαf → ∃ζ∃t∃f(Zζ & Tt & Ff &At(αf,ζ,t,f))) & ∀ζ(Zζ → ∃αf∃t∃f(AFαf & Tt & Ff &At(αf,ζ,t,f))) & ∀t(Tt → ∃αf∃ζ∃f(AFαf & Zζ & Ff &At(αf,ζ,t,f))) & ∀f(Ff → ∃αf∃ζ∃t(AFαf & Zζ & Tt &At(αf,ζ,t,f))) A region αf is a closed ball or neighborhood in a Euclidean space, and object or event ζ is within it just in case every point of ζ is a point in the interior of αf. We can equally well speak of a region of a scene αṡ, a 2-D space, into which objects or events are projected.
Cinerama Semantics
439
As subspaces and frames of reference are otherwise empty moving targets, landmarks prove necessary to fix their contours and to get a fix on any one of them f, if one is to make determinate reference to it. Consider an airborne squadron. Its formation is a subspace and frame of reference with coordinates (of relative position) in use in pilot communications. As landmarks are defined above, when one aircraft peels away from the squadron it leaves behind a new, different frame of reference with one fewer landmark. This is too stringent to the extent that one judges that the same squadron in the same formation remains in flight minus one aircraft. This is just to remark that squadrons, formations, subspaces, and frames of reference inherit all the problems of gestalt object recognition and perseverance under change, which I have no intention of addressing. But let it be introduced that in frames of reference with multiple landmarks, some may serve as such for different periods of time: (132) Moment(ṫ ) ↔def ∀t∀t′((overlap(ṫ,t) & overlap(ṫ,t′)) → overlap(t,t′)) (133) Landmark(ζ,t,f ) iff ∀αf ∀ṫ ∀ζ′( (Moment(ṫ) & overlap(ṫ,t) & ∀ζ″(overlap(ζ″, ζ′) → overlap(ζ″, ζ)) & At(αf,ζ′,ṫ,f)) → ∀ṫ((Moment(ṫ) & overlap(ṫ,t)) → At(αf,ζ′,ṫ,f ))) An object or event ζ is a landmark for f for the duration t just in case it occupies a fixed address αf in f for all of t and fixed orientation to f. The latter condition ensures that a frame of reference is in rotation with respect to a frame of reference that contains it only if its landmarks are also in motion with respect to the larger frame of reference. That is to say that her egocentric frame of reference cannot spin without the ballerina herself pirouetting. With the overt spoken vocabulary, recall (section 9.1.0) that its use expresses no determinate thought in (134)–(136) without definite reference to a frame of reference, either by demonstration (134) or description (135), and that such reference is grammatically constrained to be the same across the sentence. (134) The ballerina will pirouette with her partner on the left to the middle of the ice, do a triple somersault, and pirouette with her partner on the left the rest of the way. (135) The left chair and (the) right chair on Jorge’s visit to the library are not the left chair and (the) right chair on Luis’s. Recall from section 9.1.1, too, that the spatial vocabulary is also temporal and must support a usage in (136) that in effect means while left and while right: (136) The left (skater) and (the) right skater are zigzagging/zigzag at decreasing frequency.
440
Chapter 9
In addition to these parameters, natural language left proves to be a comparative that means “more leftwards”, for which an additional parameter ζ’ for the comparison class is provided in ‘left(e, ζ, f, ζ’)’: (137) left(e, ζ, f, ζ′) iff W(e, ζ) & ∃αf ∃αf′(At(αf,ζ,t(e),f) & At(αf′,ζ’,t(e),f) & r(αf) * cosine(alt(αf)) * cosine(az(αf)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) Within a Cartesian coordinate system, a is to the left of b just in case a’s x-coordinate is less than b’s x-coordinate. I hope (137) gives the correct translation from celestial coordinates to Cartesian coordinates.39 The comparative meaning of the directional becomes apparent in a featureless landscape absent determinate celestial meridian, and the infelicity that then results when there is nothing as imagined for (138) to be to the left of: (138) M (139) #The left object. #The object on the left. (140) The right-leaning object. In contrast, the same featureless landscape, given the orientation of the observer, supports reference to the more leftward object even absent an absolute left side: (141) M1 M2 (142) The left object is Messier Number 1. A point of grammar also urges assimilation of the directionals in (144) to the comparatives in (145), where the comparison class for the inner adjectives is {M1, M2, M3, M4} and {M3, M4} for the outer adjectives: (143) M1 M2 M3 M4 (144) The left right object is Messier Number 3. (145) The large small object is Messier Number 3. If, without comparison to other locations, the meaning of a directional were simply location within the appropriate hemisphere of an understood frame of reference, the felicitous reference in (144) to M3 would require that right refer to a frame of reference for all four objects with M1 and M2 in its left hemisphere and M3 and M4 in its right. In turn, left zooms to a smaller frame of reference including only M3 and M4 in different hemispheres. Yet this variation in frame of reference from one token of a directional adjective to the next is exactly what is proscribed in (134). Rather, then, the frame of reference is constant in (144), and it is the comparison class that varies, as (145) already attests it must. In (137), there is (concealed) quantification over the subspaces αf of frame of reference f, and it is this same frame of reference that aligns leftness as fixed by the
Cinerama Semantics
441
translation of its celestial coordinates into Cartesian coordinates. Some arcane examples suggest that two distinct frames of reference may be in play. One has in mind to locate object or event within a subspace αf of f but one designates subspaces of f according to their location within another frame of reference, the egocentric one in most cases. If so, (137) should be modified for the additional parameter: (146) left(e, ζ, f, ζ′, g) iff W(e, ζ) & ∃αf ∃αf′∃αg ∃αg′( At(αf,ζ,t(e),f) & At(αf′,ζ’,t(e),f) & (At(αg,αf,t(e),g) & At(αg′, αf′,t(e),g) & r(αg) * cos(alt(αg)) * cos(az(αg)) < r(αg′) * cos(alt(αg′)) * cos(az(αg′)) The suggestive examples start from the following contrast, in which locations are fixed in different frames of reference via their leftness in the (superimposed) egocentric frame of reference. (147) These are the adventures of Supermensch, able to move tall buildings in a single bound: “Now it’s on my left. [Pirouetting 180°.] Now it’s on my right.” (148) #These are the adventures of Supermensch, able to move tall buildings in a single bound: “Now it’s on the left. [Skating to the opposite side of the square.] Now it’s on the right.” The joke in (147) is that the literal and felicitous truth it reports is no superhero feat. Yet the implicature of a change of state from now to now is satisfied in that Supermensch has rotated his own egocentric frame of reference to relocate the building within it. My left and my right refer to different regions within the same space under the same coordinate system and partition into left and right hemispheres. In contrast, (148) is neither synonymous, despite the same understanding of leftness and rightness, nor felicitious out of the blue. In (148), the frame of reference in which the building is located is one in which all the buildings and urban fixtures are landmarks—that is, always at the same address within it—rather than the frame of reference for (147), where Supermensch is its only landmark. In the urban frame of reference, the implicature of a change of state (now here … now here) is disappointed, and it is furthermore somewhat perverse to refer to the same location under different description especially when orienting attention to that location: (149) #Now the lawn mower is in the shed close to the bicycle. [The bicycle is moved.] Now the lawn mower is in the shed far from the bicycle. Out of the blue, the left and the right refer to sections of the urban frame of reference, but of course nothing precludes after a pause the conclusion that the speaker intended the egocentric frame of reference, in which case (148) comes to be heard to say the same thing as (147). Just so, the left and the right can certainly be understood to refer into the urban frame of reference, stumbling into an
442
Chapter 9
infelicity, and my left and my right cannot, referring only into Supermensch’s egocentric frame of reference. Of course, leftness and rightness are throughout egocentric to Supermensch. These observations do not yet prompt accommodation of an extra parameter in the meaning of directionals. The left and my left are nominal expressions. Glossed “the left region there [the urban frame of reference]” and “the left region of mine,” the tacit noun is occasion for the extra parameter. It is then these regions that are the objects or events ζ said to be left of Supermensch: (150) a. left(e, αSupermensch, fSupermensch, Supermensch) b. left(e, αNYC, fSupermensch, Supermensch) The question is whether the contrast seen in (147) and (148) recurs when an overt nominal, the left building, my left building, fixes the locatum to be objects or events other than regions of space. So, suppose that Supermensch is beset on both sides by Good Humor ice cream trucks, his attackers, set to destroy his diet: (151) “Good Humor is on board my left attacker, and [spinning around] OMG, it’s now on board my right attacker.” (152) #“Good Humor is on board the left attacker, and [spinning around] OMG, it’s now on board the right attacker.” The locatum ζ is a truck. It is only the concealed quantification over subspaces αf in the meaning of left in (137) and (146) that affords reference to subspaces of the egocentric frame of reference on the one hand and to subspaces of the urban frame of reference on the other. In the latter instance, as this diverges from the frame of reference fixing leftness, it would suggest the extra parameter g in (146). Note that for any naive notion of absolute space, the building and ice cream truck have not moved, although it happens that absolute space is variously parsed into hemispheres at a moment of utterance according to the speaker’s perspective on them. The contrasts observed require for their explanation location in frames of reference that are themselves in rotation or translation within absolute space. It is only with respect to such a rotation that the building and ice cream truck change position in the examples that satisfy the implicature of such a change. For later developments in later sections, several variants of (137) and (146) will be convenient, introduced here so they may be compared while the meaning of spatial vocabulary is under discussion. Spatial position is temporary—while left—and so (137) and (146) have it that ζ is on the left while participating in e for the duration of e. Let’s just say that e itself is on the left, and if there is a future need to say it about ζ too, it will be said that ζ participates in a left-placed event e:
Cinerama Semantics
443
(153) left(e, f, ζ′) iff ∃αf ∃αf′(At(αf,e,t(e),f) & At(αf′,ζ′,t(e),f) & r(αf) * cosine(alt(αf)) * cosine(az(αf)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) (154) left(e, f, ζ′, g) iff ∃αf ∃αf′∃αg ∃αg′( At(αf,e,t(e),f) & At(αf′,ζ′,t(e),f) & (At(αg,αf,t(e),g) & At(αg′, αf′,t(e),g) & r(αg) * cos(alt(αg)) * cos(az(αg)) < r(αg′) * cos(alt(αg′)) * cos(az(αg′)) Henceforth, I will suppress this arcane context and its two parameters, f and g, for frames of reference, relying only on (153). There is also to be an absolute notion of leftness, not only the comparative leftward, designating location in the left hemisphere: (155) left(e, f) iff ∃αf ∃αf′(At(αf,e,t(e),f) & αf′ = f & r(αf) * cosine(alt(αf)) * cosine(az(αf)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) So far there are only reductions in the number of open parameters. It will however be convenient to refer to the region α that e occupies on the left of f: (156) left(e, α, f) iff ∃αf′(At(α,e,t(e),f) & αf′ = f & r(α) * cosine(alt(α)) * cosine(az(α)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) Given some understanding, such as the above, of frames of reference and orientation within them, visual navigation within them depends on a relation to scenes. For any moment t, observatory o with center π and egocentric frame of reference fo, a panorama is the projection with depth perception (2.5-D) at t of 3-D space onto the interior surface of o, the result, as one might imagine, of an instantaneous, omnidirectional survey of one’s surroundings from point of view π. Projection is meant to be a dated event, and projection at a given moment t—that is, at a moment in universal, absolute time (utc; see (96))—is determinative of the scene projected. A scene such as might result from actual visual experience at time t and oriented with respect to a frame of reference fo is a partial view cropped from a panorama at t for fo. Scenes inform spatial orientation even for just the egocentric frame of reference fo only if their location is known with respect to determinate t and fo. Imagine an atlas of scenes (from memory and experience) that inventories views of everything there is from a certain point of view. Included in these scenes could be encyclopedic knowledge of the distance of objects from the point of view and knowledge of their relative position along lines of sight that an object is close behind or far ahead of another. Yet without their location in the frame of reference, this atlas would no more assist spatial orientation than a world atlas from which have been erased all signs of latitude and longitude.
444
Chapter 9
A (still-life) scene ṡ at a moment utc t, a fragment of a panorama, is the projection at t from point of view π for egocentric frame of reference fo, determined by a line of sight l = fo for some altitude and azimuth relative to fo, and the observer’s optics o including in particular field of view and resolution, about which more is said below. All visual experience, it is supposed, is concurrent with some awareness of the line of sight’s altitude and azimuth in the observer’s egocentric frame of reference (e.g., a telescope is known to be at 90° azimuth with respect to the 0°-position etched on its mount, whether or not the mount is aligned with true north). Within a scene, location may be addressed with coordinates in “local time”, as positive or negative declination from the scene’s line of sight, . It can then be said of an object that appears in two distinct scenes ṡ0 and ṡ1 of far-flung events that it is in the “same location” in both in that its local-time coordinates are the same—that is, for some m,n, the object is at ,lṡ 0fo> at t(ṡ0) and at ,lṡ 1fo> at t(ṡ1). This proves useful in the comparison of scenes said to arrange their subjects in the same way (see sections 15.0.0–15.0.1). Projection is a function from the above parameters to scenes: (157) ṡ = projection(t,π, fo,l,o) For all scenes ṡ projected at utc t for an egocentric frame of reference fo, π = fo. It suffices for spatial orientation and location within the egocentric frame of reference to be aware of the line of sight l = fo of any projection. For allocentric frames of reference, the coordinates for π may vary with the time t that ṡ is projected. Note that any event of observation is projection for many frames of reference, among which is always the egocentric frame of reference for the observer o at π, with respect to which line of sight l = fo is specified. For allocentric frames of reference within which the event of observation occurs, the projection is determined by two parameters, the coordinates f for π at time t within the allocentric frame of reference and the orientation at that time of the egocentric frame of reference for o. The latter is fixed by a compass on board o that indicates the heading fo in the egocentric frame of reference fo to the allocentric frame of reference f (i.e., the origin for its coordinate system). This heading, abbreviated N of , suffices to orient the egocentric fo to the allocentric f. In the case that f = fo—that is, when only orientation to the egocentric frame of reference itself is at issue—stipulate that N ofo = 〈 90, 0 〉 fo , identifying egocentric N ofo with the zenith of the egocentric frame of reference. Then, for any arbitrary frame of reference f and an event of observation that occurs at utc t with egocentric frame of reference fo, (158) ṡ = projection(t,πf, f, lfo, N of , o), where πf are the coordinates for π in f and lfo are the coordinates for l in fo, and N of is the heading in egocentric coordinates fo toward f’s center.
Cinerama Semantics
445
Location, line of sight, optics, and optical field are sufficient to determine a scene. The cinematic conditions, the anticonvergence condition in particular, are however about the spaces that scenes subtend, an explicit parameter for which now joins those already listed: (159) ṡ = projection(t, αf, πf, f, lfo, N of , o), where πf are the coordinates for π in f, lfo are the coordinates for l in fo, N of is the heading in egocentric coordinates fo toward f’s center, and αf are the coordinates in f for the points of the space that ṡ subtends. With a view toward the path integration of scenes into allocentric frames of reference, the coordinates αf for the space subtended are those of allocentric f, although of course that space also has coordinates in the egocentric frame of reference fo. It will be useful to have in hand functions that extract a scene’s projection parameters for a given frame of reference f: (160) t( sɺ) = t ↔ df ∃π f ∃α f ∃f ∃l fo∃N of ∃o sɺ = projection(t, α f , π f , f , l fo, N of , o) σ f ( sɺ) = α f ↔ df ∃π f ∃l fo∃N of ∃o sɺ = projection(t, α f , π f , f , l fo, N of , o) π f ( sɺ) = π f ↔ df ∃t∃α f ∃l fo∃N of ∃o sɺ = projection(t, α f , π f , f , l fo, N of , o) l( sɺ) = l fo ↔ df ∃t∃α f ∃π f ∃f ∃N of ∃o sɺ = projection(t, α f , π f , f , l fo, N of , o) o( sɺ) = o ↔ df ∃t∃α f ∃π f ∃f ∃l fo ∃N of sɺ = projection(t, α f , π f , f , l fo, N of , o) (161) frame( sɺ, f ) ↔ df ∃t∃α f ∃π f ∃l fo∃N of ∃o sɺ = projection(t, α f , π f , f , l fo, N of , o) As remarked above, any event of observation is projection for many frames of reference. Hence no function in (160) refers to the frame of reference for the scene; there is instead the relation in (161) that holds between a scene and the frames of reference for which it is projected. Note that the time of a scene, its observer, and the observer’s line of sight in projecting that scene are not contingent on the frame of reference. Visual navigation through a frame of reference f, pilotage, depends on the inverse projection from scene ṡ to utc time t and coordinates πf for point of view, to determine one’s current position in f, and to Nf to get a fix on one’s heading within f. As a scene is always coincident with a particular event of observation, the inverse of projection is also a function relative to the choice of f, projection–(ṡ,f) = (t(ṡ),σf(ṡ),πf(ṡ), f, l(ṡ), N of , o(ṡ)). Many numerically distinct scenes so-called may however be indiscernible, such as all those from inside a rubber ball (represented by the gray bar below): (162)
446
Chapter 9
The navigator is in trouble triangulating her position at t inside a rubber ball f since she finds no scenes S in her visual experience and no point of view πf for which she knows that (163) ∀ṡ∃αf∃lfo(Sṡ → ṡ = projection(t, αf,πf, f, lfo, N of , o)) The perceptual experience crucial for navigation is of course cinematic, comprising many scenes at various times from various points of view and lines of sight under varying conditions of observation, all of which is subject to continuity conditions on the flow of time and the observer’s positions and lines of sight. A cinematic scene sɶ may change at any of the moments LTsɶ( sɶ) in its duration start (δt0 sɶ ) to finish at (δtFINISsɶ ) (see section 9.3): (164) Moment(ṫ) ↔def ∀t∀t′((overlap(ṫ,t) & overlap(ṫ,t′)) → overlap(t,t′)) (165) ∀tɺ (LTsɶ ( sɶ)(tɺ) ↔ def Moment(tɺ) & δt0 sɶ ≤ tɺ ≤ δtFINISsɶ ) A cinematic scene sɶ is a continuous function from LTsɶ( sɶ) to scenes ṡ coinciding with the (imaginary) visual experience of an observer o in motion navigating through any frame of reference f, of which it can be said that (166) i. The cinematic scene is the perception of a single observer, λt. LTsɶ ( sɶ)(t ): o( sɶ(t )) = λt. LTsɶ ( sɶ)(t ): o; ii. A cinematic scene is the projected visual experience of continuous, uninterrupted navigation, λt. LTsɶ ( sɶ)(t ): t( sɶ(t )), λt. LTsɶ ( sɶ)(t ): π f ( sɶ(t )) , and λt. LTsɶ ( sɶ)(t ): l( sɶ(t )) are continuous, differentiable functions.40 Editing a cinematic scene, one has occasion to refer to extended excerpts from it, for which the above functions are extended to temporal intervals: (167) ∀t (LTsɶ ( sɶ)(t ) ↔ def ∀tɺ ((Moment(tɺ) & overlap(tɺ, t )) → δt0 sɶ ≤ tɺ ≤ δtFINISsɶ ) (168) For any t such that LTsɶ ( sɶ)(t ) & ¬Moment(t ), sɶ(t ) = def λtɺ.Moment(tɺ) & overlap(tɺ, t ): sɶ(tɺ) It will be equally useful to have in hand functions that extract a cinematic scene’s projection parameters for a given frame of reference f: (169) t( sɶ) = λ t. LTsɶ ( sɶ)(t ): (℩tUTC)(∃α f ∃π f ∃f ∃l fo∃N of ∃o sɶ(t ) = projection(t UTC, α f , π f , f , l fo, N of , o)) σ f ( sɶ) = λt. LTsɶ ( sɶ)(t ) : (℩αf)(∃π f ∃l fo∃N of ∃o sɶ(t ) = projection(t, α f , π f , f , l fo, N of , o)) π f ( sɶ) = λt. LTsɶ ( sɶ)(t ): (℩πf)(∃α f ∃l fo∃N of ∃o sɶ(t ) = projection(t, α f , π f , f , l fo, N of , o)) l( sɶ) = λt. LTsɶ ( sɶ)(t ): (℩lfo)(∃α f ∃π f ∃f ∃N of ∃o sɶ(t ) = projection(t, α f , π f , f , l fo, N of , o)) o( sɶ) = λt. LTsɶ ( sɶ)(t ): (℩o)(∀t∀αf∀πf∀f ∀lfo ∀N of ∀o′ ( sɶ(t ) = projection(t, α f , π f , f , l fo, N of , o′) → o = o′))
Cinerama Semantics
447
(170) frame( sɶ, f ) ↔ df ∀t(LTsɶ ( sɶ)(t ) → ∃α f ∃π f ∃l fo∃N of ∃o sɶ(t ) = projection(t, α f , π f , f , l fo, N of , o)) The temporal function ‘t( sɶ)’ tracks the cinematic scene’s internal clock to the actual time of its projection. Similarly, the other functions track the cinematic scene’s internal clock variously to the addressing αf for the space subtended in the frame of reference f, the addressing in f for the points of view πf within it from which the cinematic scene is projected, and the lines of sight lfo in the observer’s egocentric frame of reference. For a scene ṡ, a frame of reference f from which it is projected, ṡ = projection(t(ṡ),σf(ṡ),πf(ṡ), f, l(ṡ), N of , o(ṡ))), and some things X there (i.e., At[αṡ,X,t(ṡ),ṡ] & At[αf,X,t(ṡ),f]), the relation between their location αṡ in the scene and their location αf in 3-D f is nontrivial, dependent not only on the quality of resolution and its possible variation across the visual field but also on the distortion characteristic of any particular kind of perspectival projection. Yet (an approximation to) that relation must be known for any projection in the observer’s cognitive repertoire; a scene aids spatial orientation and navigation only when αf is recovered from αṡ. Recall from section 9.2 and its sentences about diamond solitaires still being on the left or again being on the left that the semantics for still and again41 included a relation en-scène (171), which can now be defined as in (172)–(173): (171) en-scène[E, Σ,A,F] ↔df For frames of reference F and addresses A, events E are at A in F & scenes Σ are (projections) of A in F. (172) en-scène(e, sɶ , αf,f) ↔df At(αf, e, t(e), f) & ∀ṫ((moment(ṫ) & overlap(ṫ, t(e))) → f ɶ ɺ sɶ(tɺ) = Projection(ṫ, αf, π f ( sɶ)(tɺ) , f, l( sɶ)(tɺ), No( sɶ ) , o( s )(t ) ))) ɶ F,F] ↔df (173) en-scène[E,S,A ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ∀e(Ee → ∃sɶ ∃αf ∃f( Ss ɶ ɶ → ∃e∃αf ∃f( Ss ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ∀sɶ ( Ss ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ɶ ∀αf(AFαf → ∃e∃s ∃f( Ss ɶ ∀f(Ff → ∃e∃sɶ ∃αf ( Ssɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) The en-scène cinematic scene sɶ is allowed according to (172) a moving camera, provided there is a space αf always in view on which event e is staged. Thus, the en-scène relation happens to be about observation of a fixed location, and its pluralized version is about several such scenes, their fixed locations, and events that occur there. Navigation will of course import different relations between scenes and the spaces projected. 9.4.1
Path integration and anticonvergence
To this point in the discussion, scenes, defined as projections of spaces under certain conditions of projection, have implied no scene analysis. A scene, as it were, is just
448
Chapter 9
that array of light projected at time t from point of view πf in frame of reference f subtending space αf . For such scenes to hold any import for semantics requires something more intentional: the observer of a scene apprehends that it is a scene of some things and not others. (174)
For the visual experience that constitutes the Necker cube flicker (174), there are (at least) two scenes and two different times, ṡ1 = projection(t(ṡ1), αf,πf, f, lfo, N of , o) and ṡ2 = projection(t(ṡ2), αf,πf, f, lfo, N of , o), such that ṡ1 is of one cube (and not the other) and ṡ2 is of the other (and not the one). There is usually at least some overlap between the space subtended in a scene and the space occupied by that which it is a scene of. An extreme close-up of a hair follicle on an eyebrow may not be recognized as such or as anything else.
From there, one may zoom out to a scene that is of Groucho Marx and recognized as such although the scene contains little of him and perhaps only eyebrows.42 The same can be said of events: scenes of battles can be taken to be scenes of the war although no scene is the full pageant. But for this discussion, the relevant scenes typically contain what they are scenes of. Recall that scene analysis is the concurrent, inverse projection of the streaming, incoming visual and somatosensory scene onto an egocentric frame of reference for the ambient environment. Recall also that path integration (Aloimonos 1997; Gallistel 1990; Golledge 1999a, 1999b; Redish 1999) is the subsequent, often partial translation of the egocentric frame of reference and
Cinerama Semantics
449
the features perceived therein to allocentric frames of reference, as a result of which transient perception ends in a map for an environment that is invariant to the conditions of observation. I see new construction several blocks ahead of me (scene analysis) and recognize that Manhattan has a new tower at 620 Eighth Avenue (path integration). Momentarily disoriented emerging from the subway, the scenes before and after this hiatus differ only in that the former is orienting for only the egocentric frame of reference, and the latter, for both the egocentric frame of reference and Manhattan. Navigating with restricted “viewpoint” as when navigating narrow streets, path integration is the process by which local translations and rotations, and the perspective views encountered, whether continuous or discrete, are integrated to provide a current estimate of position and orientation within a larger spatial frame of reference (Gallistel 1990; Loomis et al. 1999). Navigational scenes are annotated with addresses in coordinates for the frames of reference for which they are orienting. At a minimum, a navigational scene is annotated for the egocentric frame of reference under which it is projected and subject to scene analysis. In much point-to-point wayfinding, through narrow streets or subways, scenes are further integrated into a proper succession of frames of reference, street to street, subway station to subway station, without their integration into a single comprehensive frame of reference. As a reflex of path integration, further annotations enrich the scenes with coordinates for the frames of reference for which they have become orienting, such as they are at any given moment along the way. It will be convenient to regard as distinct, scenes that are distinguished only by their annotation for this or that frame of reference. Subjects of the same visual experience will then be said to perceive the same scenes just in case the subjects have path-integrated these scenes to the same frames of reference. As an illustration, consider an interior space and any exterior space that properly contains it. Of course any interior for which there is an address in a frame of reference for that space also has an address in any frame of reference for any of the exterior spaces containing it. It is not a fact of projection that a scene is orienting for an interior that a subject finds herself in but not for the exterior. It is rather knowledge about visual navigation and its cognition that informs us that the scenes from within an interior that a narrator refers to are likely to be orienting for her for that interior and not for an exterior one that contains it. This relation of a scene’s orientation for a space or frame of reference was crucial in section 9.2 for the semantics of still and again. And this is always the question with any scene referred to or evoked in discourse—For whom is it orienting and for which frames of reference? Even where a discourse is obviously not the narration of a navigation in progress, an original authorship under such conditions is assumed. As in section 9.3, the historical present tense refers to the time displaced from the present present that yokes the original authorship to a concurrent visual experience. The original
450
Chapter 9
author’s presence is felt twice: first, assertion in the historical present tense is felicitous only if it could have been said then with what was known then, and second, the narrative is again infelicitous if temporal-frame adverbials modifying the historical present tense imply that the distant author is in two different places at once. Scenes referred to in discourse are presumed to cohere as the visual experience of some navigating agent, even if she is not present, for whom the scenery was input to her visual navigation and thus orienting for the frame of reference that was the target of her path integration: f (175) nav-orient(ṡ,f) ↔def ṡ = projection(t(ṡ),σf(ṡ),πf(ṡ), f, l(ṡ), No( sɺ ) ,o(ṡ)) & f o(ṡ) sees at t(ṡ) that ṡ = projection(t(ṡ),σf(ṡ),πf(ṡ), f, l(ṡ), No( sɺ ) ,o(ṡ))
Absent profound pathology, I assume that nav-orient(ṡ, fo(ṡ)), i.e., that visual experience is at least always orienting for the egocentric frame of reference and ambient environment, the city block ahead and behind, even if one is otherwise disoriented with respect to the city. As remarked above, some navigation—point to point through narrow streets at oblique angles to each other or in a blur of hot pursuit—affords at best path integration and orientation for a succession of frames of reference without their integration into a comprehensive one. In the privileged condition that a cinematic scene sɶ is orienting throughout its duration for a given frame of reference f, with annotations throughout in coordinates for that frame of reference, it is said to be a nav-topographical survey for that frame of reference: (176) nav-orient( sɶ, f ) ↔ def ∀t(LTsɶ ( sɶ)(t ) → nav-orient( sɶ(t ), f )) nav-topo-survey( sɶ, f ) ↔ def nav-orient( sɶ, f ) A topographical survey of f affords a perspective-free representation in that all that is surveyed can be addressed using only the allocentric coordinates for f, without reference to the observer’s positions in f. Several surveys may survey (different regions of) the same frame of reference: ɶ ɶ → nav-topo-survey( sɶ, f )) (177) nav-topo-survey[Sɶ , f ] ↔ def ∀sɶ(Ss As much as visual navigation is a benchmark condition for language design, not all narrated visual experience guides concurrent navigation. As in section 9.3, the scene now experienced, if relived from memory or in a cinema, reprises its original projection, when it was navigational. Moreover, as in cinerama generally, it may be one scene among many culled from the reportage of many navigators. Still, as it is presented with the purpose to represent and archive How-the-World(-of Manhattan)Is, it affords some orientation in it, for which the above notion of orientation should be generalized: f © (178) orient(ṡ,f, n) ↔def ṡ© = projection(t(ṡ©),σf(ṡ©),πf(ṡ©), f, l(ṡ©), No( sɺ ) ,o(ṡ )) & n sees at t(ṡ) that ∃πf ∃l ∃N of ∃o ṡ© = projection(t(ṡ©),σf(ṡ©),πf, f, l, N of ,o)43
Cinerama Semantics
451
(179) orient( sɶ, f , n) ↔ def ∀t(LTsɶ ( sɶ)(t ) → orient( sɶ(t ), f , n)) topo-survey( sɶ, f , n) ↔ def orient( sɶ, f , n) ɶ ɶ → topo-survey( sɶ, f , n)) (180) topo-survey[Sɶ , f , n] ↔ def ∀sɶ(Ss A scene s is orienting in a frame of reference f for a narrator n narrating it just in case she knows where and when in f to place what she sees in s. Notice that a scene may be orienting for a frame of reference despite ignorance of the conditions under which it was originally projected, ignorant, for example, of whether it was zoomed in from long distance or from a wide angle at short range. Thus, transmitted orientation, unlike navigational orientation in real time, is not assumed to be a fix on the observer’s position in f. It suffices that the events and objects en-scène are at known position. The above makes no room for partial orientation, for the recognition of where without recall for when or for when without knowing quite where. But this just amounts to knowledge de re of even fewer parameters than (178) requires for full orientation. One may, as with ignorance of the original conditions for projection, see only that for some unknown time, the events en-scène happened at known locations in f. Navigational orientation, which is de se in character, and public orientation de re are related as follows: (181) orient(ṡ©, f, o(ṡ©)) ↔ nav-orient(ṡ©,f) Note that the stronger (182) is false, for the usual reasons that lapses of memory can dissociate knowledge de re from its counterpart de se knowledge: (182) F orient(ṡ, f, o(ṡ©)) ↔ nav-orient(ṡ©,f) For the solo navigator, distinct scenes deriving from distinct events of projection are of distinct spatiotemporal regions, which is trivially true if distinct moments in her visual experience are always of distinct moments in real time: (183) nav-orient( sɶ, f ) → ∀t∀t ′((LTsɶ ( sɶ)(t ) & LTsɶ ( sɶ)(t ′) & t ≠ t ′) → t( sɶ(t )) ≠ t( sɶ(t ′)) Moreover, at any given moment, any difference of location in the (spherical) panorama then projected is a difference of location in the topography being charted. How else could a scene be a scene of what it represents and orienting for it? How else could pointing at a scene be at the same time pointing into the space it is a scene of? Of visual navigation, the visual experience and the records of such, whether atlases, surveys, or memories, conform to this condition on representation: what looks different from here is different out there. This condition of course holds of the scene analysis of proximate visual experience, that is, the scenery of visual navigation. It persists, too, as a condition on the visual representation, archived from such a survey, of the allocentric frame of reference that is the target of path integration—a perspective-independent, 3-D holographic model, as it were, of
452
Chapter 9
How-the-World(-of Manhattan)-Is, which can be rotated, traversed, or circumnavigated in memory. As much as this characterizes intrapersonal cognitive mapping, it is also a mission of interpersonal communication to settle once and for all How-the-World(-of Manhattan)-Is as reflected in the negotiated construction of just such a 3-D model. Imagine that twelve drones, NCC-1700 … NC-1711, cooperate in a topographical survey of Manhattan under blackout conditions. Their transmitted imagery, streaking through Manhattan, might be thought of as the moment-by-moment illumination of the space subtended, as if the lights inside the darkened apartments were momentarily switched on as the drones passed through them. Even if there is a delay in the inverse projection from the scene transmitted to the three dimensions of the space it subtends, their transmissions are so organized that at any given moment, the twelve scenes transmitted adorn a memory palace isomorphic to the space and allocentric frame that is the target of their survey, Manhattan: what was seen then there, pointing to any one of twelve scenes, was there, the same place, in Manhattan, and to know what in Manhattan was there, look there at the same place. Notice that any two simultaneous transmissions that are scenes of the same space, placed side by side in competition for the same address (absent their fusion into a holographic projection), breach this condition on visual representation. Surveyors cooperating to transmit imagery that conforms to this condition on visual representation will thus at any given moment find themselves charting different terrain. Removed from a navigation in progress, cinerama in reprise is, by definition, archive. If not the narrator’s own eyewitness, it draws on, reports, and defers to the documentary eyewitness of others, from which is assembled a single, veridical representation of How-the-World-Is. It aspires to an authoritative, enduring representation of some invariances in the world that can be revisited independent of their source and original perspective, making it a public resource and record of how things allegedly really were. As a condition on its veridicality as a visual representation of what is, distinct scenes must be of distinct subject matter, where the distinction in subject matter can be represented in the archival language mentioned above that is free of perspectival-dependent language such as now-en-scène. What looks different was different, as this cinerama is not to be a Rashomon-like compilation of multiple views of the same goings-on.44 For the special case of the topographical survey above, it is a substantive condition on its cineramic presentation that the scenes culled from multiple eyewitnesses conform as above as a visual representation of the frame of reference into which they are all integrated. A cineramic presentation serializes, as it must, what would otherwise be the simultaneous transmissions and simultaneous illuminations of How-the-World(-of Manhattan)-Is in 3-D. Thus, of any two such scenes in succession, either narrative time advances historical time forward so that the second scene
Cinerama Semantics
453
is witness to later events or backward in a flashback, or the successive scenes, overlapping in historical time (“Meanwhile, in …”), must be scenes of different spaces. In a topographical survey, there are not multiple scenes of the same spatiotemporal region. So much seems plain enough for scenery organized as a topographical survey. Alongside topographical survey, and of particular interest here, is documentary, which bears witness to a history by tracking its disparate cast across the world’s stage. The twelve spies sent to scout out the land of Canaan (Numbers 13.1–14.9) conduct their topographical survey so as not to be in the same place at the same time, and neither will they track the same actors if their mission is a chronicle of local history and custom. For distinct scenes, the eyewitness of distinct scouts, oriented to a global frame of reference to which the scenes of this history are pathintegrated, it should not turn out that lives presumed distinct that these scouts have been assigned to follow are suddenly discovered to coincide. There thus appear to be two conditions, depending on the scouts’ mission. In a topographical survey, their charts partition the surveyed landmarks. In tracking, they instead partition the actors whose actions they will bear witness to. Tracking them, it may very well happen that the scouts follow them to a confrontation, their scenes so converging on the same time and place in what would therefore fail to conform to a survey of that terrain. These rather different missions—survey and tracking—can be brought into closer alignment, recalling that a tracking scene is one that maintains itself in constant orientation to a mobile, allocentric frame of reference in which the thing tracked is a landmark. Anticonvergence can then generalize across the two conditions—survey and tracking—in saying that scenes integrated into the same frame of reference are scenes of landmarks that the other integrated scenes are not scenes of. That is to say that the scenes will be a spatiotemporal partition of the landmarks of interest, even if they do not always amount to a partition of an entire spatiotemporal region per se. Let us first define that arbitrary scenes Sɶ that are projections in frame of reference f are integrated to f just in case they never provide alternative views of the same landmarks at the same time: ɶ ɶ → frame( sɶ, f )) & (184) integrated(Sɶ , f ) ↔ def ∀sɶ(Ss ɶ ∃F ∀sɶ(Ssɶ → ∃f (Ff & frame( sɶ, f ) & ∃ζ (ζ ≠ o( sɶ) & Landmark(ζ , t( sɶ), f )) & ɶ ɶ′ & Ff ′ & sɶ ≠ sɶ′ & ∀ζ (Landmark(ζ , t( sɶ), f ) → ¬∃t0 ∃sɶ′∃t∃t ′∃f ′(Ss frame( sɶ, f ′) & t( sɶ(t )) = t( sɶ′(t ′)) = t0 & Landmark(ζ , t( sɶ′), f ′) & ∃α sɶ ( t ) At(α sɶ ( t ),ζ , t( sɶ(t )), sɶ(t )) & ∃α sɶ′ ( t ) At(α sɶ′ ( t ),ζ , t ( sɶ′(t )) , sɶ′(t )))))) If, as in a topographical survey, the scenes are oriented and orienting for all the same frames of reference, in particular, to the global one of their path integration, then they find their different landmarks only by dividing up the landmarks of the topography surveyed, and the spaces the scenes subtend will more or less never overlap. If, on the other hand, the scenes are orienting and oriented for several
454
Chapter 9
different frames of reference, those that they severally track, then it suffices that the landmarks for these frames of reference are not seen to converge. Note that any two still frames from any two of the scenes Sɶ cannot be of the same landmarks at the same time. The point of anticonvergence is that veridical, integrated scenes can be relied on at the end of the day to be of different subject matter, whatever the mission’s interest, topographical survey, or documentary history.45 Any cineramic experience π assembles46 scenes Sɶπ at an arbitrary distance from their original events of projection. As with orientation, it must be allowed that scenes now in progress might be in reprise. Recall (see section 9.3) that an authorship function © secures reference to a scene’s original projection, here generalized to the plural: ɶ ɶ & s = sɶ ) (185) Sɶ s ↔ def ∃sɶ(Ss Arbitrary scenes Sɶ are then said to be well integrated to a frame of reference f just in case their original scenes of projection are. Revising (184) accordingly: (186) integrated(Sɶ , f ) ↔ def ∀sɶ(Sɶ sɶ → frame( sɶ, f )) & ∃F ∀sɶ(Sɶ sɶ → ∃f (Ff & frame( sɶ, f ) & ∃ζ (ζ ≠ o( sɶ) & Landmark(ζ , t( sɶ), f )) & ∀ζ (Landmark(ζ , t( sɶ), f ) → ¬∃t0 ∃sɶ′∃t∃t ′∃f ′(Sɶ sɶ′ & Ff ′ & sɶ ≠ sɶ′ & frame( sɶ, f ′) & t( sɶ(t )) = t( sɶ′(t ′)) = t0 & Landmark(ζ , t( sɶ′), f ′) & ∃α sɶ ( t ) At(α sɶ ( t ),ζ , t( sɶ(t )), sɶ(t )) & ∃α sɶ′ ( t ) At(α sɶ′ ( t ),ζ , t( sɶ′(t )), sɶ′(t )))))) A cineramic experience π is an accurate rendition and fair witness to history only if its scenes Sɶπ are well integrated for that history’s frame of reference. As a general condition on cinerama, anticonvergence presupposes that its scenes are well integrated for any frame of reference that a narrator may intend. Moreover, intending a veridical, visual representation of history, it must be that the narrator is not oriented to any frame of reference in which she would discover that the scenes are not well integrated. (187) Anticonvergence cinerama(π) → (∀n∀f ∀sɶ(¬integrated(Sɶπ , f ) → (Sɶπ sɶ → ¬ orient( sɶ, f , n))) Thus the documentary narrator who has assembled the scenes of many witnesses represents herself as disoriented in those frames of reference to which the assembled scenes are not well integrated. The state of the narrator’s (dis)orientation is called on to discriminate the felicitous (188) from the infelicitous example (189) of event counting 3000 clients who are only 1000 lonelyhearts serially speed dating: (188) Three thousand clients that attended speed-dating events last month exchanged phone numbers at the end of the date. (189) ???(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers on the fridge.
Cinerama Semantics
455
How can it be known either by visually tracking the 3000 clients (or by inference from the known facts) that they do as reported without it being seen that some clients converge on the same lonelyheart? One could imagine transmissions from personal cameras that have been implanted at the speed-dating events, with a field of vision just wide enough for client and refrigerator. If, on the other hand, the scenes in witness of (189) visit the clients’ homes, the documentary narrator in uttering (189) represents herself as suffering a lapse in the path integration that would put some of these homes at the same address containing the same fridge and the same client, since the 3000 scenes could not be path-integrated to the same frame of reference without them failing to be orienting for it in the sense of (184). Thus, (189) is rejected or at least suspect in that either the scenes needed in evidence for it are too arcane out of the blue or the speaker of (189) has to be taken to represent herself as not in full command of her faculties in uttering it. In contrast, all the scenes needed for (188) are scenes of the speed dates and only the speed dates, and these scenes satisfy anticonvergence for all frames of reference. Any narration pairs narrative discourse Δ to visual experience Π synchronizing their clocks (section 9.3). No matter that it might be composed from a scatter of original eyewitness and commentary, a narration is a singular event that imposes a standard of coherence on its utterances Δ and its scenes Π. In particular, it is not a narration unless it is a cinerama. That is, for some π, cinerama(π) & Π = Sɶπ . Accordingly, the fundamental relation en-scène is revised and restricted to the scenes that cohere in a cinerama (cf. (172)–(173)): (190) en-scène(e, sɶ , αf,f) ↔df ∃π(cinerama(π) & Sɶπ sɶ ) & At(αf, e, t(e), f) & ∀ṫ((moment(ṫ) & overlap(ṫ, t(e))) → sɶ(tɺ) = projection(tɺ, α f , π f ( sɶ)(tɺ), f , l( sɶ)(tɺ), Nof ( sɶ ), o( sɶ)(tɺ))))
ɶ ɶ → Sɶπ sɶ)) & (191) en-scène[E,Sɶ ,AF,F] ↔df ∃π(cinerama(π) & ∀ sɶ(Ss ɶ ∀e(Ee → ∃sɶ ∃αf ∃f( Ssɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ɶ ɶ → ∃e∃αf ∃f( Ss ∀ sɶ (Ss ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ∀αf (AFαf → ∃e∃sɶ ∃f( Ss ɶ ɶ ɶ ∀f(Ff → ∃e∃s ∃αf ( Ss & AFαf & Ff & en-scène(e, sɶ , αf,f))) For its use in any narrative tale retold or live communication for concurrent navigation, the language of cinerama semantics provides the fundamental (unspoken) predicate, now-en-scène: (192) ⌜now-en-scène(ei)⌝ abbreviation for ⌜ ∃sɶ ∃αf∃f(now( sɶ ) & en-scène(ei, sɶ , αf,f))⌝ (193) ⌜now-en-scène[E]⌝ abbreviation for ⌜ ∃S∃A ⌝ ɶ ɶ ɶ F∃F(now(S) & en-scène[Ei,S ,AF,F]) (194) (See (103)) For discourse Δ and token μ of ⌜now(vi)⌝ in Δ such that ltΔμ©(μ) = δt, Σ satisfy μ ↔ Within(nowΔμ©(δt), Σ(vi))
456
Chapter 9
If now is the only present-tense indexical morpheme in the object language, then now-en-scène abbreviates an object language expression as in (192)–(193). If instead the vocabulary of the object language is expanded to include now-en-scène as a native indexical, (195)–(196) are its definition: (195) For discourse Δ and token μ of ⌜now-en-scène(ei)⌝ in Δ such that ltΔμ©(μ) = δt, Σ satisfy μ ↔ ∃sɶ ∃αf∃f (Within(nowΔμ©(δt), sɶ ) & en-scène(Σ(ei), sɶ , αf,f)) (196) For discourse Δ and token μ of ⌜now-en-scène[Ei]⌝ in Δ such that ltΔμ©(μ) = δt, © ɶ ɶ ɶ Σ satisfy μ ↔ ∃S∃A F∃F(Within(nowΔμ (δt), S ) & en-scène[Σ(Ei), S ,AF,F]) Embedded in a narration to a concurrent navigation, the meaning of now-en-scène is as expected. In addition, as defined, it affords that any narrative, visual navigation in progress or not, may be cineramic via apostolic succession from original reports of visual navigation. The scenes so narrated, recalled, and brought together from distant eyewitnesses must however cohere as a cinerama, conforming in particular to condition (187) on anticonvergence. This condition will be the explanation for the sequence-of-events condition so-called on event counting (section 11.0.2). 9.4.2
Resolution and reticulation for visual counting
The conditions of observation determine the optical resolution of the scenes projected. A grain of sand on Orchard Beach and the northern coastline of the Long Island Sound overlap in absolute space and are both addressable () from points of view 40,000 m above the Sound, and yet no natural scene from any point of view resolves both grain of sand and coastline. Any given projection, ṡi=projection(t(ṡi),σf(ṡi),πf(ṡi), f, l(ṡi), N of ,o(ṡi)), is scaled to resolve some things and not others. Distinct projections, ṡi=projection(t(ṡi),σf(ṡi),πf(ṡi), f, l(ṡi), N of , o(ṡi)) and ṡj=projection(t(ṡj),σf(ṡj),πf(ṡj), f, l(ṡj), N of , o(ṡj)), may coincide in time, point of view, and space subtended and differ only in their resolution, o(ṡi) ≠ o(ṡj), that is a difference in their optics.47 A projected scene guides spatial orientation and navigation for the space and frame of reference from which it is projected to a precision that depends on what is resolved in the scene and the accuracy of the projection’s inverse mapping from regions in the 2.5-D scene to regions of 3-D space. Resolution constrains the precision of the positions discriminated. Learning that the Great Red Spot is a Jovian storm rather than a topographical feature, it can be said with one resolution in mind that its molecules like the Spot itself have been in the same place since Cassini and Hooke (1665) first observed it, and with another resolution in mind, that its molecules are never in the same place twice.48 Let s be the Great Red
Cinerama Semantics
457
Spot, and S, the storm’s molecules. At Cassini and Hooke’s resolution, they are in the same place—then and now: (197) At[α,s, 1665, fch] & At[α,S, 1665, fch] & At[α,s, 2006, fch] & At[α,S,2006,fch] Resolutions fine enough to resolve the storm’s molecules resolve regions Aṡ so that ∀x∃αṡ(Sx → (Aṡαṡ & At[αṡ, x,t(ṡ),ṡ] & ∀y((Sy & At[αṡ,y,t(ṡ), ṡ]) → x=y)) & ∀x∀y((Sx & Sy & x≠y) → ℜ[ sɺ, x, y]). At fine resolution, unless something goes awry with the second law of thermodynamics, we do not expect to find t(ṡ)≠t(ṡ′), with the storm’s molecules in the same place Aṡ. As remarked earlier, for any scene ṡ, a frame of reference f from which it is projected, ṡ = projection(t(ṡ),σf(ṡ),πf(ṡ), f, l(ṡ), N of , o(ṡ)), and some things X there (i.e., At[αṡ,X,t(ṡ),ṡ] & At[αf,X,t(ṡ),f]), the relation between their location αṡ in the scene and their location αf in 3-D f must be known for any projection in the observer’s cognitive repertoire, since a scene aids spatial orientation and navigation only when αf is recovered from αṡ. There is thus an intimate connection among the resolutions of scene, proximate scene analysis (i.e., projection of the ambient space, city block), and distal path integration (city island). They are scaled mutually, so that what follows about the resolution and reticulation of projected scenes applies equally mutatis mutandis to the resolution and reticulation of frames of reference for the spaces through which the scenery navigates. A scene optically resolves objects or events ζ from other objects or events ξ only if ζ are within some region of the αṡ and none of the ξ is. The boundary of αṡ in that case completely encloses ζ and excludes ξ. But this is not a sufficient condition for their resolution as there may be a stretch of the boundary separating them, limit points for them both, to which ζ and ξ come infinitesimally close, so that it were as if there were a seamless continuity between the nearly contiguous ζ and ξ, wherein the scene has obviously failed to resolve ζ and ξ. A higher degree of separation is required: a scene resolves ζ from ξ only if they share no limit points. (A point p is a limit point for points α just in case any open ball centered on p with arbitrarily small radius contains points in α.) A scene ṡ is said to optically resolve ® some objects or events ζ from some objects or events ξ: (198) ®[ṡ, ζ, ξ] ↔df ∃αṡ(At[αṡ,ζ,t(ṡ), ṡ] & ∀x(ξx → ¬∃y(overlap(x,y) & At(αṡ,y,t(ṡ),ṡ))) & ¬∃p(At(p,p,t(ṡ), ṡ) & Limit point(p, ζ, ṡ) & Limit point(p, ξ, ṡ)) A scene ṡ resolves ζ from ξ just in case there is a region that encloses all of ζ and excludes all of ξ, and there is no point on a boundary between them to which they both come arbitrarily close. That is, there is everywhere daylight between them. So understood, a scene of lovers holding hands does not resolve them until they let go. So understood, no scene ever resolves the mountains in a contiguous range. For lovers or mountains that are joined, counting them means counting synecdochic
458
Chapter 9
proxies, noses or peaks, that can be clearly discerned. That is, there are some noses or peaks that a scene resolves each from the others according to (198)’s stringent conditions. The lovers in the scene are as many as the noses it resolves just in case lovers and noses align one to one. If one were to have relied on a show of hands rather than noses, this bijective condition is needed to warrant that the hands indeed match the lovers in number. For lovers, whether two hands is one or two lovers is organic to their constitution. On the other hand, the resolution of mountains in a mountain range according to their peaks, large or small, or according to any other aspect of their form is rather more ambiguous and arbitrary, decided ad hoc by a protocol for measurement or counting that fixes scale and granularity. For any scene, one may imagine the further superimposition of a grid or reticule that delineates scale, granularity, and the regions within which are resolved whatever events or things are to be resolved and then optically counted. For convenience, it is assumed that a scene s already comes reticulated with a reticule s (anticipating no need to talk further about the same scene under different reticules). Relativizing optical resolution to reticule and to whatever conditions make a good grid for marking off measurement, (198) is revised: (199) ®[ṡ, ζ, ξ] ↔df ∃α sɺ (sɺα sɺ & At[α sɺ,ζ , t( sɺ), sɺ] & ∀x(ξx → ¬∃y(overlap(x,y) & At(αṡ,y,t(ṡ),ṡ))) & ¬∃p(At(p,p,t(ṡ), ṡ) & Limit point(p, ζ, ṡ) & Limit point(p, ξ, ṡ)) It will also prove expedient to distinguish reticules according to the neighborhoods or addresses they designate to delineate the events or things they resolve. A reticule, for example, for a scene that resolves Central Park and the Public Garden will contain exactly two addresses, nonoverlapping regions, that divide Central Park and the Public Garden between them. Under one reticule, the addresses may coincide with the park boundaries, and yet under another, the addresses may be New York and Boston, and so on for other reticules for the same scene. No reticule will designate both the park boundaries and New York as its addresses for Central Park, as the addresses overlap and each addressee is to have a unique address under the addressing the reticule adopts: (200) ®[ṡ, ζ, ξ, Aṡ] ↔df ∀α sɺ ( Asɺ α sɺ → sɺ α sɺ ) & ∀αṡ∀αṡ′((Aṡαṡ & Aṡαṡ′ & overlap(αṡ,αṡ′)) → αṡ=αṡ′) & ∀αṡ∀αṡ′((Aṡαṡ & Aṡαṡ′ & At[αṡ,ζ,t(ṡ), ṡ] & At[αṡ′,ζ,t(ṡ), ṡ]) → αṡ=αṡ′) & ∃αṡ(Aṡαṡ & At[αṡ,ζ,t(ṡ), ṡ] & ∀x(ξx → ¬∃y(overlap(x,y) & At(αṡ,y,t(ṡ),ṡ))) & ¬∃p(At(p,p,t(ṡ), ṡ) & Limit point(p, ζ, ṡ) & Limit point(p, ξ, ṡ)) It may then be said that a scene ṡ resolves some things or events ζ—lovers, mountains, or conflagrations—despite their massive overlap and entanglement and the
Cinerama Semantics
459
vagueness of their boundaries, just in case they have equinumerous proxies ξ that are optically well separated in ṡ (by addresses Aṡ): (201) ℜ( sɺ, ζ , Asɺ ) ↔df ∃ξ(∀x∀y((ξx & ξy & x ≠ y) → ®[ṡ, x, y,Aṡ]) & ∀x(ζx → ∃y(ξy & overlap(x,y) & ∀z((ζz & overlap(z,y)) → z = x))) & ∀x(ξx → ∃y(ζy & overlap(x,y) & ∀z((ξz & overlap(z,y)) → z = x))) More exacting is to say that a scene ṡ resolves some things or events ζ just in case it sorts the equinumerous proxies ξ into the equinumerous addresses Aṡ, which is the effect of saying below that the Aṡ are a minimal class of addresses to optically separate ζ: (202) ℜ[ sɺ, ζ , Asɺ ] ↔ df ℜ( sɺ, ζ , Asɺ ) & ∀Asɺ ′((ℜ( sɺ, ζ , Asɺ ′) & 49 ∀α sɺ ( Asɺ α sɺ → Asɺ ′α sɺ )) → Asɺ ′ = Asɺ ) Counting motions, events, landmasses, breezes, fires, custards, times, regions, areas, etc.—domains the density of which threatens to undermine any finite measure— must rely on some notion of a maximal, contiguous, or uninterrupted stretch of what is to be counted. But if one were to count just those things that are the maximal, contiguous, or uninterrupted examples of what they are, the dissected plateau that is the Catskill mountains could only count as one mountain (as noted above), and moreover, a forest fire of scattered blazes could not count as one fire: (203) There are many mountains in the Catskills. (204) There is a fire in the Catskills. Counting by reticule and proxy, on the other hand, imposes countability on the dense domain while still leaving room for the likes of (203) and (204). Recall that counting one or two greens (shown as gray below) is dependent on the scene being narrow or wide and the resulting presentation of what is green: (205)
(206) Two greens flank white. (207) *Two greens surround white. (208) One green surrounds white.
460
Chapter 9
The scene chosen determines the count of what is counted, in that the narrow scene resolves two greens and the wider one only one green. Events or objects are countable in a scene ṡ just in case it resolves them: (209) countable[ṡ, ζ] ↔ ∃Asɺ ℜ[ sɺ, ζ , Asɺ ] Many an event of counting or measurement is, however, extended in time and not of objects or events countable in a single still life ṡ. It may even be that at the moment it is counted, object or event occupies the very same spot that anything else does in its turn to be counted. Resolution and countability in a cinematic scene sɶ go as follows: (210) ℜ( sɶ, ζ , Asɶ ) ↔ df ∃ξ(∀x∀y((ξ x & ξ y & x ≠ y) → ∃t [ sɶ(t ), x, y, Asɶ ]) & ∀x(ζx → ∃y(ξy & overlap(x,y) & ∀z((ζz & overlap(z,y)) → z = x))) & ∀x(ξx → ∃y(ζy & overlap(x,y) & ∀z((ξz & overlap(z,y)) → z = x))) (211) ℜ[ sɶ, ζ , Asɶ ] ↔ df ℜ( sɶ, ζ , Asɶ ) & ∀Asɶ ′((ℜ( sɶ, ζ , Asɶ ′) & ∀α sɶ ( Asɶ α sɶ → Asɶ ′α sɶ )) → Asɶ ′ = Asɶ ) (212) countable[ sɶ, ζ ] ↔ df ∀x(ζ x → ∃t∃α sɶ (At[α sɶ , x, t( sɶ(t )), sɶ]) & ∀x∀y∀t∀α sɶ ∀α sɶ ′((ζ x & ζ y & x ≠ y & At[α sɶ , x, t( sɶ(t )), sɶ]& At[α sɶ ′, y, t( sɶ(t )), sɶ]) → ∃ξ(ξ x & ξ y & ∃Asɶ ( t ) ℜ[ sɶ(t ), ξ, Asɶ ( t ) ])) The ζ are countable in cinematic scene sɶ just in case none of them ever appear in it together unresolved. An event of visual counting is a single cinematic scene: (213) count(eµ ) → ∃sɶ∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) (214) count(eµ, ζ ) → ∃sɶ(∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) & countable[ sɶ, ζ ]) An event of counting n-many ζ fits them under a reticule of n-many addresses, with a standard definition of n-many ((216)–(218)): (215) Visual counting count[eµ , ζ , n] ↔ df ∃sɶ∃Asɶ (∃α sɶ At(α sɶ , eµ , t( sɶ), sɶ) & ℜ[ sɶ, ζ , Asɶ ]& card[ Asɶ , n]) (216) For ordered pairs θ, injective[θ] ↔df ∀x∀y∀z(((θ() & θ()) → y = z) & ((θ() & θ()) → y = z)) (217) card[X,Z] ↔df [∃θ: injective[θ]](∀x(Xx → ∃z(Zz & θ())) & ∀z(Zz → ∃x(Xx & θ())) (218) card[X,n] ↔df [℩Z: N(n) & ∀z(Zz → (N(z) & z < n))] card[X,Z] Thus, ‘count[eμ, ζ, n]’ is the defining relation among an event of measurement, the scene or conditions of observation for that measurement, what is measured, and the number measured.
Cinerama Semantics
461
What an event of counting purports to count must be countable in the scene in which it is counted, (214): (219) Three hundred thousand Soviet troops flowed through Red Square in the May Day parade. (220) One hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. (221) The 300,000 Soviet troops that flowed through Red Square in the May Day Parade were 100,000 Soviet troops streaming around Central Moscow. The CIA field officer at the reviewing stand in Red Square counts 300,000 (219), and the spy plane overhead unmasks the deceptive tally (220). Given (212) and (213), these must be distinct events of counting (with different results) in distinct cinematic scenes, and (221) asserts the identity of what has been counted under these distinct scenes and conditions.50 The literature, in calling (219) event counting and (220) object counting, presumes that the former needs apology and theory while the latter is always unproblematic and freely available. Yet the CIA field officer confronts a parade of faceless troops qualitatively identical, and he is unable to recognize the head of the procession when it recycles through Red Square. As a phalanx of 250 passes the reviewing stand and crosses the field of view of his optical counter in 30 seconds, these 250 images are 30-second proxies for a 30-second event of measurement e30sec that counts 250 events E30sec (i.e., count[e30sec, E30sec, 250]), events E30sec of being a Soviet troop for 30 seconds (i.e., ∃X Soviet troops[E30sec, X]). As no one is in two places at once, it is inferred that these 250 30-second images are also optical proxies for equinumerous events that are lifetime careers as a Soviet troop.The 30-second event of measurement e30sec counts them too, count[e30sec, ESoviet military career, 250]. Suppose the parade moves at constant velocity and it takes a phalanx 10 minutes to appear at one end of Red Square and disappear across it. Letting the optical counter run for 10 minutes is then a 10-minute event of measurement e10min in which 5000 30-second images are proxies for an event counting 5000 events of being a Soviet troop for 30-seconds (count[e10min, E30sec, 5000]), also for an event counting 5000 events of being a Soviet troop in Red Square (count[e10min, E10min, 5000]) and for an event counting 5000 careers (count[e10min, ESoviet military career, 5000]), as can all be inferred from the conditions of observation. The CIA officer then has in hand a measure of the troop strength needed to occupy Red Square. On the other hand, letting the optical counter run for the 10-hour duration of the parade does register 300,000 30-second proxies and amount to a 10-hour event of measurement counting 300,000 events of being a Soviet troop parading across Red Square, count[e10hours, E10min, 300,000], as (219) reports. But without an estimate of the rate at which the troops cycle through Central Moscow to reappear in Red Square, total troop strength can be neither inferred nor directly observed, which must be left to the spy plane.
462
Chapter 9
9.4.2.0 Summation of visual counts
The above example is meant as vivid illustration of the solitary event of visual counting and what (215) is intended to codify. Yet clearly some sentences rest on the addition of several such measurements. Below, I count people on board 7-hour transatlantic flights, relying on a 20-minute event of counting that clicks a counter during the boarding process and trusts that security prevents anyone from exiting the aircraft and reboarding once she has been counted: (222) Thirty people on American Airlines 100 departing JFK at 6:00 p.m. needed thirty special meals. (223) Thirty people on American Airlines 104 departing JFK at 8:00 p.m. needed thirty special meals. (224) Therefore, sixty people (on AA 100 and 104) needed sixty special meals (on their flights). The inference to (224) does not imply a single event of visually counting to sixty but reports instead the summation of two visual counts to thirty. The inference is valid event counting, equivalent to counting passengers, events of being a person for a 7-hour flight. The premises and conclusion then all conform to a narrow sequencing of events (section 8.3 and chapter 15) in that the meals are contemporary with the events counted. Contrast reports of future events: (225) Thirty people on American Airlines 100 departing JFK at 6:00 p.m. needed thirty special meals and hold return tickets from Heathrow on American Airlines 107. (226) Therefore, thirty people (from American Airlines 100) will need thirty special meals on American Airlines 107. (227) Thirty people on American Airlines 104 departing JFK at 8:00 p.m. needed thirty special meals and hold return tickets from Heathrow on American Airlines 107. (228) Therefore, thirty people (from American Airlines 100) will need thirty special meals on American Airlines 107. The events counting thirty separately warrant anticipation of thirty special meals on American Airlines 107, as in (226) and (228). The protocol for counting passengers at boarding counts proxies that are also good proxies for persons (i.e., lifetimes), and it may be inferred that these thirty persons will need thirty special meals on the next flight, too. But an inference to sixty meals is unsound unless it is also known that flights AA 100 and AA 104 depart the same day, so that someone on the first could not also be on the second 2 hours later: (229) #Therefore, sixty people from AA 100 and 104 will need sixty special meals on American Airlines 107.
Cinerama Semantics
463
It is the same as at Red Square. There the protocol affords measurement of the troop strength at Red Square, but no measure of the parade’s total troop strength is to be had from the summation of such measurements. Similarly, counting at boarding counts people on board that flight, but the summation of such counts does not reliably count people. If, for the sake of the sound inference in (224), logical form allows for the summation of measurements, like the summation of measurement in general, it must be restricted to exclude overlapping measures. Rather than subtract a measure of the overlap, as is common (230), summation is here defined only for disjoint measures, (231): (230) μ(A∪B) = μ(A) + μ(B) − μ(A∩B) (231) a. sum(eμ) = n ↔df ∃ζ count[eμ, ζ, n] b. ¬Eμeμ → (∀e′μ(E′μe ′μ ↔ (Eμe′μ ∨ e′μ = eμ)) → sum(E′μ) = sum(Eμ) + sum(eμ)) Then for the sake of (224) and the like, the logical form for sixty people on AA 100 and 104 allows for the summation of disjoint visual counts Eμ: (232) [∃X: ∃E(∃Eμ count[Eμ, E, 60] people[E,X] …)] (233) count[Eµ, ζ , n] ↔ df ∃sɶ countable[ sɶ, Eµ ]& n = sum(Eµ ) & [∀eμ : Eμeμ][∃ζμ : [∀v: ζμv] ζv] ∃n count[eμ, ζμ,n] & [∀v: ζv][∃ζμ : ζμv & [∀v: ζμv] ζv][∃eμ : Eμeμ] ∃n count[eμ, ζμ,n] & [∀v: ζv][∀eμ: Eμeμ][∀e′μ: Eμe′μ][∀ζμ : ζμv & [∀v: ζμv] ζv] [∀ζ′μ: ζ′μv & [∀v: ζ’μv]ζv]((∃n count[eμ, ζμ,n] & ∃n count[e′μ, ζ’μ,n]) → eμ = e′μ) The summation of measurements—two measurements, each measuring the passengers boarding—sums determinate, discrete acts of measurement according to a protocol that determines in advance when to start counting and when to end. That is, the events of measurement a sentence like (224) may intend to sum are only those that can be the objects of plural reference to events, some events of measurement. For a single flight boarding in 20 minutes, note that the 20-minute counter clicking as passengers surrender their boarding passes is a single measurement that does not readily parse into discrete measurements even though the sum of the first 10 minutes’ count and the sum of the last 10 minutes’ count could be well defined. The explicit restriction to countable events in (233), ‘∃sɶ countable[ sɶ, Eµ ]’, risks harmless redundancy if the restriction to plural events of measurement in (224) should derive from the general nature of (second-order) quantification and reference to events. More might be said about when a protocol for visual counting is an effective measurement of what is to be measured. Consider an arcade of photo booths and an assessment of its business, for which one might wish to know for the month assessed: the number of its photographic subjects (revenue), proxies for which could
464
Chapter 9
be photo-booth flashes (assuming no group photos), the number of its photo-booth visitors (i.e., entrance-and-exits, to measure impulse buying in a ratio with revenue), and the number of the arcade’s customers (to estimate the size of its advertising target). Although a walk through an aircraft is a way to count passengers, a walk through this photo-booth arcade is no way to count anything. It does not reliably count customers, the more enthusiastic of which run from booth to booth only to be multiply counted on any walk-through. Nor does it reliably count photographic subjects or photo-booth visitors. For while the walk-through is measuring the activity in one section of the arcade, elsewhere it lets escape uncounted photographic subjects getting flashed and visitors entering and exiting booths. Instead of a walkthrough, optical counters should be left in place at the photo booths to count photographic subjects or visitors by the summation of their month-long counts. Even so, after a walk-through, it is not false that thirty photographic subjects said “Cheese!,” as these thirty were certainly among the “Cheese!”-saying photographic subjects in the arcade, despite a protocol that could not have provided a reliable count of them all. Rather the question is whether the gloss for Thirty photographic subjects said “Cheese!” should reflect that the speaker has in mind to consider only those protocols robust enough in principle to address the global how-many question lurking in the background. If so, then the count to thirty is either part of a larger base count or conducted under an identical protocol, fixing all the conditions for observation, scene, frame of reference, and adverbialization canvassed above:51 (234) [∃X: ∃E([∃Ebase: [℩E: ∃X photographic subjects[E,X]] ∃n count[Ebase, E, n]] [∃Eμ : ident-protocol[Ebase, Eμ]] count[Eμ, E, 30] photographic subjects[E,X] …)] This refinement, if that is what it is, will be suppressed from now on, leaving aside overt expression of the standards for robust measurement it is meant to enforce. 9.4.2.1 Visual counting and spatial orientation
Any measurement event or a summation of them preserves the equinumerosity between what is counted ζ (‘count[Eμ, ζ, n]’) and their visual proxies ξ under the reticules for visual counting (‘®[ṡ, ζ, ξ, Aṡ]’; see (200) above). If at Red Square one means only to count visitors to Red Square in uniform, who are more numerous than the troops, one 10-hour event of measurement for the duration of the parade determines n = 300,000. If one means to deploy the same counter and protocol to measure Soviet troops, or military careers, this cannot be done to measure the troop strength that May Day. Counting the larger career-length events, these are equinumerous with the visual proxies counted only for those measurements during which the proxies initially counted remain in view—that is, only for so long as it takes a phalanx to cross Red Square. As remarked above, no summation of these 10-minute
Cinerama Semantics
465
measurements measures the May Day troop strength. Of course, the behavior of the optical counter itself is indifferent to what is to be measured. When and where it clicks away is entirely determined by visual presentation and reticulation, which can be subjectively defined for the observer’s point of view and egocentric frame of reference. In counting the passengers on AA 100 and AA 104, described as two events of measurement, I had in mind turning a counter on at the start of boarding and off at its end for one aircraft and then doing the same again for the other aircraft. I might just as well have turned the counter on at the start of boarding of the first aircraft and left it on, although I am transported unconscious or sedated to the other. Given the visual conditions that must obtain in my egocentric frame of reference for the counter to click, this single, prolonged run will equal the sum of the two former counters, at least as far as the number of clicks goes, with a long silence between two clicks. Either protocol suffices when what is counted is just the number of passengers, without implication for the number of longer-lived people they happen to be, as suffices for the inference to (224). Recall, however, that reasoning to the conclusion in (229) requires the further premise that no one could be on both flights, for which I would need to know the time between their departures, difficult if unconscious. If alert to it, the time on my egocentric clock (“local time”) of the counter clicking and the duration of the silence measured would have sufficed. In general, however, the judgment that what is counted is indeed equinumerous with its visual proxies under a given protocol depends on more than just egocentric orientation. To judge whether distinct images in the egocentric frame of reference could or must be projections from distinct bodies, it matters, for example, whether the observer is in rotation around a fixed point or the observed and counted revolve around her. The scenery and its locations in the egocentric frame of reference are the same in both conditions—all the images that trigger the optical counter flash at the observer’s . The conditions are discriminated only with some knowledge of the observer’s and observed’s positions in a common allocentric frame of reference. For another example, imagine a VIP counting the hands of the adoring throngs as she rushes by. The optical counter she has on board, trained on the optic flow and given the local resolution of its field of view for its mobile egocentric frame of reference, accurately counts hand waves—any hand wave is clearly discrete from any other hand wave nearby and no hand wave is at risk of being counted twice.52 If the VIP is in a railcar and the hand waves belong to admirers stationed along the track, this count of hand waves is also a count of hands and therefore of admirers. If, on the other hand, her vehicle follows a random serpentine path through crowds that are themselves in motion and clamoring for her repeated blessings, nothing more persistent than the hand waves themselves is counted. Under both conditions, the behavior of the optical counter is the same, especially if we imagine that the backgrounds are sufficiently vacant and the optics sufficiently poor that the optic flow is
466
Chapter 9
identical under the two conditions. In these toy examples, the observer’s vestibular sense of her own acceleration might suffice to distinguish among her own rotation, linear motion, and stationary position, but the judgment that just so many images project as many bodies is, I imagine, complex, open-ended, and may rely on special knowledge about the things measured, the measurement instruments, and so on. Even counting passengers qua passengers requires knowledge about boarding procedures and aircraft exits if one is to infer that the brief check-in at the gate is a count equinumerous with the 7-hour passenger events it presumes to count. For this, I assume there must be projection to the allocentric frame of reference of gate and aircraft, if the general knowledge of procedure, aircraft design, and human locomotion is to apply to these particulars on this occasion. It is safe to assume that reticulation, its topology, and the coordinate system for its addresses are all defined with respect to the observer’s egocentric frame of reference. Judgment about what is therein counted defers, however, to the integration of the egocentric scenery to allocentric frames of reference. This is no more than to acknowledge once again that all the scenery here, whether counting is in progress or not, is spatial orientation and navigation, with proximate concurrent projection and integration into the ambient environment (scene analysis), the city block, and distal often partial integration into larger allocentric frames of reference, the city island. Recall from section 9.4.0 that all scenes are taken to be annotated with coordinates for the frames of reference for which they afford the observer orientation. If so, then the counting at two boardings must be regarded as two separate measurements, as they are scenes different in their frames of reference. There are the allocentric frames of reference, each an aircraft and its gate, to which I am oriented during boarding, but neither of these is one to which I am oriented while at the other. Nor is there a third to which I am oriented, given the unconscious hiatus between them, that comprehends both of the aircraft and their gates. 9.4.2.2 Anomalous counts
The judgment whether or not two distinct visual proxies are proxy for the same event—enduring a lifetime of frequent flying or no longer than a 7-hour flight depends in part on a judgment whether the same body could have been in both scenes, which in turn depends on the spatiotemporal distance between the scenes and knowledge thereof. A lapse in consciousness between the scenes thwarts estimation of their spatiotemporal distance. In fact, between scene analysis and path integration, the apprehension of spatiotemporal distance is at best a sometime thing. Much visual navigation, pilotage, block to block, turn by turn with reference to visible landmarks does not integrate to a common frame of reference, say, Manhattan.53 The allocentric frame of reference into which the cinematic scene in progress projects at one moment, the city block, is too small to include later scenes, turning
Cinerama Semantics
467
a corner, and it is ignorance of the relative positions of these sequential frames of reference under a large one that defeats distance measures across them. This is true of disparate moments in the course of one continuous journey and a fortiori of two scenes from memory without a continuous experience connecting them. Given what we know about human navigation and spatial orientation and human memory, more often than not, to invoke scenery, connected or not, that sweeps through such multiple frames of reference or cuts from one to another in montage is to invoke visual experience for which there is no known path integration to a single larger frame of reference, again a fortiori if the cinerama sources several witnesses. Note that frames of reference and (sub)spaces come in all sizes, but those that annotate the observer’s scene are at least large enough for the entire space that the scene subtends. It has been supposed that all nominal phrases are AdrPs with addresses in frames of reference at which are located the events or things referred to: (235) *Unsolved murders were a cluster near the Green River in Washington. ‘Some unsolved murders en-scène in various frames of reference were in that frame of reference/one of those frames of reference a cluster near the Green River in Washington.’ (236) Some unsolved murders were a cluster near the Green River in Washington. ‘Some unsolved murders en-scène in a frame of reference were in that frame of reference a cluster near the Green River in Washington.’ If, as paraphrased in (235), the null determiner in a bare plural merely flags reference to many frames of reference (see section 13.1), and these are also said to be en-scène and the frames of reference for the projection of that scenery, then for there to be plural frames of reference, it must be a cinematic scene rather than still life, either sweeping or a montage of unsolved murders, the very conditions that more often than not thwart integration to a single frame of reference. But absent their integration to a single frame of reference in order to reveal that the spatiotemporal distances among them conform to the configuration of a cluster, it cannot be seen that the unsolved murders are indeed a cluster. In contrast, the overt article some in (236) supports reference to a single frame of reference with addresses for all the unsolved murders and recognition that they are a cluster. It is easy to get a counter counting, clicking away at just about nothing. They are robust measures only of the very small or brief visual stimuli that trip them. It is a complex, fragile judgment to deploy them to measure anything else. The CIA field agent in Red Square reliably and robustly counts the troopery across his field of view and reliably infers the number of troops filling Red Square, and he is bereft of any measurement of troop strength in Moscow that May Day. The VIP counting hand waves streaming by in the optic flow counts more than hand waves only under
468
Chapter 9
the special conditions under which her scene analysis and projection into her ambient egocentric frame of reference afford her orientation in the allocentric frame of reference where what else is to be counted persists. Counting is anomalous in a sentence when it is understood to taken on epistemic conditions that are inappropriate for what it purports to measure—arithmetically, or geometrically in the case of a cluster. This inverts expectations. It is the event counting, so-called, that is easier, and the object counting is that for which the epistemic conditions of observation and the background knowledge need to be just right. But how could it be otherwise when holding an actual counter and intending to use it to count passengers, faces in a crowd, photo-booth subjects, and so on?
10
Adverbialization in Logical Form
The analytic truths sentences (1)–(6) express do not commit their subjects to an active life beyond retirement nor deprive anyone of an alternative means of transportation despite their generic force: (1) (2) (3) (4) (5) (6)
The The The The The The
teacher teaches. president presides. dissenter dissents. motorist motors. cyclist cycles. pedestrian is on foot.
The generic quantification over events had thus better be restricted to the subject’s tenure in the classroom, in office, or on the barricades or to their locomotion by particular means: the teacher, whenever such, that is, whenever a teacher, teaches. The nominal is in effect a phase sortal (Burge 1975; Larson 1983, 1998; Enç 1986; Higginbotham 1987; Parsons 2000) concealing reference to events (or times) that is itself accessible to quantification when the nominal copied restricts the generic quantification over events. To render an analytic truth, which speakers recognize as such, say that the teacher teaching teaches. What serves the representation of these analytic truths also serves descriptive anaphora in (7) under the favored treatment (among others, Evans 1977, 1980; Davies 1981; Heim 1990; De Swart 1991; Ludlow 1994): (7) No teacher of a science postmodernizes it. ‘No teacher of a science, in any event of teaching a science, postmodernizes the science in that event.’ (8) Every fugitive from a federal prison is soon apprehended near it. ‘Every fugitive from a federal prison, in any event of being a fugitive from a federal prison as of some time is soon after the time of that event apprehended near the federal prison in that event.’ (After Enç 1986)
470
Chapter 10
Here too the descriptive content of the subject DP should restrict quantification over events, in order to describe the events that support singular reference to the science in such an event. Thus certain analytic truths and descriptive anaphora are evidence that nominal content joins adverbial content. In the logical forms for (1)–(7), as in the above paraphrases, adverbial modification is quantification over events, as I have assumed throughout. If so, recall from chapter 2 that there is a caution against implying in the analysis of (9), for example, that the parasite’s living is the same as the host’s dying: (9) The parasite living, the host dies. *[℩E: the parasite living[E]][∀e: Ee] the host dies[e] In general if p,q should not be rendered ‘whatever events that p are events that q’, as there need not be p-events identical to q-events. Rather, adverbial modification is mediated by some sort of accessibility relation,1 ‘N(e,e′)’, to be read “in the neighborhood of”: (10) The parasite living, the host dies. [℩E: the parasite living[E]][∀e: Ee][∃e′: N(e,e′)] the host dies[e′] Thus a logical form for (7), suppressing supermonadicity, looks like (11): (11) [No x: ∃e(teacher(e,x) & [A y: science(y)]Theme(e,y))] [℩E: ∀e(Ee ↔ (teacher(e,x) & [A y: science(y)]Theme(e,y)))] [∀e: Ee] [∃e′: N(e,e′)] … postmodernize(e′) … [℩y: science(y) & Theme(e,y)] Theme(e′,y) ‘No teacher of a science, in the neighborhood of any event of teaching a science, postmodernizes the science in that event.’ In the current setting with supermonadicity, the logical form for (1) will be something along the lines of (12), which remains an analytic truth to the extent that it is analytic that any event e is in its own neighborhood, N(e,e): (12) [the x : ∃E teacher[E,x]][℩E : teacher[E,x]] [∀e: Ee] [∃E0 : N[e,E0]] W[E0,x] [℩E0:pro0] Present[E0] O[E0,E1] ∃XAgent[E1,X] Cause[E1,E2] teach[E2]. ‘The teacher is such that in the neighborhood of any being a teacher of his he teaches.’ Supermonadicity delivers the second line in (12); the first line reflects adverbialization. The analytic truths in (1)–(3) and the descriptive anaphora in (7) put in evidence a relation between the content of a subject DP and a matrix generic event quantifier in what happens to be a generic assertion in these particular cases. Nevertheless, adverbialization applies with full generality to any token of nominal
Adverbialization in Logical Form
471
quantification, so that the events, Ej in (13), whose participants the nominal describes, are confined to the neighborhood where the nominal description finds them, (14). Adverbialization (13) … [Dα : ∃ENP[E,α]] (… θ[Ej,β] Ψ) ⇒ (14) … [Dα : ∃ENP[E,α]][℩E : NP[E,α]][∃Ei : N[E,Ei]](… θ[Ej,β] Ψ) The logical form for (15) is then as in (16), continuing to suppress event pronouns: (15) The teacher taught the student. DP-W-Tense-O-Agent-Cause-DP-W-O-Patient-teach (16) [the x : ∃E teacher[E,x]][℩E : teacher[E,x]][∃E0 : N[E,E0]] W[E0,x] Past[E0] O[E0,E1] ∃XAgent[E1,X] Cause[E1,E2] [the x: ∃E student[E,x]] [℩E : student[E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E2] ∃XPatient[E2,X]teach[E2]
(i) (ii) (iii) (iv) (v)
‘The teacher while (in the neighborhood of) being a teacher participated in some action (i–ii); the student while (in the neighborhood of) being a student participated as a patient in some teaching (iv–v); the action caused the studying (iii).’ That is, ‘The teacher teaching taught the student studying.’ The logical form of the open sentence x taught y, prior to closure by DPs, corresponds to lines (ii), (iii), and (v) in (16). Its closure, articulated in lines (i) and (iv), involves more than the prefixation of objectual quantifiers. A nominal quantifier, [D: NP], surely tokens quantification over the objects NP describes, but at the same time it tokens concurrent quantification over the events NP describes, and like any other adverb confines the matrix events to that neighborhood. Observe however that given supermonadicity, a nominal description circumscribes the neighborhood only for the events of the thematic relation associated with its host DP: only the actions in (16) need to be in the neighborhood of the teacher being a teacher, and only the student’s participation is fixed by this logical form to be in the neighborhood of the student’s being a student. This observation proves crucial for all the problems of subsequent sections and their solution. 10.0
The dependence of Tense and temporal reference on the descriptive content of nominal quantifiers
Supermonadicity revises the basic clause structure offered in (neo-)Davidsonian analysis so that even a simple sentence speaks of more than one event, at least as many as there are arguments.2 The analysis of kill as cause-die is especially vivid in that cause and effect are easy to tell apart, but here it is merely a special case of a
472
Chapter 10
more general formal point. It’s subevents all the way down—with every term in its own subevent. Life couldn’t be lonelier, spinning around in a subevent solo. Supermonadicity then invites further causal/topological relations to mediate among the subevents described, and the vocabulary of thematic relations is itself supplemented by ‘W[E,X]’, beyond what is invoked solely for the classification of who did what to whom (Agent, Theme, and so on.) Now, with so many events afoot, when it comes to adverbs or to any morpheme that describes events, such as Tense— to be discussed shortly—it must be asked which of the sentence’s several events it describes. It turns out that these morphemes discriminate among the events that supermonadicity delivers and thus support the revision of basic clause structure. Prior to the arcane temporal phenomena to be introduced, an elementary observation about Tense alone should in any case predispose an earnest Davidsonian toward supermonadicity: (17) The Big Bang is still propelling the universe’s expansion. (18) A protogalaxy 13.2 billion light-years ago illuminates the Hubble Space Telescope now that it is pointed in the right direction. (19) Supernova GRB 090423 at the most distant edge of the universe 13 billion years ago is bombarding the Swift space-based observatory with gamma radiation, which it was designed to detect. (20) Supernova 1987A 160,000 light-years ago in the Large Magellanic Cloud is triggering [February 1987] neutrino detectors at the Kamioka Observatory. (21) Eli Yale’s 1718 donation of ₤800 still supports scholarship students due to its prudent investment. (22) Eli Yale still supports scholarship students due to the prudent investment of his 1718 donation of ₤800. (23) The Chernobyl disaster is still causing birth defects in farm animals. If it is affirmed that an agent exists for the action it participates in, the events reported to be in the present in (17)–(23) must be distinct and at some remove from the actions their agents participate in. Even without action sentences, among statives, a similar point holds: (24) Franz Kafka is famous. (25) Gregor Samsa, the protagonist of The Metamorphosis and Kafka’s alter ego, is famous. (26) Vulcan, phlogiston, and luminiferous aether are all famous for not existing when they were supposed to. To regard as literal truths (25) and (26) does seem to engage some commitment to possible worlds or possibilia to warrant their reference to fictional and nonexistent objects. But, the literal truth of (24) does not demand a possible-Kafka in the present
Adverbialization in Logical Form
473
to bear the property of being presently famous if it is recognized that events entirely within the past in which an actual Kafka actually participated, his writing, made an imprint that is the subject of present fame from which he himself is absent (see section 10.0.8): (27) Agent(e, k) & Cause(e, e′) & Present(e′) & be famous(e′) Cf. Agent(e,k) & About(e, e′) & Present(e′) & fame(e′) There is little to tell apart talk of Eli Yale (22) from talk of Franz Kafka (24). Such talk sidesteps the nonactual and their ghosts only if the participation of theirs reported does not imply their participation in the present, from which a supermonadic logical form does indeed excuse them. The above observations favor supermonadicity without preference for adverbialization. Sections 10.0.1–10.0.5 argue that adverbialization and supermonadicity are both necessary to manage temporal reference. Under close scrutiny will be implicatures called the selective lifetime effect (Kratzer 1995; Musan 1997), where, uttered out of the blue, (29) in the past tense implicates that Lenin is dead and (31), that he is alive, both while leaving alone Queen Victoria’s current condition: (28) Lenin resembles Queen Victoria. (29) Lenin resembled Queen Victoria. (30) Lenin dead resembles Queen Victoria. (31) Lenin dead resembled Queen Victoria. Supermonadicity and adverbialization are necessary to explain why the implicature is a comment on the subject rather than the object—that is, where the selectivity comes from—and to explain the shift under the influence of the adverb from an implicature of death to one of resurrection. The lifetime effect is shown in turn to be a special case of the more general interaction of adverbialization and supermonadicity, according to which the derived adverbial temporally locates the events and only those events that are related under supermonadicity to the adverbialized DP. 10.0.0
Symmetric relations under Figure and Ground
A defense of supermonadicity takes an interest in discriminating between a sentence’s arguments, in finding cases where, say, the subject is off doing something that an object is not, and where this observed asymmetry neither reflects the chosen lexical items nor the intrinsic nature of the events they describe. Such cases, were they found, would suggest that sentences conceal other means—other structures and relations—to say one thing of one argument and something different of another. It would be welcome evidence of supermonadicity in being unrelated, unlike that of earlier chapters, to the syntax and semantics of conjunction and independent of any move to accommodate adverbialization.
474
Chapter 10
In section 2.5.1, in explaining the meaning of partial number agreement in Lebanese Arabic, I enlisted Talmy’s (1978, 1983) and Gleitman et al.’s (1996) studies of symmetric lexical relations: (32) The humblest citizen is equal to the President. (33) The President is equal to the humblest citizen. If ever there was a symmetric relation, it is is equal to, yet the sentences (32) and (33) grasp an asymmetry, with (32) intending to elevate the humblest citizen and (33), to diminish the President. It may appear from these examples that a truly symmetric relation is therefore beyond human expression, but the same lexical item manages to express as much in (34), which erases the asymmetry between the participants, reporting their equality without suggestion as to how they got there (although protocol may dictate to some who should be mentioned first). (34) The President and the humblest citizen are equal(s). The humblest citizen and the President are equal(s). If it is the same lexical item, equal, in (32)–(34), then nothing in its meaning contributes toward an asymmetry. If not the lexical item, neither will pragmatic context induce the asymmetry contrasting (35) and (36), which, as Gleitman et al. (1996, Experiment 5) show, survives the absence of real-world knowledge: (35) The ZUM is identical to the GAX. (36) The ZUM and the GAX are identical. Rather, Talmy and Gleitman et al. conclude that the asymmetry derives instead from the meaning of the construction surrounding the lexical item in (32) and (33). And, of course, the same constructional meaning conveying an apparent asymmetry is there to envelope any lexical item expressing an otherwise symmetric relation: (37) The button matched the shirt. (38) #The shirt matched the button. (39) The shirt and the button matched. (40) The bicycle is near the garage. (41) #The garage is near the bicycle. (42) The garage and the bicycle are ?(now) near each other. In the present context, constructional meaning amounts to the projection in phrase structure of fresh thematic relations, which I took to be perspectival relations drawing out the asymmetry between subject and object that the word equal does not itself express. In Talmy’s terms, the subject is the Figure in the event as presented, and the object, the Ground. Accordingly, the humblest citizen is the figure on the move in (32), the body whose position the assertion of (32) fixes against the heavenly background that is the President. The analysis of the perspectival relations invoked
Adverbialization in Logical Form
475
will depend on facts about perception, frames of reference, perceptual fields, the conditions under which something is taken to be a figure against something taken to be background, and so on. Their intentional character distinguishes these relations from the metaphysical bedrock—what’s invariant along with the relations deployed in describing that invariance. Identity is symmetric identity, nearness is symmetric nearness, no matter how we may frame the participants or come to understand how they arrived in such a symmetric condition.3 Perspective and its asymmetries also govern the interpretation of Tense. Sentence (43) implicates and, given special knowledge about transposition, implies that the prelude and the étude are no longer identical: (43) The prelude was identical to the étude until a few notes were transposed. But which has changed? Changes of relative position, as it were, are changes to the Figure against a constant Ground. That is, we understand the prelude to have undergone the transposition.4 As before, the asymmetry is effaced in (44), which gives no hint as to where the transposition was made: (44) The prelude and the étude were identical until a few notes were transposed. Similarly, both (45) and (46) implicate that Soul and Kaddish ceased to be equally dead—but (45) says that Ray Charles revived Soul, and (46) that he revived the Jewish mourners’ prayer: (45) Soul was as dead as Kaddish until Ray Charles recorded “What’d I say?” (46) #Kaddish was as dead as Soul until Ray Charles recorded “What’d I say?” Again, as Nixon never knew from Jimi Hendrix, it could not have been Nixon’s version of “The Star-Spangled Banner” that changed, contrary to (48): (47) “The Star-Spangled Banner” resembled “The Star-Spangled Banner” as Nixon knew it until Jimi Hendrix played at Woodstock in 1969. (48) #“The Star-Spangled Banner” as Nixon knew it resembled “The StarSpangled Banner” until Jimi Hendrix played at Woodstock in 1969. (49) “The Star-Spangled Banner” (as American youth knew it) and “The StarSpangled Banner” as Nixon knew it resembled each other until Woodstock 1969. Identity is, after all, identity. Why should it care which term in (43) topples out of it into nonidentity? Well, in fact, it doesn’t in (44). Likewise for death and the equative as … as in (45) and (46). Death as measured by degree is a grim numeric comparison the specific content of which cannot escape involving at bottom a fundamentally symmetric notion. Yet on top of this, sentences (45) and (46) manage to convey more. If Tense interacts differentially with subject and predicate, but nothing in the lexical predicate expresses an asymmetry, then that expression must reflect the
476
Chapter 10
meaning of the surrounding structure in some way. I should like to think that the temporal effects just noted are merely the temporal analog and a special case of Figure and Ground, just as I have presented it. The formal development that will follow nevertheless takes up Tense on its own, without taking on perception and perspectival relations. 10.0.1
The selective lifetime effect
In reaching for the temporal asymmetry in (43)–(49), one starts with the observation that the use of the past tense out of the blue5 implicates that what was is no longer. As Musan (1997) emphasizes, this is a Gricean implicature: first, given what is known about the longevity and stability of the attributed states and given that time is dense, it is known that these present-tense assertions imply their past-tense counterparts: (50) The prelude resembles the étude. → The prelude resembled the étude. [∃t: Present(t)] the prelude resemble(t) the étude → [∃t: Past(t)] the prelude resemble(t) the étude Surely if the prelude now resembles the étude, it also did so some moments ago in the past. The converse implication does not hold, and so for the speaker to settle on the weaker assertion is for her to implicate, assuming she would know better and would want us to know as much, that the present-tense assertion is not true—in other words, that what was is no longer, as in (52): (51) The prelude resembled the étude. (52) Asserted: [∃t: Past(t)] the prelude resemble(t) the étude Implicated: ¬[∃t: Present(t)] the prelude resemble(t) the étude Now repeated here are the fundamental observation and problem. In order for what was to be no longer, it suffices that either the prelude or the étude fall out of the resemblance that (51) reports to have held in the past. But (51) carries the further implicature that how things were with the subject is no longer, and thus that it is the prelude that has apparently changed. How then to derive the stronger implicature that privileges the condition of the subject? Kratzer (1995) and Musan (1997) call the special relationship between Tense and subject the lifetime effect. Like (43)–(49), the use of the past tense in (53)–(69) implicates that what is described is over, and, crucially, as the result of some change in the condition of the subject, be it death or disfigurement (54)–(58), decomposition (68), or resurrection (69): (53) (54) (55) (56) (57) (58)
Ted Kennedy resembles JFK.6 #Ted Kennedy resembled JFK. JFK resembled Ted Kennedy. Ted Kennedy is JFK’s youngest brother. #Ted Kennedy was JFK’s youngest brother. JFK was Ted Kennedy’s second-oldest brother.
Adverbialization in Logical Form
(59) (60) (61) (62) (63) (64)
477
Jim Belushi resembles John Belushi. #Jim Belushi resembled John Belushi. John Belushi resembled Jim Belushi. Jim Belushi is John Belushi’s brother. #Jim Belushi was John Belushi’s brother. John Belushi was Jim Belushi’s brother.
(65) Gorbachev is from the USSR. (66) #Gorbachev was from the USSR.
(Musan 1997) (Musan 1997)
(67) Lenin dead resembles Queen Victoria. (68) Stalin dead resembled Queen Victoria. (69) Lazarus dead resembled Uncle Ephraim napping. What cannot be entertained is that an analogous change in the condition of the object put an end to it. Thus, the dissolution of the Soviet Union does not rescue (66) from the implicature that Gorbachev has died. Similarly, to whatever extent (57) and (58) describe a state of affairs that is no longer, it can only reflect a change of state in the subject. The lifetime effect as it preoccupies Kratzer and Musan is called so from consideration of a limiting case, which will prove to be important in fixing the account to be offered here. A so-called individual-level attribution takes it to be common ground that the underlying relations are eternal at least with respect to the time under consideration: (70) resemble(x,y,t) → resemble(x,y,t’) (71) be-from(x,y,t) → be-from(x,y,t’) If a resemblance holds at all, it holds throughout, and one’s origins are also steady throughout. But if so, it is with individual-level attributions that the use of the past tense starts to look especially problematic. The use of the past tense implicates as above that what was is no longer. But common ground (70) and (71) says that what was resemblance or origins is forever, and so if something is no longer, there had better be something else (or else the past tense will be as groundless as it is in 1 was not equal to 0). So the resemblance lingers, but the subject dies, to fit the facts.7 Musan proposes that the meaning of the lexical item is itself asymmetric and stipulates that its subject is alive: (72) resemble(x,y,t) is true of iff Alive(x,t) & resemble(x,y,t) The implicature of an individual-level attribution then proceeds as follows: (73) #Ted Kennedy resembled JFK. (74) Asserted: Implicated:
[∃t: Past(t)] Ted Kennedy resemble(t) JFK ¬[∃t: Present(t)] Ted Kennedy resemble(t) JFK
478
Chapter 10
As before, the use of the past tense implicates that what was is no longer, (74). What was, now construing the lexical item as in (72), was that Ted Kennedy was both alive and resembling JFK. As the resemblance is itself eternal (70), if what was is no longer it can only be that Ted Kennedy is no longer. Hence, sentence (73) implicates that Ted Kennedy is no longer alive. Without any mention of JFK’s being alive, his condition cannot affect for better or worse the times at which the sentence holds, distinguishing subject and object as desired. In locating the lifetime effect inside the meaning of the lexical item, the proposal immediately runs into trouble in (75): (75) Lenin dead resembles Queen Victoria. The sentence is true, as any recent visitor to the Kremlin will confirm, but it should rather imply the contradiction that Lenin is both dead and alive if resemble had the meaning in (72). But anyway, all this is unresponsive to the fundamental problem that does not confine itself to individual-level attributions of an eternal resemblance. As we now know, the bloom of youth fades. Having viewed recent to March 2003 a televised interview, Henk Verkuyl then took (73) to be felicitous and true in that the portly septuagenarian senator no longer resembles the sporty JFK cut down in his prime and frozen in memory. This understanding of (73) nevertheless comports with the implicature to be derived. What was is no longer in that how things were with the subject is no longer: Ted Kennedy has changed and dissipated the resemblance. Removed from the common ground (70) that resemblance is eternal, lexical emendation will do little to derive the implicature that privileges the condition of the subject. Returning to (51) and (52), the use of the past tense implicates as before that the prelude’s resemblance to the étude is no longer. That is, either the prelude is dead, now allowing for (72), or the resemblance itself is no longer. Yet, again, it suffices for the latter that either the prelude, the subject, or the étude, the object, has changed. The logical form of (74) cleaves to the idea that Tense indicates the time of the resembling, as one might have hoped, and thus the above represents Musan’s attempt to escape from Kratzer’s (1995) essentially correct yet more radical insight that the implicature privileging the subject derives from a privileged association of Tense with the subject DP. I will present that insight in a formulation that allows easy access to some confirming evidence as well as to the reasoning that how things are with the subject matters. For the evidence, consider Lauren Bacall, the slinky screen siren of To Have and Have Not (1944),8 and Lauren Bacall, the stately dame of some recent Broadway cabaret, to whom is compared Linda Fiorentino, a contemporary film noir femme fatale (The Last Seduction, 1994), who can handle a gun without a
Adverbialization in Logical Form
479
torch song. Emerging from Bacall’s recent revue, it would be felicitous but false to say (76) or (77), having in mind the performance at hand:
(76) Linda Fiorentino resembles Lauren Bacall. (77) Linda Fiorentino resembles Lauren Bacall now. But it rings true to compare the film noir stars:
480
Chapter 10
(78) Linda Fiorentino resembles Lauren Bacall then [To Have and Have Not, 1944]. (79) Linda Fiorentino now resembles Lauren Bacall then. As in (78) and (79), the present tense is indifferent to Lauren Bacall then or now. Tense appears rather to agree, so to speak, with the subject, so that present tense is Linda Fiorentino now, and past tense fixes Linda Fiorentino then (anomalously, if we do not know when then is): (80) Linda Fiorentino resembled Lauren Bacall now. (81) Linda Fiorentino resembled Lauren Bacall then. (82) Linda Fiorentino then resembled Lauren Bacall now. (83) Linda Fiorentino then resembled Lauren Bacall then. (84) *Linda Fiorentino now [as I speak]9 resembled Lauren Bacall now. (85) *Linda Fiorentino now [as I speak] resembled Lauren Bacall then. The syntax of (76)–(85) makes it obvious in what respect there could be a privileged relationship between Tense and the subject, which can be taken up directly in logical form as follows. Let ‘ resemble ’ be a (timeless) relation between temporal slices or stages of objects, and let Tense apply to the temporal index of the subject and not to the object. Given what is known about the longevity and stability of resemblance and given that time is dense, it is known that these present-tense assertions imply their past-tense counterparts: (86) Linda Fiorentino resembles Lauren Bacall then. → Linda Fiorentino resembled Lauren Bacall then. [∃t: Present(t)] resemble → [∃t: Past(t)] resemble The converse implication does not hold, and so for the speaker to settle on the weaker assertion is for her to implicate, assuming she would know better, that the present-tense assertion is not true—in other words, that what was is no longer, as in (87): (87) Asserted: Implicated:
[∃t: Past(t)] resemble ¬[∃t: Present(t)] resemble
That is, there is implicated to be no present slice of Linda Fiorentino that resembles Lauren Bacall then. But, if present slices of Linda Fiorentino resemble her past slices, then her present slices would also resemble Lauren Bacall then, contrary to what is implicated. So, Linda Fiorentino’s present slices do not resemble her past slices, which is to say she has changed. (88) Implicated: [∃t: Past(t)][∃t: Present(t′)]¬ resemble
Adverbialization in Logical Form
481
Note that it does not matter whether or not Lauren Bacall changes. It does not matter what resemblances hold of Lauren Bacall now or at any time other than then. Given the truth of the past-tense assertion, there is always Lauren Bacall then to be related by resemblance to Linda Fiorentino past or present. Kratzer’s point is that the Tense morpheme locates the subject’s temporal slices and says nothing of the object’s.10 As the example illustrates, this derivation privileging the subject survives resemblances that fall short of eternal, individual-level attributions, holding only of things at certain stages or temporal slices. Tense must be explicit in applying to how things are with the subject to the exclusion of how they are with an object. Thus, despite Musan’s efforts and despite it being tokened as verbal morphology, it is not enough for Tense to locate the event of Fiorentino resembling Bacall. Taking to heart the logical forms in (86)–(88) that serve to make the point vivid, do we then concede that Tense is a slicer of objects, taking one in as its argument and returning a stage or temporal slice of that object? If we do it here, do we then take Tense to always be a nominal operator trapped in the body of a verbal morpheme? 10.0.2
Adverbial restriction and shift in the lifetime effect
If Tense is trapped into adnominal modification, it presses many an adverb into that service too. Consider the intervention of adverbs, which restrict or shift, the pragmatics of the lifetime effect: (89) Gorbachev resembles Khrushchev. (90) #Gorbachev resembled Khrushchev. (91) Gorbachev bald resembles Khrushchev. (92) Gorbachev bald resembled Khrushchev. (93) Gorbachev in a hairpiece resembles Brezhnev. (94) Gorbachev in a hairpiece resembled Brezhnev. As before, the past-tense assertion implicates the falsity of its present-tense counterpart, and it further implicates that how things were with the subject is no longer. But what is no longer in (92), for example, is the state of Gorbachev being bald, from which he has now been relieved either by death or by hair restoration (it being unclear which is the worse fate). The intervention of the adverb vacates the implicature that he is dead tout court. As we have seen earlier, an adverb accomplishes its most striking shift in (97): (95) Lenin resembled Queen Victoria. (96) Lenin dead resembles Queen Victoria. (97) Lenin dead resembled Queen Victoria.
482
Chapter 10
What is implicated is that how things were with Lenin dead are no longer: either the resemblance itself has faded, perhaps from Lenin’s stalled but progressive decomposition, or Lenin has joined the undead by resurrection. Even adverbs that are better dressed as such intervene with the same effect: (98) Lenin while resting dead resembles Queen Victoria. (99) Lenin while resting dead resembled Queen Victoria. (100) Lenin during his big sleep in the Kremlin resembles Queen Victoria. (101) Lenin during his big sleep in the Kremlin resembled Queen Victoria. Nevertheless, this reversal of the lifetime effect drives a conclusion that all of these adverbs are adnominal despite appearances, if Tense is itself adnominal. Attempting to forestall this conclusion, suppose to the contrary that adverbs remain adverbial while Tense is adnominal (granting the force of Kratzer’s argument for the latter). Recall the basic lifetime effect as it concerns simple sentences without adverbs, such as (102): (102) Lenin resembled Queen Victoria. (103) Asserted: Implicated:
[∃t: Past(t)] resemble ¬[∃t: Present(t)] resemble
Let it be as a matter of fact that Lenin resembled Queen Victoria throughout his lifetime. As Kratzer and Musan observe, it nevertheless satisfies what is implicated that Lenin is now dead. That is, it must be, for example, that ¬ resemble , which is expected to the extent that there is no 2003 temporal slice or stage of Lenin.11 Now an adverb that is not adnominal plays no role in describing the relevant temporal slice of Lenin. Rather, it qualifies what the predicate says about such a slice, along the lines of (105): (104) Lenin bald resembled Queen Victoria. (105) Asserted: Implicated:
[∃t: Past(t)] ( resemble & bald()) ¬[∃t: Present(t)] ( resemble & bald())
What is then implicated for is that ¬( resemble & bald()), that is, either the resemblance or baldness fails to hold of . Were Lenin still alive, the resemblance would still hold, but the implicature could still be satisfied provided that he were no longer bald in 2003, all of which taken together renders (104) correctly. Now the same reasoning applies mutatis mutandis for an adverbial occurrence of dead:
Adverbialization in Logical Form
483
(106) Lenin dead resembled Queen Victoria. (107) Asserted: Implicated:
[∃t: Past(t)] ( resemble & dead()) ¬[∃t: Present(t)] ( resemble & dead())
What is implicated for that same is that ¬( resemble & dead()). But, in satisfying (102)’s implicature (103), it has already been established that ¬ resemble , which now undermines the intended implicature for (106). For what is intended for (106) is that how things were with Lenin dead are no longer. That is, either Lenin has since joined the undead or the resemblance has withered. The implicature should fail if Lenin remains dead and is well enough preserved to keep the resemblance going—exactly the conditions that would verify the present-tense assertion that Lenin dead resembles Queen Victoria. Under these conditions, the implicature represented in (107) would however turn out true simply in virtue of Lenin being dead in 2003. What has gone wrong here is that the adverbial treatment of the adverbs bald and dead allows the same temporal slices to be presented as arguments to resemble in both (104) and (106). We should rather have that the temporal slices under consideration in (104) are just those of Lenin being bald (and alive), and these no longer resemble Queen Victoria if there are no longer temporal slices of Lenin being bald (and alive). Opposite (104), the temporal slices entering into a resemblance in (106) are just those of a being dead, and these no longer resemble Queen Victoria if there are no longer temporal slices of the being dead. But this is as much as to say that the adverbs bald and dead are themselves adnominal, restricting the argument delivered to Tense. Adnominal Tense means adnominal adverbs, too. 10.0.3
Semantic innocence regained
If Lenin, Lenin-s, Lenin-ed, Lenin bald-s, Lenin bald-ed, Lenin dead-s, and Lenin dead-ed all refer to different things, special axioms must then be invoked to explain when a speaker thinks what is true of one is also true of another.12 And if, for the sake of this construction, a Tense or an adverb is sometimes adnominal and sometimes adverbial, can it ever turn out to mean the same thing and partake of the same syntax, remaining itself unambiguous, whether tokened as adnominal or as adverbial?13 Recall what has driven this retreat from semantic innocence: there can be no account of the implicature that how things are with the subject is no longer—that what has changed is the prelude rather than the étude in (51) and Linda Fiorentino rather than Lauren Bacall in (80) or (81)—if Tense in its ordinary way is left to modify the (entire) resembling e as in ‘[∃e: Past(e)] x resemble(e,x,y)’ or in ‘[∃e: Past(e)](Theme(e,x) & resemble(e) & TO(e,y))’. So, with no place else to go,
484
Chapter 10
Tense squeezes onto the subject, as Kratzer (1995) concludes, taking adverbs with it. But under supermonadicity, there is plenty elsewhere to go, and a privileged association emerges between Tense and how things are with the subject without violence to Tense or adverbs as they are ordinarily understood. Each argument relates to its own events, and number agreement, ‘[℩E0:pro0]’, as a matter of fact (see chapter 2), already denotes what events the subject participates in alone, ‘W[E0,α]’: (108) W[E0,α] [℩E0:pro0]14 Past[E0] O[E0,E1] ∃XTheme[E1,X] … resemble[Ei] …15 Given the close company that Tense and number agreement keep, it is straightforward for Tense to apply to these events, as in (108), while those in which an object participates remain remote and apart.16 This realizes, I believe, Kratzer’s (1995) insight: the Tense morpheme locates what the subject is up to and pointedly says nothing (direct) about what involves the object. If Tense and adverbs are to remain adverbial, then the supermonadic logical form is necessary, as just argued, for subject and object in Ted Kennedy resembles JFK to go their separate ways, the one alive and the other dead. This logical form is necessary but not sufficient as the lexical item itself has something to say in the matter— resemble differing, for example, from be in Ted Kennedy is JFK, where now Ted Kennedy’s lifetime and JFK’s must coincide. I have yet to spell out how such differences enter into a logical form such as (108) varying the lifetime effects observed. So far all I have are some skeptical remarks banishing the lifetime effect from the meaning of resemble. So I will first take up in sections 10.0.4 and 10.0.5 the representation of the lifetime effect itself and the interaction with adverbial modification that prompted the earlier skepticism. Section 10.0.6 returns to the lexical differences that emerge when there are multiple lifetime effects to be coordinated—that is, whenever a sentence has more than one nominal argument. After having settled the questions of logical form, there follows in section 10.0.7 a walk through the targeted implicatures. 10.0.4
Adverbialized logical form
The “lifetime effect” requires a further articulation of logical form, relating the condition of the subject to the time of what is described by the matrix sentence. (109) Lenin resembled Queen Victoria. (110) Lenin bald resembled Queen Victoria. (111) Lenin dead resembled Queen Victoria. There are of course resemblances that engage not only the dead but also the abstract and immaterial—Triangle A resembles triangle B, Your conduct resembles the divine. If it is the same lexical item resemble that reports on all these, a lifetime effect could
Adverbialization in Logical Form
485
hardly be the burden of this lexical item. Yet Musan (1997) is certainly correct that something must be said somewhere if the mere use of the Past Tense in (109) and the implicature that what was is no longer further implicates that Lenin has gone to the grave. When individual-level attribution presupposes that resemblance is eternal, it is in fact only Lenin’s death that rescues (109) from infelicity. Compare the infelicitous Past Tense in (112) where numbers never die and nonidentity is eternal: (112) #1 was not equal to 0. Whatever is then said to secure the lifetime effect in (109), it must be said, as we have seen, in such a way as to admit both qualification, as in (110) where what is no longer now implicates death or hair restoration, and total reversal as in (111), where what is no longer implicates new life. As these three sentences (109)–(111) all embed (108), differing in their logical syntax only with respect to adverbial quantification, (113) Lenin … (108) … (114) Lenin bald … (108) … (115) Lenin dead … (108) …, nothing that (108) (or anything else taken to be the analysis of resembled Queen Victoria) says can impose the lifetime effect without running into the contradiction in (111) that Lenin is both dead and alive, the very objection to Musan’s lexical version of the lifetime effect ((72), above). If so, the lifetime effect derives from structures other than those such as (108) that supermonadicity alone supplies. Seeking elsewhere for a lifetime effect, recall that adverbialization in the logical form for (116) looks like (117), suppressing for the moment supermonadicity: (116) No teacher of a science postmodernizes it. ‘No teacher of a science, in any event of teaching a science, postmodernizes the science in that event.’ (117) [No x: ∃e(teacher(e,x) & [A y: science(y)]Theme(e,y))] [℩E: ∀e(Ee ↔ (teacher(e,x) & [A y: science(y)]Theme(e,y))] [∀e: Ee] [∃e′: N(e,e′)] … postmodernize(e′) … [℩y: science(y) & Theme(e,y)] Theme(e′,y) ‘No teacher of a science, in the neighborhood of any event of teaching a science, postmodernizes the science in that event.’ With supermonadicity, an adverbialized logical form for (118) will be something along the lines of (119), which remains an analytic truth to the extent that it is analytic that any event e is in its own neighborhood, N(e,e): (118) The teacher teaches.
486
Chapter 10
(119) [the x : ∃E teacher[E,x]][℩E : teacher[E,x]] [∀e: Ee] [∃E0 : N[e,E0]] W[E0,x] [℩E0:pro0] Present[E0] O[E0,E1] ∃XAgent[E1,X] Cause[E1,E2] teach[E2]. ‘The teacher is such that in the neighborhood of any being a teacher of his he teaches.’ Supermonadicity delivers the second line in (119) (cf. (108)); the first line is a further articulation of logical form. It applies to every token of nominal quantification, so that the events, Ej in (120), whose participants the nominal describes, are confined to the neighborhood where the nominal description finds them, (121). (120) … [Dα : ∃ENP[E,α]] (… θ[Ej,β] Ψ) ⇒ (121) … [Dα : ∃ENP[E,α]][℩E : NP[E,α]][∃Ei : N[E,Ei]](… θ[Ej,β] Ψ) Recall that the adverbialized logical form for (122) is then as in (123), continuing to suppress event pronouns: (122) The teacher taught the student. DP-W-Tense-O-Agent-Cause-DP-W-O-Patient-teach (123) [the x : ∃E teacher[E,x]][ιE : teacher[E,x]][∃E0 : N[E,E0]] W[E0,x] Past[E0] O[E0,E1] ∃XAgent[E1,X] Cause[E1,E2] [the x: ∃E student[E,x]] [℩E : student[E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E2] ∃XPatient[E2,X]teach[E2]
(i) (ii) (iii) (iv) (v)
‘The teacher while (in the neighborhood of) being a teacher participated in some action (i–ii); the student while (in the neighborhood of) being a student participated in as a patient in some teaching (iv–v); the action caused the studying (iii).’ That is, ‘The teacher teaching taught the student studying.’ Recall also that a nominal description circumscribes the neighborhood only for the events of the thematic relation associated with its host DP: only the actions in (123) need to be in the neighborhood of the teacher being a teacher, and only the student’s participation is fixed by this logical form to be in the neighborhood of the student’s being a student. 10.0.5
The lifetime effect reconfigured
To know what it means to be a teacher, president, or dissenter is to know among other things that (124) teacher(e,x) → alive(e,x) president(e,x) → alive(e,x) dissenter(e,x) → alive(e,x)
Adverbialization in Logical Form
487
So it is with Lenin (and many another name if not all) that one knows that (125) Lenin(e,x) → alive(e,x) All DPs are quantifiers with some content, which adverbialized logical form joins to an adverbial quantifier over events. Thus the DP Lenin is a definite description, ‘[℩x : ∃ELenin[E,x]],’ containing a phase sortal predicate.17 The adverbialized logical forms for (109)–(111) (repeated below) are (126)–(128) respectively, where (108) (also repeated below) translates resembled Queen Victoria. According to (126), the past events that amount to resembling Queen Victoria are found in the neighborhood of Lenin-izing; (127) and (128) rather put them in the neighborhood of being bald or being dead respectively, which is in turn in the neighborhood of Lenin-izing. (109) Lenin resembled Queen Victoria. (110) Lenin bald resembled Queen Victoria. (111) Lenin dead resembled Queen Victoria. (108) … W[E0,α] [℩E0:pro0] Past[E0] O[E0,E1] ∃XTheme[E1,X] … resemble[Ei] … (126) [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]][∃E0 : N[E,E0]] ((108) … Past[E0] …) ‘Lenin Lenin-izing resembled Queen Victoria.’ (127) [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]] [ιE-1 : N[E,E-1] bald[E-1]]18[∃E0 : N[E-1,E0]]((108) … Past[E0] …) ‘Lenin Lenin-izing bald resembled Queen Victoria.’ (128) [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]] [℩E-1 : N[E,E-1] dead[E-1]][∃E0 : N[E-1,E0]]((108) … Past[E0] …) ‘Lenin Lenin-izing dead resembled Queen Victoria.’ As argued earlier, nothing in the translation of resemble Queen Victoria implies that its subject is alive. The lifetime effects felt in the sentences (109)–(111) will derive instead from the tokens of ‘Lenin[E,x]’ in (126)–(128) and the knowledge (125) that to Lenin-ize in some events is to live them. Crucially, the tokens of ‘Lenin[E,x]’ that carry this implication occur outside the scope of the adverbials, bald and dead, that shift the lifetime effect. The intervening adverbials can thus distance the Leninizing from the subject’s participation in the resembling, and this will be the basis for discriminating among (109)–(111). The variation in the lifetime effect and, in particular, its reversal in (111) rest unsurprisingly on a difference in meaning between the adverbs bald and dead. It must be according to the meaning of bald that for a balding to be in the neighborhood of Lenin-izing, the participant in the balding is the same as the participant in the Lenin-izing, and the participant in the balding is as alive there as he is in the Lenin-izing. The same cannot be said for dead, and the difference between bald and
488
Chapter 10
dead is also apparent in the contrast between the robust judgments in (129) and the quicksand in (130): (129) T Lenin bald is still Lenin. (130) Lenin dead is still Lenin.
F Lenin bald is not Lenin. Lenin dead is not Lenin.
A being dead in the neighborhood of Lenin-izing may involve the same participant (cf. note 19), but the participant in the being dead is clearly not as alive as the Leninizer. Recall that the subject’s participation in the resemblance, E0 in (127) and (128), is then located in the neighborhood of the balding (127) or in the neighborhood of the being dead (128). Neighborhoods, I assume, are monotonic with respect to the qualia of their participants that have been antecedently established, so that the participant in a resemblance in the neighborhood of a balding is as alive as the participant in the balding, and correlatively the participant in the neighborhood of a being dead is as alive as the participant in the being dead. This last observation is then what will effectively allow the lifetime effect to shift or reverse under adverbial iteration (see section 10.0.7).19 It is important that what is observed here is nothing not already exemplified when other conditional modifiers iterate, If Φ0; if Φ1, Ψ: (131) If the match were struck, it would light. (Stalnaker 1968) (132) If the match were struck; then, if it had been soaked in water overnight, it would not light. (133) If Lenin is a hero, he fights the oppressors of the working class. (134) If Lenin is a hero, then (even so) if he is dead, he does not fight the oppressors of the working class. (135) If Lenin is a hero, then if he stands bald at the vanguard, he fights the oppressors of the working class. Reverting to more familiar talk about possible worlds (rather than events), note that (133) is true only if in the closest worlds where Lenin is a hero, he is alive—else he would not be fighting in these worlds. But these worlds cannot be among the closest worlds in (134) where he is dead, the internal conditional having, as it were, shifted the neighborhood. On the other hand, the worlds in which he is a hero and alive remain among the closest in (135) where he stands at the vanguard. Beyond the proposal that the nominal contributes its content to an (absolutive, conditional) adverbial, Lenin-izing, occurring in the position indicated in adverbialized logical form, I intend no semantic innovation or special provisions for this construction. Interpretation proceeds as it does elsewhere when adverbials are iterated, whether or not such adverbial modification is couched in terms of event quantification as it is here.20
Adverbialization in Logical Form
489
Reviewing those aspects of logical form contributing toward the lifetime effect, recall first that supermonadicity in providing each argument a clause to describe its own events allows (and even compels) Tense and adverbial modifiers to describe one argument’s participation rather than another’s, thus realizing Kratzer’s (1995) insight that it is a matter for logical form that Tense describes how things are with the subject, E0 in (126)–(128). In all this, Tense and the adverbial modifiers appear innocently, interpreted here as they are elsewhere. So too is the lexical item resemble innocent, univocal and unencumbered by any sense that does not seem faithful to its meaning elsewhere: (136) Lenin’s embalmed corpse resembles Queen Victoria. (137) Lenin’s statue resembles Queen Victoria. (138) Lenin’s likeness resembles Queen Victoria. (139) Triangle A resembles triangle B. (140) The unattainable resembles the divine. The lifetime effects as such in (109)–(111) derive from tokens of ‘Lenin[E,x]’ in (126)–(128) and the knowledge (125) that to Lenin-ize in some events is to live them. Crucially, adverbialized logical form makes an adverb out of ‘Lenin[E,x]’ and fixes its position outside the scope of the adverbials, bald and dead, which then shift the lifetime effect as above. Were it enough only to discriminate the effects of adverbial modification in (109)– (111), it would suffice, without Lenin-izing and without treating the name Lenin as anything other than a term referring directly to Lenin, to introduce being alive as an adverb on its own, in place of ‘[℩E : Lenin[E,x]]’, as in (141)–(143): (108) … W[E0,x] [℩E0:pro0] Past[E0] O[E0,E1] ∃XTheme[E1,X] … resemble[Ei] … (109) Lenin resembled Queen Victoria. (141) λx([℩E : alive[E,x]][∃E0 : N[E,E0]] (108))(l) (110) Lenin bald resembled Queen Victoria. (142) λx([℩E : alive[E,x]] [℩E′ : N[E,E′] bald[E′]] [∃E0 : N[E′,E0]] (108))(l) (111) Lenin dead resembled Queen Victoria. (143) λx([℩E : alive[E,x]] [℩E′ : N[E,E′] dead[E′]] [∃E0 : N[E′,E0]] (108))(l) Of course the appearance of this particular adverb here would be entirely adventitious, and its disappearance in (136)–(140), equally mysterious. Rather, the lifetime effect seems to arise just in case the underlying concept (e.g., being Lenin) happens to entail being alive. This co-occurrence restriction, the existence of the lifetime effect and its dependence on the choice of nominal, follows only from the adverbialization of nominal content as adverbial content.21
490
10.0.6
Chapter 10
Multiple lifetime effects and multiple adverbs, asynchronous but coordinated
Given supermonadicity, a nominal description under adverbialized logical form circumscribes a neighborhood only for the events of the thematic relation associated with its host DP. The lifetime effects governing one thematic relation need not coincide with those of another. As Musan (1997) remarks, (144) says that Gorbachev and his being from the USSR continue into the present, without it saying the same for the USSR itself. But a further observation is that the prepositional argument may impose its own lifetime effects, synchronized in some fashion with the others but not necessarily coincidental. One cannot, after all, say (145), precisely because there has been no overlap between Gorbachev’s lifetime and the last czar’s. (144) Gorbachev is from the USSR. (145) #Gorbachev is from czarist Russia. Moreover, mere overlap in the lifetimes of the principals would not be sufficient for the being from. Suppose that, with his departure from office, Gorbachev had abandoned Russian soil, relinquishing his residences and all his personal ties there for self-imposed exile in Geneva. In that case, one could not say (146): (146) #Gorbachev is from post-Soviet Russia. (147) I never made it to Leningrad, but I visited St. Petersburg last week. #I never made it to Leningrad, but I visited Leningrad last week. (Saul 1997a) Although the lifetimes of Gorbachev, the post-Soviet era, and post-Soviet Russia all overlap, that overlap does not contain a relevant being from, so it seems.22 Authority has already addressed the question as it concerns the lifetimes of Lenin, Petersburg, Petrograd, and Leningrad: On reading over what I had written I found that I had called Leningrad in my recollections either “Petrograd” or “Petersburg,” while many other comrades call the Petrograd of old times “Leningrad.” This seems to me wrong. Can one say, for example: “Lenin was imprisoned in Leningrad”? It is clear that Lenin could not be imprisoned in Leningrad. Still less can one say: “Peter I founded Leningrad.” Perhaps in the course of years or decades the new name of the city—as all proper names in general—will lose its actual historical meaning.23 But for the present we still feel too clearly and acutely that Petrograd is called “Leningrad” only since the 21st of January 1924, and could not be called so before. Therefore, in these recollections of Leningrad, I keep to the name by which it was known in the time of the events described. (Leon Trotsky [1924] (1971), a contemporary of Prince Nikolai Troubetzkoy and Roman Jakobson)24
The occasion for the lifetime effects of subject and (prepositional) object to diverge is a matter that the lexical items in the construction decide. Gorbachev’s
Adverbialization in Logical Form
491
current state, as reported in (148) and (149), is the result of there having been within his extended now certain events of birth or education in the lifetime of the USSR. (148) Gorbachev is from the USSR. (149) Gorbachev is Stavropol born and Moscow educated. Absent this partitive aspect, the state in which Gorbachev participates temporally coincides with that in which the USSR participates. Thus (150), in contrast to (148), runs up against the shortened lifespan of the USSR. Gorbachev cannot now be in what is no longer: (150) #Gorbachev is in the USSR. Some coordination is always implied among the lifetime of the subject, the lifetimes of the objects, and the time of the events relating them, but it need not be overlap or coincidence as in the preceding examples. In fact, meaning may dictate that there is no overlap among the event or state reported and the time the nominal descriptions hold true: (151) a. Alfonso XIII succeeded Alfonso XII to the Spanish throne. b. Alfonso XIII was the successor of Alfonso XII to the Spanish throne. The successor cannot be Alfonso XIII before Alfonso XII is no longer. Despite the lack of overlap, the times when the nominal descriptions hold and the time of the succession are related to each other by strict prescription, and they can become rather entangled. Suppose, for example, that Alfonso XII is also Alfonso XI, having earlier abdicated the throne for a career at the opera. After the death of his successor Boffo XI½ and his operatic ambitions thwarted, the Spanish law of succession reinstates him as Alfonso XII. Accordingly, (151)–(154) faithfully record the Spanish succession (152), but (155)–(157) do not. (152) Alfonso XI (1857–1868)*—Boffo XI½ (1868–1874)—Alfonso XII (1874– 1885)—Regency (1885–1902)**—Alfonso XIII (1902–1931). *Alfonso XI abdicates to sing the leading role in La Cenerentola and is reinstated according to the law of succession in 1874 as Alfonso XII. **Alfonso XII dies in 1885 during his consort’s pregnancy. Maria Christina of Austria gives birth in 1886 and remains regent until 1902 when Prince Fonsy is installed as Alfonso XIII. (From Monty Python’s History of the Spanish Seat) (153) a. Boffo XI½ succeeded Alfonso XI to the Spanish throne. b. Boffo XI½ was the successor of Alfonso XI to the Spanish throne.
492
Chapter 10
(154) a. Alfonso XII succeeded Boffo XI½ to the Spanish throne. b. Alfonso XII was the successor of Boffo XI½ to the Spanish throne. (155) a. *Boffo XI½ succeeded Alfonso XII to the Spanish throne. b. *Boffo XI½ was the successor of Alfonso XII to the Spanish throne. (156) a. *Alfonso XI succeeded Boffo XI½ to the Spanish throne. b. *Alfonso XI was the successor of Boffo XI½ to the Spanish throne. (157) a. *(In 1886,) Alfonso XIII succeeded Alfonso XI to the Spanish throne. b. *(In 1886,) Alfonso XIII was the successor of Alfonso XI to the Spanish throne. Surely what informs the judgment against (155) is that no Boffo XI½-ing succeeds Alfonso XII-ing and that, on the contrary, the reverse is true—the Alfonso XII-ing succeeds the Boffo XI½-ing. Whether it is temporal overlap, coincidence, or disjointness, two aspects of logical form make possible the variation in lifetime effects: that nominal descriptions circumscribe the neighborhood for only the events described by the associated thematic relation and that thematic relations under supermonadicity come to describe different events even within the same sentence. Consider how to represent the lifetime effects of (158), where the lifetimes of the USSR and MSNBC Online do not overlap. (158) Gorbachev went from (a boyhood in) the USSR to (his desk at) MSNBC Online. (159) ∃e (Theme(e, g) & Past(e) & go(e) & from(e,USSR) & to(e, MSNBC Online)) The event that a simplified logical form (159) would describe covers the entire trajectory of Gorbachev’s career, only proper parts of which can be located within the lifetime of the USSR and within the lifetime of MSNBC Online. If the lifetime effect is to be made explicit, the preposition from should rather express a relation ‘from(e,e′)’ between the course of his career and an initial segment in which the USSR participates during its lifetime, and the preposition to, ‘to(e,e″)’, should similarly relate his career to a final segment within the lifetime of MSNBC Online:25 (160) Gorbachev went from the USSR to MSNBC Online. DP-W(e)-Past(e)-Theme(e,α)-go(e)-DP-W(e′)-Source(e′,β)-from(e,e′)-DPW(e″)-Goal(e″,γ)-to(e,e″)26
Adverbialization in Logical Form
493
(161) [℩x : ∃EGorbachev[E,x]][℩E : Gorbachev[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0 : pro0]Past[E0]O[E0,E1] ∃XTheme[E1,X]O[E1,E2] go[E2] [the x : ∃EUSSR[E,x]][℩E : USSR[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E3]∃XSource[E3,X]from[E2,E3] [℩x : ∃E MSNBC Online[E,x]][℩E : MSNBC Online[E,x]] [∃E0 : N[E,E0]] W[E0,x]O[E0,E4]∃XGoal[E4,X]to[E2,E4]
(i) (ii) (iii) (iv) (v) (vi)
Gorbachev, Gorbachev-ing, in the past, participated (as Theme) in some going (i–ii); the USSR, USSR-ing, participates (as Source) in some events (iii–iv); those goings were from those USSR-ings (iv); MSNBC Online, MSNBC Online-ing, participates as Goal in some events (v–vi); those goings were to those MSNBC Online-ings (vi). In (161), adverbialized logical form imposes parallel treatments on Gorbachev, the USSR and MSNBC Online (and on any other DP, see (120)–(121)): the USSR’s participation is in the neighborhood of USSR-ing and MSNBC Online’s participation is in the neighborhood of MSNBC Online-ing, just as Gorbachev’s participation is in the neighborhood of Gorbachev-ing. Thus, if participation while Gorbachev-ing implies Gorbachev’s live participation, so does participation while USSR-ing imply the USSR’s live participation, and similarly for MSNBC Online. Of course, even if they are alive when they participate in their respective events, these events are not all coextensive with the going itself as sections 10.0.1. and 10.0.2 took pains to argue in the case of resembling Queen Victoria. Rather, the relations among what Gorbachev does, what the USSR does, and what MSNBC does are to be settled by the choice of lexical items, go, from, and to in (160). The prepositions ‘from[Ei,Ej]’ and ‘to[Ei,Ek]’ put some distance between the events they relate, but not so the preposition ‘in[Ei,Ej]’, where its meaning is all that tells apart (148) from (150): (148) Gorbachev is from the USSR. DP-W-Tense-Theme-Be-DP-W-Source-from (162) [℩x : ∃EGorbachev[E,x]][℩E : Gorbachev[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0 : pro0]Present[E0]O[E0,E1]∃XTheme[E1,X]O[E1,E2] be[E2] [the x : ∃EUSSR[E,x]][℩E : USSR[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E3]∃XSource[E3,X]from[E2,E3]
(i) (ii) (iii) (iv)
494
Chapter 10
(150) #Gorbachev is in the USSR. (163) [℩x : ∃EGorbachev[E,x]][℩E : Gorbachev[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0 : pro0]Present[E0]O[E0,E1] ∃XTheme[E1,X]O[E1,E2] be[E2] [the x : ∃EUSSR[E,x]][℩E : USSR[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E3]∃XLocation[E3,X]in[E2,E3]
(i) (ii) (iii) (iv)
As a result of the intervention of ‘from[E2,E3]’ in (162) the events E3 from the lifetime of the USSR need not endure as long as those present events E2 from Gorbachev’s.27 In contrast, (150) runs up against the shortened lifespan of the USSR. It is sufficient to derive the anomaly that the meaning of in[Ei,Ej] requires Ei and Ej to coincide temporally, thus imposing the present on the lifetime of the USSR.28 Similar considerations drive the account of what goes wrong in (155). Our understanding of succession puts some distance between what Boffo XI½ participates in and what befalls Alfonso XII. The former precedes the latter, and logical form articulates that the former happened while Boffo XI½-ing and the latter, while Alfonso XII-ing, contrary to the Spanish succession in (152): (155) *Boffo XI½ succeeded Alfonso XII to the Spanish throne. DP-W-Tense-Theme-(isuc)ceed-DP-W-Source-SUBi-DP-W-Goal-to29 (164) [℩x : ∃EBoffo XI½[E,x]][℩E : Boffo XI½[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0 : pro0]Past[E0]O[E0,E1] ∃XTheme[E1,X]O[E1,E2] (suc)ceed[E2] [℩x : ∃EAlfonso XII[E,x]][℩E : Alfonso XII[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E3]∃XSource[E3,X]SUB[E2,E3] [the x : ∃E Spanish throne[E,x]][℩E : Spanish throne[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E4]∃XGoal[E4,X]to[E2,E4] The same is to be said for resemble in (165) and (166), the symmetric, stative predicate at the center of the discussion about how to derive the implicature that how things were with the subject is no longer: (165) Lenin resembled Queen Victoria. (166) Ted Kennedy resembles JFK. Generalizing the adverbialization of logical form to all argument positions, to both Ted Kennedy and JFK in (166), the logical form for (166) develops along the lines of (167):
Adverbialization in Logical Form
495
(167) [℩x : ∃E EMK[E,x]][℩E : EMK[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0:pro0]Present[E0]O[E0,E1]∃XTheme[E1,X]O[E1,E2] resemble[E2] [℩x : ∃E JFK[E,x]][℩E : JFK [E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E3] ∃XGoal[E3,X]TO[E2,E3]
(i) (ii) (iii) (iv)
‘Ted Kennedy (in the neighborhood of) EMK-ing participates as Theme in some events of resembling (i–ii); JFK (in the neighborhood of) JFK-ing participates as Goal in some events (iii–iv); that resemblance is to those events (JFK’s) (iv).’ Again, the treatments of Ted Kennedy and JFK are parallel: JFK’s participation is in the neighborhood of JFK-ing just as Ted Kennedy’s participation is in the neighborhood of EMK-ing. Thus, if participation while EMK-ing implies Ted’s live participation in E1, so does participation while JFK-ing imply JFK’s live participation in E3. The discussion of the lifetime effect has of course emphasized that E1 and E3 do not coincide—that the subject is alive now and the resemblance ought to hold now but JFK is not now. Thus the relation that the structure must provide to relate what Ted Kennedy does to what JFK does, ‘TO[E2,E3]’ in (167), must not be one of overlap or coincidence. In fact, any relation will do—even a transitive ‘resemble[E2,E3]’ (omitting ‘TO[E2,E3]’)—intruding anywhere in the structure to sever the coincidence of the events E1 and the events E2, provided that the present tense in (167) continues to confine the subject’s participation, E1. Recall that one may, as Henk Verkuyl did, judge (166) false in that the portly septuagenarian no longer resembles the JFK of his youth. The earlier resemblance in that case should not turn out to verify the present assertion. The logical form should rather imply that both the EMK-ing and the extent of Ted Kennedy’s participation in the resemblance had better be now, as (167) implies. The displacement in time can enter only when the analysis turns in lines (iii–iv) to JFK’s participation.30 In (168), overt and independent modification of both the subject’s and object’s participation steps in to corroborate the logical form (see also (76)–(85) above). (168) The former Soviet republics, struggling against the EU economy, resemble Imperial Russia (when (she was)) besieged by Napoléon. [the X : ∃Eformer Soviet republic[E,X]] (i) [℩E : former Soviet republic[E,X]] [℩E′ : N[E,E′] struggling[E′] against the EU economy][∃E0 : N[E′,E0]] (ii) W[E0,X] Present[E0] O[E0,E1]∃YTheme[E1,Y]O[E1,E2] resemble[E2] (iii) [℩x : ∃EImperial Russia[E,x]][℩E : Imperial Russia[E,x]] [℩E′ : N[E,E′] besieged[E′] by Napoléon][∃E0 : N[E′,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3]
(iv) (v) (vi)
496
Chapter 10
‘The former Soviet republics, while republicizing while struggling against the EU economy, in the extended now participate as themes in some events of resembling (i–iii); Imperial Russia while imperial while besieged by Napoléon participates as Goal in some events (iv–vi); the resembling is to those Imperial Russian events (vi).’ (169) #The former Soviet republics, struggling against the EU economy, resemble Imperial Russia (when (she was)) besieged by the Third Reich. #The former Soviet republics, struggling against the EU economy, resemble Bolshevik Russia (when (she was)) besieged by Napoléon. As the present tense dictates, the resemblance holds of the former Soviet republics’ current condition, described as a current struggle against the EU economy. On the other hand, Imperial Russia is not now besieged by Napoléon. But if ‘(when (she was)) besieged by Napoléon’ is an adverbial phrase as ordinarily understood and no different really from ‘struggling against the EU economy’, it too describes events, which must be—as supermonadicity provides—other than those overlapping the present. If so, there must be in the structure some relation, such as ‘TO[E2,E3]’, as a bridge between the events that the one adverb describes and those that the other describes, and the content of this relation must allow the events to span long distances in time. The overt adverbial phrases then dovetail with those that adverbialized logical form supplies to further restrict participation to the neighborhoods that nominal content describes, (168). At this point, rather than go blind from the likes of (160)–(169), one may be tempted to say with less fuss that there are (at least) six distinct objects, czarist Russia, Romanov Russia, post-Soviet Russia, the Soviet Russian Republic, Stalinist Russia, and Russia, and aver that Gorbachev is from three of them, those x that ∃e from(e,g,x). It is to admit that any two time slices of Russia are different objects to the extent that Russian demographics finds someone who is from the one and not from the other. Maybe political entities are so fickle, but the contrast between (153) and (155), that Boffo XI½ succeeds Alfonso XI but not Alfonso XII, will also yield proof that Alfonso XI is not the same object as Alfonso XII if one is so intent on purging logical form of talk about Alfonso XI-ing and Alfonso XII-ing. If many Russias and Spanish kings who outnumber their own persons is not already more ontological angst than one can bear, adverbialized logical form has already been defended on grammatical grounds. It is but a generalization of what is already required for the logical syntax representing the analytic (118) (see (1)–(3)), and it alone delivers the distribution of the lifetime effect while assimilating further
Adverbialization in Logical Form
497
adverbial modification. It tames the metaphysics and at the same time explains that (i) the existence of the lifetime effect is dependent on the choice of nominal, (ii) adverbial modifiers (successively) narrow or shift the lifetime effect, and (iii) the lifetime effect, derived from nominal descriptions standing in as adverbial modifiers, like all adverbial modifiers, governs local participation in small subevents.31 10.0.7
That how things were with the subject is no longer
Supermonadic, adverbialized logical form and analysis of resemble deliver the logical forms for (109)–(111) displayed in (170)–(172) respectively and differing only in adverbial modification: (109) Lenin resembled Queen Victoria. (170) [℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]][∃E0 : N[E,E0]] W[E0,x] Past[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2]
(i) (ii)
[℩x : ∃EQueen Victoria[E,x]][℩E : Queen Victoria[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3]
(iii) (iv)
‘Lenin (in the neighborhood of) Lenin-izing participated as Theme in some resembling (i–ii); Queen Victoria (in the neighborhood of) QV-ing participated as Goal in some events (iii–iv); the resembling is to Queen Victoria’s events (iv).’ (110) Lenin bald resembled Queen Victoria. (171) [℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]] [℩E′ : N[E,E′] bald[E′]]32 [∃E0 : N[E′,E0]] W[E0,x] Past[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2]
(i) (ii) (iii)
[℩x : ∃EQueen Victoria[E,x]][℩E : Queen Victoria[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3]
(iv) (v)
‘Lenin (in the neighborhood of) Lenin-izing, balding, participated as Theme in some resembling (i–iii); Queen Victoria (in the neighborhood of) QV-ing participated as Goal in some events (iv–v); the resembling is to Queen Victoria’s events (v).’ (111) Lenin dead resembled Queen Victoria.
498
Chapter 10
(172) [℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]] [℩E′ : N[E,E′] dead[E′]] [∃E0 : N[E′,E0]] W[E0,x] Past[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2]
(i) (ii) (iii)
[℩x : ∃EQueen Victoria[E,x]][℩E : Queen Victoria[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3]
(iv) (v)
‘Lenin (in the neighborhood of) Lenin-izing, dead, participated as Theme in some resembling (i–iii); Queen Victoria (in the neighborhood of) QV-ing participated as Goal in some events (iv–v); the resembling is to Queen Victoria’s events (v).’ As before, given what is known about the longevity and stability of the attributed states, here stretched out to include what is known about neighborhoods and so on, and given that time is dense, it is known that these present-tense assertions imply their past-tense counterparts. Surely if Lenin, simple, bald, or dead, now resembles Queen Victoria, he also did so some moments ago in the past. As before, the converse implication does not hold, and so for the speaker to settle on the weaker assertion is for her to implicate, assuming she would know better and would want us to know as much, that the present-tense assertion is not true—in other words, that what was is no longer. Thus, (109)–(111) implicate that their present-tense counterparts are false, as in (173) for example: (109) Lenin resembled Queen Victoria. (173) Asserted: [℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]][∃E0 : N[E,E0]] W[E0,x] Past[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2] [℩x : ∃EQueen Victoria[E,x]][℩E : Queen Victoria[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3] Implicated: ¬[℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]][∃E0 : N[E,E0]] W[E0,x] Present[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2] [℩x : ∃EQueen Victoria[E,x]][℩E : Queen Victoria[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3]
(i) (ii) (iii) (iv) (i) (ii) (iii) (iv)
To show the further implicature that how things were with the subject is no longer, suppose the contrary. Suppose that Lenin lives on, in all relevant respects unwavering since his resemblance to Queen Victoria was first established during her sitting
Adverbialization in Logical Form
499
for the Diamond Jubilee portrait of 1897. Queen Victoria, on the other hand, lapses from the resemblance, suffering from the rigors of Empire. What was is no longer, but this change from how things were with the object does not satisfy the implicature in (173). For, if Queen Victoria at her 1897 sitting, as captured on film, verifies Lenin’s asserted past resemblance to her, and if Lenin is constant, his present condition continues to bear the resemblance to her of that 1897 sitting. According to the logical form on offer, this resemblance to her at that sitting suffices in turn for Lenin to resemble in the present Queen Victoria.33 What is implicated by (173) is thus false, as desired when only the object has changed. So it is with all the implicatures for (109)–(111). Now to the lifetime effect proper. To expose it, assume, as in Kratzer’s and Musan’s discussion of individual-level attributions, that the resemblance between Lenin and Queen Victoria has been steady throughout, without either one ever departing from it in relevant respects. As before, to assert (109) is to implicate that its present-tense counterpart is false, as shown in (173). Now if there were Leninizing in the present—that is, if some present events were in the neighborhood of Lenin-izing—then given the individual-level attribution, it would continue to bear the resemblance to Queen Victoria, contrary to what is implicated. The only way the present-tense counterpart could fail to be true under the circumstances is if there is no Lenin-izing in the present. If Lenin is no longer alive, there is no Lenin-izing in the present and presumably therefore no present events in the neighborhood of Lenin-izing.34 These observations repeat themselves mutatis mutandis in (110) and (171), where how things were with the subject is further restricted by adverbial modification. Present tense puts Lenin’s present participation in the resemblance in the neighborhood of his being bald, which is itself in the neighborhood of Leninizing. Given the individual-level attribution, if there is both Lenin-izing and his being bald in the present, there is a present resemblance to Queen Victoria. The implicature that present-tense assertion is untrue is satisfied only if there is either no Leninizing or no being bald in the present, which reflects the qualified lifetime effect induced by the adverb. Altogether reversing the lifetime effect in (111) unsurprisingly requires further discrimination between the adverbs bald and dead. As remarked earlier in section 10.0.5, for a balding to be in the neighborhood of Lenin-izing, the participant in the balding is the same as the participant in the Lenin-izing, and the participant in the balding is as alive there as he is in the Lenin-izing. In contrast, the participant in the being dead is clearly not as alive as the Lenin-izer. The subject’s participation in the resemblance is then located in the neighborhood of the balding in (171) or in the neighborhood of the being dead in (172). In the neighborhood of a balding, the participant in a resemblance is as alive as the participant in the balding, and correlatively the participant in the neighborhood of a being dead is as alive as the
500
Chapter 10
participant in the being dead. This last observation reverses the lifetime effect under this adverb. That is, the present-tense counterpart puts Lenin’s present participation in the resemblance in the neighborhood of his being dead. Given the individual-level attribution and the assertion that there was a past resemblance in the neighborhood of his being dead, if there remains his being dead in the present, there is also a present resemblance. The implicature that present-tense assertion is untrue is satisfied only if there is no longer a way to pass from the neighborhood of Lenin-izing to the neighborhood of the dead and on into the present, which obtains just in case Lenin has now joined the undead, which is, as was to be shown, what this usage of the past tense in an individual-level attribution implicates.35 10.0.8
Lifetime effects and intentional events
Lifetime effects obtain when nominal descriptions restrict the descriptions of events, as provided for under supermonadicity and adverbialized logical form. I have, however, not yet considered the interaction of tense and reports of intentional events, which mangles the lifetime effects that have been the basis for the preceding discussion, as Musan (1997) observes. Thus (174) is a truth of modern literature, but Kafka’s fame and his lifetime do not overlap. Virtually unknown at his death, Kafka left behind instructions that his manuscripts be destroyed. (174) Kafka is famous. (175) [℩x : ∃EKafka[E,x]][℩E : Kafka[E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E1] Present[E1] ∃Y Theme[E1,Y] O[E1,E2] Be[E2] ∃E0∃YW[E0,Y]O[E0,E3]∃Y Theme[E3,Y]O[E2,E3]famous[E3] (176) Kafka is in Prague. (177) Kafka is from Prague. If (175) as written is correct, it is implied that there is something x that participates in the present and has Kafka-ed, ∃x(∃EKafka[E,x] & ∃E0 ∃E1(W[E0, x] & O[E0,E1] & Present[E1])), and yet, as I understand it, there is no present Kafka-ing, ¬∃E∃x(Present[E] & Kafka[E,x]). If one were so inclined, I suppose this thing x could be the possible-Kafka that does not Kafka at the present coordinate although it is the object of present fame. Alternatively, first recall (section 10.0) that supermonadicity in (178) affords a distance as great as physical time itself between the action while a supernova, which the subject participates in, and its effect, the event in the present that Tense describes: (178) Supernova GRB 090423 at the most distant edge of the universe 13 billion years ago is bombarding the Swift space-based observatory with gamma radiation, which it was designed to detect.
Adverbialization in Logical Form
501
(179) [℩x : ∃E Supernova[E,x]][℩E : Supernova[E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E1] ∃Y Agent[E1,Y] Cause[E1,E2] Present[E2] bombard[E2] … Accordingly, Tense as shown in (179) is yet more deeply embedded. And yet Tense may not be deeply embedded enough in (179), which says merely that the effect at the observatory occurs in the present. Surely (178) is report of a chain reaction that ends in the present in some ongoing causing-to-be-bombarded and being bombarded.36 Section 10.3 finds that this displacement of Tense to describe events or states at some remove from those in which the subject directly participates holds even of identity statements such as Prince Hal is King Henry V. Here, as an alternative to present possible-Kafka, entertain that (174) may be a report of present fame at some remove from any events or states that Kafka participated in while Kafka. There must, after all, be a sense in which a present thought, memory, or representation can be of a thing although that thing is not itself present. Distinguishing perception from hallucination or imagination, there is some causal connection between the present thought and what it is a thought of, which should not however imply that the thing thought of survives until this thought. If it is mistaken to say that what a thought is about causes the thought, say instead that the causal connection—that original imprint—is a relation of about-ness joining the thought to what it is about. The logical form of action sentences contain ‘Cause[Ei,Ej]’; statives, ‘O[Ei,Ej]’; and, reports of intentional events, ‘About[Ei,Ej]’: (180) [℩x : ∃EKafka[E,x]][℩E : Kafka[E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E1] ∃Y Theme[E1,Y] About[E1,E2] Present[E2] Be[E2] ∃E0∃YW[E0,Y]O[E0,E3]∃Y Theme[E3,Y]O[E2,E3]famous[E3] The literal truth of (174) does not demand a possible-Kafka in the present to bear the property of being presently famous if it is recognized that events entirely within the past in which the actual Kafka actually participated, his writing, made an imprint that is the subject of present fame from which he himself is absent. So it is with fame that one can discern it in the surrounding buzz and know that it is of x while x is dead, gone, and out of sight, but not out of mind, as fame is itself an intentional event. As Musan (1997) says, the lifetime effect depends on lexical and real-world knowledge, which can be realized without ontological dereliction if ‘About[Ei,Ej]’ is introduced into a supermonadic logical form to distinguish (174) from (176) and (177) where the lifetime effect and the presence of a present Kafka are felt. In contrast to (174), to regard as literal truths (181) and (182) does seem to engage some commitment to possible worlds or possibilia to warrant their reference to fictional and nonexistent objects. (181) Gregor Samsa, the protagonist of The Metamorphosis and Kafka’s alter ego, is famous.
502
Chapter 10
(182) Vulcan, phlogiston, and luminiferous aether are all famous for not existing when they were supposed to. But none of that need intrude on talk about the real Franz Kafka. 10.0.9
When that which was and is no longer is perspectival
Recall from section 10.0.0 that the prelude in (43) has undergone the transposition, according to the implicature that how things were with the subject is no longer. (43) The prelude was identical to the étude until a few notes were transposed. The Figure changes against a constant Ground here and elsewhere, as in the contrast that Talmy (1978, 1983) and Gleitman et al. (1996) find between (32) and (33), the former elevating the citizen to the station of the President and the latter lowering the President to the citizen’s station: (32) The humblest citizen is equal to the President. (33) The President is equal to the humblest citizen. It is then to be explained why in (183) and (184), in contrast to (43), there is no obvious implicature favoring the subject, no suggestion that the subject is the thing deviating from the object’s constant station. Come the revolution, who knows? (183) The humblest citizen was equal to the President until the revolution. (184) The President was equal to the humblest citizen until the revolution. Preliminary to an account of the contrast, recall that how things were with the subject is described by the phrase that precedes Tense (see note 35). In the PastTense Lenin resembled Queen Victoria, with subject and no adverb, it is asserted that one can get from the neighborhood of Lenin-izing to a past state that bears the resemblance to Queen Victoria, and it is implicated that there is no way from that neighborhood to a present state bearing the resemblance. If the resemblance would persist, it must be the Lenin-izing that has not. With further modification, Lenin bald resembled Queen Victoria or Lenin dead resembled Queen Victoria, one entertains that perhaps Lenin-izing bald or Lenin-izing dead is no longer, perhaps by hair restoration or by resurrection. Any term in the expanded description may be the one that keeps the neighborhood from reaching the present. Notice now that the logical form of (43) and (183)–(184), like every logical form, includes a token of ‘W’ preceding Tense: (185) Asserted [℩x : ∃E NP[E,x]][℩E : NP[E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E1] Past[E1] ∃Y Theme[E1,Y]O[E1,E2]Be[E2] … Implicated [℩x : ∃E NP[E,x]][℩E : NP[E,x]]¬[∃E0 : N[E,E0]] W[E0,x] O[E0,E1] Present[E1] ∃Y Theme[E1,Y]O[E1,E2]Be[E2] …
Adverbialization in Logical Form
503
Among other things, it is asserted that in the neighborhood of NP-ing, there is some past W-ing, and it would suffice for what is implicated if that neighborhood did not include any present W-ing. Yet this last observation has no bite to the extent that I have been (harmlessly) glossing ‘W’ as participation. Surely, if there is Lenin-izing in the present, there are present events in which Lenin participates. Thus, without some other term failing to hold of the present, a failure to W could not be what matters for the implicature. In (43) and (183)–(184), the prelude-ing, citizen-ing, and president-ing all continue into the present. But if it should turn out that ‘W’ has more content, then even in the neighborhood of present NP-ing, a failure to W may yet turn out to verify that how things were is no longer. In judging (43), one does not imagine that there could have been any change in the rules of the game for metaphysical identity. The space over which Figure and Ground are defined and its metric are absolute from one moment to the next. If it is implicated that how things are with the subject in this space is no longer, it can only reflect a change to the prelude’s intrinsic constitution, as indeed is the case. In contrast, the equality reported in (183)–(184) is no measure of intrinsic worth or constitution. The space defining the relative position of Figure and Ground is as fragile as the social fabric from which it is constructed. In this context, the perspectival content of ‘W’ cashes out as if (183) were to say that the humblest citizen democratically (or as located in a democratic space) was equal to the President. It suffices for the implicature that after the revolution, the subject can no longer be located under the same perspective for Figure and Ground. The Earth he stands on has moved, the social structure has collapsed, and that old perspective on his position is no longer. But once it is recognized that in this sense the W-ing does not survive into the present, nothing further is implicated as to how the citizen and President end up in the new order after the revolution. Both (43) and (183)–(184) as instances of (185) implicate that how things were with the subject is no longer. It may be, as in (43), that the subject has changed its constitution, but it could also be, sense and context permitting, that the subject is no longer Figure to the same Ground, as in (183)–(184). So it also is in (186). After enough havoc to the original scene, there is no further implicature that it was the bicycle rather than the garage that the tornado carried off, although, within any normal scene, bicycles are the Figures whose positions vary with respect to garages as stationary Ground, (40) vs. (41): (40) The bicycle is near the garage. (41) #The garage is near the bicycle. (186) The bicycle was near the garage until the tornado struck. As supermonadicity dictates, ‘W[E0,x]’ describes a scene for the subject alone, a Figure-Ground-ing centered on x. It is then a matter of inference that this
504
Chapter 10
perspective positions others when it overlays what is later described as involving other participants. The exact content of ‘W[E0,x]’, apart from the role I have assigned it here, remains to be uncovered, as I do not have much to say about perceptual fields or frames of reference, about when a perspective is centered on x, or about the conditions under which x is taken to be a figure against something taken to be a background. Yet some such content for ‘W[E0,x]’ seems to be all that tells apart the implicatures of (183)–(184) and (186) from (43). In fact, (43)’s implicature that the prelude rather than the étude has changed can be suspended, even in this case of metaphysical identity if the context for (43) brings it more in line with (183)–(184) and (186). Imagine that the étude is being composed or edited and compared side by side with the prelude, and it is the composer’s intention to preserve yet distinguish the prelude by making changes to the étude with which it is in direct rapport. So it is with problems of concurrent mapping and navigation that a figure in motion deforms the terrain in the course of its progress, as when a massive ball rolls across a sand dune. Figure and ground are no less clearly apprehended, and the changes to the ground are recognized as a consequence of the figure’s momentum and reflect a change in its position. Thus the change in the étude comes in fact to reflect a change in how things are for the prelude.37,38 10.1
Mode-of-presentation effects: Substitutivity failures in simple sentences
Conditional, adverbial modification is not confined to temporal relations: (187) A triangle, this polygon has angles that add up to two right angles. (188) Being a triangle, this polygon has angles that add up to two right angles. (189) If a triangle, this polygon has angles that add up to two right angles. Accordingly, whether formalized as quantification over possible worlds or as event quantification, temporal relations are not the only relations one conceives of to relate what the adverbial phrase describes and what the matrix describes.39,40 Temporal relations have occupied the discussion to the extent that we have been preoccupied with the logical syntax underlying the interaction of Tense, a temporal relation, and the lifetime effect, also temporal. But I should first begin to worry if temporal effects were all that betrayed the presence of the covert Lenin-izing and the like, since adverbialized logical form proposes that concealed adverbial modification is no different from the overt kind. Better not to discover that these event quantifiers have to be stipulated ad hoc to co-occur with only temporal instantiations of ‘N[E,E0]’. As witness to lurking atemporal effects, suppose, recruiting an example from Moore 1999, with an improvisation from Zimmermann 2005, that Superman arranges in advance to confirm his anniversary date with Lois, as he does every year, with a
Adverbialization in Logical Form
505
telephone call and a murmur “Here’s looking at you, kid.” Superman also arranges an appointment with his nemesis Lex Luthor at a predetermined location, to be confirmed with a telephone call and coded message, “Here’s looking at you, kid.” Clark Kent, despondent and hoping to double his chances for a date on Saturday night, decides to try men. Clark Kent enters the Good Luck Bar, dials up on his mobile phone a conference call to Lois and to Lex Luthor, tips his drink, leers at the fellow on the adjacent bar stool, and murmurs for all to hear “Here’s looking at you, kid.” In the space of a few words, Superman makes two dates with destiny, and Clark Kent, who can’t even manage one, strikes out again. That is, (190) Clark Kent may be Superman, but Clark Kent doesn’t resemble Superman at all. As befits a disguise, it is often so that Superman-izing does not coincide with Clark Kent-ing, but at the Good Luck Bar, events of Superman-izing and Clark-Kent-ing coincide with the utterance “Here’s looking at you, kid.” If (190) is to be rescued from contradiction by the disclosure of hidden adverbial modifiers, their force must be more than spatiotemporal (see also chapter 8, note 8). Example (190) presents a special case of Saul’s (1997a) substitution puzzle for extensional sentences41—how to hold the (a) examples in (192)–(195) true and the (b) examples false, while holding the identity (191):42 (191) Clark Kent is Superman. (192) a. Superman leaps tall buildings more often than Clark Kent does. b. Superman leaps tall buildings more often than Superman does. (193) a. Clark Kent went into the phone booth, and Superman came out. b. Clark Kent went into the phone booth, and Clark Kent came out. (194) a. Clark Kent always arrived at the scene just after one of Superman’s daring rescues. b. Superman always arrived at the scene just after one of Clark Kent’s daring rescues. Superman always arrived at the scene just after one of Superman’s daring rescues. (195) a. He hit Clark Kent once, but he never hit Superman. b. He hit Clark Kent once, but he never hit Clark Kent. Adverbialized logical form endorses Forbes’s (1999) solution to the puzzle (after a prompt from Moore 1999 to emend Forbes 1997) and blesses it as a systematic and constant feature of natural language quantification. Forbes’s solution has Superman and Clark Kent referring to the same thing, but it recognizes that to act Supermanishly may be a very different event from slinking off Clark Kent-wise. Sentence (192)
506
Chapter 10
escapes contradiction only because of such a difference between the events if not between their participants: there are more Superman-ish leaps than Clark Kent-ish ones. In (192)–(195) (and in nearly all the examples discussed in the surrounding literature43), it happens that the coreferring names are tokened in overtly distinct clauses,44 which under almost any analysis afford description of distinct events. But the puzzle takes in substitutions into the same simple clause as well: (196) a. Superman outscored Clark Kent. b. Superman beat Clark Kent. (197) a. F Superman outscored Superman. b. F Superman beat Superman. Sentences (196) can be taken as accurate reports either of the aerial contest over Metropolis or of what passed in the Good Luck Bar. Yet if the sentences never engage more than the description of a single event, an outscoring or a beating, (196) cannot fail to imply the contradictory (197). All that (196), for example, would say is that an outscoring is Superman-ish and the same event, that outscoring, is also Clark Kentish, which implies what (197) says—that an outscoring with the same participants is Superman-ish (saying it twice over). Adverbialized logical form is a general solution to the substitution puzzle only in conjunction with supermonadicity, so that (196) (and (190)) describe Superman-ish (sub)events that are not the same as the Clark Kent-ish ones: (198) [℩x : ∃ESuperman[E,X]][℩E : Superman[E,x]][∃E0 : N[E,E0]] W[E0,x] Past[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2](out-)score[E2] [℩x : ∃EClark Kent[E,x]][℩E : Clark Kent[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] OUT[E2,E3]
(i) (ii) (iii) (iv)
‘Superman (in the neighborhood of) Superman-izing participated as Theme in some outscoring (i–ii); Clark Kent (in the neighborhood of) Clark Kent-ing participated as Goal in some events (iii–iv); Superman’s events outdid Clark Kent’s (iv).’45 (190) … Clark Kent does not resemble Superman. (199) [℩x : ∃EClark Kent[E,x]][℩E : Clark Kent[E,x]]¬[∃E0 : N[E,E0]] W[E0,x] Past[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2] [℩x : ∃ESuperman[E,x]][℩E : Superman[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3] ‘Clark Kent (in the neighborhood of) Clark Kent-ing did not participate as Theme in some resembling (i–ii) such that Superman (in the neighborhood of) Superman-izing participated as Goal in some events (iii–iv), and the resembling is to Superman’s events (iv).’
(i) (ii) (iii) (iv)
Adverbialization in Logical Form
507
In the Good Luck Bar, if the Superman-izing and the Clark-Kent-ing there happen to coincide spatiotemporally, the events of Superman’s, E2 in (198), and those E3 of Clark Kent’s that they outdo cannot be distinguished by locating them with respect to the same spatiotemporal neighborhood. Rather, here the (covert) adverbial modification allows us to compare what are the effects of Clark Kent-ing under a causal explanation with what are the effects of Superman-izing, as expected if adverbial modifiers are ever more than temporal or locative expressions. The same remarks apply to (190)’s report of a failed resemblance triggered by the utterance “Here’s looking at you, kid.” The consequences of Clark Kent-ing, striking out yet again, are nothing like the ever-manly consequences of Superman-izing. Adverbialized logical form, so it appears, introduces conditional adverbial modifiers subject to the same semantic variability as overt tokens of such modifiers and consonant with an underlying quantification over possible worlds or events rather than just (spatio)temporal locations. Recall that every nominal quantifier occasions an event quantifier restricted by its nominal content. It is thus important that the mode-of-presentation effects just canvassed reach beyond names to all types of quantifiers: (200) At DC Comics, a superhero (always) leaps tall buildings more often than a civilian does. (201) At DC Comics, no civilian leaps tall buildings as often as a superhero does. (202) At DC Comics, a civilian goes into a phone booth, and a superhero comes out. (Never the other way around.) (203) At DC Comics, no superhero ever goes into a phone booth and a civilian come out. (204) At DC Comics, a civilian always arrived at the scene just after a superhero’s daring rescue. (205) At DC Comics, no civilian arrived at the scene at the same time as a superhero’s daring rescue. (206) He hit a civilian once, but he never hit a superhero. (207) He hit no superhero and got away with it, but he hit many a civilian with impunity. Back at the Good Luck Bar, Peter Parker and Bruce Wayne strike out under exactly the same conditions as Clark Kent does, and Spiderman and Batman succeed like Superman. That is, (208) In the Good Luck Bar, a civilian may be a superhero, but a civilian never resembles a superhero when it comes to dating. A civilian never outscores a superhero.
508
Chapter 10
(209) In the Good Luck Bar, civilians may be superheroes, but no civilian resembles a superhero when it comes to dating. No civilian ever outscores a superhero. Civilian-izing is not the same as superhero-izing, and what rescues the above from contradiction is that the effects of the one are not the same as the other. Similarly, although every creature cordate is renate, cordate-ing and renate-ing, as introduced by the (nonspecific) indefinite descriptions in (210), go their separate ways: (210) Inspecting the laboratory, Holmes discovered that three cordates had been on the dissection table, and two renates, too. [Holmes finds three ovine hearts, and four bovine kidneys.] Discussion in the literature,46 the fault of the examples under consideration, has been preoccupied with the semantics and pragmatics of names proper, neglecting that any solution to the substitution problem should apply with full generality to all quantifiers similarly afflicted.47 Adverbialized logical form provides such a general solution. If (211) and (212) are true, it is no eccentricity of names that makes them so: (211) Yesterday, Phosphorus shone brightly, but clouds obscured Hesperus. (Moore 1999) (212) Hesperus is brighter than Phosphorus. (Barber 2000) It is rather that names are like other DPs: (213) The pond the temperature of which was recorded early this morning was colder than the pond the temperature of which was recorded early this evening. (214) Every pond the temperature of which was recorded early this morning was colder than every pond the temperature of which was recorded early this evening. (215) Walden Pond early this morning was colder than Walden Pond early this evening. (216) #Walden Pond was colder than Walden Pond (was). In neither (213) nor (214) is there an implication that the morning ponds are other than the evening ponds. If, in fact, Walden Pond is among them both, (213) and (214) imply (215). Insofar as adverbial qualification is necessary to escape contradiction (cf. (216)), its only source in (213) and (214) are the nominal phrases themselves, which according to adverbialized logical form, do supply adverbial phrases to modify the local thematic relations.48,49
Adverbialization in Logical Form
509
The modification, as adverbialized logical form assumes, of a local, thematic relation per se rather than a tensed clause is essential to the following: (217) (218) (219) (220)
The man is balder than the boy. The man is balder than the boy was. #The man is balder than the boy is. #The man is balder than the man. #The man is balder than the same man. (221) The man is balder than the man was. The man is balder now than the same man was a year ago. (222) #The man is balder than the man is. #The man is balder than the same man is. If an unspoken tense morpheme is ever elided in a comparative clause, it must be the same as the matrix tense—otherwise, (220) could as sensibly compare a man to his younger self as (221) does, rather than being no better than (222). But then any unspoken tense morpheme in (217) should render it equivalent to (219) rather than (218). It must rather be that the comparative clause in (217) is small enough to omit tense altogether,50 and the comparison of the adult to the younger self is accomplished by adverbial restriction of the local thematic relation to when the subject was boy-ing. Thus, we see the nominal ‘boy(e,x)’ fixing temporal reference for a matrix predicate independent of tense or any other (verbal) morpheme. Given ‘Hesperus(e,x)’ and ‘Phosphorus(e,x)’, little remains to distinguish the logical forms of (212) and (217). 10.2
Analytic, nonlogical substitutivity
The truths that (191) and (223) tell are not of the form ⌜a=b⌝. That ⌜a=b⌝ is unattested in natural language (known all along and repressed along with Tense) and that the simple sentences of natural language are rather complex would seem to render substitution into simple sentences more perilous than it is despite its occasional failures. A logical form surrendered to the representation of those failures, to rendering as literal truths (190)’s second clause, (196), (224) from Moore 1999, and Barber’s (2000) report (225) that Venus is hotter when it is dusk at the speaker’s location than when it is dawn there, seems to shortchange the observation that (191) and (223) are true unambiguously and unequivocally.51 (190) Clark Kent is Superman, but Clark Kent didn’t resemble Superman at all. (191) Clark Kent is Superman. (196) Superman outscored Clark Kent. (223) Hesperus is Phosphorus.
510
Chapter 10
(224) Yesterday, Phosphorus shone brightly, but clouds obscured Hesperus. (225) Hesperus is hotter than Phosphorus. Despite reference to phases that are not coincident, (191) and (223) remain undeniably true even to the speaker preparing to assert (196), (224), or (225). Why isn’t she made aware by her own phase-dependent assertions that the identities might also be false if construed analogously? Pace argument to the contrary, the speaker’s blindness to anything of the kind for (191) and (223) could be taken prima facie to recommend ⌜a=b⌝ for their translation. In answer, the translation that section 10.3 will offer to conform to adverbialized logical form is also judged true and could not be judged otherwise by any speaker who understands it although it relies on nonlogical vocabulary. Whatever (191) and (223) really do say, here let us consider what for a speaker discriminates between two equally simple and even structurally identical contexts, into one of which remarks such as (191) and (223) seem to license a substitution and the other not.52 A context opaque to substitutivity enters into an inference such as (226) unperturbed by (191): (226) i. If one is more successful with women than another, then the one dates women more often than the other. ii. Superman is more successful with women than Clark Kent. iii. Therefore, Superman dates women more often than Clark Kent. The inference goes through without comment, keeping in mind that (i) is underlyingly “If … onei onei-ing … anotherj anotherj-ing …, then the onei onei-ing … the otherj otherj-ing …,” and that (i) and (iii) are instantiations with respect to both nominal and adverbial content. Whatever (191) means, it crucially does not imply that Clark Kent-ing is Superman-ing, which spares (ii) and (iii) contradiction. On the view here, (ii) is a literal truth and the reasoning in (226) to the literal truth (iii) is sound and representative of the speaker’s competence. If ⌜a=b⌝ were all there were to (191) and the semantics of its constitutents, there would be no terms for Superman-ing and Clark Kent-ing and thus no way to finesse the truth of (ii) and (iii), let alone to represent apparently sound reasoning the conclusion of which the speaker accepts fully aware of (191). In contrast, the same speaker does however reject (227): (227) i. If one is more successful with women than another, then the one is happier than the other. ii. Superman is more successful with women than Clark Kent. iii. *Therefore, Superman is happier than Clark Kent. What frustrates this inference is indeed the knowledge that Superman and Clark Kent have the same mental life. The conclusion would rather be accepted if Super-
Adverbialization in Logical Form
511
man and Clark Kent were the split personalities of a single troubled soul. For the sane superhero with memory intact, states of happiness are in the neighborhood of Superman-izing just in case they are also in the neighborhood of Clark Kent-ing:53 (228) happy[e,t] → ((Φ[e0,x] & Ψ[e1,x] & Memory[e0,x,t] & Memory[e1,x,t]) → (Nmens(e0,e) ↔ Nmens(e1,e))) In rejecting (227), (i) is understood to redraw the neighborhood to fit claims about mental life—if in the mental neighborhood of one’s one-ing there is more success with women than in the mental neighborhood of another’s other-ing, then there is more happiness in the one than in the other. Now (ii) fails to instantiate (i)’s antecedent clause so understood. The mental neighborhoods of Superman-izing and Clark Kent-ing being the same, there cannot be more success with women in the one than in the other. What then tells apart (226) and (227) is only a difference in the neighborhood relation seized on in understanding what is intended. The minor premise (ii) that (226) and (227) share is true in the more local, episodic neighborhood and false in the mental, phase-invariant one. As just related, the variability arises in the course of resolving ambiguity in the neighborhood relation. Given that a speaker does not reject (i) in (227) out of hand as internally inconsistent—as she must if the speaker were to choose the episodic neighborhood for the antecedent clause and the mental one for the consequent—the speaker, in recognizing that the consequent is an assertion about mental health, takes the relevant neighborhood for the antecedent clause also to be the mental one. After disambiguation of N, the conditional (i) is literally true. Consider also an alternative for (226) and (227) that makes no appeal to disambiguation as such. Suppose that the only neighborhood relation is the episodic one in evidence in (226), so that (227i) is literally not true. Nevertheless, a charitable, pragmatic accommodation allows for its implicit restriction: (229) If one could be happier than another, then—if one is more successful with women than another, then the one is happier than the other. The premise (i) may then be accepted without quarrel, and the inference (227) is subsequently rejected on the grounds that Superman could not be happier than Clark Kent.54 Whatever alternative proves correct in explaining the variation in substitutivity, the variation is also in play in (230)–(233), with the judgments indicated for a speaker who is fully aware of the identity. Only such a speaker could advance (233), which she delivers without contradiction in her first conjunction contrasting all Superman-izings with all Clark Kent-ings: (230) Superman is successful, but Clark Kent is not successful. (231) #Superman is happy, but Clark Kent is not happy.
512
Chapter 10
(232) #Superman is successful and therefore proud of himself, but Clark Kent is not successful and therefore ashamed of himself. (233) Superman always looks happy, and Clark Kent always not, but since Superman is Clark Kent, it cannot be that the one is and the other isn’t. Supermonadicity and adverbialized logical form directly countenance the failure of substitution to preserve truth, as in (226) and (230). Its apparent success in (227) and (231) is strictly nonlogical, the difference between dating and happiness. There are also more structural factors under which a sentence becomes a context where substitution preserves truth. Sentence (234) does not imply that the hitters are not the same (see (213), (214), and note 48). In fact, as a law of baseball, it is confirmed by statistics that contrast the same individual’s performance against the two kinds of pitchers. Nor do the descriptions of particular hitters in (235) or in (236) preclude that they describe the same hitter (see note 41). (234) The hitter that faces an opposite-handed pitcher outperforms the hitter that faces a like-handed pitcher. (235) That hitter who faced an opposite-handed pitcher 365 times during the season outperformed that hitter who faced a like-handed pitcher 248 times. (236) The hitter now facing an opposite-handed pitcher will outperform the hitter who earlier faced a like-handed pitcher. In contrast, as an effect of the tenses and aspects chosen, (237) and (239), despite the formal similarity to (234)–(236), do imply that the hitters facing different pitchers are themselves different: (237) #The hitter now facing an opposite-handed pitcher outperformed the hitter who earlier faced a like-handed pitcher. (238) The hitter now facing an opposite-handed pitcher outperformed last inning the hitter who earlier faced a like-handed pitcher in the first inning. (239) #The hitter now facing an opposite-handed pitcher outperforms the hitter who earlier faced a like-handed pitcher. Asserting (236), the speaker believes that facing an opposite-handed pitcher confers enough of a marginal advantage at any appearance at the plate to warrant a favorable expectation from this next one. A difference between two performances, the next one and a previous one, implies nothing about the identity of the performers. In contrast to (236) and (238), (237) and (239) are not episodic and rather assert generalizations about average career performance. Crucially, there are no true
Adverbialization in Logical Form
513
(lawful) conditionals that relate the individual episode to career history or career forecast: (240) F If a hitter is now facing an opposite-handed pitcher, he (out)performs … F If a hitter is now facing an opposite-handed pitcher, he (has) (out) performed … Such conditionals contrast with those behind (234) and (235) that relate a career facing an opposite-handed pitcher to generalizations about that career. In terms of adverbialized logical form, the difference cashes out as follows. In (234) and (235), the subject relative clause that is to be adverbialized, describes in accordance with its tense and aspect the (many) events scattered throughout a career in which the hitter faces an opposite-handed pitcher. These in turn are related by a local, causal neighborhood relation to many performances at the plate that themselves constitute a career extended in time according to the requirements of the matrix tense and aspect. The same can be said for the performances in the neighborhood of the events described by the direct object’s relative clause. The performances so related to the subject’s description and those related to the direct object’s description are (necessarily) distinct and remain so even if they involve the same hitter. Thus the sentence’s assertion that the performances of the one average better than the performances of the other carries no implication for the hitters’ identity. In (237) and (239), in contrast, there is no known neighborhood relation that relates the single appearance at the plate, which the present progressive in the relative clause describes, to the many events, a career, that span the temporal interval required by the matrix tense. No known causal relation relates single appearance to career. There is only the defective (phase-invariant) relation that puts any two events in the same neighborhood just in case they have the same participants. Defaulting to that relation, the performances so related to the subject’s description in (237) or (239) will be exactly the same as those related to the direct object’s description if the hitters are the same. In that case, the assertion of unequal performance does imply different hitters. Thus, given a background causal theory or lack thereof and an interaction with the particular tense and aspect chosen, sentences (237) and (239) come to be contexts for substitutivity, in contrast to (234)–(236) and (238), and only under such nonlogical assumptions are they so. Given the nearly identical logical syntax that subsumes (234)–(239), it cannot be that substitutivity in (237) and (239) is a consequence of a logical form they share with (234)–(236) and (238), which do not license the substitution of the F for a coreferential the G. It reflects instead a conspiracy of nonlogical, tacit knowledge. A final respect in which sentences that do not support substitutivity may be coaxed into it touches all of Saul’s (1997a) original sentences and later examples. Demonstrative contexts that counterdemonstrate the nominal descriptive content
514
Chapter 10
undermine the intuitions that drive Saul’s puzzle. Sentence (241), previously a thirdparty narrative report, may also be used, even to an audience well aware of the identity, as an introduction (242) for the superhero presented in his cape and tights:55 (241) In hot dates, Superman outscores Clark Kent (or any other civilian not in uniform). (242) In hot dates, Superman (here) [pat on the shoulder] outscores Clark Kent (or any other civilian not in uniform). (243) Superman (here) [pat on the shoulder], as powerful as he is, just can’t keep them away, can he? If, however, Clark Kent stands before the audience in glasses, coat, and tie, looking un-Superman-ly, it is not that his alter ego cannot be addressed without the audience being misled, as in (243), but sentence (242) in such a context can only be heard to imply incoherently that he outscores himself. The same observation holds of overt descriptions: (244) In hot dates, the superhero who bends steel, stops bullets, and leaps over tall buildings outscores the mild-mannered reporter with foggy glasses. (245) In hot dates, this superhero (here) [pat on the shoulder] who bends steel, stops bullets, and leaps over tall buildings outscores the mild-mannered reporter with foggy glasses. Sentence (245) ends up implying Superman’s outscoring himself if the demonstration presents him as much qua mild-mannered reporter as superhero.56 The other examples suffer similarly. Any truth that (246) conveys collapses into the contradiction that yesterday, Venus shined brighter than Venus if uttered in a context that counterdemonstrates the nominal content, such as (247): (246) Yesterday, Phosphorus shined brighter than Hesperus. (247) Yesterday, Phosphorus there [pointing in the evening] shined brighter than Hesperus. To gain some purchase on a solution, recognize that we often use demonstratives, that liar, this cheat, those backstabbers, truthfully and felicitously without ever catching anyone red-handed in the context of the utterance. The deverbal nouns uncontroversially express a relation to events but to none presently in evidence. The person demonstrated is, if not in the scene demonstrated, at least and at best a sometime liar or a sometime cheat (cf. that once-and-future ruler): (248) [℩x: ∃E (that[u,E] & W[E,x] & ∃E′( [E, E′] & … lie[E′] …))] … Suppose that in any demonstrative DP or in any DP used demonstratively the nominal content is taken to describe the scene demonstrated, which among other things must meet conditions imposed by the demonstrative itself, for example of
Adverbialization in Logical Form
515
being proximate (this) or distal (that). According to (248), that liar refers to an x who is the participant or subject of scene E now demonstrated in the context of utterance u and that current scene and its subject bear no uncertain relation (see section 10.3 for ) to some lying some time. If, in general, a demonstrative DP includes description of the scene demonstrated, then when referring demonstratively with this superhero or Superman to a sad scene of only Clark Kent, one means this sometime superhero or this sometime Superman who is presented now as being in the condition of a some (other) time superhero or some (other) time Superman. But given the identity of Superman and Clark Kent, to now be a sometime Superman is to now be a sometime Clark Kent, whether one is now being Superman or now being Clark Kent.57 If so, the events in the neighborhood of now being a sometime Superman are exactly those in the neighborhood of now being a sometime Clark Kent and there can be no sense in which the events of one neighborhood outscore the events of the other, which accounts for the nonsense of the sentences in counterdemonstrative contexts. Here too, in (241)–(247), the problem has been to derive the appearance of substitutivity under just this condition by means that cannot be strictly logical given the common logical form modulo demonstrative elements. 10.3
Neighborhood watch: The view from symmetric predicates
In the examples used so far to illustrate lifetime effects, the temporal relations imposed among lifetimes and (sub)events emerge from the following constraints: participation in the local subevent occurs within a “lifetime” as circumscribed by the associated nominal description and any other intervening adverbial modifiers, the local subevent in which the subject participates is anchored to the time that Tense refers to, and the meanings of lexical items (succeed, precede, in, to, from, SUB) put distance—forward, backward, or remaining in place—between the various subevents described. One then infers what one can about the temporal position of the lifetimes and subevents associated with nonsubjects. Whatever skepticism will greet this picture, I take it that if this is the picture, it is uncontroversial that these lexical items have the force they do, with the effects noted on the relative position of the subevents they describe. What else, for example, could succeed mean? On the other hand, other lexical items and phrases, symmetric predicates in particular, do not express concepts that are so determinative of temporal relations. The relations “x resembled y,” “x shared ancestors with y,” “x was made of the same stuff as y” say little in this regard. Granted that the resembler’s resembling occurs within the subject’s lifetime and within the past, the mere idea of resemblance does not hinder moving either way to reach a being resembled. Similarly, a common family tree does not fix relative chronology for the leaves who share it. If so, it should be possible
516
Chapter 10
when using a symmetric predicate to locate an object’s lifetime either in the subject’s future or past. The temporal relations in sentences embedding symmetric predicates are not, however, so unconstrained. The constraints observed must derive from the construction itself if lexical meaning will not provide them (see sections 10.0.0 and 2.5.1). To elicit the latent constraints, consider assertions of personal identity through personal transition. Sentences (250)–(266) are about a man who has undergone surgery to become a woman. Now Julia Bagwell, once Jules Bagwell. (249)
Jules Bagwell Carnaval, Rio de Janeiro, 1995.
Julia Bagwell Halloween, Hollywood, CA, 2005.
Now
Contrasts in temporal orientation emerge, such as the one illustrated in (250) and (251), if, as supposed, no phases of Julia and Jules overlap: (250) Julia Bagwell was Jules Bagwell. (251) #Jules Bagwell was Julia Bagwell. There has been no moment at which there is both Julia Bagwell-ing and Jules Bagwell-ing, and due to the complexity of surgical procedures, there may even be a vacant margin that overlaps neither of their phases.58 This lack of overlap is crucial in eliciting the contrast (and remains so below). Had Jules and Julia been two halves of a split personality cohabiting the same body until psychiatric intervention or exorcism, both (250) and (251) could be used felicitously to describe their past, untreated condition. But with their separation in time, a contrast surfaces. There can be movement from the temporal location of the subject’s phase back in history to phases asserted to be of the same person, as in (250), but no analogous movements forward in time. Since Julia is Jules’s future and not his past, (251) is anomalous.59 At play is an interaction among Tense, the neighborhood relation, overt adverbial phrases, and perhaps the lexical meaning of the copula. To begin, suppress interference from the lifetime effect, which contributes much to the contrast between (250) and (251), and fix independent reference to some past events: Jules attended a masquerade at Carnaval 1995 in Rio de Janeiro and Julia, a Halloween costume ball in Hollywood in 2005.60 It is quickly discovered that being so-and-so at some time t strictly locates some so-and-so-ing at t: (252) T At Halloween, Jules Bagwell was (already) Julia Bagwell. (253) F At Halloween, Julia Bagwell was Jules Bagwell.
Adverbialization in Logical Form
517
(254) F At Carnaval, Jules Bagwell was Julia Bagwell. (255) T At Carnaval, Julia Bagwell was Jules Bagwell. Sentence (253) is false in that there was no Jules-ing at Halloween in 2005, and (254), in that there was no Julia-ing at Carnaval 1995. Time travel, of which there must be some to secure any truths when Julia and Jules do not overlap, is however permitted to the subject, either forward in time as in (252) from Jules-ing to the more recent Julia-ing at Halloween or backward as in (255) from present Julia-ing to the Jules-ing at Carnaval.61 Returning to examples from the preceding section, the same remarks apply: (256) T At the mayor’s press conference, Superman was Clark Kent. (257) F At the mayor’s press conference, Clark Kent was Superman. If the reporter covered the press conference, it must be reported as in (256) rather than as in (257). Being Clark Kent at the mayor’s press conference implies Clark Kent-ing then, while being Superman there falsely implies Superman-ing then. Similar considerations distinguish (258) and (259): (258) In June, Venus is Hesperus. (259) #In June, Hesperus is Venus. An independent reference to past events, again sidestepping the lifetime effect, also allows talk about resemblances (or any other symmetric relation) to move forward or backward from the subject’s location. Suppose Jules was a fraternal triplet along with Julius and Juliana, bearing an unmistakable resemblance to Julius and not to Juliana before becoming Julia when a resemblance to Juliana takes over. (260) T At Halloween, Jules Bagwell resembled Juliana Bagwell. F At Halloween, Jules Bagwell resembled Julius Bagwell. (261) T At Halloween, Julia Bagwell resembled Juliana Bagwell. F At Halloween, Julia Bagwell resembled Julius Bagwell. (262) F At Carnaval, Jules Bagwell resembled Juliana Bagwell. T At Carnaval, Jules Bagwell resembled Julius Bagwell. (263) F At Carnaval, Julia Bagwell resembled Juliana Bagwell. T At Carnaval, Julia Bagwell resembled Julius Bagwell. At Halloween, the resemblance that holds is only to Juliana, and at Carnaval, it is only to Julius. In reporting either, one may shift indifferently from either Julius or Julia, forward or backward. A present tense construed independently of the subject also allows for time travel: (264) T Jules Bagwell now resembles Juliana Bagwell.62 F Jules Bagwell now resembles Julius Bagwell.
518
Chapter 10
(265) T Julia Bagwell now resembles Juliana Bagwell. F Julia Bagwell now resembles Julius Bagwell. Prior to making explicit the structure that will allow shifts away from the subject’s phases (‘…’ in (254), line (ii)), the temporal orientation observed forces a certain gloss on the neighborhood relations ‘N[Ei,Ej]’ tokened in the structure inherited from previous sections:63 (254) F At Carnaval, Jules Bagwell was (already) Julia Bagwell. [∃E: at Carnaval[E]]64,65 [℩x : ∃EJules Bagwell[E,x]][℩E : Jules Bagwell[E,x]] … [∃E0 : N[E,E0]]W[E0,x] [℩E0:pro0]Past[E0] O[E0,E] O[E0,E1]∃XTheme[E1,X]O[E1,E2]Be[E2] [℩x : ∃EJulia Bagwell[E,x]][℩E : Julia Bagwell[E,x]] [∃E0 : N[E,E0]]W[E0,x] [℩E0:pro0]O[E0,E3] ∃Y Theme[E3,Y]O[E2,E3]
(i) (ii) (iii) (iv) (v)
For if Julia’s neighborhood ((254), (iv)) could reach into her distant past, it would overtake Carnaval 1995 and Jules’s past, rendering (254) true mistakenly. It must rather be that this relation that occurs with every nominal relates to a local neighborhood as close to spatiotemporal coincidence as modification generally expresses absent contrastive tense.66,67 As in the earlier contrast between Lenin and Lenin dead, where further modification may look beyond the local neighborhood of Leninizing in which he lives, the adverbs in (266), qualifying Julia’s participation and only hers, explicitly reach into her past to find where she had indeed been Jules. (254) F At Carnaval, Jules Bagwell was (already) Julia Bagwell. (266) T At Carnaval, Jules Bagwell was Julia Bagwell gender (still) uncorrected. T At Carnaval, Jules Bagwell was Julia Bagwell gender not yet corrected. T At Carnaval, Jules Bagwell was Julia Bagwell before the operation. As remarked earlier, (255) allows a shift from the subject’s more recent phases back to a past of being Jules Bagwell, where Jules Bagwell can be described further as in (267) according to his condition then. Adding a description of his future state as in (268) only falsifies the report. (255) T At Carnaval, Julia Bagwell was (still) Jules Bagwell. (267) T At Carnaval, Julia Bagwell was Jules Bagwell gender (still) uncorrected. T At Carnaval, Julia Bagwell was Jules Bagwell gender not yet corrected. T At Carnaval, Julia Bagwell was Jules Bagwell before the operation. (268) F At Carnaval, Julia Bagwell was Jules Bagwell gender (already) corrected. F At Carnaval, Julia Bagwell was Jules Bagwell after the operation.
Adverbialization in Logical Form
519
Without independent reference to a past event, sentence (250) is subject to the “lifetime effect”: (250) Julia Bagwell was Jules Bagwell. The implicature that how things were with Julia is no longer is satisfied in this case without her death as the result of the operation that ends the Jules phase. That suffices because the local neighborhood circumscribed in being Jules Bagwell coincides with Jules Bagwell-ing. Adverbial modification of the object position may, however, extend that neighborhood so that (269) clashes with the lifetime effect: (269) #Julia Bagwell was Jules Bagwell gender corrected. (But she still is!) #Julia Bagwell was Jules Bagwell after the operation. Contrary to the implicature, how things were with Julia still is. Although she may no longer be Jules Bagwell pro se, she is still him gender corrected. As for overt adverbial phrases modifying the subject position, their interaction with the allowed temporal shift away from the subject’s phase needs careful stipulation. (255) T (At Carnaval,) Julia Bagwell was Jules Bagwell. (270) F (At Carnaval,) Julia Bagwell gender corrected was Jules Bagwell. F (At Carnaval,) Julia Bagwell after the operation was Jules Bagwell. Since Julia Bagwell’s phase in her gender-corrected condition coincides with her entire life as Julia Bagwell, there should be no contrast between (255) and (270) if the temporal shift allowed the subject is understood to shift the neighborhood from Julia Bagwell-ing in (255) and from the materially equivalent Julia Bagwell-ing while gender corrected in (270). It is rather that the overt adverbs in (270) describe the neighborhood into which the past Tense reaches, as in (271): (270) F (At Carnaval,) Julia Bagwell gender corrected was Jules Bagwell. (271) [∃E: at Carnaval[E]] [℩x : ∃EJulia Bagwell[E,x]][℩E : Julia Bagwell[E,x]] … [℩EN : N[…, EN] gender corrected[EN]][∃E0 : N[EN,E0]]W[E0,x] [℩E0:pro0]Past[E0] O[E0,E] O[E0,E1]∃XTheme[E1,X]O[E1,E2]Be[E2] [℩x : ∃EJules Bagwell[E,x]][℩E : Jules Bagwell[E,x]] [∃E0 : N[E,E0]]W[E0,x] [℩E0:pro0]O[E0,E3] ∃Y Theme[E3,Y]O[E2,E3]
(i) (ii) (iii) (iv) (v)
Julia’s participation in that past condition of being Jules Bagwell at Carnaval is qualified to be during her gender-corrected phase, which it was not. As (271) hints at, [D NPi]– NPi-ing- … -(AdvP-W) …, the AdvP gender corrected remains below to modify Julia’s participation as (first) Theme in the being, modifying only the local and lowest thematic relation associated with the subject. Meanwhile the DP Julia
520
Chapter 10
Bagwell moves to a higher perch where it interacts with whatever relations or operators express the temporal shift away from Julia Bagwell-ing allowed in (255) but subsequently withdrawn when it meets the condition that gender corrected imposes in (270). (272) [∃E: at Carnaval[E]] [℩x : ∃EJulia Bagwell[E,x]][℩E : Julia Bagwell[E,x]] [∃E0 : N[E,E0]]W[E0,x] O[E0,E1]∃XTheme[E1,X] [E1, E2 ] [℩E: gender corrected[E]][∃E0 : N[E,E0]] ∃X W[E0,X] [℩E0:pro0]Past[E0] O[E0,E] O[E0,E1]∃XTheme[E1,X]O[E1,E2]Be[E2] [℩x : ∃EJules Bagwell[E,x]][℩E : Jules Bagwell[E,x]] [∃E0 : N[E,E0]]W[E0,x] [℩E0:pro0]O[E0,E3] ∃Y Theme[E3,Y]O[E2,E3]
(i) (ii) (iii) (iv) (v) (vi)
In (272), a dislocative morpheme − appears in (ii) like a causative morpheme as a higher verb or projection, expressing a relation between events E1 and events E2 just in case they are events in the history of the same objects. ‘[E1, E2 ]’ entails of course that their participants are the same (i.e., ∀X(W[E1,X] ↔ W[E2,X]), but common participants might not be sufficient for events to be related by a history. It might be that narrative or continuity conditions on a coherent history prevent arbitrary scenes from a life to be so related. It is enough here that coherent histories, whatever their other requirements, traverse historical time.68 In summary, with independent Tense (see note 60), referring in these examples to Carnaval or to Halloween, to events that are past (255) or future (252) with respect to the subject’s phase, the events verifying identity or resemblance may be found anywhere in the subject’s history that Tense demands. The identity or resemblance holds of the subject’s condition then. In an identity statement, there is nothing other than a token of to shift away from the local neighborhood, so that wherever one lands in the subject’s history, there is Julia Bagwell-ing (be Julia Bagwell) or Jules Bagwell-ing (be Jules Bagwell) unless overt adverbial phrases modifying object position shift the neighborhood again. This distribution, a solitary token of as higher verb and tokens of ‘N[Ei, Ej]’ everywhere else, alongside every adverbialized description, derives the observed asymmetry between subjects and objects, namely, that the subject’s phase but not the object’s phase may be at some remove from when the identity is said to hold. Similar remarks hold for resemblances. The present Julia Bagwell must timetravel to her past Jules Bagwell-ing in order to resemble her male fraternal triplet, according to (273): (273) T (At Carnaval,) Julia Bagwell resembled Julius Bagwell.
Adverbialization in Logical Form
521
(274) F (At Carnaval,) Julia Bagwell gender corrected resembled Julius Bagwell. F (At Carnaval,) Julia Bagwell after the operation resembled Julius Bagwell. Adverbial phrases modifying the subject position modify her destination. Thus (274) are false in that there is no Jules Bagwell-ing of hers in her gender-corrected state after the operation. In contrast to a being so-and-so, which implies being so-and-so while so-and-so-ing, recall that for a resemblance to something (167), the thematic relation ‘TO[Ei, Ej]’ puts distance between the resembler’s Jules Bagwell-ing at the time of (273)’s resemblance and the resembled’s Julius Bagwell-ing. For all that is known, Julius Bagwell may have died long before Carnaval 1995. With independent reference to Carnaval or to Halloween, the temporal shift from the subject’s phase may proceed in either direction, as (255) and (252) have already attested, and has been defined accordingly without any temporal orientation. Absent reference to Carnaval or Halloween as in (250) and (251), an orientation appears. It appears that there can be movement from the temporal location of the subject’s phase backward as in (250), but no analogous movements forward as in (251). (250) Julia Bagwell was Jules Bagwell. (251) #Jules Bagwell was Julia Bagwell. Yet the apparent temporal orientation under the particular circumstances imagined may be passed off as the lifetime effect. That is, if Jules Bagwell ever was Julia Bagwell, he still is, as Julia Bagwell still is; (251) falsely implicates that he is no longer. Although the contrast is attenuated, it however survives without an assist from the lifetime effect. Suppose that Bagwell died last year, silencing the effect, without any other adjustment in Bagwell’s biography: (275)
Jules Bagwell Carnaval, Rio de Janeiro, 1995.
Julia Bagwell Halloween, Hollywood, CA, 2005.
It remains odd to say (251) out of the blue without independent indication of the temporal frame, as if under those conditions one really can only search backward in time from the subject’s phase.69 Time’s arrow similarly imposes itself on reports
Now
522
Chapter 10
of symmetric kinship. Franz Kafka was the eldest of six children, two of whom died in infancy before the birth of his three sisters: (276) Franz Kafka 1883–1924 Georg Kafka 1885, d. 15 mo. Heinrich Kafka 1887, d. 6 mo. Elli Kafka 1889–1941 Valli Kafka 1890–1942 Ottla Kafka 1892–1943 Like Jules and Julia Bagwell, the lifetimes of Georg and Ottla Kafka do not overlap, and they are of course each other’s siblings. Yet one says (277), Ottla having had the same parents as her historical predecessor in the cradle, and one cannot say (278) as there had been no Ottla at the time of Georg’s birth.70 (277) Ottla Kafka was related to Georg Kafka. Ottla Kafka was a sibling of Georg Kafka’s. (278) #Georg Kafka was related to Ottla Kafka. #Georg Kafka was a sibling of Ottla Kafka. That their lifetimes did not overlap is, however, no matter when they are both subjects (see (32)–(34), (43)–(44), and examples in section 2.5.1): (279) Georg Kafka and Ottla Kafka were related. Georg Kafka and Ottla Kafka were siblings. Ottla Kafka and Georg Kafka were related. Ottla Kafka and Georg Kafka were siblings. Granting that history is time-dependent, let us say that events E1 and E2 are historically related as of the time of some reference event or events ER (after Reichenbach 1947), just in case they are events in the history of the same objects as it has unfolded up to and including ER, [E1, E2 ] & ∀e((E1e ∨ E2e) → ∃eR (ER eR & e ≤ eR)). A temporal-frame adverbial mentioning Carnaval 1995 or Halloween 2005 refers to these events within a frame of reference that includes the time of utterance—that is, the speaker’s now. What is then said about Bagwell’s phases concerns temporal relations among events all of the past from the perspective assumed, the speaker’s own: (252) At Halloween 2005, Jules Bagwell was Julia Bagwell. [℩ER: now]-PP-DP-W-Theme [[W-Past-Theme] Be [DP-W-Theme]]
Adverbialization in Logical Form
523
(280) [℩ER : now[ER]][∃E: at Halloween 2005[E]]71 [℩x : ∃EJules Bagwell[E,x]][℩E : Jules Bagwell[E,x]] (i) [∃E0 : N[E,E0]]W[E0,x] O[E0,E1]∃XTheme[E1,X] [E1, E2 ] ∃E0∃X W[E0,X] (ii) [℩E0:pro0]Past[E0,ER] O[E0,E] O[E0,E1]∃XTheme[E1,X]O[E1,E2]Be[E2] (iii) [℩x : ∃EJulia Bagwell[E,x]][℩E : Julia Bagwell[E,x]] [∃E0 : N[E,E0]]W[E0,x] [℩E0:pro0]O[E0,E3] ∃Y Theme[E3,Y]O[E2,E3]
(iv) (v)
Absent explicit adverbial, there is however a default to a perspective in (250), (251), and (277)–(279) contemporary with the subject. For contemporary observers of Julia Bagwell and Ottla Kafka, their respective predecessors, Jules Bagwell and Georg Kafka, are figures from the known past ((250), (277)), but for observers contemporary with Jules Bagwell and Georg Kafka, whose nows are their nows, Julia Bagwell and Ottla Kafka are beyond the event horizons for their frames of reference and thus out of reach of the nonmodal assertions (251) and (278):72 (251) #Jules Bagwell was Julia Bagwell. DPi-[℩ER: theni]-W-Theme [[W-Past-Theme] Be [DP-W-Theme]] (281) [℩x : ∃EJules Bagwell[E,x]][℩E : Jules Bagwell[E,x]] [℩ER : then[ER,E]] [∃E0 : N[E,E0]]W[E0,x] (i) O[E0,E1]∃XTheme[E1,X] [E1, E2 ] ∃E0∃X W[E0,X] (ii) [℩E0:pro0] Past[E0,ER] O[E0,E] O[E0,E1]∃XTheme[E1,X]O[E1,E2] (iii) Be[E2] [℩x : ∃EJulia Bagwell[E,x]][℩E : Julia Bagwell[E,x]] (iv) [∃E0 : N[E,E0]]W[E0,x] [℩E0:pro0]O[E0,E3] ∃Y Theme[E3,Y]O[E2,E3] (v) A syntax like that of the verb-second construction is stipulated for ⌜[℩ER: Φ]⌝. Leftmost, it is left to context to fix Φ, defaulting to now. Movement of a DP to leftmost position imposes an anaphoric relation between Φ and the content of DP.73,74 The observation from the beginning of this section now follows—that the contrast between (250) and (251) disappears when there is an overlap between Jules-ing and Julia-ing, as when split personalities cohabit. If, on the one hand, the speaker has in mind such a period and makes explicit reference to it, either demonstratively or with an explicit adverbial, movements forward or backward in time, as licenses, confirm both (250) and (251). If, on the other hand, Jules Bagwell commands the frame of reference for ⌜[℩ER: Φ]⌝ as in (281), some Julia Bagwell-ing still falls within his event horizon, also confirming the identity and eliminating contrast with (250). Evaluated in other contexts, the contrast between (250) and (251) and contrasts between other sentences embedding symmetric lexical relations such as be identical
524
Chapter 10
to and resemble (see (260)–(265)), which do not themselves tamper with temporal relations, have allowed a structural bias to shine through in the form of and ⌜[℩ER: Φ]⌝ and the enveloping syntax. In this and previous sections (see sections 10.0.0 and 2.5.1), it has been observed that the sentence frames of natural language do not express symmetric relations even where most expected. Neither [_is identical to_] nor [_is_] does, not for lack of a lexicon of symmetric relations, but because all sentences embed such lexical items in frames with asymmetric twists, so it seems. The caricature of as time travel, as motion of a Figure against a Ground in time, invites the speculation that all the asymmetries observed in these sections derive from the application of the same perspectival relations (see note 68) in different contexts. If not, itself is to be stipulated in the basic clause structure, where perhaps it is just one of several asymmetric relations that basic clauses draw on. Its universality depends in part on the lexical relations it allegedly co-occurs with and whether the differences among the resulting sentences can be ascribed to lexical variation while remains constant. 10.3.0
Trans-scene identity
Just to compare in this connection statements of resemblance and identity, recall that in a single utterance at the Good Luck Bar of “Here’s looking at you, kid,” it became apparent to all that (190) Clark Kent is Superman, but Clark Kent didn’t resemble Superman at all. In rendering as literal truth (190)’s second clause, it might seem that sections 10.1– 10.2 and this section shortchange the observation that (190)’s first clause, “Clark Kent is Superman,” is true unambiguously and unequivocally. Despite the presumed adverbialization of all DPs, it shares no interpretation with either (282) or (283): (282) #Clark Kent Clark Kent-ing is Superman. (283) F Clark Kent Clark Kent-ing is Superman Superman-ing. (284) T Clark Kent Clark Kent-ing is not Superman Superman-ing. Sentence (283) is plainly false and its denial in (284) true in that no Clark Kent-ing coincides with Superman-ing. Sentence (282) might be literally true, but it is pragmatically defective in implicating that Clark Kent in circumstances other than Clark Kent-ing might not be Superman. It is as defective as an assertion that Tuesdays, Clark Kent lives and breathes implicating otherwise for other days of the week, and as such, (282) shares no interpretation with (190)’s true and felicitous first clause. The interpretation of (282)–(284) is already accounted for under the remark (see (270)–(272)) that explicit adverbial phrases modify the local and lowest thematic relations, so that Clark Kent participates only while Clark Kent-ing and similarly
Adverbialization in Logical Form
525
Superman, only while Superman-ing. A misspoken adverbial phrase falsifies the historical identities (252) and (255) as well: (252) T At Halloween 2005, Jules Bagwell was Julia Bagwell. (285) F At Halloween 2005, Jules Bagwell Jules Bagwell-ing was Julia Bagwell. (255) T At Carnaval 1995, Julia Bagwell was Jules Bagwell. (286) F At Carnaval 1995, Julia Bagwell Julia Bagwell-ing was Jules Bagwell. Moreover, as Jules was already Julia in 2005 and Julia still Jules in 1995, it is true that (287) At Halloween 2005, Jules Bagwell was not Jules Bagwell Jules Bagwell-ing. (288) At Halloween 2005, Jules Bagwell was not still Jules Bagwell. (F At Halloween 2005, Jules Bagwell was still Jules Bagwell.) (289) At Carnaval 1995, Julia Bagwell was not Julia Bagwell Julia Bagwell-ing. (290) At Carnaval 1995, Julia Bagwell was not yet Julia Bagwell. (F At Carnaval 1995, Julia Bagwell was already Julia Bagwell.) Absent an explicit adverbial phrase, the truth of simple identity statements such as (190)’s first clause, (250)–(252) and (255), relies on to relate disparate scenes from the same life just because they are scenes from the same life. Without , (190), given the adverbialization of the subject, would be equivalent to (282) and so taken to be false and (291) taken to be true insofar as Clark Kent-ing does not coincide with Superman-ing. (291) Clark Kent is not Superman. But as there is no denying the identity of Clark Kent and Superman, the occurrence of must be obligatory in identity statements (see (281)): (292) DPi-[℩ER: theni]-W-Theme (-not)- [[W-Past-Theme] Be [DP-W-Theme]]75 (293) *DPi-[℩ER: theni]-W-Theme (-not)-[[W-Past-Theme] Be [DP-W-Theme]] Recall also that the dislocative morpheme as placed in (292) induces an asymmetry between subject and object in displacing only the former, with the result that (256) and (257) diverge: (256) T At the mayor’s press conference, Superman was Clark Kent. (257) F At the mayor’s press conference, Clark Kent was Superman. Without a token of intervening to vacate the adverbial phrase derived from the object nominal, (257) does imply Superman-ing at the press conference, like (294): (294) F At the mayor’s press conference, Clark Kent was Superman Superman-ing.
526
Chapter 10
A comparable effect has been observed in statements of resemblance (see (260)–(265)): (295) Before retirement, Batman resembled Superman. (296) Now Batman resembles Clark Kent. The past resemblance is to Superman-ing and the present to Clark Kent-ing, a difference represented only if the resemblance is qualified by the nominal-derived adverbials without interference. Also, absent present Jules Bagwell-ing, time travel must displace the subject in (264) to report Jules’s present resemblance, now that he is Julia, to his triplet sister (see also (273)–(274) above): (297) (264) (298) (299) (300)
Jules Bagwell is Julia Bagwell. Jules Bagwell (now) resembles Juliana Bagwell. Jules Bagwell does not resemble Julius Bagwell. Jules Bagwell resembled Julius Bagwell. Jules Bagwell did not resemble Juliana Bagwell.
The token of indicated for (264), however, threatens to undermine (190)’s denial of a resemblance. The Clark Kent-ing, which never resembles Superman-ing, looks as if might transport its participant into a resemblance with Superman, just as Jules Bagwell in (264) is transported into a resemblance with Juliana Bagwell: (301) DPi-[℩ER: theni]-W-Theme (–not)- [[W-Past-Theme] resemble …] Removing the difference in tense from (190)’s sentences puts the problem in sharper relief. Both (302) and (303) are unequivocally true given the standards by which a disguise is judged effective:76 (302) Clark Kent is Superman. (303) Clark Kent does not resemble Superman (enough to be instantly recognized).77 Relating scenes from the same life to support the truth of (302), should not thereby undermine the truth of (303). One might at this point despair of ’s universality, never to be derived from Figure and Ground perspectival relations, its occurrence stipulated for identity statements (see (292), (293), and (302)) and withheld from some sentences with resemble (see (303)) and yet not others ((264) and (296)). istory finds reprieve in the observation that the Bagwells divide it in two epochs marked by their transition but the superhero and civilian share the times. A precision in the interpretation of these sentences that exploits this difference preserves as a constant for at least all stative sentences. Resembling Superman is not resembling Clark Kent, which, as remarked, can only be a consequence of the nominal adverbials fixing on scenes of Superman and of Clark Kent to fail at resemblance.
Adverbialization in Logical Form
527
As (190) is a comment on the events in the Good Luck Bar, its tense refers to the time then, and since tense occurs within the scope of , any travel from Clark Kenting is to events then. The resemblance denied is thus one between how Clark Kent was at the Good Luck Bar and Superman-ing.78 Of course the example has been contrived so that the same time and place hold both Clark Kent-ing and Supermaning (and similarly for the present in (303)), only the former of which can be denied to resemble Superman-ing. As there is also Superman-ing then and there, what is to prevent time travel to such events, which would render (190) and (303) mistakenly false? The answer rests on the conditions for successful nonmaximal reference (appendix 1) that apply to descriptive event pronouns such as ‘[℩E2:pro2]’ in (305), which refers to what events Clark Kent participates in at the Good Luck Bar: (304) DPi-[℩ER: theni] -not- W-Theme - [[W-Tense-Theme] resemble …] (305) [℩x : ∃EClark Kent[E,x]][℩E : Clark Kent[E,x]] [℩ER : then[ER,E]]79 ¬[∃E0 : N[E,E0]] W[E0,x]O[E0,E1]∃XTheme[E1,X] [E1, E4 ] ∃E2∃X W[E2,X] [℩E2:pro2] Present[E2] O[E0,E3] ∃XTheme[E3,X] O[E3, E4] ∃E4 resemble[E4] [℩x : ∃ESuperman[E,x]][℩E : Superman[E,x]] [∃E5 : N[E,E5]] W[E5,x] O[E5,E6] ∃X Goal[E6,Y] to[E4,E6]
(i) (ii) (iii) (iv) (v)
Reference to less than everything Clark Kent has ever done is accomplished in the first instance by spatiotemporal localization. Thus reference to whatever he did then and there at the Good Luck Bar on this occasion is reference to some Superman-ing and some Clark Kent-ing, given the contrivance of that occasion. But then it is judged false that this Januslike behavior resembles steady Superman-ing or it violates a presupposition that intends a comparison between two steady behaviors that persist throughout the scenes compared. To refer to less than all the Clark Kent-ing and Superman-ing at the Good Luck Bar, one might allow ‘[℩E2:pro2]’ to copy more content than ‘W’ from antecedent description but that could at best come to include mention of Clark Kent-ing and not Superman-ing, and indeed the Clark Kent-ing at the Good Luck Bar does not resemble the Superman-ing there, as (190) and (303) say. For the Bagwells divided between two epochs, spatiotemporal localization easily finds a time of steady Jules Bagwell-ing and a time of steady Julia Bagwell-ing, which variously resemble their fraternal triplets. and the descriptive event anaphora in (305) time-travel Clark Kent to his Clark Kent-ing or to his Clark Kent-ing and Superman-ing at the Good Luck Bar, neither of which resembles his Superman-ing there and thus a resemblance can be denied. It remains, however, that the basic identity statement (302) is unequivocally and unambiguously true despite the difference and failed resemblance between the events referred to. It should not be surprising to turn at this point to lexical
528
Chapter 10
differences between resemble and the copula construction to distinguish the sentences of (190) or the minimal pair (302) and (303). The identity construction involves a relation as in (307) and (308), sparing be from taking on a meaning that will not fit its use elsewhere (e.g., Superman is resembled by Clark Kent): (306) DPi-[℩ER: theni]-W-Theme (-not)- [W-Tense-Theme resemble …] (307) DPi-[℩ER: theni]-W-Theme (-not)- [W-Tense-Theme Be [DP-W-Theme ]] (308) [E1, E2 ] ↔def ∀X(W[E1,X] ↔ W[E2,X]) & ∀T(time[T] → (At[E1,T] ↔ At[E2,T])) Departing from Clark Kent Clark Kent-ing, continues at the Good Luck Bar only with more of Clark Kent Clark Kent-ing (or with Clark Kent-ing and Supermaning). Superman participates there Superman-ing. The Clark Kent-ing (or the Clark Kent-ing and Superman-ing) is not the Superman-ing, and the identity that (302) asserts is affirmed only by the intervention of to relate these barroom events. They are related according to (308) just in case their participants are the same and they are contemporaneous. The latter condition keeps the baby from being discarded with the bathwater. If it related any events from a shared history, would mistakenly render (254) true, allowing Julia Bagwell-ing to be related to Jules Bagwell at Carnaval, when he had as yet no plans for his future gender correction: (254) F At Carnaval, Jules Bagwell was (already) Julia Bagwell. Such considerations, recall, compel the neighborhood relation N relating Julia Bagwell-ing to the local events in which she participates to be local, which should not now undermine. For (302), it suffices for to occur anywhere within (307). But scope interaction with overt adverbials discloses it to be part of the analysis of the predicate nominal’s contribution. The relevant considerations rehearse those relied on earlier to locate the position of overt adverbials. Recall that despite the adverbialization of DPs, (302) shares no interpretation with either (282) or (283): (282) #Clark Kent Clark Kent-ing is Superman. (283) F Clark Kent Clark Kent-ing is Superman Superman-ing. In (282) and (283) (see (270)–(272)), the overt adverbial phrases modify a local, lower thematic relation. Crucially, Clark Kent-ing falls within the scope of (position relative to not yet at issue) so that Clark Kent participates at the time referred to only while Clark Kent-ing without being displaced elsewhere and similarly Superman, only while Superman-ing: (309) DP [℩E: Clark Kent]-[℩ER : then[ER,E]]-W-Theme [Clark Kent-ing [W-Tense-Theme] … Be …]
Adverbialization in Logical Form
529
Consider now alongside the unequivocal and true (302) the contrast elicited when overt adverbials modify the predicate nominal:80 (310) T Clark Kent is Superman in coat and tie. (311) F Clark Kent is Superman in cape and tights. Rather than the placement shown in (307), the operator appears in (312) within the scope of the predicate nominal and therefore entirely within the description of Superman’s participation. In (313), the denominal adverbial occurs adjacent to its DP, and as before the overt adverbial modifies the lower thematic relation within the scope of : (312) DP [℩ER : then[ER,E]] W Theme [W Tense [Theme Be [DP W Theme ]]] (313) DP [℩E: Clark Kent] [℩ER : then[ER,E]] W Theme [W Tense [Theme Be [DP [℩E: Superman] W Theme in c. & t.]]] As before, under and descriptive event anaphora, Clark Kent is considered only while Clark Kent-ing (or while Clark Kent-ing and Superman-ing), which isn’t while Superman-ing. The Superman-ing with which the predicate nominal’s description of events begins is related by to any other events in which the man participates provided that they are as the overt adverb describes them. So related to events in coat and tie, (310) is true as these are events of Clark Kent-ing. But passing from Superman-ing to events of his in cape and tights falsifies (311), since none of these is Clark Kent-ing. The overall syntax that we are left with in (312) and (313) does not seem out of line, with one relation for perspectival relations such as any sentence might engage placed high and indifferent to the lexical content of the predicate below, and another relation, to the extent it belongs to the analysis of identity statements, associated with the predicate nominal. In identity statements, overt adverbial modification of subjects and that of predicate nominals differ in ways not uncongenial to the asymmetries already noted or the structure under consideration. Sentences (302) and (314) are unequivocally true and, as just noted, the former is spoiled by the modification in (311): (302) Clark Kent is Superman. (314) Superman is Clark Kent. (311) F Clark Kent is Superman in cape and tights. In contrast, the response to (315) is more ambivalent, admitting the defense that Superman in fact is always Clark Kent: (315) Superman in cape and tights is Clark Kent.
530
Chapter 10
The ambivalence attaches to (315) only out of the blue. With a specific occasion in mind, as in (256), (316) is straightforwardly false since the reporter in coat and tie and not the superhero in cape and tights attended: (256) T At the mayor’s press conference, Superman was Clark Kent. (316) F At the mayor’s press conference, Superman in cape and tights was Clark Kent. The rejection of (316) follows directly from what has been observed throughout the recent sections. Tense and its temporal-frame adverbials, at the mayor’s press conference, locate the being Clark Kent and Superman’s participation in it. The overt adverbial in cape and tights, occurring within the scope of , must hold of Superman’s participation then and there, which it does not: (317) [℩ER : now[ER]] [∃E: at the mayor’s press conference] (DP [℩E: Superman] W Theme [in cape and tights [W Past[E, ER] Theme [Be [DP [℩E: Clark Kent] W Theme ]]]) Out of the blue, (315) comments on Superman-ing in cape and tights in the present, of which there is (318) (DP [℩E: Superman] [℩ER : then[ER,E]] W-Theme [in cape and tights [W-Present[E] Theme [Be [DP [℩E: Clark Kent] W Theme ]]]) Any such Superman-ing shares its participant with Clark Kent-ing, which due to the late intervention of in (318) suffices for the truth of (315). Anyone who knows (315) must also know the briefer and more general (314), which renders (315)’s assertion pragmatically defective in some way. As far as it goes, this remark should apply to (319) as well, which however becomes felicitous in suggesting to the hearer the circumstances when she might recognize that Superman is Clark Kent: (319) Superman in coat and tie is Clark Kent. I will not, however, attempt here to represent the difference between (315) and (319). 10.3.1
Identity without entity
That ⌜a=b⌝ is unattested in natural language and that the simple sentences of natural language are rather complex has rendered substitution salva veritate into simple sentences nonlogical (see section 10.2). Worse, a logical form surrendered to the representation of substitution failures neglects even to validate sentences habit translates as instances of ⌜a=a⌝. Spoken with flat intonation,81 it has been observed that (320) and (321) are unequivocally and unambiguously true, as well they might be, bearing distant resemblance to ⌜a=a⌝:
Adverbialization in Logical Form
531
(320) At Halloween 2005, Jules Bagwell was Jules Bagwell. (321) At Carnaval 1995, Julia Bagwell was Julia Bagwell. But these self-identities then enter a dilemma opposite the false historical identities (253) and (254): (253) F At Halloween 2005, Julia Bagwell was Jules Bagwell. (254) F At Carnaval 1995, Jules Bagwell was Julia Bagwell. Recall that (253) is false in that there was no Jules-ing at Halloween in 2005, and (254), in that there was no Julia-ing at Carnaval 1995. Being Jules Bagwell at Halloween entails Jules Bagwell-ing at Halloween, without regard for the sentence’s subject. Thus, the circumstances that falsify (253) should equally falsify (320), assuming that the sentences are identical in logical form mutatis mutandis. Similarly, what falsifies (254) should also falsify (321). Why then does a precision about the phases in Bagwell’s life matter for (253) and (254), for the historical identities, but not for the self-identities, (320) and (321), unless these in turn are under adverbials as in (287)–(290)? The self-identities lapse not only under temporal adverbials as shown there, but apparently under any modification as the following brings out. Recall again that the false historical identities (253) and (254) can be made true with a temporal adverbial shifting the local neighborhood: (322) T At Halloween 2005, Julia Bagwell was Jules Bagwell gender (already) corrected. T At Halloween 2005, Julia Bagwell was Jules Bagwell after the operation. (266) T At Carnaval 1995, Jules Bagwell was Julia Bagwell gender (still) uncorrected. T At Carnaval 1995, Jules Bagwell was Julia Bagwell gender not yet corrected. T At Carnaval 1995, Jules Bagwell was Julia Bagwell before the operation. There was no Jules Bagwell-ing at Halloween 2005, falsifying (253); but verifying (322), there was at that time Jules Bagwell gender corrected-ing and Jules Bagwell after the operation-ing. Modification by the explicitly temporal, before and after, and the implicitly so, the participle of a verb of change of state, manages to shift the local neighborhood in a way that other modifiers do not, even if knowledge of realworld conditions would warrant an inference that such has occurred. Consider that the Bagwells have always been punctilious in their social attire—men in black tie and women in basic black and pearls—and, therefore, Jules Bagwell attended Carnaval 1995 in black tie and Julia Bagwell, Halloween 2005 in basic black and pearls.
532
Chapter 10
In contrast to (322), which in fact makes a report consistent with Julia’s appearance on that occasion, (323) is a false report of a drag queen at Halloween: (323) T At Halloween 2005, Julia Bagwell was Jules Bagwell in basic black and pearls. That is, the neighborhood remains local to Jules Bagwell-ing and the modifier only narrows in on an occasion in basic black and pearls. Halloween was such for Julia, but not as Jules then, falsifying (323). Unsurprisingly, the stative locative in basic black and pearls fails to shift the temporal neighborhood, even though we have every reason to believe that it should, knowing that a Bagwell is not caught dead in basic black and pearls unless she is a woman. The locative PPs thus resist construal as a temporal modifier rather robustly. Even so, their intrusions in self-identities are as effective as the time-shifting adverbials in (287)–(290) in demanding precision about the phases described: (320) T At Halloween 2005, Jules Bagwell was Jules Bagwell. (324) F At Halloween 2005, Jules Bagwell was Jules Bagwell in basic black and pearls. (321) T At Carnaval 1995, Julia Bagwell was Julia Bagwell. (325) F At Carnaval 1995, Julia Bagwell was Julia Bagwell in black tie. The unequivocal and unambiguous truth (320) gives way to falsehood in (324)—not for lack of basic black and pearls, Bagwell’s attire that evening—but because (324), in contrast to (320), implies her entrance as Jules, and therefore in drag, which a Bagwell never does. Similarly, although it counts little against the truth of (321) that Julia was not yet Julia in 1995, it is just this fact that makes (325) false and so odd, no matter that her person in 1995 did attend Carnaval in black tie. In summary, the flatly intoned, self-evident self-identities of (320) and (321) are fragile and break under modification, of which (326) and (327) are further example (see section 9.2 on still, already, and again in identity statements): (320) T At Halloween 2005, Jules Bagwell was Jules Bagwell. (326) F At Halloween 2005, Jules Bagwell was still Jules Bagwell. (321) T At Carnaval 1995, Julia Bagwell was Julia Bagwell. (327) F At Carnaval 1995, Julia Bagwell was already Julia Bagwell. These are stark minimal pairs, for which the logical forms, with a modifier and without, are presumably the same mutatis mutandis. With the semantics delivering falsehoods for (326) and (327), one might expect then to do the same for (320) and (321) at least for some reading of even the flatly intoned sentences, and it wants explanation how it escapes notice and a different reading obtains instead.
Adverbialization in Logical Form
533
Recall that the problem now faced concerns temporal reference that is dependent on a temporal frame independent of the subject. The unequivocal truth of (291) already derives from its structure in (292), and the contrast with (257), which is false when Clark Kent covers the press conference, has also been derived in that being so-and-so at t has implied so-and-so-ing at t. (291) Clark Kent is Superman. (292) DP [℩E: Clark Kent] [℩ER : then[ER,E]] W Theme [W Present Theme Be [DP [℩E: Superman] W Theme ]] (257) F At the mayor’s press conference, Clark Kent was Superman. (328) T At the mayor’s press conference, Superman was Superman. (329) [℩ER : now[ER]] [∃E: PP[E]] DP W Theme [W Past[E,ER] Theme] Be [DP W Theme ]] It is this last remark that runs up against the perhaps not untrue report in (328) of the same press conference at which only Clark Kent was in attendance. But the implication from being so-and-so at t to so-and-so-ing at t rests on the interpolation of a local, phase-dependent neighborhood relation when the DPs are adverbialized in passing from (329) to (330): (330) [℩ER : now[ER]]- [∃E: PP[E]]-[DX : ∃E′ NP[E′,X]] [℩E′: NP[E′,X]] [∃E0 : N[E′,E0]]-W-Theme [[W-Past[E]-Theme] Be [[DX : ∃E′ NP[E′,X]] [℩E′: NP[E′,X]] [∃E0 : N[E′,E0]]-W-Theme] [E]] Following the example of overtly pronounced conditionals and adverbials and what has been shown necessary for the interpretation of the adverbialization, events where there is no Superman-ing are not in the neighborhood of Superman-ing (257), and similarly events where there is no Jules Bagwell-ing are not in that neighborhood either (253). Yet recall from section 10.2 that counterdemonstrative contexts manage demonstrative reference in contexts that do not include demonstration to the events that the adverbialized nominal apparently describes: (331) Superman (here) [patting Clark Kent on the shoulder], as powerful as he is, just can’t keep them away, can he? But it was pointed out that many a demonstrative, that liar, this cheat, those backstabbers, is used felicitously without catching anyone red-handed in the context of the utterance. In such usage, tacit event quantification in the nominal description delivers that sometime liar, this sometime cheat, and those sometime backstabbers and in (331) the name describes this sometime (but not now) Superman.82 Next observed was that given the identity of Superman and Clark Kent, to now be a sometime
534
Chapter 10
Superman is to now be a sometime Clark Kent, whether one is now being Superman or now being Clark Kent. If so, the events in the neighborhood of now being a sometime Superman are exactly those in the neighborhood of now being a sometime Clark Kent. Given the variability in names—now Jules Bagwell, now Julia Bagwell, sometimes Jules Bagwell, sometimes Julia Bagwell—the dilemma that juxtaposes historical identities with the (few) self-evident self-identities is not a problem of finding representation for one or the other interpretation. The self-identities all find representation when selecting sometimes Jules Bagwell and sometimes Julia Bagwell, and the historical identities, when selecting now Jules Bagwell and now Julia Bagwell. The problem is rather an embarrassment of riches: Why are the selfevident self-identities not ambiguous, without a reading that would be false if the name without existential closure of its event description were selected, and why do the historical identities resist interpretation with descriptions of being sometimes so-and-so? For the historical identities, a sanction against elegant variation83 seems at play (see Schlenker 2004 for discussion). A speaker would not use distinct expressions F and G to refer to the neighborhood of F-ing and the neighborhood of G-ing knowing that the neighborhoods are the same. That is, knowing that Clark Kent and Superman are the same, the speaker uses Clark Kent and Superman only if she does not believe that Clark Kent-ing and Superman-ing are the same, which is to say that she does not believe the nouns are coextensive. For a speaker eschewing elegant variation, the variation in the name is itself a sufficient cue that sometimes Clark Kent and sometimes Superman are not intended, ruling out that (257) and the like could ever have meant something true. This of course says nothing about the selfidentities, to which I now turn, where the speaker must just as clearly be taken to have intended sometimes NP and only such for her assertion to be true and unambiguous. As remarked, assertion of a self-evident self-identity about a Bagwell is fragile in circumstances where he or she is not there, (320) and (321)—dependent on a flat intonation and broken under modification, (326) and (327). Similarly, such an assertion about Superman (328) is, even if true, still misplaced as a remark about an event from which he was absent qua Superman. To explain (or perhaps merely to apologize for) how these self-identities come to be self-evident under the restricted conditions noted, first observe the effect of intonation in the resolution of ambiguous demonstrative reference, which will be extended to names. New Year’s Eve, while some disorderly guests are rounded up, I say to you: (332) Last New Year’s Eve too, that guest who bit the hostess was that guest who bit the hostess. [Or, Last New Year’s Eve too, the guest who bit the hostess was the guest who bit the hostess.]
Adverbialization in Logical Form
535
Granting that I intend with the first demonstrative a demonstration into the current scene, the frame of reference for the second demonstrative still remains, uncontroversially, ambiguous.84 On one interpretation, it repeats demonstration into the current scene or, to the same end, it is demonstrative to a linguistic antecedent, the first demonstrative, in the current context; on the other interpretation, it is demonstrative to a scene from a shared memory of last year’s New Year’s Eve or, equivalently, to some earlier reference grounded in such an earlier frame of reference (see (248)): (333) [∃Ef: last New Year’s Eve[Ef]][℩x: ∃E (that[u,E] & W[E,x] & guest who bit the hostess[E])] … [℩x: ∃E (that[u,E] & W[E,x] & guest who bit the hostess[E])] … (334) [∃Ef: last New Year’s Eve[Ef]][℩x: ∃E (that[u,E] & W[E,x] & guest who bit the hostess[E])] … [℩x: ∃E (that[Ef,E] & W[E,x] & guest who bit the hostess[E])] … Here, to intend (332) to assert a self-evident self-identity is to intend that the second demonstrative be disambiguated the first way, (333).85 It seems however that if I do intend to continue, without pointing, with the current scene as the frame of reference for the second demonstrative, I had better muster the logic instructor’s bored, flat intonation to underline that the second demonstrative repeats the first and is intended to be understood as the first was, evaluated with respect to all the same parameters. The deaccentuation indicates that the second token is not contrastive in any respect from the first and so means the same (Tancredi 1992). Understanding the demonstratives the same way in turn prejudices how their content is cashed out. Both tokens with exactly the same content describe the same events with exactly the same neighborhood. So to locate the event in the current scene at this New Year’s Eve party in the same neighborhood as remote events from last New Year’s Eve party, that neighborhood must be described as the events while in a state of having or being about to have just bitten this New Year’s Eve hostess.86 Given so large a neighborhood, the only interpretation for (332) with flat intonation is one that is true without implying anything at all about the subject’s behavior last New Year’s Eve, as desired.87 The discussion of counterdemonstrative contexts in section 10.2 explored one respect in which names behave and should be analyzed as if they are complex demonstratives. Barber (2000) has also already noted that names display context dependence on frames of reference. Should it be discovered that Venus is 10 degrees hotter when it is dusk in North America than when it is dawn there, accurate reports of the discovery would be (335) addressed to a North American conference of
536
Chapter 10
astronomers, (336) addressed to a conference in New Zealand, and (337) addressed to anyone:88 (335) Hesperus is hotter than Phosphorus. (336) Phosphorus is hotter than Hesperus. (337) Venus at 22:00 GMT is hotter than Venus at 10:00 GMT. To his observation, it may be added that (338) is also an accurate report anywhere: (338) In North America, Hesperus is hotter than Phosphorus. (339) #In North America, Venus at 22:00 GMT is hotter than Venus at 10:00 GMT. The logical form of PPs is fixed by Davidsonian analysis to engage at least one event variable, which should not occur vacuously (cf. (339)). Since the PP does not modify the matrix predicate in (338), which is about the heat on Venus rather than in North America,89 something else is said to be in North America, something on which nominal reference is contingent for the rest of the sentence to be felicitous ((338) vs. (339)): (340) [∃Ef: In North America[Ef]][℩x : ∃E Hesperus[E, Ef, x]] …90 In this example, the semantic denotation of the nouns Hesperus and Phosphorus varies with frame of reference—not in x, which is invariantly Venus, but in the events E of Hesperus-ing or Phosphorus-ing—which distinguishes these nouns from Superman, Clark Kent, Julia Bagwell, and Jules Bagwell. As such, utterances of (335) and (336), in contrast to (337), express no proposition at all except with reference tacit or overt as in (338) to a frame of reference.91 Such an example turning on the intended phases of x joins more familiar examples from Burge (1973), where a name typically denotes many objects x, Two George Bushes have been President, and singular reference with George Bush or that George Bush depends on a contextually supplied frame of reference. At the Philosophy Department of the University of Walamaloo,92 one philosopher named “Bruce” may call the attention of a second to remark that a third disagrees with a fourth: (341) Bruce, Bruce rejects Bruce’s solution to the problem of the many. Each token of the name prompts a change in the frame of reference only the narrow confines of which secure singular reference, requiring some cue from the speaker. In this exaggerated example, speaker and hearer turn some especially tight corners to navigate through the intended frames of reference,93 but it illustrates the general point that every name and demonstrative is understood with respect to an intended one, which may require more or less explicit indications (cf. (248)):
Adverbialization in Logical Form
537
(342) [℩x: ∃E (N94[u1,E] & W[E,x] & (… Bruce[E,x] …))], [℩x: ∃E (N [u2,E] & W[E,x] & (… Bruce[E,x] …))] … [℩x: ∃E (N [u3,E] & W[E,x] & (… Bruce[E,x] …))] … If so, there is little to tell apart the logical forms of (320), (321), and (328) from those of (332) in (333) and (334), and the explanation is the same for when these sentences are taken to express self-evident self-identities. It takes a flat intonation, deaccentuation of the second token of the name, to cue that it is intended to be interpreted with respect to the same frame of reference as the first, despite the available alternative represented by the adverb. The result is a self-evident truth but not one without pragmatic defect in the use of the adverb, which is as out of place here as in Yesterday, zero equaled zero or At IBM, zero equals zero (see note 87). Nevertheless, the power of prosody to disambiguate a speaker’s intentions points to just such an anomalous interpretation. It was observed that the self-evident self-identities were also fragile enough to disappear under further adverbial modification, contrasting (320) and (321) with (326) and (327): (320) T At Halloween 2005, Jules Bagwell was Jules Bagwell. (326) F At Halloween 2005, Jules Bagwell was still Jules Bagwell. (321) T At Carnaval 1995, Julia Bagwell was Julia Bagwell. (327) F At Carnaval 1995, Julia Bagwell was already Julia Bagwell. Certainly, the speaker represents herself as asserting something that could be temporally contingent, and a true self-identity would now threaten to strand two adverbs. It happens in (343), which I take to be literally true although pragmatically defective: (343) Yesterday, zero still equaled zero. At IBM, zero already equals zero. A defect may be enough to render any other interpretation more salient, as in (326) and (327), but a lacuna remains if the report is accurate that a true self-identity is beyond the reach of these sentences. Why couldn’t the speaker rely here as well on a deaccentuated second token of the name to signal her intention to assert the pragmatically defective truth? Perhaps, with (343) in highlight, it will be conceded that she may, although the result would be no more cogent than (343). If not, there should be more to the grammar than I have let on—perhaps the position or focus of the second adverb itself signals a shift in the frame of reference.95 The problem can, however, be left aside for purposes of this project—a defense of the logical syntax called on to represent the salient, false interpretations of (326) and (327), which can be no instantiation of ⌜PP, a=a⌝. If such translation is always in error, then all the better that (326) and (327) never express anything equivalent—if those
538
Chapter 10
are the facts—even if it is not understood why the equivalent is not remotely possible. The invariant logical syntax for all these sentences is that delivered by adverbialized, supermonadic logical form; in result, expression of a self-evident self-identity equivalent to ⌜a=a⌝ is only a sometime thing. 10.4
Further consequences of adverbialization and supermonadicity: NPs as event descriptions
Adapting Schwarzschild’s (1996) discussion to the present setting, it can be said that the speaker who intends (344) to be true of the table in (347) but false of the table in (348), which fails to match any fiction title with complementary nonfiction titles, has in mind that the tables both demonstrate five events or states each corresponding to a single row. It suffices for this speaker that the domain of events quantified over is restricted as in (345) to those demonstrated. These events are complementings just in case each is (see (346)), and in fact no row of (348) is a complementing in any meaningful sense—a science fiction voyage through the human body, Fantastic Voyage, surveying creation rather than capitalism, has nothing to do with Das Kapital and The Wealth of Nations. (344) The (seven) fiction books in the table complement the (seven) nonfiction books. (345) [∃E: as displayed[E]] the (seven) fiction books in the table complement[E] the (seven) nonfiction books. (346) complement[E] ↔ ∃eEe & ∀e(Ee → complement(e)) (347) Fiction Alice in Wonderland Fantastic Voyage David Copperfield, Hard Times Oedipus Rex, Agamemnon Richard III
Nonfiction Aspects of the Theory of Syntax, Language Gray’s Anatomy Das Kapital, The Wealth of Nations Introduction to Psychoanalysis The Prince
(348) Fiction Alice in Wonderland Fantastic Voyage David Copperfield, Hard Times Oedipus Rex, Agamemnon Richard III
Nonfiction The Prince Das Kapital, The Wealth of Nations Introduction to Psychoanalysis Gray’s Anatomy Aspects of the Theory of Syntax, Language
A problem is joined, however, when a speaker utters (349) intending just like (344) for it to be true of (347) but false of (348):
Adverbialization in Logical Form
539
(349) The (seven) fiction books in the table are listed to the left of their (seven) complementary nonfiction books. Seven fiction books are listed to the left of seven complementary nonfiction books. The (seven) fiction books in the table sit opposite their (seven) complementary nonfiction books. Seven fiction books sit opposite seven complementary nonfiction books. The tables demonstrate the same events as before, but each row is undeniably a listing-to-the-left-of or a sitting-opposite, and it is now not obvious how the adjective complementary embedded in a plural, collective description of some other things manages to apply to the events described by the matrix sentence, discriminating between those of (347) and those of (348). Such usage is pervasive—surely (350) has a sense that would be a false report if nations thwarted only their enemies and gave comfort only to themselves. Similarly, it is not sufficient for the intended reading of (351) that each lovebird is as good as itself even though one or the other in each of three pairs is a rather better domestic partner than its mate. Sentence (352) is meant to rule out chaos in the kindergarten where no pupil wears her assigned name tag or sits in her assigned seat, and (353) clearly has one reading that implies that each cowboy escorted his bride to his dream homestead and another that he escorted her to hers. Similarly, if Alex and Janet did not go to the same movie, (354) can be understood to imply that each took her own children. (350) The nations thwarted their allies, and they gave comfort to their enemies. The nations thwarted allied nations, and they gave comfort to enemy nations. (351) Those six lovebirds are as good in the nest as their six better halves. (352) Twenty very obedient pupils wore their twenty assigned name tags and sat in their twenty assigned seats. (353) Seven cowboys personally escorted their seven mail-order brides to their seven personally selected dream homesteads. (354) Alex and Janet took their respective children to the movies. (Gawron and Kehler 2002) In contrast, these circumstances falsifying (350)–(354) verify the corresponding sentences of (355)–(359). In the case of (355), for example, if every one of these nations is an enemy to some among them and if these nations have thwarted their enemies, then these nations have all been thwarted, verifying (355). (355) These nations thwarted these nations, and these nations gave comfort to these nations. (356) Those six lovebirds are as good in the nest as those six lovebirds.
540
Chapter 10
(357) Twenty very obedient pupils wore these twenty name tags and sat in these twenty seats. (358) Seven cowboys personally escorted these seven brides to these seven homesteads. (359) Alex and Janet took their children to the movies. (Gawron and Kehler 2002) As Gawron and Kehler (2002) observe in a discussion of respective, if the only contribution of nominal description to meaning is to fix reference to what is described (or quantified over), then it is not obvious how sentences (350)–(354) manage to convey narrower truth conditions than their counterparts in (355)–(359) about the same things. Biting a bullet, Gawron and Kehler (2002) propose that any appearance to the contrary distinguishing (354) and (359) is thanks to an unspoken adverb respectively: (360) Alex and Janet respectively took their respective children to the movies. The adjective respective spoken in (354) or (360) is held to be as ineffectual as its compositional semantics cranks it out to be, the reference of the definite description containing it being no different from the one in (359) without it. (Something else, presumably, clues the hearer to the presence of the unspoken respectively and tells apart (354) and (359) to the extent that (359) seems to be a weaker vessel for the narrower truth conditions.)96 Sentences (354) and (359) are claimed to converge, as expected since the adjective itself doesn’t mean much, on those occasions when the unspoken adverb is absent, as when (354) (or (359)) is used as a true report of Alex and Janet going together with their children to the same movie. Such a context—so it is argued—shows that the adjective respective occurs in (354) without real teeth. There is to be sure something about this convergent context to be explained: both (359) and (354) are, despite the adjective respective, true in it and thus contrast with overt expression of the adverb as in (361): (361) Alex and Janet respectively took their children to the movies. On the other hand, contrary to the convergence claimed, if Alex and Janet each take the other’s offspring (and not their own) to different cinemas uptown and downtown, (354) is plainly false under any construal of the sentence, while (359) is true with no scruples against swapping children. Moreover, if the interpretation and contribution of the adjective respective is, as alleged, decided entirely within the DP and without itself commanding any relation to the events described by the matrix, it is unexpected that (362) and (363) are worse than (354) and that (364) and (365) should then improve them: (362) *Alex and Janet’s respective children sat in the first row.
Adverbialization in Logical Form
541
(363) *Alex and Janet’s respective children are smarter than the average bear. (364) Alex and Janet’s respective children sat in their laps. (365) Alex and Janet’s respective children are smarter than they were at their age. Nor is it expected, since the reference of their (two) children and their respective children is the same, that (367) and not (366) should lapse incoherent if each parent has only one child: (366) Waiting on line, Alex and Janet stood between their (two) children. (367) *Waiting on line, Alex and Janet stood between their respective children. It should have rescued (367) if it could ever mean what (366) means, but in fact, respective, the adjective, does not occur without itself relating in some way to events described outside its host DP, resulting in this contrast, the contrast between (362)– (363) and (364)–(365), and the contrast in truth between (354) and (359) in the event that Alex and Janet take each other’s children to different cinemas. As already hinted at in the contrast between respective in (354) and respectively in (359), the scope of the adjective need not coincide with what is otherwise available to adverbs. Scope that includes something from the matrix description of events is indicated by the contrast between (368) and (369), and a scope small enough to exclude the matrix verb itself is indicated by the contrast between (370) and (372).97 The teenagers may be participating within the make-out session from behind their respective chaperones (implicating that the one’s participation is not behind the other’s chaperone) but the make-out session itself cannot be so described. (368) The (two) teenagers made out in the backseat between their (two) chaperones. (369) *The (two) teenagers made out in the backseat between their respective chaperones. (370) The (two) teenagers made out in the backseat (directly) behind their respective chaperones (in the front seat). (371) ?The (two) teenagers made out in the backseat respectively behind their chaperones (in the front seat). (372) *The (two) teenagers respectively made out in the backseat (directly) behind their chaperones (in the front seat). In this case, the presence of an overt preposition behind affords a position for the adverb respectively in (371) to paraphrase the intended interpretation. Absent overt preposition, the adjective respective may acquire a scope within the matrix sentence smaller than any available to respectively: (373) At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer disentangled/separated their champions.
542
Chapter 10
(374) At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer disentangled/separated their respective champions. (375) *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer respectively disentangled/separated their champions. *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer disentangled/separated respectively their champions. *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer disentangled/separated their champions respectively. Sentence (374) requires that Ali’s trainer grab hold of Ali and Frazier’s trainer grab hold of Frazier. If, instead, they each go for the other’s champion, (374) becomes false while (373) remains true. That is, the trainers acted respectively on their champions, as (374) asserts in contrast to (373). But since they did not disentangle or separate them respectively, the adjective cannot be paraphrased away by any placement of the adverb in (375), which all end incoherent. There are further scope effects concerning the adjective to be noted. To my ear, (376) is equivalent to (374), and (377) is no better than (375): (376) At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer pulled their respective champions apart. (377) *At the Thrilla in Manila, Muhammad Ali’s trainer and Joe Frazier’s trainer pulled apart their respective champions. Where the resultant state is expressed by a freestanding morpheme, apart in (376) and (377), word order is disambiguating. In (377), their respective champions occupies a higher position that captures pull apart within its scope. The position of their respective champions in (376) must be low enough to exclude apart from its scope. Where the resultant state is either unpronounced or perhaps pronounced as a morpheme incorporated into the verb, as in (374), a logical form similar in structure to (376) must be assumed to derive the coherent interpretation. Of course, under the adverbialization of logical form, the scope of respective can be as small as the smallest structure available as the scope of its host DP. The adverbialization of logical form denies that the only contribution of the restriction inside DP is to fix reference to what the DP describes or quantifies over. It is also restriction to an adverbial quantifier over events, with the narrow scope afforded it under supermonadicity: (13) … [Dα : ∃ENP[E,α]] (… θ[Ej,β] Ψ) ⇒ … [Dα : ∃ENP[E,α]][℩E : NP[E,α]][∃Ei : N[E,Ei]](… θ[Ej,β] Ψ) On this view, not only the adjective respective but also complementary, allied, enemy, better-half, assigned, mail-order, and personally selected in (349)–(354) are all as spoken and on their own merit constituents of adverbial phrases, whose scope is as
Adverbialization in Logical Form
543
local as their allied DPs and unrelated to the syntactic positions of -ly adverbs, respectively, alliedly, and so on. Although the nations, their allies, and their enemies are all the same members of the UN, the adverbialization of logical form gains purchase on a semantic distinction between (350)–(354) and their counterparts in (355)–(359) to the extent that nations and allied nations, or nations and enemy nations, afford descriptions of different events when embedded in the adverbial event quantification: (378) The nations thwarted their allies DP-W-O-Agent-Tense-Cause—DP-W-O-Patient-thwart (379) [the X : ∃E nationsΦ[E,X]][℩E : nations[E,X]][ ∃E0 : N[E,E0]] W[E0,X] O[E0,E1] ∃XAgent[E1,X] Past[E2] Cause[E1,E2] [the Y: [[℩X:Φ] ∃E allies[E,Y,X]] [℩E : [[℩X:Φ]allies[E,Y,X]][ ∃E0 : N[E,E0]] W[E0,Y] O[E0,E2] ∃XPatient[E2,X]thwart[E2]
(i) (ii) (iii) (iv) (v)
According to (379), lines (iv–v), the events of the allies’ being thwarted are in the neighborhood of their alliances. The neighborhood is understood in context— perhaps a being thwarted is in the neighborhood of an alliance just in case it is a causal consequence of a betrayal within it, or even more remotely, the one is in the neighborhood of the other just in case their juxtaposition is delicious confirmation of one’s cynicism about affairs of state. Anyway, the neighborhood relation implies enough of a common history to infer that this being thwarted would not be in the neighborhood of that alliance unless the participants in the thwarting were also those in the alliance:98 (380) N[E,E0] ↔ ∀e0(E0e0 ↔ ∃e(Ee & Φ[e, e0])) ∀x(Φ[e, e0] → (W(e,x) ↔ W(e0,x))) The schema (380) characterizes those neighborhood relations that happen to be distributive, relating according to some Φ each being allied e, in this example, to a being thwarted e0. Such a distributive relation is necessary to express the interpretations at issue for (349)–(354), but the sentences may also be used to express weaker truth conditions as in (381) with possibly a nondistributive neighborhood relation in mind: (381) You could say that twenty pupils sat in their twenty assigned seats, but so unruly they were, that not a one sat in the seat assigned to her herself rather than the one assigned to her best friend. A sentence such as (382) is thus ambiguous between an interpretation implying that each driver stood beside his rented automobile and another that is vague about which drivers stood beside which automobiles. The knowledge that a single
544
Chapter 10
automobile cannot be stood between in turn disambiguates (383) in favor of the vague, nondistributive neighborhood: (382) The (two) drivers stood beside their (two) rented automobiles. (383) The (two) drivers stood between their (two) rented automobiles. The possibility of (381) and (383) points to one way in which the adjectives of (349)–(354) deviate from respective, which rejects altogether a nondistributive neighborhood, as in (385) (and in (367), (369), and (377)): (384) The (two) drivers stood beside their (two) respective automobiles. (385) *The (two) drivers stood between their (two) respective automobiles. Short of stipulating restrictions on the various relations ⌜N[E,E0]⌝ that may cooccur with respective and the adjectives of (349)–(354), it may be more promising to look to more basic differences between them. The adjectives of (349)–(354) do after all derive from the primitive Davidsonian vocabulary of event concepts—there are rentings, complementings, alliances, assignments, and so on, whereas respective does not—there are no respectings. Like many event concepts, rent, complement, ally, and assign apply to both solitary and collective events. If one has in mind in understanding (382) that there is a renting by them of the two automobiles, then a neighborhood relation that is according to (380) distributive with respect to events will look for a standing-beside in which participate the two drivers and two automobiles, which indeed leaves vague the arrangement of individual driver and automobile. If, on the other hand, as supposed for the interpretations of (349)–(354) at issue, there are taken to be two rentings, each of a single automobile by a single driver, then each renting will look in its neighborhood for a standing beside with the same solitary participants. What has so far been argued for respective is that in the translation of their respective automobiles, it contributes in some way to a phrase, the restricting NP, which describes events: (386) … [the Y: [[℩X:Φ] ∃E respective-automobiles[E,Y,X]] [℩E : [[℩X:Φ] respective-automobiles [E,Y,X]] [∃E0 : N[E,E0]] … The unacceptability of (385) (and similarly (367), (369), and (377)) is derived if the phrase modified by respective cannot describe a single, collective event or state (say, of possession—cf. their respectively owned automobiles) in which the two automobiles participate jointly. If respective forces essentially plural reference to events, (385) ends up demanding incoherently that each event of only a single automobile has in the neighborhood an event of its being stood between. It seems that respective rejects not only singular reference to events but also mass reference. Those events referred to meet the (nonoverlap) conditions for plural count reference99 as the following illustrates. Suppose three thermos bottles contain steam, water, and ice,
Adverbialization in Logical Form
545
respectively. The descriptions in (387)–(389) relating containings to noncontainings are mass terms denoting such events: (387) The content of these three containers is insulated by its containers from (all) the noncontaining containers. (388) The contents of these three containers are insulated by their containers from (all) the noncontaining containers.100 (389) The contents of this thermosware are insulated by their containers from (all) the noncontaining containers. The same thoughts can be expressed with respective only to the extent that such events are corralled as in (391) into three discrete events, each the maximal containing by a single thermos bottle: (390) *The content of these three containers is insulated by its respective containers from (all) the noncontaining containers. (391) ?The contents of these three containers are insulated by their respective containers from (all) the noncontaining containers. ?The contents of these three containers are insulated by their respective containers from (all) their respective noncontaining containers. (392) ?*The contents of this thermosware are insulated by their respective containers from (all) the noncontaining containers. ?*The contents of this thermosware are insulated by their respective containers from (all) their respective noncontaining containers. Although I will not take up the syntax and semantics of respective and respectively,101 if the nominal descriptions that embed them are essentially plural descriptions of events, there may be nothing further to be said about the sense in which these event descriptions modify matrix phrases. Their neighborhood relations are just those invoked in adverbial modification elsewhere. The contrast between (383) and (385), and similar contrasts, descend solely from the fact that (383) and not (385) contains an event description that may be taken to denote a single collective event. 10.5
Summary
Of a wedding picture, it can be said that (393) The two newlyweds flank the wedding cake, beside their respective fathers. (394) The two newlyweds flank the wedding cake, behind their respective fathers. (395) The two newlyweds flank the wedding cake, between their fathers. But not that (396) *The two newlyweds flank the wedding cake, between their respective fathers.
546
Chapter 10
The reference of the two newlyweds and their (respective) fathers is plain. If nominal expressions contribute only their referents to meaning, a semantic explanation for the contrast between (395) and (396) is forfeit. Yet it cannot be a point of syntax to banish the co-occurrence of the preposition between and an adjective respective embedded within its complement, as they do co-occur, meaning permitting: (397) The two newlyweds flank the wedding cake, between their respective parents. It is rather that an adverb respective(ly) with scope outside the nominal description, as Gawron and Kehler (2002) propose, enforces a difference between (396) and (393)–(394) and between (396) and (397). It is however not an adverb that includes the verb within its scope, as the newlyweds do not respectively flank anything: (398) *The two newlyweds respectively flank the wedding cake, beside their fathers. (399) *The two newlyweds respectively flank the wedding cake, behind their fathers. There is a single flanking—not one for each. In it, the newlyweds are respectively beside (or behind) their fathers. What then explains (396)—its contrasts with the other prepositions and with (397)—is that the newlyweds cannot respectively be between their fathers, while they can be respectively beside or behind their fathers and they can be respectively between their own parents. Adverbialization is the rule of grammar by which the adjective respective, embedded within the nominal description, nevertheless comes to modify the local thematic relation, and supermonadicity delivers the clause structure that allows modification of the thematic relation alone to the exclusion of the verb, so that (393)–(394) are not confounded with (398)– (399). It is a small corner—the paradigm (393)–(399) and what is to be said about respective(ly). It delivers a compelling, compact argument for two Eventish standard-bearers, adverbialization and supermonadicity. As the chapter began, adverbialization puts the analyticity in the analytic truth (400) when it is understood to say that the teacher teaching teaches, and it is again put to good use in the service of descriptive anaphora as paraphrased in (401): (400) The teacher teaches. (401) No teacher of a science postmodernizes it. ‘No teacher of a science, in any event of teaching a science, postmodernizes the science in that event.’ For the lifetime effect (section 10.0) (Kratzer 1995; Musan 1997), supermonadicity and adverbialization are necessary to explain why the past tense in (403) draws out an implicature about the subject’s condition that what was is no longer rather than
Adverbialization in Logical Form
547
about the object’s condition—that is, wherefrom the selectivity that joins Tense to the subject. Supermonadicity and adverbialization are also necessary to explain the shift under the influence of the adverb in (404) from an implicature of death to one of resurrection: (402) Lazarus resembles Uncle Ephraim. (403) Lazarus resembled Uncle Ephraim. (404) Lazarus dead resembled Uncle Ephraim. When the lifetime effect is suspended, as it is with intentional predicates (405) (Musan 1997) and elsewhere (406), adverbialization with its scope confined to the local thematic relation is how it can be that live participation while Franz Kafka or while Eli Yale of the now dead has effects that survive as present scholarship support or as present fame, without the ghosts of Kafka and Yale also present: (405) Franz Kafka is famous. (406) Eli Yale still supports scholarship students due to the prudent investment of his 1718 donation of ₤800. The lifetime effect is a special case of the more general interaction of adverbialization and supermonadicity, according to which the derived adverbial temporally locates the events and only those events that are locally related under supermonadicity to the adverbialized DP. The interaction derives the temporal coordination among while being Gorbachev, while being the USSR, while being Putinate Russia, the present, the past, being-from, and being-in that results in (407)–(414), for a Gorbachev born in Stavropol in 1931 and visiting Moscow today: (407) Gorbachev is from the USSR. (Musan 1997) (408) #Gorbachev was from the USSR. (Musan 1997) (409) #Gorbachev is in the USSR. (410) Gorbachev was in the USSR. (411) # Gorbachev is from Putinate Russia. (412) #Gorbachev was from Putinate Russia. (413) Gorbachev is in Putinate Russia. (414) Gorbachev was in Putinate Russia. Moreover, adverbialization and supermonadicity uphold the literal truth and felicity of (415) and (417), without encroaching on the personal identity of the man and the boy or of the city before and after the revolution: (415) Lenin was arrested in St. Petersburg and not in Leningrad. (Saul 1997a; Trotsky [1924] 1971)
548
Chapter 10
(416) #Lenin was arrested in St. Petersburg and not in St. Petersburg. (417) The aged don Michael Corleone has surpassed the boy Michael Corleone in murder and mayhem and not much else. (418) #Michael Corleone has surpassed Michael Corleone in murder and mayhem and not much else. Adverbialization and supermonadicity thus make room for a tame metaphysics while explaining that (i) the existence of the lifetime effect is dependent on the choice of nominal, (ii) adverbial modifiers (successively) narrow or shift the lifetime effect, and (iii) the lifetime effect, derived from nominal descriptions standing in as adverbial modifiers, like all adverbial modifiers, governs local participation, small subevents. The logical syntax of (419) and (420) hardly tells them apart: (419) The hitter now facing an opposite-handed pitcher will outperform the hitter who earlier faced a like-handed pitcher. (420) #The hitter now facing an opposite-handed pitcher outperforms the hitter who earlier faced a like-handed pitcher. Yet (420) means either that the hitters are not the same or, incoherently, that the same hitter’s overall performance is better than itself, whereas (419) can be understood to say that this hitter, like all others, does better against opposite-handed pitchers and will do so now. As in the discussion surrounding (234)–(239) above, the contrast hangs on an interaction among the background laws of baseball, the particular tense chosen, and adverbialization. Adverbialization and supermonadicity are universals of clause structure. All the above, the selective modification of local thematic relations and the asymmetries it induces in the temporal coordination of the local events, recur in sentences that assert identities and resemblances—symmetric relations par excellence (section 10.3). But they are embedded in a canonical logical form ((424), (423)) that both allows the observed asymmetries to emerge and shows how it is possible in the first place for identity statements and the like to be modified by temporal adverbials and secondary predicates: (421) For some evenings in 1892, Venus was Hesperus (aligned with a crescent moon). For some evenings in 1892, Venus was the evening star (aligned with a crescent moon). (422) #For some evenings in 1892, Hesperus was Venus. #For some evenings in 1892, the evening star was Venus. (423) DPi-[℩ER: theni]-W-Theme (-not)- [W-Tense-Theme Be [DP-W-Theme ]]
Adverbialization in Logical Form
549
(424) DPi-[℩ER: theni]-W-Theme (-not)- [W-Tense-Theme resemble …] (425) [E1, E2 ] ↔def ∀X(W[E1,X] ↔ W[E2,X]) & ∀T(time[T] → (At[E1,T] ↔ At[E2,T])) (426) [E1, E2 ] ↔def ∀X(W[E1,X] ↔ W[E2,X]) And it has been explained how the identity in (427) is unequivocally true and yet fails to license a substitution that infers (429) from (428): (427) Clark Kent is Superman. (428) Clark Kent does not resemble Superman (enough to be instantly recognized). (429) #Clark Kent does not resemble Clark Kent (enough to be instantly recognized). When substitutivity under identity appears valid in an inference from (427) to (430), it must however be conceded (sections 10.3.0–10.3.1) that it is so only via the intercession of enthymemes as Parsons (2000) foresaw (see note 53): (430) Clark Kent is no happier than Superman is.
11
Naive Reference for the Cinéaste
Without logical form supermonadic and adverbialized, and thus without other recourse, the corruption of nominal reference to refer to phases, stages, or temporal slices of things despite mention only of the things themselves seems inevitable when something else is counted (Barker 1999; Carlson 1982; Doetjes and Honcoop 1997; Moore 1994):1,2 (1) National Airlines served at least two million passengers in 1975. ⊬ National Airlines served at least two million persons in 1975. (Gupta 1980, 23) (2) Four thousand ships passed through the lock last year. (Krifka 1990) (3) 57,000,760 liters of water passed through the pump last month [in a closed system with only 1,000,000 liters in circulation]. (Moore 1994) As in section 10.1, it would be a mistake to think that time alone parses what is counted: (4) Four hedge fund managers are under investigation in four jurisdictions. (5) Four thousand city officials have not improved on their predecessors’ records in office. (6) According to the unpaid citations, three violators broke a health code, a fire code, and a building code. In (4), the same crook managing four hedge funds, never having managed one without the other three, is the elusive target of independent investigations into the four funds. It is enough for (5) that 4000 city offices languish—4000 venues at which nobody does much of anything—although the dogcatcher may concurrently be the tax collector and the fire marshal, and so on. As much as what one knows about what (6) says, the sale of one miserable pushcart hot dog may be all that is behind it, the same act cited as three violations.3
552
Chapter 11
Whatever is counted, the lifetime effects and the interaction among descriptive content, adverbial modifiers, and Tense remain mutatis mutandis as observed in section 10.1: (7) Last year, 4000 ships dismantled by Customs resembled an iron junkyard and reassembled passed through the lock minus 400,000 tons of contraband. (8) (9) (10) (11)
Four Four Four Four
superheroes superheroes superheroes superheroes
resemble their archenemies. resembled their archenemies. unmasked resemble their archenemies. unmasked resembled their archenemies.
Both (9) and (11) uttered out of the blue imply that how things were with the superheroes is no longer—(9), that the superheroes are no longer, and (11), that their being unmasked is no longer. It doesn’t matter that the four are one who moonlights in four cities with a weakness for wardrobe changes. The argument of section 10.0 against nominal reference to temporal slices or aspects has been that it falls short of an account of such lifetime effects and the interaction with Tense and adverbial modification, and that providing a true account makes nominal reference to temporal slices or aspects superfluous. I would not like counting to now undermine the argument or its implications for logical syntax. If, to the contrary, the ambiguity in what is counted in (1)–(3) and (4)–(6) and below in (12)–(13) is taken at face value to directly reflect what the (in)definite descriptions denote, we should wonder why one cannot comment directly on the identity of what is purportedly referred to. If, for example, (14) and (15) are true, it is unambiguously and unequivocally so, despite the false identities in (16) obtained when descriptions refer transparently to what those in (12)–(13) are purported to (see section 11.1 for further discussion).4 (12) Three million passengers crowded National Airlines routes last year. (13) ⊬ Three million frequent fliers crowded National Airlines routes last year. (After Gupta 1980, 23; Moore 1994) (14) The three million passengers who crowded National Airlines routes last year were the one million frequent fliers loyal to it. (15) The three million passengers who crowded National Airlines routes last year were the frequent fliers loyal to it. (16) (The) three million (pairs) of passenger and passage that crowded National Airlines routes last year are not (the) one million frequent fliers loyal to it. (The) three million temporal slices/phases each of a passenger in passage crowding National Airlines routes last year are not (the) one million frequent fliers loyal to it.
Naive Reference for the Cinéaste
553
Naive reference will out: (the) three million passengers refers to passengers (i.e., persons simpliciter) and never really to anything else, just as the one million frequent fliers refers naively to such, the same such as it turns out. The unwavering truth of (14) and (15) says as much. (17) [(the) ∃X : ∃E 3,000,000 passengers … [E,X]] … [(the) ∃X : ∃E 1,000,000 frequent fliers … [E,X]] … If so goes the reference of these descriptions, it cannot be that the arithmetic expressions they contain express the arithmetic properties determined by the numeric identity of what is referred to, since what is three million is not what is one million: (18) (3,000,000(X) & 1,000,000(Y)) → X ≠ Y Instead, three million translates as something like “now counted to three million”, abbreviated as ‘∃eμ count[eμ, E, 3,000,000]’, a description of an event of measurement (the counter clicks, as it were) under an explicit protocol (section 12.1): (19) [(the) ∃X : ∃E (∃eμ count[eμ, E, 3,000,000] passengers[E,X] …)] … [(the) ∃X : ∃E (∃eμ count[eμ, E, 1,000,000] frequent fliers[E,X] …)] … Who is counted and who the (in)definite descriptions (the) one million frequent fliers and (the) three million passengers refer to is who there is, persons, who are counted and under some protocols such as the one in effect for (12) recounted. As the very same X are the passengers and the frequent fliers, it must be that one protocol captures them in three million events E of passenger-ing and the other in one million states of being a frequent flier, coincident with their lifespans as travelers and coincidentally equinumerous with the X themselves. Sentences (12), (13), and (15) are among the puzzles of extensional substitutivity (section 8.1), where an identity statement (15) fails to license the substitution and inference from (12) to (13). Such puzzles, recall, are resolved via adverbialization, without denying either extensional substitutivity or the semantic innocence of (the) three million passengers: (20) [(the) ∃X : ∃E (∃eμ count[eμ, E, 3,000,000] passengers[E,X] …)] [℩E : ∃eμ count[eμ, E, 3,000,000] passengers[E,X] …)] [∃E0 : N[E,E0]] Φ[E0] The (in)definite description refers naively, but a coreferent description substitutes for it salve veritate only if their derived, adverbial descriptions of events are also coreferent. The events while three million passengers have to be the same as the events while frequent fliers, which the failed substitution shows them not to be. Of course a naked arithmetic property of the passengers affords no such discrimination among events: the passengers are ever three million for as long as they ever are. So for them to do something while they are three million is merely to do it while they
554
Chapter 11
be. If, on the other hand, what is counted is events, then it matters for what events are described that they have been counted to be three million. Given the logical form of adverbial modification, the second line of (20), it can in turn be expected that what events are counted and how so might condition in some way (via the neighborhood relation ‘N[E,E0]’) the matrix events to which they are related. Moore (1994) and Doetjes and Honcoop (1997, 291ff.) observe such conditioning, which the latter call sequencing of events (evoking sequence of tense), in the contrast between (24) and (25): (21) Three million passengers flew National Airlines last year. (After Gupta 1980, 23) (22) Three million passengers crowded National Airlines routes last year. (After Moore 1994) (23) The three million passengers who flew National Airlines routes last year were the one million frequent fliers loyal to it. (24) Three million passengers had three million opinions about the food on National Airlines. (25) *Three million passengers have three million opinions about the food on National Airlines. (26) One million frequent fliers loyal to it flew National Airlines last year. (27) One million frequent fliers loyal to it crowded National Airlines last year. (28) One million frequent fliers had opinions about the food on National Airlines. (29) One million frequent fliers have opinions about the food on National Airlines. Despite (23), both (21) and (22) can still be followed by (24), recounting the passengers’ experience on board and continuing to count as three million the one million. In the present tense, however, (25) cannot count more passengers than persons who fly, despite the fact that these persons still have three million opinions about flights they remember all too well. The relation between the count and present tense wants explanation. Instead of the bare ‘3,000,000(_)’, three million translates as above as “now counted this way to three million.” With the counting entered into the description of events, adverbialization will derive in effect (30) and (31): (30) Three million passengers while counted to three million had three million opinions about the food on National Airlines. (31) *Three million passengers while counted to three million have three million opinions about the food on National Airlines.
Naive Reference for the Cinéaste
555
Clicking the counter as passengers go by frames or locates opinion recorded contemporaneously as in (30), but there is no counting to three million that frames current conditions, which, according to (31), is what defeats (25). The speaker is secure in her constant reference to naive, familiar objects by a sleight of hand that manipulates her epistemic conditions, in this case, conditions of measurement. It remains to section 11.0 to spell out the conditions on sequencing of events that while counted to three million glosses over, and to section 12.1, to argue for and define a semantics for three million in which events of measurement are indeed denoted. Section 11.1 revisits the identity statements about counting, such as (23), that launch the current puzzle of extensional substitutivity and provide them with a univocal, unambiguous semantics free of ontological dereliction, while conforming to and providing more evidence of the canonical clause structure that adverbialization and supermonadicity impose on them. Prior to the details, there is already in a few brushstrokes the highlights of an argument for some of the more important conclusions. First, take notice, as Moore (1994) trenchantly has, that all the descriptive generalizations and conditions that attach to (32) attach with equal force to (33), which replaces the verb with the collective crowd: (32) Three million passengers flew National Airlines last year. (After Gupta 1980, 23) (33) Three million passengers crowded National Airlines routes last year. (After Moore 1994) But if the events counted to three million frame three million events of a passenger participating, these are not three million crowdings but must be other events in which the passengers participate—hence, supermonadicity tout de suite. These three million events E collectively constitute (‘O[E,E′]’) National Airlines’ crowds E′ of last year, which are of course fewer in number. Naive reference holds that the things referred to, the passengers and the frequent fliers, are the very same. To put the skeptic in more desperate straits, (34)–(37) recruit from the language of aviation the same noun soul to substitute for passenger and frequent flier throughout the paradigm. Good luck to whoever would deny naive reference and so deny that the souls on board last year were the same things as the souls ever on board last year: (34) The three million souls on National Airlines manifests last year are one million Social Security numbers in the TSA database. (35) The three million souls on board National Airlines flights last year were the one million souls ever on board National Airlines flights last year.
556
Chapter 11
(36) *Three million souls on board crowded National Airlines routes last year. They are flying again this year. (37) One million souls crowded National Airlines routes last year. They are flying again this year. If naive reference is not to be denied, it must be instead that there is a difference in the events described afforded by counting souls to three million rather than to one million, if there is to be any account at all of why the former fails in (36) a condition on sequencing of events that somehow the latter complies with in (37). To count the events or states e of being a soul, ∃x soul(e,x), rather than the souls x, ∃e soul(e,x), themselves does not however suffice for a difference to be counted. After all, the events or states that there are of being a soul and the events or states there are of such sharing participants with events of flying are all exactly the same whether reported as in (37) or as (36) attempts. It is unclear, if the facts are the same, why or how their measurement should ever differ, and if allowed to differ, it is equally unclear how their measurement could be determinate. The same spatiotemporal goings-on measure different only as the result of different conditions for measurement. There is an essential relationship between measurement and a frame of reference for what is measured. As remarked on earlier, if instead of events or states of being a soul, one is to count green regions, in the narrow frame of reference in (38), the measurement is 2, exactly and determinately so; just as determinate, the measurement is 1 under the wider frame of reference: (38)
(39) Two green regions are all the green there is. (40) One green region is all the green there is. Then, to count 2 with any true and felicitous utterance of (39) commits the speaker to a perspective on the goings-on coincident with the narrow frame of reference, and correlatively, to count 1 must be a report from the wider one. Similarly, it may have been a ruse to thwart accurate measurement of troop strength that the Soviet troops parade through Red Square and circle back out of sight to parade through it again:
Naive Reference for the Cinéaste
557
(41) Three hundred thousand Soviet troops flowed through Red Square in the May Day parade. (42) One hundred thousand Soviet troops flowed through Red Square in the May Day parade. Yet even if it is ultimately a false measurement of troop strength, the field officer in Red Square (or a traffic counter stretched across the parade path in front of the reviewing stand) measures 300,000 under conditions that can only measure 300,000. The spy plane above, under conditions just as determinate, measures 100,000, and both field officer and spy plane have set out to count events or states of being a Soviet troop on parade, parts of a life, as it were. Those the field officer counts to 300,000 are indeed different from and more numerous than those the spy plane counts to 100,000. To tell apart (36) and (37) while preserving naive reference, there is to be a difference in the events the derived adverbials describe, which three million and one million secure only if field officer, spy plane, or some others’ frames of reference are thrown into the mix so that the events of being a soul on board that the one frame of reference fixes at three million are indeed different from the same such events that under the other frame of reference measure one million. Then, in turn, if the derived adverbials do indeed describe different events, the adverbials may very well not be appropriate frames for the same events: (43) #Three hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. (44) One hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. Despite a field officer on the ground and even with a traffic counter at the reviewing stand to provide a determinate measure of 300,000, it is of no help to (43), which, in contrast to (41), remains infelicitous counting more than bodies. The field officer with the narrow frame of reference for a count to 300,000 is in position to observe a flow of troops through Red Square (41), but that frame of reference is too narrow a space for a stream around Central Moscow (43), which fits only the spy plane’s frame of reference, from which troop strength is determinately measured 100,000 (44). This spatial dimension to adverbialization came up earlier (section 8.3) in the contrast between (45) and (46), where the frame of reference necessary to count to 2 in counting the morning star and the evening star is too narrow to encompass their circling the sun: (45) The morning star and the evening are playing hide-and-seek in the twilight, season after season. (46) #From the vantage point of Jupiter, (you can see that) the morning star and the evening star are circling the sun. From the vantage point of Jupiter, (you can see that) the morning star and the evening star is circling the sun.
558
Chapter 11
Measurements with different results are of course incoherent divorced from their frames of reference. If so, one does not grasp the thought intended with any of these sentences and judge it true or false without also grasping the intended frame of reference or at least description of it precise enough to converge on the same measurement in all its instantiations. It should be apparent that to count one or two green regions, three million or one million fragments of a life, soul on board or Soviet troop on parade, is as much spatial judgment as it is temporal. It is then a short step to recognize in spatiotemporal judgments rendered under intended frames of reference, judgments that derive from a perspective that affords the intended frame of reference and a scene from that perspective parsing the spatiotemporal into the events to be counted. To explain the contrast between (36) and (37) is thus to take on board a cinematic semantics, according to which there is always a scene in progress the narration of which the discourse advances. What is to be said about (36) and (37) is that the cinematic setting under which souls on board are counted three million proves not to be appropriate to frame the present events reported of this year’s flying. In contrast, a cinematic setting under which the souls count to one million suffers no such limitation in framing present events. If narrative’s cinematic structure is at issue, it can be expected that sequencing events will engage viewpoint aspect as much as it does the actual spatiotemporal location of events. Preliminary to this engagement, consider the opposing viewpoints of simple past and present perfect.5 Let a single cell in a Petri dish be generation 1 and each subsequent generation be a division of the preceding generation’s cells in two. At generation 2, the Petri dish contains two cells, and it contains four cells at generation 3. Having observed two sets of divisions in three generations and now staring at four cells in a Petri dish, a lab rat may say (47) in the simple past to report an event that explains the current number of cells, generation 2 dividing into generation 3: (47) Two (parent) cells (just) divided. The simple past unmodified cannot however report generations 1 and 2 becoming generation 3 (48). The intended interval requires description as in (49): (48) #Three (parent) cells (just) divided. (49) In two generations, three (parent) cells (#just) divided. (50) Two (parent) cells have (just) divided. (51) Three (parent) cells have (#just) divided. The structuring of event space into episodes is context- and interest-dependent enough that explicit instruction in (49) to regard what happens across two generations to be an episode suffices for the simple past’s report that there was one. The remaining anomaly in using just is just that the adverb is contrastive, implying that
Naive Reference for the Cinéaste
559
the reported event is the more recent, which must fail when it spans all the time under consideration. Absent explicit instruction to the contrary, one falls back on the salient and obvious perception that witnesses two episodes of division, none of which involves three cells, (48). The present perfect in (50)–(51) does not tamper with episodic perception—what counts as an episode of one or more cells dividing remains the same and is the basis for the surviving contrast in the use of just. The perfect, however, is not committed to a report of a solitary episode, as (51) contrasts with (48). In this history of cell division, there are exactly three divisions with absolute spatiotemporal locations and three cells that exist for lifetimes that do not all overlap. The temporal relations that discriminate simple past and present perfect are rather relations to scenes of divisions (i.e., episodes) contrasting an episode of divisions (simple past) and episodes of divisions (present perfect). The three cells parenting in the past have severally contributed to the present condition of there having been some divisions. In the present perfect, there is no expectation that these three divisions in two episodes should be taken as one, as in the simple past. The opposing viewpoints are again on display in (52)–(53): (52) Last year, four centenarians were the oldest person alive. (53) Last year, four centenarians have been the oldest person alive. Sentence (52) in the simple past seems to imply that the four centenarians are the same age, insofar as it is held that the four states of being the oldest person alive belong to the same single episode or scene, and a state endures throughout its episode. In contrast, the present perfect in (53) is a felicitous report of centenarians of different ages, each reigning before dying off as the oldest person alive for a season. The present perfect allows the four states of being the oldest person alive to span serial episodes or scenes. Viewpoint aspect, the cinematic conditions on measurement and sequencing of events next find close interaction in the following. In a brief Argentine-tango revue, a couple first appears on stage to dance a tango, exits, and is succeeded by two more couples who share the floor as both also dance tangos before their exit. There are thus six dance partners—the two in the first set and the four in the second set. As the production is known for its low budget and small company, it may be asked of the six dance partners in identical costume how many dancers they were. The answer is four: the dance partners of the first set reappear in the second set with different partners. Reports from the revue may say: (54) Two dance partners tangoed once and exited. (55) Four dance partners tangoed once and exited. (56) a. Six dance partners tangoed once and exited. b. The six dance partners tangoed once and exited.
560
Chapter 11
In using the simple past, the speaker may have in mind a single episode or scene corresponding to the first set (54), the second set (55), or the revue itself (56) (see (49)). The same reports may be transposed into the present perfect as a comment on the behavior of partners during the revue, with the first partnership in mind (57), the second partnerships (58), or all three (59): (57) Two dance partners have tangoed once and exited. (58) Four dance partners have tangoed once and exited. (59) a. Six dance partners have tangoed once and exited. b. The six dance partners have tangoed once and exited. Of those counted four, the dancers, two of whose dancing spans the two sets, felicitous reports differ. The simple past may be report of the first set (60) or the second (61): (60) Two dancers tangoed once and exited. (61) a. Four dancers tangoed once and exited. b. The four dancers tangoed once and exited. But as a report of the entire revue, it is simply false that the four dancers tangoed once, since two tangoed twice. Reports of the four dancers and six dance partners differ at (62) and (63): (62) a. F In that revue, four dancers tangoed once and exited. b. F In that revue, the four dancers tangoed once and exited. (63) a. In that revue, six dance partners tangoed once and exited. b. In that revue, the six dance partners tangoed once and exited. In the present perfect, a viewpoint surveying as many episodes as there are within the allowed temporal interval, the contrast persists. The six dance partners have in the course of the revue tangoed no more than once; two dancers have done so twice: (64) a. F Four dancers have tangoed once and exited, in two sets. b. F The four dancers have tangoed once and exited, in two sets. (65) a. Six dance partners have tangoed once and exited, in two sets. b. The six dance partners have tangoed once and exited, in two sets. Of the four dancers, unless the simple past warrants reference to something readily taken to be a single episode or scene, such as the second set for (61), any reference to the entire revue as a single episode (62) or as a serial two (64), straightforwardly falsifies in that two of the four dancers tangoed twice there. But is it then to be said of the six dance partners that no one of them exists for longer than one set—that dance partners die when they part just as cells die when they divide ((47)–(51))—lest (63) and (65) be falsified too, and for which reason it also happens that (66) is false?
Naive Reference for the Cinéaste
561
(66) a. F Six dance partners tangoed in the last set. b. F? The six dance partners tangoed in the last set. Is it to be that six dance partners are four dancers, but they do not quite coexist? One should hope for an account that escaped such absurdities while explaining the robust interaction between counting and the simple past and present perfect—how it is that while being a dancer among four states of being a dancer, there are nearby two tangos, for two of the dancers, but while being a dance partner among six such states, there is nearby only one tango. 11.0
A certain sequence of events
If a prenominal cardinal counts events, it must matter for what events the NP describes that they are counted one way rather than another. Given the NP’s adverbialization, it can be expected that what events are so counted and described will condition the matrix events the description of which the derived adverbial modifies. As remarked above, Moore (1994) and Doetjes and Honcoop (1997, 291ff.) observe such conditioning, which the latter call sequencing of events (with sequence of tense in mind): (67) (The) 4000 ships that passed through the lock arrived in the port of Rotterdam 3 hours later. (Adapted from Doetjes and Honcoop 1997) (68) (The) 4000 ships that passed through the lock had picked up their cargo in the port of Rotterdam. (Adapted from Doetjes and Honcoop 1997) (69) #(The) 4000 ships that passed through the lock are now in the port. (Adapted from Doetjes and Honcoop 1997) In (67) and (68), the 4000 passages counted correspond to 4000 subsequent arrivals and 4000 loadings, respectively. In contrast, in (69), the 4000 passages do not correspond to 4000 beings in port if fewer than 4000 ships have made those passages, and the sentence thus fails to have a reading that counts passages rather than ships. It is not, as a glance at (67)–(69) might suggest, a coarse opposition between episodic and stative predicates in the matrix. As Doetjes and Honcoop remark, (70) does not imply that a red mast is a transitory feature. Rather, if 4000 passages are to be counted, it suffices that the passages are related to 4000 observations of this permanent fixture. Similarly, (72)–(73) do not imply that registration took place near a lock in Panama. It suffices for the reading that counts 4000 passages that papers with proof of registration were checked with each such passage.6
562
Chapter 11
(70) (The) 4000 ships that passed through the lock had a red mast. (Adapted from Doetjes and Honcoop 1997) (71) #(The) 4000 ships that passed through the lock have a red mast. (72) (The) 4000 ships that passed through the lock last year were registered in Panama. (73) Four thousand ships passed through the lock last year. They were registered in Panama. (74) #Four thousand ships passed through the lock last year. They are registered in Panama. (75) #The 4000 ships that passed through the lock last year are registered in Panama. In contrast, in the present-tense examples, (71) and (74)–(75), 4000 passages are to be related either to 4000 present states of having a red mast or of being registered in Panama or to 4000 present observations of such states, and it is not understood how 4000 passages are so related if there are fewer than 4000 ships at present. Moore (1994) offers striking minimal pairs encapsulating the effect: (76) Two thousand ships were more than 50 meters long. (77) Two thousand ships are more than 50 meters long. (78) Four thousand ships passed through the lock last year. Two thousand were more than 50 meters long. (79) #Four thousand ships passed through the lock last year. Two thousand are more than 50 meters long. Neither (76) nor (77) in isolation counts anything other than ships. Yet, in the context of (78), (76) reports that 2000 passages corresponded to 2000 measurements en route each of greater than 50 meters. A ship that has passed through more than once may very well have been measured as often. In contrast, (79) excludes repeated present measurements of the same ships. Even if a ship is measured only as often as it passed through the lock, say twice, one fails to grasp in what respect a present measurement is contingent on, say, its first passage rather than its second and the next measurement contingent on the second rather than the first. In relating the passages counted to being red-masted, registered in Panama or 50 meters long, it is plain that sequencing of events demands neither that the one episodic event, the passage, frame another nor that the time of the episodic event frame the time of an otherwise enduring state such as being 50 meters long. What these examples demand is that the time of the 4000 passages times observations that bear witness to the enduring state. Yet it would be mistaken to move from these examples to the general conclusion that sequencing of events, so-called, demands a concord in tense between the time of the events counted and the time of either the matrix events or observa-
Naive Reference for the Cinéaste
563
tions thereof. Such tense concord is neither necessary nor sufficient, as will be illustrated throughout in the subsequent discussion of what constitutes sequencing of events, descriptively and then formally. As it is a convenient shorthand, I will often follow Doetjes and Honcoop 1997 in speaking of sequencing of events as a condition on event counting—the counting above that returns 4000—and as one that exempts object counting, the counting that returns 1000 (assuming that each ship made four passages through the lock). Of course, for me, all counting counts events, some as brief as passages through the lock and others that last a lifetime. Formally, event counting and object counting are to be identical in logical form, and so there can be no rule of grammar formulated to apply to the one and not the other. Appearances to the contrary that would distinguish them, the sequencing of events that appears to constrain only event counting, will have to emerge from other considerations. 11.0.0
Witness to a consecution
Suppose that by international convention no passport is valid for more than 15 years and on expiration, if renewed, it is altogether replaced under a new number. Suppose also that passports are inked in indelible ink, and that their entry into the Homeland Security database is also indelible. Then, indifferent to multiple entries, one may felicitously event-count passports as they pass through Passport Control, relating such past event counting to present conditions without tense concord: (80) Forty thousand passports passed through Passport Control at JFK in January 1995. They are expired. (81) (The) 40,000 passports that passed through Passport Control at JFK in January 1995 are expired. (82) Four thousand passports were briefly detained at Passport Control in 1995 and suspected counterfeit. They are certainly expired by now. (83) (The) 4000 passports briefly detained at Passport Control in 1995 and suspected counterfeit are certainly expired by now. (84) Forty thousand passports passed through Passport Control at JFK in January 1995. They are stamped with the date and time of entry. (85) (The) 40,000 passports that passed through Passport Control at JFK in January 1995 are stamped with the date and time of entry. (86) Four million passports passed through JFK last year. They are (now) in the Homeland Security database. (87) (The) four million passports that passed through JFK last year are (now) in the Homeland Security database.
564
Chapter 11
(88) Four thousand passports were briefly detained at Passport Control at JFK in 1995, suspected counterfeit, stamped with the date and time of entry, and entered into the Homeland Security database. (I was there and saw it for myself.) They are all expired now (since it has been 17 years), but they are still stamped with the point of entry and still an entry in the Homeland Security database (since these never expire). It may very well have been and may even be presumed that many of the passports were recounted in the course of their multiple entries through JFK. But given the long-ago past events at JFK reported and counted, what is known and asserted in these examples about the passports’ present condition is known strictly by inference from civil and natural law. The usage of the present tense in these examples carries no implication of a present presentation to the speaker of the passports. Despite an inclination to understand the present tense to convey exactly such, (80)–(87) in effect abbreviate the reasoning elaborated in (88), which escapes commitment to a present presentation. But, absent civil or natural law and subject to accidental fate at the hands of their bearers, much else of what has happened to the passports can be known only by tracking their histories to a presentation of their present condition: (89) a. #Forty thousand passports passed through Passport Control at JFK in January 1995. They are renewed for another 15 years or less. b. #Forty thousand passports passed through Passport Control at JFK in January 1995. They have been renewed for another 15 years or less. (90) a. #(The) 40,000 passports that passed through Passport Control at JFK in January 1995 are renewed for another 15 years or less. b. #(The) 40,000 passports that passed through Passport Control at JFK in January 1995 have been renewed for another 15 years or less. (91) #Four thousand passports were briefly detained at Passport Control in 1995 and suspected counterfeit. They are now for sale on the black market. (92) # (The) 4000 passports briefly detained at Passport Control in 1995 and suspected counterfeit are now for sale on the black market. In contrast to (80)–(88), (89)–(92) exemplify the sequencing of events that rules out counting more passports than there are passport numbers. Also with Doetjes and Honcoop’s (1997) original examples, a context that grounds the assertion in epistemic necessity suspends the sequencing of events. When one agrees with them that (69) rejects counting events rather than ships, one has on reflection tacitly granted that (69) is intended as a report of the current scene at the port: (69) #(The) 4000 ships that passed through the lock are now in the port.
Naive Reference for the Cinéaste
565
Imagine instead that the harbormaster at the lock is far away from the port. Commenting on the ships that have come under her watch, she merely infers that (93) (The) 4000 ships that passed through the lock are by now in the port. That is, enough time has elapsed since the 4000 passages for their ships to have reached the destination on their manifests. Such a context for (93) accepts that events rather than ships are counted. Examples (69) and (93) are a striking minimal pair that agree on the facts and events reported, varying only the position of the observer and the grounds for her assertion.7 In this example, the harbormaster’s calculation rests on the time elapsed and evidence gathered concurrently with her counting. The general case imposes no such restriction on the circumstantial evidence for the epistemic necessity of the assertion. The following example illustrates a couple of features of the interaction among event counting, eyewitness, and epistemic necessity: (94) (The) 300,000 passengers that crowded National Airlines flights last month will fly again within 6 months. (95) (The) 300,000 passengers that crowded National Airlines flights 6 months ago flew again. No defect of scene undermines event counting in (94) when none can be intended by a speaker with no claim to clairvoyance. The event counting does, however, impose that the epistemic grounds for an assertion about the future relate the particular events counted in the past to the future events described, as here when the speaker has in hand 300,000 customer satisfaction surveys completed in flight in which the passenger promises to fly again in 6 months. As in the previous example, the evidence for the assertion is collected at the time of the counting, the example varying only in being immune from confusion with eyewitness reports. Notice that it may turn out without compromising what is asserted that frequent fliers among those counted fly again just once, within 6 months of their first flight counted. Such circumstances are consistent both with the asserted content and with its epistemic grounds, namely, for each of the 300,000 passengers, in every world consistent with what is known, the passenger so counted will indeed fly again. In (95), a violation of sequencing of events again threatens unless epistemic necessity steps in. Imagine8 that the warrant for (95)’s assertion is 300,000 promotional ticket vouchers distributed on board 6 months ago, all of which have been used and returned to the airlines. Here the circumstantial evidence to relate the distant event counted to the more recent one reported is itself acquired more recently than the circumstances under which passengers were first counted. Circumstantial evidence recently acquired may also reclaim (69). Suppose suspicious behavior during each passage has launched a customs agent to track the ship’s subsequent movements. The agents, unaware of
566
Chapter 11
each other, report to me. Receiving 4000 reports just now of the ships’ current whereabouts, I may say (69) without implying that the 4000 reports are in fact of 4000 different ships. Remark that each of these 4000 reports is proof de eventu about an event counted that it relates to a present state of being in port, the same present state for many a four of the counted events. Epistemic necessity in providing grounds for assertion other than eyewitness exempts the speaker from any conditions on the scenes of eyewitness reports. If, for whatever reason, the speaker still intends to refer to a scene, the exemption is withdrawn. As before, (96) smoothly counts events rather than persons—3000 passages resulted, as observed or recorded then, in 3000 opinions: (96) Three thousand passengers had (3000) opinions about their flights last year. (97) (And so,) 3000 passengers still have (3000) opinions about their flights 1 year later. Opinions, as much as travelogues, survive into the present and support an inference from past observation (96) to present conditions that are not directly observed (97). Imagine however that I, CEO of National Airlines, now address a focus group of all of last year’s passengers, whom I trust are frequent fliers and are now gathered together at obviously fewer than 3000, being 1000 persons: (98) Three thousand passengers have (3000) opinions about their flights last year. In reporting to you your activity in the past year, I could very well say (96), but it would be bizarre to utter (98) looking you all in the eye. The problem is that the presentation to which I make indexical reference, the scene before me just now, does not itself resolve into 3000 presentations. The 3000 separate opinions that past experience has generated now converge in presentations of their havers, and nothing presented to me now can be thought to be of one opinion and not of another held by the same person. Removed, however, from these passengers and once again alone in my office, I may, thumbing through the 3000 customer surveys collected in flight last year, say (98) to myself or to anyone else in earshot, reasoning from the circumstantial evidence available to me (see (95)), which eschews a scene of the passengers. Absent clear and present epistemic necessity, eyewitness testimony is presumed. It is however left largely unexplained why one quick to reject (69) and (75) out of the blue is so quick in failing to conjure their more felicitous contexts. Perhaps the simple present, which appears so often in illustrations of sequence-of-events violations, favors evaluation with respect to the present scene. If the present tense is normally taken to be indexical to the context of utterance, then that may well include assumptions about the perspectives and frames of reference available to speaker and hearer. Should customs agents plant a camcorder on every vessel
Naive Reference for the Cinéaste
567
passing through the lock, one could very well utter (69) gesturing to 4000 monitors, each of which shows a view that looks out onto the port. The 4000 images, although live in real time, are not projections from a single point of view under a single frame of reference. Presumably, without it being made explicit, one assumes that the context of utterance for (69) comes without such prosthetic enhancements to the speaker’s and hearer’s point of view. Assertions out of the blue come with rich assumptions about a neutral observer, neutral point of view, and what are the most likely grounds for the assertion. Under the presumption that eyewitness testimony is being reported, it is then as if the camera tracking passports for (91) and (92) or ships at sea for (69) is a mobile analog to the stationary camera for (99) on board the spy plane above Central Moscow (see (43)): (99)
#Three hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. (100) One hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. The frame of reference necessary to track the action in (89)–(92) and (69), like the one to comprehend the scene in (99), is too wide in which to count 4000 discrete events of being a passport (cf. (38)) or of being a ship. In (80)–(88), it was seen that tense concord is not a necessary condition for event counting if civil or natural law bridges past events to present condition. Nor is it a sufficient condition if again eyewitness tracking undermines the event count: (101) #(The) 40,000 passports that passed through Passport Control at JFK in January 1995 were renewed for another 15 years or less. (102) # (The) 4000 passports briefly detained at Passport Control in 1995 and suspected counterfeit were for sale on the black market. (103) #(The) 4000 ships that passed through the lock were in the port (6 months later). With the passport renewals and appearance on the black market far removed from events at Passport Control and with passage through the lock 6 months from port, it hardly matters for the frames of reference under which the passports or ships are counted and tracked that these distant, later events happen also to be in the past. Whatever fails event counting in (89), (90), and (69) fails just the same in (101), (102), and (103). What goes wrong, it seems, is that anything counted n under an initial frame of reference should not be tracked to a frame of reference where it would be counted n-k for k > 0. In effect, what is counted n and then tracked should never be seen to converge on fewer than n.
568
Chapter 11
This reduction of the sequence-of-events effect to what amounts to a continuity condition on well-formed cinematic scenes offers two ways out from the effect. The first, just seen, is when epistemic necessity preempts eyewitness. With natural or civil law in hand, snapshot events at Passport Control while being 4000 passports suffice, as they may be related by inference to their remote events, and similarly for snapshots at the lock while being 4000 ships. What is counted 4000 at Passport Control or at the lock remains 4000 for the snapshot’s brief duration. Absent natural or civil law, eyewitness report launches a cinematic scene while being 4000 at Passport Control or at the locks, and it tracks passport or ship to a climax on the black market or in port. The event counting fails in that what is counted 4000 could not be so counted 4000 later in the scene. Besides epistemic necessity, the second way out from the sequence-of-events effect works even with long eyewitness tracking if it does not track what has been counted: (104) Four thousand butterflies visiting Kinkaju-ji in the spring triggered a fall tornado in the Golden Gate. The Lorenz Butterfly Effect may advertise that butterfly turbulence in Kyoto could cause tornadoes on the other side of the Pacific, but there is no inference from natural law to the particulars reported in (104), which must rather be grounded in eyewitness weather tracking as small eddies induced by 4000 butterfly visits snowball to tornado strength in San Francisco. The count in (104) is event counting by butterfly visit, indifferent to the habit of butterflies to revisit food sources, and indeed it is the turbulence of 4000 such visits of butterflies aflutter—not 1000—that have caused the tornado. Yet the butterflies so counted leave the scene, and tracking their turbulence to verify their effect is not tracking them. Their appearance in the scene is no longer than that of the participants in the snapshot events that earlier suffice for an inference to remote events from natural or civil law. What is counted n is not seen to converge on fewer than n despite the transpacific scene here. What else other than such a constraint on scenes could distinguish the accepted event counting in (104) from its rejection in (101)–(103)? Note that (104) itself embodies a minimal pair if contexts are contrasted. Event counting fails if it is imagined that the 1000 butterflies counted 4000 in their spring visits to Kinkaju-ji fly themselves or find transport to the Golden Gate where their fluttering on site triggers a fall tornado. Parallel to the citation examples of sequence of events, it carries across sentences, too: (105) Four thousand butterflies visited Kinkaju-ji in the spring. They triggered a tornado in the Golden Gate in the fall. (106) One thousand butterflies visited Kinkaju-ji in the spring. They triggered a tornado in the Golden Gate in the fall.
Naive Reference for the Cinéaste
569
Event counting in (105) is acceptable only if the speaker intends report that 4000 springtime actions at Kinkaju-ji caused the fall Golden Gate tornado, whereas (106), object counting, is fair report either of like action or of later action at the Golden Gate that causes the tornado. As much as this characterization of sequence of events so-called is a more accurate survey of its scope, little yet amounts to an analysis. There ought be no rule that stipulates dual processes, event counting and object counting, such that the former is then subjected to special constraint. There is also little hope for a principle to directly relate counting (occasioned by the cardinal predicates within NP) to conditions on how the scenes animating the rest of the sentence unfold. It will rather be that a general condition on cineramic narration happens to be breached when eyewitness tracking fails to preserve the count. Of the scenes constituting such an eyewitness documentary, it holds that distinct scenes oriented to the same frame of reference (e.g., for the Pacific Ocean) are of distinct subject matter, where the distinction in subject matter can be represented in a language free of egocentric perspectival-dependent vocabulary such as now-en-scène, referring only to coordinates for the allocentric frame of reference, the Pacific Ocean. In short, what looks different was different and not a Rashomon-like compilation of multiple views of the same goings-on. A sentence is rejected in apparent violation of sequence of events just in case no eyewitness scenes conform to this condition on cineramic narration and there is no other warrant for the assertion, in particular, none in an inference from epistemic necessity. Relying on witness, if there happens to have been an event count—that is, a count of events of being so-and-so under a frame of reference narrow enough to frame as discrete episodes from the same life—there is inherent risk that the unfolding scene will engage a larger frame of reference within which events of being so-and-so, first counted n, are seen to converge on n-k for k > 0. In contrast, object counting, merely counting at the onset from a frame of reference large enough that a discrete event of being so-and-so coincides with a life as such, faces no such risk that an even larger frame of reference later in the scene will count the events different.9 Object counting never runs afoul of this general condition on scenes and so it appears as if sequence of events is special to event counting. If this is the program for an analysis, a couple of steps remain before it is realized. The first is to deliver the condition on cineramic narration that implies that its scenes are count-preserving, and the second is to show how reference to and quantification over such scenes and frames of reference enter (or not) into the logical form of such minimal pairs as (69) and (93) (repeated here): (69) #(The) 4000 ships that passed through the lock are now in the port. (93) (The) 4000 ships that passed through the lock are by now in the port.
570
Chapter 11
The logical form for (69) represents a thought with scenes in witness of what is reported. Sentence (93) is an inference without the benefit or burden of such scenes. These steps are deferred to section 11.0.2, while section 11.0.1 gives a fuller picture both of the conditions under which the scenes bearing witness fail to preserve the event count and of the conditions that warrant assertion by inference without witness and are therefore a way out from the sequence-of-events effect. 11.0.1
Local cardinality preservation
The linguistic behavior that dissociates event counting and object counting so-called is broader than what can be characterized as a condition on scenes, with which it should not be confounded. All of it, as listed below, rests however on the same semantic bedrock: that different counts of events of NP-ing are measurements under different conditions, which imposes via adverbialization that the events that the NPs could have participated in while counted under the one condition are different from those that could have been participated in while counted under the other condition: (107) [∃X : ∃E (count[E,X, 3,000,000] Soviet troops[E,X] …)] [℩E : count[E,X, 3,000,000] Soviet troops[E,X] …)] [∃E0 : N[E,E0]] Φ[E0] … (108) Three hundred thousand Soviet troops flowed through Red Square in the May Day parade. (109) One hundred thousand Soviet troops flowed through Red Square in the May Day parade. (110) #Three hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. (111) One hundred thousand Soviet troops streamed around Central Moscow in the May Day parade. To have counted 300,000 discrete events of being a Soviet troop, many a three of which share their participant, is to have done so from a narrow frame of reference in Red Square, large enough to frame a troop goose-stepping past the reviewing stand but not so large as to frame his movement around Central Moscow or to reveal that events of being a Soviet troop counted discrete are continuous within the lifetime of the same troop. If, in (107), counting events E of being a Soviet troop is to count participation in the May Day parade, it falls to the Neighborhood relation ‘N[E,E0]’ to enforce attendance in as many E0 as there are E. It is especially vivid for (112), the point of which is that 4000 butterfly visits and no fewer triggered the tornado, that the Neighborhood relation entail that every counted event E of being a butterfly frames its own butterfly action at Kinkaju-ji:
Naive Reference for the Cinéaste
571
(112) Four thousand butterflies visiting Kinkaju-ji in the spring triggered a fall tornado in the Golden Gate. (113) One thousand butterflies visiting Kinkaju-ji in the spring triggered a fall tornado in the Golden Gate. The circumstances imagined, where 4000 butterfly visits intersect just 1000 butterfly lifetimes, also make (113) true. In this case, it is not that 1000 visits trigger the tornado, the force from which would have been too weak for one. It is rather the same 4000 visits of theirs, the 1000. So it must not be that ‘N[E,E0]’ entails that E and E0 are equinumerous but rather that the E induce via the Neighborhood relation an equinumerous partition of the E0. Each of the thousand butterfly lifetimes does contribute its own four actions at Kinkaju-ji: (114) N[E0,E1] →.∀e0∃e1(E0e0 → (E1e1 & N(e0,e1))) & ∀e1∃e0(E1e1 → (E0e0 & ∀e(N(e,e1) ↔ e = e0))) The events counted and the events in their neighborhoods that are at least as numerous include, as Moore (1994) observes, neither the flow through Red Square, the stream around Central Moscow, nor the tornado at the Golden Gate. Thus, the events with troops and the events with butterflies that are counted stop short at the proximate events, the W-ing that supermonadicity introduces: (115) [∃X : ∃E (count[E,X, 3,000,000] Soviet troops[E,X] …)] [℩E : count[E,X, 3,000,000] Soviet troops[E,X] …) [∃E0 : N[E,E0]] W[E0, X] ∃E1 Φ[E1, E0] (116) [∃X : ∃E (count[E,X, 4000] butterflies[E,X] …)] [℩E : count[E,X, 4000] butterflies[E,X] …) [∃E0 : N[E,E0]] W[E0, X] ∃E1 Φ[E1, E0] This is just another instance of the conclusion from earlier chapters that universal adverbialization must confine the derived adverbial while NP-ing to a local thematic relation. After all, counted 4000 or 1000 (see (104)–(106)), the butterflies are long since dead and dust by the time their actions at Kinkaju-ji stir up a tornado at the Golden Gate, none of which occurs while they be butterfly-ing. Where the semantic bedrock gets you is just from counting n NPs to multiple n W-ings. These may, as just remarked, be remote from the subsequent events the sentence reports, but they are the proximate effect that counting NPs has on the sentence’s meaning and gatekeeper to when it is that the subsequent events are too remote, the sequencing-ofevents effect. Any elicitation of an effect from counting that lingers into the rest of the sentence must emerge from something yet unseen in the semantics or pragmatics for Φ in (115)–(116). Future tense, for example, as noted above (see (94)), comes with
572
Chapter 11
epistemic requirements for its assertion that are dependent on the W-ings that event counting has framed: (117) (The) 300,000 passengers that crowded National Airlines flights last month will fly again within 6 months. Further effects of counting derive from Φ’s constituent relations such as ‘Cause[Ei, Ej]’: (118) [∃X : ∃E (count[E,X, 4000] butterflies[E,X] …)] [℩E : count[E,X, 4000] butterflies[E,X] …)] [∃E0 : N[E,E0]]W[E0, X] O[E0, E1] ∃XAgent[E1, X] Cause[E1,E2] … trigger[E2]… The meaning of ‘Cause[Ei, Ej]’ discriminates the successful event counting in (119) from its failure in (121), where what is counted are visits to the park bench or screen door without regard that some visitors make repeat visits: (119) Four thousand pigeons coated the park bench in 6 hours. Four thousand pigeons covered the park bench in 6 hours. Four thousand pigeons splattered the park bench in 6 hours. Four thousand pigeons splotched the park bench in 6 hours. Four thousand pigeons speckled the park bench in 6 hours. Four thousand pigeons slickened the park bench in 6 hours. Four thousand pigeons slimed the park bench in 6 hours. Four thousand pigeons smeared the park bench in 6 hours. (120) One thousand pigeons coated the park bench in 6 hours. (121) #Forty thousand bugs coated the screen door in 6 hours. #Forty thousand bugs covered the screen door in 6 hours. #Forty thousand bugs splattered the screen door in 6 hours. #Forty thousand bugs splotched the screen door in 6 hours. #Forty thousand bugs speckled the screen door in 6 hours. #Forty thousand bugs slickened the screen door in 6 hours. #Forty thousand bugs slimed the screen door in 6 hours. #Forty thousand bugs smeared the screen door in 6 hours. (122) One thousand bugs coated the screen door in 6 hours. Sentences (121) cannot report the carnage from 1000 bugs each of which landed too hard on its 40th visit to the screen door—even if it is an actuarial certainty and natural law that 40,000 visits are 1000 deaths by screen door. It may be quickly observed that there is little sense in which visits prior to the fatal one contribute to the splatter in (121), unlike the incremental contributions of pigeons on a park bench in (119). Four thousand visits and no fewer caused the park bench to be
Naive Reference for the Cinéaste
573
coated, but only 1000 visits caused the screen door’s coating.10 If one troubles to single out causes from 40,000 neighborhoods, counting to 40,000 within the denominal adverbial, the report had better be of their effect. Albeit that the speaker’s assumed perspective or frame of reference fixes how she will count these creatures, counting them the one way or the other does not leave her free to report whatever they might have done, the count being merely a matter of her perspective on the same agents. Counting the one way or another has truth-conditional bite, sentences (112) and (119) are truths about 4000 butterflies and 4000 pigeons, respectively, and (121) is a falsehood about 40,000 bugs. Mentioning a fourth as many creatures, (113), (120), and (122) are instead all true. With cardinal predicates denoting events of counting and the adverbialization of NPs that contain them, the truth-conditional semantics reduces to the banality that different causes have different effects. On the other hand, if either adverbialization or this view of cardinal predicates is denied, it entails that the bugs that alighted on the screen door 40,000 times were not those who died a thousand deaths there, if there is to be a difference in truth between (121) and (122). Apart from the cardinality preservation inherent in counting off n W-ings then said to be the n causes of what followed, ‘Cause[E, E′]’ preempts, as already remarked in the discussion of (104), any further condition on sequencing of events. No further participation of the NPs counted is needed for their W-ings, their actions, to have the remote effects they eventually do. No such sentence need ever be narration for a scene of their continued presence. Other bridge relations similarly preserve cardinality and yet preempt the previous section’s condition on scenes, relating W-ings to distant events without the further participation of the W-ers. Recall from section 10.0.8 that ‘About[E1,E2]’ in (124) (see also ‘[E, E′] ’, sections 10.3 and 11.1) allows for the literal truth of present tense (123) without implying a present Kafka, recognizing that events E1 (and E0), his writing, entirely in the distant past in which Kafka participated, made an imprint that is the subject of present fame from which he himself is absent: (123) Kafka is famous. (124) [℩x : ∃EKafka[E,x]][℩E : Kafka[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E1] ∃Y Theme[E1,Y] About[E1,E2] Present[E2] Be[E2] ∃E0∃Y W[E0,Y] O[E0,E3] ∃Y Theme[E3,Y] O[E2,E3] famous[E3] As above, (125) allows that the same 1000 ships recycle, now with PBS film crew on board, through 4000 passages and travelogues: (125) (The) 4000 ships that passed through the lock last year were the subjects of 4000 tedious 1-hour PBS travelogues. (126) (The) 4000 ships that passed through the lock last year were the subjects of a tedious 4000-hour-long PBS travelogue.
574
Chapter 11
The same circumstances warrant present-tense assertions (assuming the travelogues survive): (127) (The) 4000 ships that passed through the lock last year are the subjects of 4000 tedious 1-hour PBS travelogues. (128) (The) 4000 ships that passed through the lock last year are the subjects of a tedious 4000-hour-long PBS travelogue. All that the semantics requires is that 4000 (or more) W-ings, passages through the lock, are the subjects of the travelogues, which they are, having been filmed.11 According to further examples of the same point, (129) and (130), the 400,000 sorties that the Flying Tigers flew to resupply China during World War II are still famous: (129) The 400,000 Flying Tigers that passed over the Hump are famous, and their lost cargoes are still guarded/worshipped by Himalayan cargo cults. (130) #The 400,000 Flying Tigers that passed over the Hump are famous and (they are) on display at the Smithsonian’s National Air and Space Museum.12 In addition to their fame, the cargo lost on these sorties establishes a causal connection between the past events and equinumerous present effects, as reported in (129). The aircraft themselves have also survived, but as there were never 400,000, the second clause of (130), without benefit of either ‘Cause’ or ‘About’ in its analysis, ends up violating sequence of events, no different in this respect from (69). Referring to the same things, event counting or object counting discriminates among the events they participate in, and these in turn amount to different causes and different subjects. With ‘Cause’ and ‘About’, what underlies the dissociation of event counting and object counting is entirely semantic, which cannot be said for the conditions on scenes of the preceding section. Event counting in (131) frames events of W-ing at Passport Control far removed from future expirations (as no expired passport would get through): (131) (The) 40,000 passports that passed through Passport Control at JFK in January 1995 are expired. (132) (The) 1000 passports that passed through Passport Control at JFK in January 1995 are expired. (133) [∃X : ∃E (count[E,X, 40,000] passports[E,X] …)] [℩E : count[E,X, 40,000] passports[E,X] …] [∃E0 : N[E,E0]]W[E0,x] O[E0,E1]∃XTheme[E1,X] [E1, E2 ] … expired[E2] … (134) [E0, E1 ] →. ∀X(participate[E0,X] ↔ participate[E1,X])
Naive Reference for the Cinéaste
575
The time-travel relation ‘[E0, E1 ]’ encountered in the analysis of identity statements (section 10.3) relates 1995 to the present. Whether event counting or object counting, there are only as many events or states in the present constitutive of expirations as there are passport numbers. The 40,000 passages at Passport Control converge on 1000 expirations. Given that (131) and (132) are both true, nothing in the meaning of discerns event counting from object counting, and we must look elsewhere for the sequencing-of-events effect acknowledged in (135): (135) a. #(The) 40,000 passports that passed through Passport Control at JFK in January 1995 are renewed for another 15 years or less. b. #(The) 40,000 passports that passed through Passport Control at JFK in January 1995 have been renewed for another 15 years or less. Also any conditions mentioning scenes and written into the meaning of would have to allow that (131) is true and felicitous absent such scenes, given its epistemic necessity under international law. The minimal pair embodied in (136) succinctly illustrates the difficulty in deriving the sequence-of-events effect from the decomposition of the shared clause structure: (136) (The) 4000 ships that passed through the lock are now in the port. Recall ((69), (93)) that (136) is true and felicitous event counting for the harbormaster at the lock who calculates from the ships’ manifests their estimated time of arrival, and the event counting is infelicitous for an assertion that relies for its grounds on eyewitness at the port. Setting aside the effects of counting just canvassed, turn finally to those that emerge specifically from the condition on scenes advertised earlier (section 11.0.0). Sentences are taken to narrate scenes when nothing else is at hand to warrant their assertion. Of such a scene, it was that any things counted n under an initial frame of reference should not be tracked to a frame of reference where they would be counted n−k for k > 0. What is counted n and then tracked should never be seen to converge on fewer than n. To delineate the illicit convergence en-scène, set aside are all those sentences in which epistemic necessity preempts eyewitness confirmation (e.g., (131)) and those sentences in which eyewitness does not track what is counted (e.g., (112), (128)). The sentences now of interest are all like (137), eyewitness report from the field (absent access to Soviet military planning) of an event, the flow through Red Square, which to participate in is to be present for, as in this case where Soviet troops in fact constitute the flow:13 (137) Three hundred thousand Soviet troops flowed through Red Square in the May Day parade. (138) One hundred thousand Soviet troops flowed through Red Square in the May Day parade.
576
Chapter 11
The exercise is to delineate the felicitous event counting here, where the eyewitness relies only on frames of reference and scenes that hold the count of Soviet troops constant, from those in which there is an illicit convergence on fewer than first counted. Felicitously event counting, (139) reports 300 passages ending in 300 near misses and allows that some poor souls may have had multiple near-death experiences. Each incident is in turn charted in an assessment of aviation safety, revealing the cluster reported in (140): (139) Three hundred passengers on DC-9s last year nearly died from hypoxia in the cabin. (140) Three hundred passengers on DC-9s last year clustered on the aviation morbidity chart. Despite the obvious causal connection, event counting is disrupted in (141) and (142): (141) #Three hundred passengers on DC-9s last year have died from accumulated/chronic radiation exposure at altitude. #Three hundred passengers on DC-9s 3 years ago died from accumulated/ chronic radiation exposure within a year of their last flight. (142) #Three hundred passengers on DC-9s last year clustered on the aviation mortality chart. Examples (141) and (142) cannot count passages that include frequent fliers, reporting 300 passages that end in, say, 100 deaths, a reading that the pragmatics of accumulated or chronic exposure should otherwise favor. As a matter of fact, any of the 100 deaths would have been averted if its victim had flown only twice rather than thrice. The 300 passages are necessary causes for their effects in 100 deaths, and thus the infelicity of (141) and (142) is a puzzle opposite (143), which, with similar causal structure, felicitously reports that 4000 pigeon visits had the collective effect of a park bench covered over (see above discussion of (119) vs. (121)): (143) Four thousand pigeons covered the park bench in 6 hours. Worse, note that all that tells apart (140) and (142) is that the former is about causing morbidity, reporting a cluster of morbidity incidents, and the latter, about mortality, a cluster of mortality incidents (i.e., deaths). So is semantics, too, a slender thread twixt morbidity and mortality? In the latter, three passengers converge— illicit for the event count—on a single death to constitute the cluster, but in the former, the cluster is constituted from the isolated experiences of passengers. A scene of a collective event in which its participants must be present is a scene of them all under a frame of reference that fixes their number. So much holds of both (140) and (142) (and of the collected events of (139) and (141)). It is just that a scene
Naive Reference for the Cinéaste
577
of a cluster of deaths numbers the passengers differently (and as being fewer) than the passages counted, but a scene of a cluster of near misses numbers them the same. Various scattered elements of a sentence in adjusting what events are described adjust what the scene must be a scene of and may thus hasten or avert an illicit convergence en-scène. The addition of out in (145), for example, rules out event counting (see (141)). In focus groups for Belle Jolie lipstick (Mad Men, “Babylon,” Episode 6 of Season 1, 2007), ladies product-test assorted shades: (144) Thirty lipsticks were smeared onto 300 lips. (145) #Thirty lipsticks were smeared out onto 300 lips. Advertising budgets and hygiene what they were in 1960, it may have been according to the felicitous event counting in (144) (see (139)) that the same lipsticks were recycled through several trials. Not all smearings are smearings-out, and it appears that eyewitness to smearings-out is witness in which lipsticks that are discrete within trials will illicitly converge. Temporal compression or dilation as a result of viewpoint aspect (see (47)–(66)) is another element that may force or advert an illicit convergence en-scène, as exemplified in the following: (146) #(The) 3000 passengers who crowded National Airlines last week are home. (147) (The) 3000 passengers who crowded National Airlines last week have been home (within 60 minutes of arrival). (148) #(The) 3000 passengers who crowded National Airlines last week overwhelm the Lost Baggage Department. (149) (The) 3000 passengers who crowded National Airlines last week have overwhelmed the Lost Baggage Department. Sentence (146) resumes the sequence-of-events effect of (136). Absent epistemic necessity, eyewitness is presumed to track 3000 passengers to a present presentation of them at home in which they are 1000, fewer in number in violation of the condition on scenes. Similarly, (148) fails in evocation of a present presentation of passengers overwhelming the Lost Baggage Department, converging there as 1000 after their 3000 counted episodes as passengers. The present perfect in (147) and (149) in describing events or states of extended duration, having been at home and having overwhelmed, affords constituent events a temporal separation that spares their simultaneous appearance in a present scene even when direct eyewitness is presumed. In the existential present perfect (so-called; see among others, McCoard 1978, McCawley 1971, 1981b, Iatridou et al. 2001, and Pancheva 2003), it may be imagined that the 3000 passengers are each tracked home within 60 minutes, where they have been at least until they depart again for another flight on National Airlines. Their current location is in fact unknown. For the contrast between (148) and
578
Chapter 11
(149), the background conditions can be taken to be the same—union rules strictly limit the work hours of the Lost Baggage Department, which have been exhausted for the week over the course of the week, as (149) successfully reports, or as (148) fails at, exhausted by one large run on it. The temporal dilation of the present perfect effectively dissociates the sentence from the scene out in front of the speaker, which of course has no temporal depth beyond the now. That dissociation in (150) rescues the speaker from the sequence-of-events violation encountered earlier with the simple present tense in (151): (150) (The) 4000 ships that passed through the lock have been in the port now for awhile if they don’t/didn’t have to be elsewhere. (151) #(The) 4000 ships that passed through the lock are now in the port if they don’t/didn’t have to be elsewhere. As before, (151) looks for 4000 in a current view of the port, and (150) makes no appeal to a perspective on the port when it is uttered. Even if uttered (150) while facing the port, the present perfect allows that the present facts asserted do not compose the scene presented to the speaker at the moment. The present perfect (150), unlike the simple present (151), implies no further arrangement of the 4000 passages’ present effects—neither among themselves nor in relation to the perspective present to the speaker at the moment of utterance. Besides viewpoint aspect, this contrast between temporal compression and dilation is also manifest in lexical semantics. For the examples below, one can be sure that the sorties in the Berlin Airlift outnumbered the aircraft deployed. The eventcounted landings may be related via ‘About’ to a newsreel present at the National Archives (see (128)), while the participants in these 3000 landings cannot be felicitously said to constitute a present squadron of fewer aircraft again violating sequence of events: (152) (The) 3000 aircraft that landed at Tempelhof in 1948 are a newsreel at the National Archives. (153) # (The) 3000 aircraft that landed at Tempelhof in 1948 are a squadron mothballed in the Mojave Desert. The aspectual contrast latent in lexical choice can be illustrated in minimal pairs that play on the ambiguity of the predicate nominals in (154) and (155): (154) (The) 3000 aircraft that landed at Tempelhof in 1948 were an airlift operation. (The) 3000 aircraft that landed at Tempelhof in 1948 were a cold war delivery service.
Naive Reference for the Cinéaste
579
(155) (The) 3000 aircraft that landed at Tempelhof in 1948 are an (ongoing) airlift operation. (The) 3000 aircraft that landed at Tempelhof in 1948 are an (ongoing) cold war delivery service. Describing a process, which implies, extended in time, that parts of it precede and others may follow, (154) and (155) felicitously event-count and say that 3000 landings were parts of a process that in the case of (155) reaches back to 1948 and continues into the present. On the other hand, if operation or delivery service is meant to refer to a company, a standing squadron of aircraft, event counting fails just like in (153) in that the 3000 (serial) landings do not correspond to any 3000 subevents that are contemporary parts of being such a company, whether it is located in the past (154) or in the present (155). The sequencing of events to which event counting, so-called, is subject reflects a cinematic condition such that no sentence is narration for a scene that counts n NP under one frame of reference and then converges into a scene that counts n-k NP for k > 0 under another frame of reference. The latter part of the section has sketched out what the illicit convergence looks like that grammar is somehow to rule out (section 11.0.2). (Earlier in the section, this cinematic condition was distinguished from other aspects of meaning that also tend toward cardinality preservation—relating n NPs to n (or more) events.) Some relation between presentation or perspective and counting, the conditions for measurement, is unsurprising. Of course the field officer on the ground in Red Square counts Soviet troops differently from the spy plane’s count, and it is unsurprising that what is counted is what participates in the action observed. But the latent puzzle is that the semantic bedrock that relates, via adverbialization, counting NPs, to the events the counted participate in is strictly local: counting to n frames n (or more) W-ings (see (114)–(116)). The number of NPs counted fixes the numerosity of W-ings and nothing more. The cinematic condition has, however, a longer-lived effect on the scene that the remainder of the sentence narrates. However it enters logical form, recall that it is an unlikely component of the basic, supermonadic decomposition of clauses, since almost any sentence may find itself in a context of epistemic necessity that preempts reference to scenes subject to the condition. Rather it will be that the addresses of AdrPs inside DPs are to scenes under frames of reference that meet the cinematic condition. But scenes from eyewitness report are concurrently scenes from the eyewitness’ visual navigation and spatial orientation. To be coherent as such, a scene must not include distinct scenes of the same spatiotemporal region, at least if a visual survey is to constitute an atlas for the space to be navigated. Note that although at first blush sequencing of events presents itself as a condition on when a DP eventcounts, the cinematic condition it reduces to makes no mention of the NPs, whether event-counted or object-counted, and is to be stated strictly in terms of an illicit
580
Chapter 11
convergence of scenes on the same spatiotemporal region. Illicit convergence enscène is grounds for the peremptory rejection of a sentence, as sampled above. But the acceptability of event counting also shows, as in the next section, a gradience that inversely matches the speaker’s implied degree of disorientation, an account of which requires the formal apparatus presented earlier in section 9.4.1. 11.0.2
Now Playing at Cinerama Wherever
Nothing in logical form distinguishes event counting and object counting, and as there is no equivocation in understanding (159), the (in)definite descriptions in (156)–(159) all refer the same, to the same Soviet troops on parade, [℩X: ∀x(Xx ↔ ∃e Soviet troop(e,x))]: (156) Three hundred thousand Soviet troops flowed through Red Square in the May Day parade. (157) One hundred thousand Soviet troops circled Central Moscow in the May Day parade. (158) #Three hundred thousand Soviet troops circled Central Moscow in the May Day parade. (159) The 300,000 Soviet troops flowing through Red Square in the May Day Parade that the field officer counted were the 100,000 Soviet troops circling Central Moscow that the spy satellite registered. What is counted throughout are events e of being a Soviet troop, [℩E: ∀e(Ee ↔ ∃x Soviet troop(e,x))], which are mistaken for objects when they coincide with a lifetime. As events of being a Soviet troop, a passenger, or a person are dense and uncountable in their plenitude, the problem of counting them is the same as counting discrete spaces, times, motions, events, and so on. Such efforts render salient that counting is contingent on the conditions of measurement and observation. Counting from the spy satellite, there are exactly 100,000 and not 300,000 discrete events of Soviet troopery. On the ground at Red Square, the field officer equipped with an electronic counter, counts exactly 300,000 discrete events of Soviet troopery. These measurements obviously count different events, since what is 300,000 is not what is 100,000, and yet their participants referred to are all the same. Counting one way or another is however not merely a different route to reference to the same things. Counting the same butterflies as 3000 implies that the energy of 3000 visits rather than 1000 visits triggered the tornado in the Golden Gate: (160) Three thousand butterflies visiting Kinkaju-ji in the spring triggered a fall tornado in the Golden Gate. What is counted frames, while so counted, the actions the butterflies participate in; these events are the causes of the tornado’s being triggered. The observed contrast
Naive Reference for the Cinéaste
581
between event and object counting has semantic bite real enough, as in this example, to have garden-pathed previous analyses to suppose a difference in the logical form of the (in)definite descriptions they deploy. Rather the adverbials derived by adverbialization, while counted 3000 … and while counted 1000 … denote different events with different neighborhoods in which the butterflies’ actions causing the tornado are fixed. Event counting 3000 butterfly-ins does imply their butterflies’ participation in 3000 events, in agreement with previous analyses. But as Moore (1994) points out, these are not 3000 of the events the verb denotes. There is only one triggering (‘trigger(e)’) that verifies (160). Similarly, (156) is report of a flow, a single event that its verb ‘flow(e)’ denotes despite the 300,000 troop movements said to constitute it. The semantic bedrock is that different counts of events of NP-ing are measurements under different conditions, which imposes via adverbialization that the events that the NPs could have participated in while counted under the one condition are different from those that could have been participated in while counted under the other condition: (161) [∃X : ∃E (count[E,X, 3,000,000] Soviet troops[E,X] …)] [℩E : count[E,X, 3,000,000] Soviet troops[E,X] …)] [∃E0 : N[E,E0]] W[E0, X] ∃E1 Φ[E1, E0] (162) [∃X : ∃E (count[E,X, 3000] butterflies[E,X] …)] [℩E : count[E,X, 3000] butterflies[E,X] …)] [∃E0 : N[E,E0]] W[E0, X] ∃E1 Φ[E1, E0] The troopery and the butterfly-ins that are counted stop short at the proximate events, the W-ing that supermonadicity introduces. The semantic bedrock gets from counting n NPs only as far as counting as many W-ings. With the semantics of counting so confined, it is a puzzle that the choice of one cardinal predicate or another within a DP should have an effect that lingers beyond the scope that the above syntax and semantics defines for it. Yet that is what is observed in the sequence of events so-called that contrasts (165) and (169): (163) (The) 30,000 passports that passed through Passport Control at JFK in January 1995 were expired. (164) (The) 1000 frequent-traveler passports that passed through Passport Control at JFK in January 1995 were unexpired. (165) (The) 30,000 passports that passed through Passport Control at JFK in January 1995 had been renewed recently. (166) (The) 1000 frequent-traveler passports that passed through Passport Control at JFK in January 1995 had been renewed recently.
582
Chapter 11
(167) (The) 30,000 passports that passed through Passport Control at JFK in January 1995 are expired by now (in accordance with international law). (168) (The) 1000 frequent-traveler passports that passed through Passport Control at JFK in January 1995 are expired by now (in accordance with international law). (169) #(The) 30,000 passports that passed through Passport Control at JFK in January 1995 have been renewed this year by their bearers. (170) (The) 1000 frequent-traveler passports that passed through Passport Control at JFK in January 1995 have been renewed this year by their bearers. Whatever the conditions for the measurement that counts to 30,000, these are met at JFK Passport Control ((163), (165), (167)), and the events, the W-ings, that the passports participate in while so counted can very well end in both expirations and renewals ((163), (165), (167)). So nothing in the meaning of the verbs expire and renew clashes with 30,000. Nor must the meaning of the present tense in (169) (despite calling it sequence of events) preclude 30,000, lest it preclude it in (167) as well. Whatever this elusive condition derives from, a context that grounds the assertion in epistemic necessity, such as that imposed by international law, suspends the sequencing of events, as in (167). If indeed the law on passports provided for their automatic renewal, (169) would earn its exemption, too: (171) (The) 30,000 passports that passed through Passport Control at JFK in January 1995 have all been automatically renewed by now several times over. So, again, all that is common to the syntax and semantics of (169) and (171) must not raise a hand against event counting, and it is identical modulo the cardinal predicate to the syntax and semantics for object counting in (170). It all looks like (161)–(162). There looks to be little room for the choice of a cardinal predicate embedded deep within the DP to engage a global condition on the sentence’s assertability. Event counting in (158) and (169) and in all previous sentences is rejected only if understanding them as cineramic report or narration, so conveyed in logical form with the use of now-en-scène, violates the anticonvergence condition as these were defined and discussed in sections 9.4.0 and 9.4.1: (172) [∃X : ∃E (∃eμ count[eμ,E, 3n] NP[E,X] …)] [℩E : ∃eμ count[eμ,E, 3n] NP[E,X] …)] [∃E0 : now-en-scène[E0] N[E,E0]] W[E0, X] [∃E1 : now-en-scène[E1]] Φ[E1, E0]
Naive Reference for the Cinéaste
583
(173) now-en-scène(e) ↔ ∃sɶ ∃αf∃f(now( sɶ ) & en-scène(e, sɶ , αf,f)) ɶ ɶ ɶ (174) now-en-scène[E] ↔ ∃S∃A F∃F(now( S ) & en-scène[E,S ,AF,F]) (175) For discourse Δ and token µ of ⌜now(vi)⌝ in Δ such that LTΔµ©(µ) = δt, Σ satisfy u ↔ Within(nowΔµ©(δt), Σ(vi)). (176) en-scène(e, sɶ , αf,f) ↔df ∃π(cinerama(π) & Sɶπ sɶ ) & At(αf, e, t(e), f) & ∀ṫ((moment(ṫ) & overlap(ṫ, t(e))) → f ɶ ɺ sɶ(tɺ) = projection(ṫ, αf, π f ( sɶ)(tɺ) , f, l( sɶ)(tɺ) , No( sɶ ) , o( s )(t ) ))) ɶ ɶ → Sɶπ sɶ) ) & (177) en-scène[E,Sɶ ,AF,F] ↔df ∃π(cinerama(π) & ∀ sɶ(Ss ɶ ∀e(Ee → ∃sɶ ∃αf ∃f(Ssɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ɶ ɶ → ∃e∃αf ∃f(Ss ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ∀ sɶ (Ss ɶ ɶ & AFαf & Ff & en-scène(e, sɶ , αf,f))) & ∀αf (AFαf → ∃e∃sɶ ∃f(Ss ɶ ɶ ɶ ∀f(Ff → ∃e∃s ∃αf (Ss & AFαf & Ff & en-scène(e, sɶ , αf,f))) (178) (Anticonvergence) cinerama(π) →. ∀n∀f ∀sɶ(¬integrated(Sɶπ , f ) → (Sɶπ sɶ → ¬ orient( sɶ, f , n))) (179) integrated(Sɶ , f ) ↔ def ∀sɶ(Sɶ sɶ → frame( sɶ, f )) & ∃F ∀sɶ(Sɶ sɶ → ∃f (Ff & frame( sɶ, f ) & ∃ζ (ζ ≠ o( sɶ) & Landmark(ζ , t( sɶ), f )) & ∀ζ (Landmark(ζ , t( sɶ), f ) → ¬∃t0 ∃sɶ ′∃t∃t ′∃fɶ ′(Sɶ sɶ ′ & Ff ′ & sɶ ≠ sɶ ′ & frame( sɶ, f ′) & t( sɶ(t )) = t( sɶ ′(t ′)) = t0 & Landmark(ζ , t( sɶ ′), f ′) & ∃α sɶ ( t ) At(α sɶ ( t ), ζ , t( sɶ(t )), sɶ(t )) & ∃α sɶ ′ ( t ) At(α sɶ ′ ( t ), ζ , t( sɶ ′(t )), sɶ ′(t )))))) Recall that the scenes of a cinerama path-integrated to a common frame of reference aspire to an authoritative, enduring representation of the goings-on in that frame of reference. As a condition toward its veridicality as a visual representation of what is, a cinerama’s distinct scenes must be of distinct goings-on. What looks different was different. Anticonvergence precludes a Rashomon-like compilation of multiple views of the same events. Furthermore, for a frame of reference to which its scenes are path-integrated, a cinerama is orienting in that ostension to distinct visual experience—to distinct scenes or fragments thereof—is always ostension to distinct spatiotemporal addresses in that frame of reference. Scenes that are presented as witness to the events reported and therefore track their participants in order to witness their participation in them may come to comply with anticonvergence in two respects. First, scenes that track participants only from distinct lifetimes—that is, as a sequel to object counting—are never at risk: distinct lifetimes never risk spatiotemporal intersection in violation of anticonvergence. Second, even when event counting so-called deploys distinct scenes to track the same lifetime, no violation of anticonvergence is encountered if the distinct scenes from the same lifetime manage to bear witness to events from that lifetime without temporal overlap. Such is the case with all felicitous event counting not grounded in epistemic necessity. The effects of this temporal condition are vividly illustrated in the
584
Chapter 11
examples from the preceding section of viewpoint aspect contrasting the present perfect and simple past. To reject event counting is in effect first to recognize that the event-counting sentence cannot be taken to express or imply the evidential claim that, for example, (180) #As (collectively) witnessed (by various witnesses), 30,000 passports that passed through Passport Control at JFK in January 1995 have been renewed this year by their bearers. What would have to have been en-scène to bear witness to what is reported (see (172)) does not amount to a coherent, orienting visual representation of what went on. So it must not be that the utterance is either itself parsed as in (172) or implies as much. It is rather an assertion ungrounded in visual experience and is not narration en-scène: (181) [∃X : ∃E (∃eμ count[eμ,E, 3n] NP[E,X] …)] [℩E : ∃eμ count[eμ,E, 3n] NP[E,X] …)] [∃E0 : N[E,E0]] W[E0, X] ∃E1 Φ[E1, E0] But if neither the speaker nor her sources can claim witness, what else could be the grounds for her assertion? The utterance is infelicitous as is, unless it is understood to find warrant in epistemic necessity such as that granted by the international conventions regulating passport expirations but not their renewals. (See chapter 9, note 5, for some mention of responses that differ according to differences in how the assertion is epistemically defective.) Neither (163) nor (169) can be parsed as in (172). But the former and only the former then finds felicitous expression in an alternative parse (181), understood as a general proposition that follows from known principles without any—or at least little—pertinent observation. In retrospect, note that the account on offer for the contrast between (163) and (169) and for the semantics and pragmatics of event and object counting in general does not confine itself to contexts of utterance that are obviously cineramic. Nor should it, as you and I feel the contrast between these example sentences with nothing but this page in view. The communication of a solitary navigator narrating concurrent visual navigation is a benchmark for language design, but cineramic semantics generalizes in several dimensions from this defining moment. First, it entertains communication and visual transmission from several navigators cooperating in a coordinated survey—a multitude of witnesses. Second, it admits that the present narrative may not be a report of present, direct witness but may be removed from such by apostolic succession. Third, rather unlike communication during concurrent navigation, it allows that the scenery visualized is subject to temporal compression, dilation, and editorial discretion as befits the organization and elicitation of memories. As a result of all this, so much more discourse, with nothing in sight,
Naive Reference for the Cinéaste
585
proves anyway to be cineramic, expounded as that which has somewhere sometime been seen en-scène. It is in fact a default in the understanding of discourse. Only when it cannot be taken on faith that what is handed down rests on its transmission from direct witness is it then questioned how else it is known.14 Cineramic semantics does presuppose the constancy of visualization—with or without present, relevant retinal activity—but doesn’t any felicitous event counting already presume as much? How else can one count 30,000 passports at the start, and accept as true any assertion that so counts them unless one has in mind the conditions, a scene and protocol that number them at 30,000 rather than any other number? The anticonvergence condition, (178)–(179), is not an absolute prohibition against the projection of convergent scenery. It is rather a condition on scenery taken to be orienting and path-integrated for its observers. Orientation and path integration must in turn be understood in terms of a given frame of reference and its (sub)space, for the simple reason that one may be oriented with a complete cognitive map of, say, the subway system or a building interior, clueless as to its orientation to the compass let alone as to its translation into coordinates for the city above or outside. Even for frames of reference and spaces, often or easily navigated and charted, there may be visual experience, the events of which are now under narration, under such extreme conditions—a blur of hot pursuit—that path integration is overwhelmed on this occasion. Frames of reference for which the scenery is said to be anticonvergent will be just those to which the observer is not oriented and as a narrator represents herself as such. Violations of anticonvergence and the resultant infelicity of event counting are then gradient according to the plausibility that an observer experiencing the scenery under consideration under the conditions imagined for that experience could remain disoriented. Suppose that the 3000 clients attending speed-dating events last month were 1000 lonelyhearts looking for love in the city of eight million stories: (182) Three thousand clients that attended speed-dating events last month exchanged phone numbers at the end of the date. (183) ?(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers in a little black book. (184) ??(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers in mind. (185) ???(The) 3000 clients that attended speed-dating events last month keep each other’s phone numbers on the fridge. Sentence (182) is a citation example of felicitous event counting, for which the relevant eyewitness scenery begins and ends with the events, the speed dating, at which 3000 are counted. Naturally enough, the eyewitness may have no memory of the
586
Chapter 11
venue for the speed-dating event, but she is oriented to the frame of reference—the catering hall hosting the speed dating—to which this scenery is path-integrated and orienting. For the rest, how can the speaker, it must be asked, have known either by visual tracking or by inference from the known facts that the 3000 clients do as reported without seeing that some clients are the same lonelyhearts in the same place and time? For (183), the scenery and frame of reference to which it is pathintegrated and orienting can remain the same, but if the present-tense assertion is not to involve more scenery that tracks the clients to an illicit convergence in a present scene, it must be that the speaker has in mind a law of little black books that what goes into a little black book, as witnessed during the speed-dating event, stays in the little black book. A similar assumption about the minds of speed daters, perhaps less obvious, spares (184) rejection outright. For (185), one might imagine that the speaker receives transmissions from personal cameras that have been implanted at the speed-dating events, with a field of vision just wide enough for client and refrigerator. If, on the other hand, the scenes in witness of (185) are visits to the clients’ homes, the speaker in uttering (185) represents herself as somehow subject to a lapse in the path integration that would put some of these homes at the same address containing the same fridge and the same client. Somehow these clients are followed home across town, presumably navigating that town, without discovery or visual representation that some of those kitchens with the phone numbers on the fridge are the same kitchens. Now that is one impaired speaker whose word is not to be trusted. It is sometimes remarked (see, e.g., Barker 1999; Carlson 1982; Doetjes and Honcoop 1997; Moore 1994; Zimmermann 2005) that the felicity of event-counting 3000 speed daters discrepant with object-counting 1000 lonelyhearts is contingent on the speaker representing herself as ignorant of or unable to recognize the identities that undermine the inflated tally (see also chapter 10, note 52), which finds its best examples in numbers involving thousands. But anyone who knows any baseball knows that players make multiple appearances as batters and that an offensive roster of twenty-seven would be illegal in the game: (186) On September 9, 1965, in the greatest game ever pitched, Sandy Koufax retired twenty-seven consecutive batters for the Chicago Cubs with fourteen strikeouts, the most ever recorded in a perfect game. It is a law of baseball that twenty-seven batters are fewer players. Koufax’s game is a perfect one in that no batter managed to get on base, and twenty-seven is their smallest number possible in a regulation nine-inning game. Here it would be infelicitous and obscure to count otherwise, even though it is true that the Cubs fielded eleven offensive players that glorious day (nine starters and two pinch hitters):
Naive Reference for the Cinéaste
587
(187) #On September 9, 1965, in the greatest game ever pitched, Sandy Koufax retired eleven Chicago Cubs with fourteen strikeouts, the most ever recorded in a perfect game. Nor could it be said that the speaker of (186) represents himself as ignorant of these facts in a sports bar, without forfeiting his credibility and risking injury. Notice rather that the scenes evoked of this ballgame are not orienting or oriented for the speaker for any frame of reference or space outside the game itself and need not be for the speaker to defend his assertion and adduce relevant eyewitness. Not much of a baseball fan, Roger Schwarzschild has nevertheless observed that there lingers something odd in a report of my attendance at Fenway Park on a less exalted occasion in which I say: (188) ???Sox had a good night. Twenty Yankees never reached on base. (189) ???Sox had a good night. Twenty batters never reached on base. Schwarzschild has further remarked that such pronouncements improve if, instead of being my report of what happened during the game I attended last night, they are broadcast sports news: (190) In baseball last night, the Red Sox in a successful outing at Fenway retired twenty Yankee batters. All of these utterances comply with the letter of anticonvergence. The variation in their felicity must then be located elsewhere. There must be, I can only presume, a further pragmatics with guidance for when it is appropriate to willingly and knowingly refer to a cramped scenery that one knows will not survive anticonvergent integration into a longer documentary for larger frames of reference—a documentary that perhaps one should always aspire to as a complete historical atlas for the world at large or at least for where one has been. It cannot be unexpected that what to narrate, how to film, from what perspective, with orientation for what frames of reference, as eyewitness for what assertions in answer to what questions, requires more directorial and editorial subtlety than has been spelled out here.15 11.1
Recounts
The puzzles of extensional substitutivity have been resolved (sections 10.1–10.3) via adverbialization: (191) (The) three million passengers who crowded National Airlines routes in 1980 were (the) one million frequent fliers loyal to it. (192) (The) one million frequent fliers who were loyal to National Airlines in 1980 are still frequent fliers. (193) #Therefore, (the) three million passengers who crowded National Airlines routes in 1980 are still frequent fliers.
588
Chapter 11
(194) As befits a clever disguise, Superman does not resemble Clark Kent. Superman is Clark Kent. #Therefore, Superman does not resemble Superman. (195) The man surpasssed the boy of his childhood in regrets and not much else. This man is that boy (grown up). #Therefore, the boy of his childhood surpassed the boy of his childhood in regrets and not much else. (196) A careless camper killed the unconscious drunk passed out in bed. The unconscious drunk was the careless camper. #Therefore, an unconscious drunk passed out in bed killed the careless camper. The events the derived adverbs denote remain distinct despite the identity of the participants. In the case of recounts, counting them to three million depicts them and what they do under different conditions than counting them to one million. If so, there is no inference. The substitution of the one adverbial for the other is invalid, whatever else can be said about counting or the structure of identity statements themselves: (197) (The) one million frequent fliers who were loyal to National Airlines in 1980 counted one million are still frequent fliers. (198) #Therefore, (the) three million passengers who crowded National Airlines routes counted three million in 1980 are still frequent fliers. As for the logical form of identity statements, it must already be far from ⌜x = y⌝ for them to be tensed or modified by secondary predicates (let alone to host adverbialization): (199) These oranges in the late Cézanne were those oranges in the early one. (200) These oranges are those oranges arranged differently. If, moreover, the DPs in identity statements adverbialize contrary adverbials—while counted three million is not while counted one million, while the man is not while the boy, etc.—it must be that their derived adverbials, as in clauses everywhere, frame their own local events. The structure that discerns separate events for the participants to an identity is independently corroborated in the asymmetric application of both Tense (section 10.3) and secondary predicate (section 10.3.0) to only the events described by the predicate nominal:
Naive Reference for the Cinéaste
589
(201) For some evenings in 1892, Venus was Hesperus (aligned with a crescent moon). For some evenings in 1892, Venus was the evening star (aligned with a crescent moon). (202) #For some evenings in 1892, Hesperus was Venus. #For some evenings in 1892, the evening star was Hesperus. In (203), recall from section 11.0.1 that the flutters the butterflies execute while counted 4000 are at a great distance from the stirring tornado, related by a relation ‘Cause[E0,E1]’ between cause and remote effect. Similarly, recall that the distance between Kafka’s creations while Kafka and eventual fame is bridged by what his fame is about ‘About[E0,E1]’: (203) Four thousand butterflies visiting Kinkaju-ji in the spring triggered a fall tornado in the Golden Gate. (204) Kafka is famous. For an identity to hold of the scattered stages of a life, section 10.3.0 introduced a dislocative relation ‘[E1, E2 ]’ to time travel between events belonging to the history of the same object, ‘[E0, E1 ]’: (205) Jules Bagwell is Julia Bagwell [gender corrected decades ago]. DPi-[℩ER: theni]-W-Theme W-Present-Theme be DP-W-Theme […] (206) [℩x : ∃E Jules Bagwell[E,x]][℩E : Jules Bagwell[E,x]] [℩ER : then[ER,E]]16 [∃E0 : N[E,E0]]W[E0,x] O[E0,E1]∃XTheme[E1,X] (i) [E1, E4 ] (ii) ∃E2∃X W[E2,X] Present[E4] O[E2,E3] ∃XTheme[E3,X] (iii) O[E3,E4] be[E4] [℩x : ∃E Julia Bagwell[E,x]][℩E : Julia Bagwell[E,x]] [∃E5 : N[E,E5]] W[E5,x] O[E5,E6] ∃X Theme[E6,X] [E4, E6 ]
(iv)
(207) [E1, E2 ] →. ∀X(W[E1,X] ↔ W[E2,X]) (208) [E1, E2 ] ↔def ∀X(W[E1,X] ↔ W[E2,X]) & ∀T(time[T] → (At[E1,T] ↔ At[E2,T])) The identity statements that happen also to be recounts have identical structure: (209) (The) three million passengers who crowded National Airlines routes in 1980 were (the) one million frequent fliers loyal to it. DPi-[℩ER: theni]-W-Theme W-Present-Theme be DP-W-Theme
590
Chapter 11
(210) [∃X: ∃E(∃eμcount[eμ,E, 3,000,000] passengers[E,X] …)] [℩E: ∃eμcount[eμ,E, 3,000,000] passengers[E,X] …] [℩ER : then[ER,E]] [∃E0 : N[E,E0]]W[E0,X] O[E0,E1] ∃XTheme[E1,X] (i) [E1, E4 ] (ii) ∃E2∃X W[E2,X] Past[E4,ER] O[E2,E3] ∃XTheme[E3,X] O[E3,E4] be[E4] (iii) [∃X: ∃E(∃eμcount[eμ,E, 1,000,000] frequent fliers[E,X] …)] [℩E: ∃eμcount[eμ,E, 1,000,000] frequent fliers[E,X] …] [∃E5 : N[E,E5]] W[E5,X] O[E5,E6] ∃X Theme[E6,X] [E4, E6 ]
(iv)
Identity statements always token and . They must, if identity statements are to have the expressive power to assert about those who have walked across the world’s remote stages just who and how many they really are among the things that persist from one stage to another. In swapping in ‘[E0, E1 ]’ for ‘Cause[E0,E1]’ or ‘About[E0,E1]’ and ‘[E4, E6 ]’ for lexical verb or preposition, the clausal structure of a recount hardly deviates from that of any other clause containing event counting, and it is equally subject to the conditions on event counting discussed in section 11.0 and to its analysis. To the extent that speed daters are lonelyhearts when they speed-date and are likely to remain in that condition for awhile, one may infer their present condition from direct observation of their past: (211) (The) 4000 clients that attended speed-dating events last month were 1000 lonelyhearts searching for soulmates in a city of eight million. (212) (The) 4000 clients that attended speed-dating events last month are 1000 lonelyhearts searching for soulmates in a city of eight million. (213) (The) 4000 clients that attended speed-dating events last month were 1000 (recent) divorcé(e)s. (214) (The) 4000 clients that attended speed-dating events last month are 1000 (recent) divorcé(e)s. In contrast, as no honeymooners or newlyweds speed-date (or so it is hoped), (215) in the past tense is plain false. And with nothing known then foretelling of future bliss, (216) in the present tense cannot report that 1000 lonelyhearts managed to marry after all, after four speed-dating events (cf. (212), (214)):17 (215) F (The) 4000 clients that attended speed-dating events last month were 1000 honeymooners/newlyweds. (216) #(The) 4000 clients that attended speed-dating events last month are (now) 1000 honeymooners/newlyweds. Again, sentence (214) by itself presents a minimal pair. Event counting is felicitous only if the divorces were recent to the speed dating and not if the clients were single
Naive Reference for the Cinéaste
591
or happily married then and are only now divorced. If not attempting an event count, the predicate nominal is otherwise free to describe a condition that did not hold while speed dating: (217) (The) 1000 clients that attended speed-dating events last month are (now) 1000 honeymooners/newlyweds. (218) (The) 1000 clients that attended speed-dating events last month are 1000 recent divorcé(e)s. Correlatively, (219) and (220) both embody a minimal pair: (219) (The) 4000 clients that attended speed-dating events last month are (now) honeymooners/newlyweds. (220) (The) 4000 clients that attended speed-dating events last month are recent divorcé(e)s. Assuming still that the predicate nominals describe what was not the case while speed dating, (219) and (220) entail that there are 4000 honeymooners/newlyweds and 4000 recent divorcé(e)s—that is, where 4000 has been understood to count objects. The event-counting interpretation of these sentences is ruled out. As in section 11.0.0, a matrix present tense when past events are event-counted is true and felicitous only via an inference that what was still is. In short, these present-tense recounts combine two dimensions: event counting (discrepant with the object count so-called) vs. object counting (where an event counts just in case it is the speed-dating history of an individual) and report of a condition that persists into the present vs. report of a novel, present condition, as schematized in graphs of a single individual’s history: t(u) (221) d1
d2
d3
d4
h
(222) D
h
(223) d1
d2
d3
d4
l
(224) D
l
t(u)
592
Chapter 11
Just one of these combinations fails, a report of (221) that 4000 speed daters last month are (now) 1000 honeymooners. Reporting honeymooners, it can only be said that 1000 speed daters last month are (now) 1000 honeymooners, where a client’s entire history last month counts as a single event, (222). Had these losers not found true love after a month on the circuit, remaining forlorn, it could be reported either that 4000 speed daters last month are (in fact) 1000 lonelyhearts (223) or, of course, that 1000 speed daters last month are 1000 lonelyhearts (224). What makes an identity statement interesting is exactly that it relates the participants in otherwise remote events—the speed daters then are the honeymooners now (217)—for which and are essential, to travel in time between events and to affirm the identity of their participants. What could then preclude event counting under like conditions since presumably (216) shares with (217) the structure in (225), disagreeing only in n? (216) #(The) 4000 clients that attended speed-dating events last month are (now) 1000 honeymooners. (217) T (The) 1000 clients that attended speed-dating events last month are (now) 1000 honeymooners. (225) [∃X: ∃E(∃eμ count[eμ,E, n000] clients[E,X] …)] [℩E: ∃eμcount[eμ,E, n000] clients[E,X] …][℩ER : then[ER,E]] [∃E0 : N[E,E0]]W[E0,X] O[E0,E1] ∃XTheme[E1,X] [E1, E4 ] ∃E2∃X W[E2,X] Present[E4] O[E2,E3] ∃XTheme[E3,X] O[E3,E4] be[E4] [∃X: ∃E(∃eμcount[eμ,E, 1000] honeymooners[E,X])] [℩E: ∃eμcount[eμ,E, 1000] honeymooners[E,X] …] [∃E5 : N[E,E5]] W[E5,X]O[E5,E6] ∃X Theme[E6,X] ℐ[E4,E6]
(i) (ii) (iii)
(iv)
Nothing, as section 11.0.2 remarks, distinguishes the logical form of infelicitous event counting in (226) from the felicitous event counting in (227) or from the object counting in (228)—nor can the verbs renew and expire or the present tense clash with event counting lest they be wrongly excluded in (227): (226) #(The) 30,000 passports that passed through Passport Control at JFK in January 1995 have been renewed this year by their bearers. (227) a. (The) 30,000 passports that passed through Passport Control at JFK in January 1995 are expired by now (in accordance with international law). b. (The) 30,000 passports that passed through Passport Control at JFK in January 1995 have all been automatically renewed by now several times over.
Naive Reference for the Cinéaste
593
(228) (The) 10,000 passports that passed through Passport Control at JFK in January 1995 have been renewed this year by their bearers. Event counting in identity statements, as in (226)–(228), is rejected only if understanding them as cineramic commentary, so conveyed in logical form with the use of now-en-scène (see sections 9.4.1 and 11.0.2), violates the anticonvergence condition: (229) [∃X: ∃E(∃eμcount[eμ,E, n000] clients[E,X] …)] [℩E: ∃eμcount[eμ,E, n000] clients[E,X] …][℩ER : then[ER,E]] [∃E0 : now-en-scène[E0] N[E,E0]]W[E0,X] O[E0,E1] ∃XTheme[E1,X] [E1, E4 ] [∃E2 : now-en-scène[E2]] ∃X W[E2,X] Present[E4] O[E2,E3] ∃XTheme[E3,X] O[E3,E4] be[E4] [∃X: ∃E(∃eμcount[eμ,E, 1000] honeymooners[E,X])] [℩E: ∃eμcount[eμ,E, 1000] honeymooners[E,X] …] [∃E5 : now-en-scène[E5] N[E,E5]]W[E5,X]O[E5,E6] ∃X Theme[E6,X] [E4, E6 ]
(i) (ii) (iii)
(iv)
Scenes as numerous as the speed daters at their start and deployed to track them to witness their honeymoons violate the anticonvergence condition if they converge on fewer honeymooners. To reject event counting is first to recognize that the eventcounting sentence cannot be taken to express or imply the evidential claim that (230) #As witnessed, (the) 4000 clients that attended speed-dating events last month are (now) 1000 honeymooners. (231) #As witnessed, (the) 4000 clients that attended speed-dating events last month are (still) 1000 lonelyhearts. What would have to have been en-scène in evidence of what is asserted does not amount to a coherent visual representation of what went on. So none of the eventcounting identity statements, neither (212) nor (216), can be parsed as in (229). But (212) and not (216) is felicitous parsed as (225) without any tokens of now-en-scène and understood to follow from known principles of the heart from what was observed and known at the speed-dating events, without further testimony from a present scene. Returning to the schematic diagrams (221)–(224), the identity statements about lonelyhearts are contingent on persistent conditions known then and there at the speed dating or on examination of the speed dater—and this holds true despite the harmless redundancy in scrutinizing a lonelyheart’s heart more than once. For identity statements about honeymooners, the speed daters’ happy endings were unforeseeable. The grounds for such an object-counting identity statement (217) might instead have been to track the 1000 to their 1000 parts in 500 honeymoons
594
Chapter 11
with their better halves (see (222)). As remarked in section 11.0.2, scenes that track participants, 1000 in this case, who are all from distinct lifetimes, never risk spatiotemporal intersection in violation of anticonvergence. That risk attaches, however, to 4000 scenes tracking 4000 speed daters to their honeymoons (see (221)). Also as remarked in section 11.0.2, the anticonvergence condition is not an absolute prohibition against the projection of convergent scenery. It is rather a condition on scenery taken to be orienting and path-integrated for its observers. Where there is a risk of illicit convergence, the speaker represents herself as disoriented in any frame of reference to which the scenes cannot be path-integrated without violating the anticonvergence condition. Out of the blue and subject to naturalistic default assumptions, one cannot imagine what grounds the speaker has for (216) that leave her in full command of her faculties, with compelling accurate eyewitness that those who were speed daters are now honeymooners, and yet unaware that she is looking on at the same goings-on from different perspectives. Recounts, identifying the 4000 with the 1000, have been the poster child illustrating that identity statements are undistinguished in their syntax, semantics, and pragmatics from any other clause—with respect to supermonadicity, adverbialization, and the vagaries of counting. But the scenes subject to the anticonvergence condition are any called on to witness that speed daters back then are honeymooners now. As then expected and observed earlier, (219) is also infelicitous event counting even without attempting a recount: (219) # (The) 4000 clients that attended speed-dating events last month are (now) honeymooners. Similarly infelicitous is a sentence that keeps up the event counting: (232) #(The) 4000 clients that attended speed-dating events last month are (now) 4000 honeymooners. If the point of many identity statements is to let us know about those who have walked across the stage—who and how many they really are in the inventory of things that persist from one stage to another on terra firma—it is unsurprising that 1000 honeymooners and other predicate nominals are more often than not used to count objects rather than events. There is nothing however in the logical form of a predicate nominal to discourage an event-counting one as in (232). All that is needed is some imagination to bridge one event count and another without violating the anticonvergence condition, for which screwball comedy comes to the rescue. Joseph, Josephine, Gerald, and Geraldine (aka Daphne) speed-date in Chicago.18 On the first date, Joseph meets Josephine at Orchestra Hall and Gerald meets Geraldine at the Art Institute, and the couples tryst and plan honeymoons in Florida. The second date, Joseph meets Geraldine at Wrigley Field and Gerald, Josephine at
Naive Reference for the Cinéaste
595
Comiskey Park, and the couples entwine and plan honeymoons in Florida. For the same weekend at the Fontainebleau, Joseph and Josephine book the Jungle Hideaway; Gerald and Geraldine, the Tiki Hut; Joseph and Geraldine, the Harem Tent; and Gerald and Josephine, the Bower of Bliss. And so (233) (The) eight clients that attended speed-dating events last month are now eight honeymooners at the Fontainebleau. This can be asserted only against a background that has drawn clear causal connections between speed dates and honeymoons: the fiancé from Orchestra Hall is the groom in the Jungle Hideaway, and the fiancé from Wrigley Field is rather the groom in the Harem Tent. Sentence (233) might be either an inference that eight speed daters must by now be eight honeymooners as per their nuptial plans or reports from several cleverly edited scenes that have managed under Billy Wilder’s direction to avoid the same place at the same time. Either way, true identities outnumber entities.19 11.1.0
The anticonvergence condition and scenes of counting
Even when compliant with the anticonvergence condition, not all scenes launched from the initial conditions for an event count track to scenery that remains compliant with the conditions on counting laid out in section 12.1. These pathological scenes manage to rescue assertion (219) that the event-counted speed daters are now honeymooners. But unlike Billy Wilder’s scenes, from which their difference is important, they fail both object and event counting in the present: (219) OK (The) 4000 clients that attended speed-dating events last month are (now) honeymooners. (234) #(The) 4000 clients that attended speed-dating events last month are (now) 1000 honeymooners. (235) #(The) 4000 clients that attended speed-dating events last month are (now) 4000 honeymooners. For the example to be developed, recall from section 11.0.0 that Doetjes and Honcoop’s (1997) report (69) of the current scene at port is infelicitous as an event count, in contrast to the harbormaster’s inference (93) and forecast derived from her examination of 4000 manifests at the lock, all of which is now expected: (69) #(The) 4000 ships that passed through the lock are now in the port. (93) (The) 4000 ships that passed through the lock are by now in the port. Recall also that in addition to the above two conditions some contexts provide prosthetic, visual enhancements that rehabilitate (69) even in the absence of inferential behavior based on prior knowledge. With customs agents having planted a
596
Chapter 11
camcorder on every vessel passing through the lock, one could utter (69) gesturing to 4000 monitors. These scenes plainly suffice as direct witness for the truth of (69) and for (236): (236) (The) 4000 ships that passed through the lock are now harbor guests in the port. Yet the 4000 scenes of harbor guests in the port remain defective grounds for counting harbor guests, even as 4000: (237) #(The) 4000 ships that passed through the lock are now 1000 harbor guests in the port. (238) #(The) 4000 ships that passed through the lock are now 4000 harbor guests in the port. With four camcorders on board each harbor guest, the scenes are 4000 views of the same spatiotemporal regions at the port. As such, the scenes would violate the anticonvergence condition if they were taken to be path-integrated and orienting to a given frame of reference, to a frame of reference for the port in particular. But this contrived context affords the speaker concurrent eyewitness evidence for her report under rare conditions where it is possible for her in her right mind to have such a visual experience without being oriented to the space from which that scenery is projected. There is, however, no well-behaved counting in that wall of scattered scenery from the port: (239) #… [∃X: ∃E(∃eμ now(eμ) count[eμ,E, n000] harbor guests[E,X])] A visual count of n000-many ζ (states of being) harbor guests must fit them to n000-many nonoverlapping addresses for the same frame of reference to which the scenes must therefore all be integrated (see sections 9.4.2 and 12.1). How else could it be concluded from n000-many scenes that there are n000-many of that which the scenes are of? Thus, the video monitors in the harbormaster’s office are witness that the 4000 ships that had passed through the lock are presently harbor guests but not to a valid count of them at present. The infelicity of (237) and (238) asserted outof-the blue does spring from the presumption that the speaker intends to count the harbor guests in the scenes just now in view, as (239) transcribes. Imagine instead (238) felicitous at the conclusion of an inference: (240) Four thousand ships passed through the lock 6 months ago. You can see there that they have reached their destination. The 4000 ships that passed through the lock are now 4000 harbor guests in the port. But here the second count of 4000 must be anaphoric to the first: (241) [∃X: ∃E(∃eμ1count[eμ1,E, 4000] ships[E,X]) …] … [∃X: ∃E [℩eμ:proμ1] count[eμ,E, 4000] harbor guests[E,X])]
Naive Reference for the Cinéaste
597
Finally, imagine that just behind the bank of 4000 monitors is the harbormaster’s window with an unobstructed view of 1000 berths below, a scene for a proper count to 1000: (242) Four thousand ships passed through the lock 6 months ago. You can see there [pointing to the monitors] that they have reached their destination. The 4000 ships that passed through the lock are now 1000 harbor guests in the port [pointing out the window]. [∃X: ∃E(∃eμ1count[eμ1,E, 4000] ships[E,X]) …] … [∃X: ∃E(∃eμ now(eμ) count[eμ,E, 1000] harbor guests[E,X])] 11.1.1
Scene changes for counting and nominal syntax
From the outset, the sentences that display in plain sight sequencing of events, Doetjes and Honcoop’s (1997) name for the judgments that contrast (67) with (69) and (72) with (75), have been juxtaposed with identity statements such as (12) that at first glance mask the effect of sequencing of events: (67) (The) 4000 ships that passed through the lock arrived in the port of Rotterdam 3 hours later. (69) #(The) 4000 ships that passed through the lock are now in the port. (72) (The) 4000 ships that passed through the lock last year were registered in Panama. (75) #The 4000 ships that passed through the lock last year are registered in Panama. (12) (The) 4000 ships that passed through the lock last year are 1000 vessels registered in Panama. (243) (The) 4000 ships that passed through the lock last year were 1000 vessels registered in Panama. Given that all these sentences instantiate a common logical form, it should be that (12) is subject to the same conditions underlying the sequencing of events, and so it is insofar as it has been shown (section 11.1), abbreviating the earlier discussion, that (12) may be asserted only as an inference from (243) that how things were still is, in contrast to an unforeseen and unforeseeable present, the contrast between (244) and (246): (244) (The) 4000 clients that attended speed-dating events last month are 1000 lonelyhearts searching for soulmates in a city of eight million. (245) (The) 4000 clients that attended speed-dating events last month were 1000 lonelyhearts searching for soulmates in a city of eight million.
598
Chapter 11
(246) a. #(The) 4000 clients that attended speed-dating events last month are (now) one 1000 honeymooners. b. #(The) 4000 clients that attended speed-dating events last month are (now) 4000 honeymooners. c. #(The) 4000 clients that attended speed-dating events last month are (now) honeymooners. The common logical form also affords to (69) and (75) contexts in which they too become felicitous as an event count if construed as an inference from how things were or also under eccentric conditions like those in the harbormaster’s office. Such contexts have been conjured up, though it is left unexplained why one is slow to conjure up these contexts out of the blue when evaluating (69) and (75) but is almost automatically inclined to do so for (12). Many present-tense assertions of episodic conditions are intended as eyewitness reports—one might have expected (69)’s speaker to continue “See—look there” (in settings without prosthetic video) or (75)’s speaker to follow with “Look here, in the registry,” which are contexts in which the event count fails. It appears that hearers presented with (69) and (75) out of the blue are biased to take it as eyewitness report. In contrast, even if hearers approach (12) with equal bias, they are quickly disabused of it by its incoherence when applied to identity statements—how could inspection of the 1000 vessels (“See, see?”) confirm the identity. Failing that, the hearer recognizes that the speaker’s grounds for the assertion must lie elsewhere—in an inference from other knowledge. All the same, the identity statement and the other present-tense assertions have a common logical form and thus the same ambiguities in context, whether or not this just-so story is why a speaker resolves their ambiguities differently. I would like now to redress a descriptive gap in the simple opposition of identity statements to other present-tense assertions. To my ear, the unmeasured identity statements in (247) and (248)–(250), with bare nominals, pattern with (75) in blocking an event count absent special context and extra effort, and thus they differ from the identity statements cited in (12) and (244) and their kin in (251)–(252), where an event count in the first nominal needs no special prompting if the second nominal is recounting: (247) #(The) 4000 clients that attended speed-dating events last month are lonelyhearts searching for soulmates in a city of eight million. (75)
#The 4000 ships that passed through the lock last year are registered in Panama. (248) #(The) 4000 ships that passed through the lock last year are vessels registered in Panama. (249) #(The) 4000 ships that passed through the lock last year are vessels that are registered in Panama. (250) #(The) 4000 ships that passed through the lock last year are vessels that were registered in Panama.
Naive Reference for the Cinéaste
599
(12) (The) 4000 ships that passed through the lock last year are 1000 vessels registered in Panama. (251) (The) 4000 ships that passed through the lock last year are 1000 vessels that are registered in Panama. (252) (The) 4000 ships that passed through the lock last year are 1000 vessels that were registered in Panama. The contrast between bare nominals and counting nominals extends beyond their occurrence within simple clauses that purport to event-count: (253) (The) 4000 ships that passed through the lock last year were 1000 vessels that are (only now) registered in Panama. (The) 4000 ships that passed through the lock last year were 1000 vessels that are (only now) under quarantine. (254) (The) 4000 ships that passed through the lock last year were 1000 merchant vessels(, and) they are (only now) registered in Panama. (The) 4000 ships that passed through the lock last year were 1000 merchant vessels(, and) they are (only now) under quarantine. In (253) and (254), 1000 (merchant) vessels occurs in a main clause respecting the sequence of events, reporting what was observed when the ships passed through the lock (see (243)), and the first clause counts 4000 events. The counting nominal 1000 (merchant) vessels, however, itself licenses a continuation in the relative clause or in the next sentence that describes present conditions. The bare nominal, in contrast, does not. That is, although (255) contains a fine event count respecting the sequence of events, a continuation into the present as in (256) or (257) undermines the felicity of the count. (255) (The) 4000 ships that passed through the lock last year were (merchant) vessels ((that were) registered in Panama). (256) #(The) 4000 ships that passed through the lock last year were (merchant) vessels that are registered in Panama. (257) #(The) 4000 ships that passed through the lock last year were merchant vessels(, and) they are registered in Panama. It seems that a bare plural, unlike its counting counterparts, cannot recount, as it were, for the benefit of later clauses, neither for relative clauses nor for later sentences. This is surprising if 1000 vessels registered in Panama and vessels registered in Panama are identical but for the cardinal predicate. Recounts manage to count events from the same lifetimes—as ships or as vessels, as speed daters or as lonelyhearts, as passengers or as frequent fliers. Given the optics of counting (sections 9.4.2 and 12.1), it must be that the scene in which the
600
Chapter 11
measure reads 4000 is not the scene in which the measure reads 1000. Any true recount implies a scene change from one counting eµ1 to the other eµ2:20 (258) [∃X: ∃E (∃eμ1count[eμ1,E, 4000] ships[E,X]) …] … [∃X: ∃E (∃eμ2 count[e2,E, 1000] vessels[E,X])] Apparently, there is nothing in the syntax or semantics of counting nominals, 1000 vessels registered in Panama, to discourage a scene change. The events counted 1000 in the new scene are equinumerous with lifetimes, and any scenes that are a continuation from there into the present are at no risk from the anticonvergence condition. Thus, any term anaphoric to 1000 vessels, such as the pronoun in (253) or (254), is itself object counting, and thus the continuations in (253) and (254) appear exempt from the sequencing of events. In contrast, the bare plurals in (256) or (257) seem unable to reorient the continuation, as if (257) were as in (259) obliged to continue the event count and thus continues to be subject to the sequencing of events as much as a nonidentity statement: (259) #(The) 4000 ships that passed through the lock last year were merchant vessels(, and) those 4000 are registered in Panama. This effect will follow if more than the omission of the cardinal predicate distinguishes the bare plural from its counting counterpart. If the null article of a bare plural in predicative position is anaphoric to prior scenes or frames of reference, the pronouns in (256) and (257) continue indeed to refer to participants in scenes where they measure 4000. It turns out that, as a class, nonidentity statements and identity statements with bare nominals are alike in obstructing the scene changes that must be taken for granted in evaluating recounts.21
12
Measuring Events
The puzzle of extensional substitutivity for counting that examples such as (1)–(3) have illustrated is resolved (section 12.1) and naive reference spared when what is counted is not what the (in)definite description denotes, which is parsed instead to count NP-ings as in (4) rather than (5): (1) Three million passengers crowded National Airlines routes in 1980. (2) The three million passengers who crowded National Airlines routes in 1980 were the frequent fliers loyal to it. (3) (1), (2) ⊬ Three million frequent fliers crowded National Airlines routes in 1980. (After Gupta 1980, 23; Moore 1994) (4) [∃X: ∃E 3,000,000[E] passengers[E, X]] (5) *[∃X: ∃E 3,000,000[X] passengers[E, X]] This syntax that divorces count morphology from nominal denotation goes one better in section 12.2, where the article a(n) in singular plurals such as a synchronized twelve lights counts as one the event or state that the adjective denotes, a synchronization, rather than the lighting. This syntax will also eventually solve the substitutivity puzzle (see section 8.1 and chapter 14) that finds that a sunrise and a sunset and a sunrise and sunset are not synonymous nor are a reporter and a superhero and a reporter and superhero: (6) A sunrise and a sunset have preceded two twilights’ bloody battle. (7) A sunrise and sunset have preceded two twilights’ bloody battle. (8) A reporter and a superhero were at the crime scene interviewing witnesses. (9) A reporter and superhero were at the crime scene interviewing witnesses. That is because when NPs rather than DPs are conjoined, the solitary article says that there is a single W-ing in which both reporter and superhero participate: (10) [∃ X: ∃E0 a(n)[E0] W[E0,X] [℩E2,3: pro2,3]O[E0,E2,3] (Φ∃X∃E2 reporter[E2,X]) and (Ψ∃X∃E3superhero[E3,X])] …
602
Chapter 12
Recall that the sentential connective and is no instrument of collective reference, which needs the intervention of a pronoun ‘[ιE2,3: pro2,3]’ referring to some events and a relation, ‘W’, between these events and their participants. The article a(n) applies to the events that the adjacent morpheme describes, the adjective in ‘a synchronized twelve lights’ or ‘W’ in ‘a W reporter and superhero’ and ‘a W sunrise and sunset’. Counting the events a noun denotes, that is, the states of being passenger, frequent flier, civilian, superhero, child, or adult, is intrinsically as unstable as counting in any dense domain—times, spaces, etc. It is entirely contingent for a determinate result on the conditions and protocol for measurement. Any numeral, now parsing (4) as (11), is thus taken to express a relation ‘count[eμ, E, n]’ to such an event of measurement eμ, the semantics of which is discussed in section 12.1: (11) [∃X : ∃E (∃eμ count[eμ, E, 3,000,000] passengers[E,X])] Under adverbialization, the chosen protocol eμ conditions the matrix events described, which are framed while so counted (section 12.0). There is furthermore independent syntactic evidence (section 12.1.0) that three million is episodic and eventive, “(now) counted to three million,” and semantic evidence (section 12.1.1) that it expresses a relation between events of measurement and whatever is measured, three million[eμ,ζ]. The extra parameter for events of measurement then proves itself essential in sorting out (section 12.3) many children, many a child, and many a one or more children, no two of which are synonymous, where for the latter two, the many are the measurements. Counting and the conditions of measurement are again implicated in a substitutivity puzzle that undermines the substitution of musicians for instrumentalists and vocalists, even though simple NP and conjoined NP describe the same participants and the same events—the musicians are the instrumentalists and vocalists, and the music is the instrumentals and vocals: (12) Many instrumentalists and vocalists are famous pairs of performers. (13) #Many musicians are famous pairs of performers. As a matter of logic, there must be more than and to distinguish the logical syntax of simple NP and conjoined NP. At this point, the reader is sufficiently brainwashed to accept that the further structure has to do with scenes, frames of reference, or addresses within, so that simple NP and conjoined NP describe distinct presentation of the same subject matter. Scenes of instrumentalists here and vocalists there segregate them in a way that scenes of musicians here and there need not. But the change of scenery does not yet come with an explanation for why it should matter to (13) (see section 14.1.1), why it matters for some collective predicates but not for others (14) (see chapter 13, especially section 13.1), why further restriction repairs
Measuring Events
603
the defect (15) (see section 13.0.1.1), or why this substitutivity puzzle generalizes to DPs with null determiners (16)–(17) and concerns only them (18) (see section 13.1): (14) Many musicians are (famously) paired up. (15) Many musicians that appeared on stage together were famous pairs. (16) Various instrumentalists and vocalists are famous pairs of performers. (17) #Various musicians are famous pairs of performers. (18) a. The/some many musicians are famous pairs of performers. b. The/some various musicians are famous pairs of performers. If and is innocent, the effort to shift suspicion elsewhere must close in on the meaning of these sentences to resolve the substitutivity puzzle without any special remarks about conjunction. 12.0
Counting with reference to events
The events antecedently described in a NP restrict the interpretation of a matrix clause under adverbialization. The analysis of number words appearing in a NP includes an event of measurement eμ, a protocol for counting events that partly determines the events referred to and their neighborhood. An interaction can thus be expected between counting and the interpretation of the matrix clause. In a context for (19) in which on a single flight 100 passengers each ordered one cocktail and paid by credit card and 80 passengers each ordered one cocktail and paid cash, the sentence is unremarkably true: (19) Most passengers who ordered a cocktail in flight paid for it by credit card. Suppose instead in an altered context passengers are divided among four flights last week, on each of which 25 passengers each paid for their cocktail by credit card and 20 passengers each paid cash for it. The sentence remains true, 100 to 80, so it would seem. Imagine next it turns out that it is the same 25 passengers paying by credit card, each flying four flights, but the passengers who pay by cash are 80 different people. All the same, the salient interpretation of the sentence remains true. If the counter clicks as a passenger orders a cocktail, payment in the local neighborhood of that order determines whether it clicks for an event that is confirming or disconfirming for (19). Counting, ordering, and paying are all from a frame of reference that tracks events as they unfold, concluding in a generalization (19) true or false about what happened in the observed history. To count otherwise, to count persons, is to click in slow motion, once per (maximal) event of being a passenger that coincides with the counted passenger’s lifetime. To choose to so count in uttering (19) is to have in mind some relation between a lifetime and ordering a cocktail in flight. Granted some such relation, for (19) to apply to the 25 passengers who each ordered a cocktail four times, the matrix clause must contain a plural event pronoun, as
604
Chapter 12
above, to assert that in each of them, the cocktail ordered is paid for by credit card. One context that favors counting a lifetime qua passenger is a longitudinal study of travel behavior. Most people, (19) concludes, have been of a disposition that whenever they were a passenger ordering a cocktail in flight, they paid for it by credit card. To this conclusion, the last scenario is a counterexample since only 25 out 105 people are that way. It presents a problem of scale to count so slow and large and expect the counted events to frame or locate ephemeral incidents.1 Of course dispositional behavior manifests itself in ephemeral incidents, and it takes a longitudinal point of view to say anything quantitative about a disposition’s prevalence in the population. That is, here is a context—even if it is not the first that comes to mind—that answers the problem of scale posed by the semantics of counting, adverbialization, and reference to events. If counting passenger events that are no longer than a flight comes to mind hearing (19) out of the blue, it may just be that its protocol for counting is on a scale with a more obvious fit to locate ordering and paying for cocktails, behaviors conditioned by one’s incidental experience as a passenger. At the scale of a longitudinal study, cocktails are usually not landmark events in the lifetime under study. The quantifier under consideration, most, counts. Under analysis (Hackl 2003, 2009), “Most Φ Ψ” says that the n-many Φ that Ψ outnumber the m-many Φ that do not Ψ (see chapter 13, note 9). That is, they all contain the counting to n exemplified earlier (section 12.1) in the logical form for 3000 passengers: (20) [∃X : ∃E(∃eμ count[eμ,E,3000] passengers[E,X])] … (21) [∃X : ∃E(∃eμ count[eμ,E,n]) Φ[E,X])] Ψ Three thousand passengers counts to 3000 some events of there being a passenger, under a protocol for a domain of events where these can be counted. Similarly, most passengers counts events of there being a passenger. The events counted under the chosen protocol are at the same time subject to all the same sequence-of-events conditions found (sections 11.0–11.1) to govern simple cardinals such as 3000 passengers. What is invariant about [DPQ NP], what is fixed by the syntax and semantics of a Determiner in construction with NP, is that what is counted are (sub)events described by NP. The protocol is always constrained by the syntax and semantics of most passengers to count passenger-ings. In this last respect, there is thus a robust and invariant formal distinction between the nominal quantifier most and the adverb usually, which counts orderings in (22)–(23) as a matter of its syntax and semantics (see chapter 10, note 31): (22) Usually a passenger who ordered a cocktail in flight paid for it by credit card. (23) Usually if a passenger ordered a cocktail in flight, she paid for it by credit card.
Measuring Events
605
As a sentence that favors event counting, (19) threatens a dilemma to the extent that it resembles sentences illustrating the “proportion problem” for descriptive anaphora (see Richards 1984, Ludlow 1994, and the references cited there). Assuming it to be such a sentence, it has been denied that (19) has an event-counting reading in the course of arguing that (19) and (22)–(23) are not synonymous. Sentences (22)–(23), with an adverb counting events, are true in the last scenario, counting passenger events, but (19) is said to be false, necessarily counting persons. Citation instances of the proportion problem include the following: (24) Most farmers who own a donkey beat it. (25) Usually a farmer who owns a donkey beats it. (26) Usually if a farmer owns a donkey, she beats it. In the same context mutatis mutandis, 25 farmers each own four donkeys and beat them all, but 80 farmers each own one and do not beat it. The observation is: first, (25) and (26) have an interpretation true in this context, counting events or states in which a farmer owns a donkey, and where beating outnumbers small mercies 100 to 80; and second, (24) is obligatorily false as the 25 farmers are a cruel minority, 25 to 105. I concur in this observation while maintaining the contrary for (19)–(23). The contrast between most passengers in (19) and most farmers in (24) turns on the pragmatic difficulties inherent in finding appropriate protocols for counting NP-ings that in turn meet the conditions on sequencing events, as just illustrated contrasting alternative interpretations of (19). Counting, each counter click clicks for an event of there being a farmer in (24) discrete from any other event counted. If the events in mind each coincide with a farmer for the duration of his lifetime, then they are as countable, as discrete, and as numerous as the farmers themselves. So counting delivers the interpretation of (24) that is false in the imagined scenario and reported to be its only one in discussions of the proportion problem. To count otherwise, to count some other events of there being a farmer, demands an alternative protocol of when, where, and for which events of there being a farmer to click the counter. A protocol is sound measurement only if it be known that it will not count twice the same event of there being a farmer. For the farmer who owns, say, two donkeys, Rucio and Platero, which of the many events of there being him is picked out by owning Rucio and which by owning Platero, and what sound protocol is thereby made available? Even if this farmer first owned Rucio and next owned Platero so that his life with Rucio and his life with Platero are distinct events, no sound protocol derives from this observation. For any farmer any of whose donkeys have been owned simultaneously, the counter will click several times for the same event of there being a farmer, failing then to properly count out events of there being a farmer. These remarks flag the hazards attempting to count farmer-ings other than those that coincide with farmers. Without a protocol in mind that escapes them, speakers reject event counting so-called in (24).
606
Chapter 12
As dependent as counting is on the intended protocol, the scaling and framing of the events described, and the conditions on sequencing events, there are contexts ready to disrupt (19)’s glib event counting. In the first context, 180 passengers on the same flight each order one cocktail—a context where, in fact, event counting and object counting coincide. In the second context clearly favoring event counting, the passengers are divided among four flights, and 100 passengers turn out to be the same 25 persons who each order one cocktail on each of the four flights. In a third context, they are now all back on the same flight, where 25 passengers each order four cocktails and pay by credit card, and 80 passengers each order one cocktail, paid for in cash. In such a context, how accessible are the object-counting interpretation, which is false 25 to 80, and the event-counting interpretation, which is true 100 to 80? Unlike the passengers on separate flights, event counting in this context risks confusion of the events counted. Event counting is encouraged if it is also known, for example, that every cocktail is paid for when ordered (or served), and every passenger orders and pays for cocktails serially, one by one. Correlatively, event counting is discouraged if it is known that passengers only settle their tabs at the gate after disembarking. That the proximity of ordering and paying matters recalls the contrast in event counting between (67) and (69), between (72) and (75), and between (96) and (98): (67) (The) 3000 ships that passed through the lock arrived in the port of Rotterdam 3 hours later. (69) #(The) 3000 ships that passed through the lock are now in the port. (72) The 3000 ships that passed through the lock last year were registered in Panama. (75) #The 3000 ships that passed through the lock last year are registered in Panama. (96) Three thousand passengers had (3000) opinions about their flights last year. (98) #Three thousand passengers have (3000) opinions about their flights last year. Recall from section 11.0.0 that sufficient context and counting by other means reclaim even the most recalcitrant cases for event counting. For (19), suppose that the sale of every cocktail and method of payment is recorded, and the speaker utters (19) examining the ledger.2 The event-counting reading is now the one the speaker must have intended. The variability and context sensitivity that attaches to event counting derives from the problem of finding which events to count of there being a passenger, a problem posed by the syntax and semantics of most passengers and 3000 passengers. As remarked above, the adverb usually counts other events, those denoted by a verb, and thus it is rather these other events that must be discrete and countable. In that case, an event counted need not coincide with an event of there
Measuring Events
607
being a passenger, as the following examples illustrate (with union workers replacing passengers): (27) Last year, most union workers who operated an assembly line assembled more than one aircraft. (28) Usually, last year, union workers who operated an assembly line assembled more than one aircraft. (29) Usually, last year, if union workers operated an assembly line, they assembled more than one aircraft. The union that once numbered itself in the thousands is now 100 rank-and-file members, who have worked many sites last year. There were 90 assembly lines each of which assembled several aircraft and was operated by 10 union workers, so that every union member worked more than one of these lines. There were also 100 assembly lines each of which assembled one aircraft operated by a solitary worker, and every union member worked one of these, too. Given this work history, the union workers who worked multiple-operator multiaircraft assembly lines are identical to the union workers who worked single-operator single-aircraft assembly lines (i.e., the entire union membership). The single-aircraft assembly lines, however, outnumber the multiple-aircraft assembly lines, falsifying (28) and (29). The events of operating an assembly line that the adverb usually counts coincide with the assembly lines themselves employing varying numbers of union workers. The object-counting interpretation of (27), counting events of there being a union worker coextensive with a worker’s collaborative history last year, is true without much comment, since every worker has worked more than one assembly line. The event-counting interpretation is also true that counts events of there being a union worker spanning only his time on a single site. Last year, a union worker on an assembly line was more likely to find himself working on the production of more than one aircraft, the odds being 9 to 1. That is, last year most assembly-line jobs were on multiple-aircraft lines. Every interpretation of (27) is true in this context, confirming, as discussions of the proportion problem conclude, that (27) is not equivalent to (28) or (29). Despite its own event counting, (27) counts (sub)events of there being a union worker, as its syntax dictates, and not the events of operating an assembly line that usually counts. 12.1
Numerals
Under dark and heavy thoughts of extensionality and individuation, who could escape the metaphysical portent of number?3 As essential and intrinsic a property as identity itself, that which is 3,000,000 in number is not that which is not: (30) (3,000,000(X) & 1,000,000(Y)) → X ≠ Y
608
Chapter 12
If, as granted, a sentence such as (31) ‘Three million passengers …’ is truly ambiguous between object counting and event counting, and yet its logical forms all begin ‘[∃X: 3,000,000(X) …] …’, it is already decided that these descriptions describe different things from one reading to the next. All that could remain is to contrive from passengers and anything else that can be shoehorned into the NP two descriptions, only one of which describes passengers while the other describes something else 3,000,000 in number. (31) Three million passengers crowded National Airlines routes in 1980. (32) (31) ⊬ Three million frequent fliers crowded National Airlines routes in 1980. (After Gupta 1980, 23; Moore 1994) Without denying a primitive, pristine concept of number, we may wonder whether to parse the natural language words three million as containing its attribution, as is assumed by ‘[∃X: 3,000,000(X) …] … ’. Perhaps instead arithmetic predicates introduce an explicit relation to events of measurement or counting, passengers counted (now) at 3,000,000, and the ambiguity emerges from how the measuring events are thought of, without metaphysical implications for the things referred to. As there is no equivocation about the truth of (33), there is no ambiguity in the reference of its definite descriptions, which refer to the same things no matter how they are counted: (33) The three million passengers who crowded National Airlines routes in 1980 were the one million frequent fliers loyal to it. Rather, number words might join other measure phrases (26 inches in 1988, 71 inches in 2014) in denoting the result of a dated event of measurement, with a logical form more like (36) and pointedly unlike (35): (34) Aaron was 26 inches in 1988 and is 71 inches in 2014. (35) *∃e(Past(e) & 26 inches(a) & In(e, 1988)), and ∃e(Present(e) & 71 inches(a) & In(e, 2014)). (36) a. ∃e(Past(e) & 26 inches(e, a) & In(e, 1988)), and ∃e(Present(e) & 71 inches(e, a) & In(e, 2014)). b. ∃e(Past(e) & Participate(e, a) & 26 inches(e) & In(e, 1988)), and ∃e(Present(e) & Participate(e, a) & 71 inches(e) & In(e, 2014)). Translating n million as “now counted to n million,” the same things once counted as passengers to number three million may then be recounted under a different protocol as frequent fliers to number one million. As much has to be said just to resolve the apparent contradiction that simple predicate arithmetic, “n million(X),” gets us into when confronted with sentences of event counting (31), object counting (32), and the identities that relate the two.
Measuring Events
609
It will allow us to affirm (33). The same things X, measured once as three million, are measured again as one million: (37) [∃X: ∃E 3,000,000 passengers[E, X]][∃Y: ∃E 1,000,000 frequent fliers[E,Y]] X=Y Some passengers counted 3,000,000 are identical to some frequent fliers counted 1,000,000. (38) *[∃X: 3,000,000(X) & passengers…(X)][∃Y: 1,000,000(Y) & frequent fliers…(Y)] X = Y Given that the passengers are the frequent fliers, that is, (39) ∃E 3,000,000 passengers[E, A] & ∃E 1,000,000 frequent fliers[E,A], counted must be the events E3 of passenger-ing and events E1 of frequent flier-ing: (40) 3,000,000 passengers[E3, A] & 1,000,000 frequent fliers[E1,A]. I.e., (41) 3,000,000[E3] & passengers[E3, A] & 1,000,000[E1] & frequent fliers[E1,A]. But as much as this is true to the meaning of number words and to the semantic innocence of the referring expressions they modify, identity statements such as (33) parsed as in (37) if left without further revision become subject to paradoxes of extensional substitutivity (sections 8.1–8.2, 10.1–10.3): (42) (The) three million passengers who crowded National Airlines routes in 1980 were frequent fliers. (43) (The) one million frequent fliers who were loyal to National Airlines in 1980 are still frequent fliers. (44) (33), (43) ⊦ #(The) three million passengers who crowded National Airlines routes in 1980 are still frequent fliers. Sentences (33) and (43) ought to felicitously entail (44), as felicitous as the entailments in (45)–(47): (45) As befits a clever disguise, Superman does not resemble Clark Kent. Superman is Clark Kent. #Therefore, Superman does not resemble Superman. (46) The man surpasssed the boy of his childhood in regrets and not much else. This man is that boy (grown up). #Therefore, the boy of his childhood surpassed the boy of his childhood in regrets and not much else. (47) A careless camper killed the unconscious drunk passed out in bed. The unconscious drunk was the careless camper. #Therefore, an unconscious drunk passed out in bed killed the careless camper.
610
Chapter 12
Adverbialization has explained away these substitutivity puzzles in that the events the derived adverbs denote remain distinct, despite the identity of their participants: (48) Superman while Superman does not resemble Clark Kent while Clark Kent. F Superman while Superman does not resemble Superman while Superman. (49) The man while a man surpasssed the boy of his childhood while a boy of his childhood in regrets and not much else. F The boy of his childhood while a boy of his childhood surpasssed the boy of his childhood while a boy of his childhood in regrets and not much else. (50) A careless camper while a careless campler killed the unconscious drunk passed out in bed while an unconscious drunk passed out in bed. F An unconscious drunk passed out in bed while an unconscious drunk passed out in bed killed the careless camper while a careless camper. Crucially, what is done while Superman is not coincident with what is done while Clark Kent, and similarly for what happens while a man and while a boy and for what befalls one while a careless camper and while an unconscious drunk. Only if number words denote events as suggested above can they in turn be recruited as spatiotemporal adverbs to join the common explanation that adverbialization provides for the substitutivity puzzles: what happens to them while three million is not what happens to the same while one million. When so recruited, the infelicity of (44) is seen to derive from the failure of while three million to denote present events: (51) #(The) three million passengers who crowded National Airlines routes in 1980 when (counted) three million … are still frequent fliers. (52) (The) three million passengers who crowded National Airlines routes in 1980 when (counted) three million … were frequent fliers. Counting the passengers at the gate, boarding, or on board, (52), they were all then frequent fliers when they surrendered their tickets. There is no counting to three million that counts them as they now are, which according to (51), is what defeats (44), violating sequencing of events so-called, which has been the preoccupation of section 11.0. About event counting in (31) and the like, Moore (1994) has made two important points. His first point is to dissociate the events counted or measured from a false identification with the reported events, contrary to what a quick glance at (2) and subsequent literature suggest: (2) Four thousand ships passed through the lock last year. (Krifka 1990)
Measuring Events
611
(53) Three hundred thousand elite Soviet troops streamed past the reviewing stand in 6 hours. (54) Much Soviet military hardware surged through the square during the parade. (55) Three thousand locusts mobbed autographs from Marilyn Monroe at the premiere of Gentlemen Prefer Blondes. (56) Three thousand callers flooded the NBC switchboard in the first 5 minutes of The Tonight Show when Julia Roberts was the guest. (57) Thirty thousand mosquitoes raked/strafed/blasted/pummeled/pounded an exposed arm in 3 minutes. (58) In the whirlwind, three billion hailstones had washed over/cascaded above the Arctic base camp in the first hour of the perfect storm. (59) In the whirlwind, much ice had washed over/cascaded above the Arctic base camp in the first hour of the perfect storm. As with all examples where the event and body counts disagree, the bodies in (53)– (59) recirculate through the events recounted. In a Potemkin-village display of Soviet strength, the parade in (53)–(54) is a circle that turns back just out of sight of the reviewing stand. The various bloodsuckers in (55)–(57) always go back for seconds and thirds, and the storm in (58)–(59) is a closed system in which the same material repeatedly strikes the Arctic camp. In each case, it is clearly understood what besides bodies is to be counted. In (53)–(54), it is every appearance of a troop or piece of hardware in front of the reviewing stand; in (55)–(57), every request for an autograph, every call to the switchboard, and every mosquito bite; and in (58)– (59), every strike of a hailstone near the Arctic camp. What is new to these examples is that they show that the events counted are not to be identified with the larger, collective event that the matrix describes them to be part of. The 300,000 counted in (53) are not 300,000 streams; there are not 3000 mobs (55), 3000 floods (56), nor 30,000 strafings or pummelings (57); and there are not three billion washes or cascades (58). A dissociation between the events measured and those subsequently related is to be expected in the present setting under supermonadicity. Recall that the adverbialization of logical form has introduced adverbial quantification relating, via a neighborhood relation, the events E described by NP to E0 described in the rest of the sentence Φ: (60) [∃X : ∃E (3,000,000[E] passengers[E,X])] [℩E : 3,000,000[E] passengers[E,X]][∃E0 : N[E,E0]] Φ[E0] … But recall from sections 11.0.0 and 11.0.2 that the expansion of Φ according to supermonadicity, as in (62), puts the E0 a long way off from the events that verbs
612
Chapter 12
denote, the stirring-up in (61) or the streams, surges, mobs, floods, blasts, washes, and cascades of (53)–(59): (61) Three thousand butterflies visiting Kinkaju-ji in the spring stirred up a fall tornado in the Golden Gate. (62) DP-W-O-Agent-Tense-Cause -DP-W-O-Patient–stir-up [∃X : ∃E 3000 butterflies … [E,X]] [℩E : 3000 butterflies … [E,X]] [∃E0 : N[E,E0]] W[E0,X] O[E0,E1] ∃XAgent[E1,X] Past[E2] Cause[E1,E2] … stir-up[E2] … Whatever conditions adverbialization imposes on E0, these are conditions on the local events or actions in which the things X referred to participate. What collective predicates then do of course is to report that these perhaps microscopic actions have a very large effect (see E2 in (62)), such as a stream, surge, mob, or flood, etc. It matters little that (53) constitutes the 300,000 events as a stream rather than as so many goose-steps: (63) Three hundred thousand elite Soviet troops goose-stepped past the reviewing stand in 6 hours. Moore’s (1994) second point about event counting is that a number word or measurement phrase within a NP is sufficient on its own to invoke an intended protocol of measurement drawn from the scientific experience that is common ground, as is implied if number words themselves denote events of counting, as suggested here. To make it vivid, 300,000 immediately is 300,000 clicks on a counter (such as large venues use to measure attendance). For what is asserted—that the counter clicked to 300,000—to have significance as a measurement, it must be understood for what the counter clicks. In (53), it is neither for stream nor troop body x, ∃e elite Soviet troop(e,x); rather, the counter clicks for certain events e of elite Soviet troopery, ∃x elite Soviet troop(e,x), which the current context and scene fix to be those before the viewing stand: 300,000[E] & elite Soviet troops[E,X] (see (41)). In arguing for NP as the locus for describing the measuring events, Moore (1994, 589) observes that further modification, different, eliminates any event count that disagrees with body count: (64) Three hundred thousand different elite Soviet troops streamed past the reviewing stand in 6 hours. (65) Three hundred thousand different elite Soviet troops goose-stepped past the reviewing stand in 6 hours. (66) Four thousand different ships passed through the lock last year. The intrusion of different blocks counting as more than one, any events with the same participant x of being an elite Soviet troop (or of being a ship)—a strange
Measuring Events
613
effect for an adjective inside the NP if event counting were the business of counting events the verb denotes, goose-step(e) past the reviewing stand or pass(e) through the lock. Rather, however the internal semantics of comparative adjectives should turn out (see Alrenga 2010), think of different ships as “ships differing (from each other)”— that is, as expressing a relation between ships and events that distinguish them: (67) different[E,X] → W[E,X] & ∀e0∀e1((Ee0 & Ee1 & ¬e0=e1) → ¬∀x(W(e0,x) ↔ W(e1,x))) (68) [∃X : ∃E(4000[E] different[E,X] ships[E,X])] … If, as in (67), that relation at least implies that the participants in distinguishing events are distinguished, it suffices to force event count and body count to coincide. Since the logical form of 4000 ships is invariant in all its occurrences, there are always events counted to 4000 even when events are not in plain sight: (69) Four thousand ships are more than 50 meters long. Every event e counted is one in which exactly one ship participates, ∃x ship(e,x), and when event counting is allowed, there can be many events counted per ship. It is, however, always a condition on counting or measurement, that the measure be determinate and that speaker and hearer understand the process that makes it so. If the ships are only a thousand and (69) is uttered out of the blue, then what measures 4000 rather than four million or infinitely many, as there are as many such events in which ships participate? Without more precise instruction from context, one defaults into counting those maximal events or states coinciding with ships, their lifetimes, which are as countable and determinate in number as the ships themselves. The view then is that the 4000 ships is parsed unambiguously as (68). There is never a “4000(X)” that applies a cardinal predicate to something other than events. It is just that—more often than not, I guess because it needs no special prompting from context—the events counted coincide with the lifetimes of their participants. The formal point the above argues for has been that three million passengers is as in (70), where what is counted, events of being a passenger, is not what the indefinite description denotes: (70) [∃X : ∃E (3,000,000[E] passengers[E,X])] … This dissociation of count morphology from nominal denotation returns with a vengeance in section 12.2 in a discussion of singular plurals such as a perfectly synchronized twelve traffic lights. Here, granting the formal point resting on (70), how is it that the same things X described as ∃E passengers[E,X] have their events of their being passengers counted one way under certain conditions of measurement
614
Chapter 12
and then counted another way under others? For the single passenger p on a single 5-hour flight, there is of course the 5-hour event e5hr such that passenger(e5hr, p), but there are also uncountably many dense events e, fragments of e5hr, that are events, passenger(e, p), of p being a passenger at e. Counting passengers is thus just like counting motions, events, landmasses, breezes, fires, custards, times, regions, areas, etc., which manages to impose finite countability on an intrinsically dense and uncountable domain. How this is done will remain speculative, but that should not disturb recognition that the problem of counting motions, events, etc., is far more pervasive—however it is eventually solved—and, in fact, takes in counting anything, passengers, ships, Soviet troops, statues, etc. What follows will suggest that 3,000,000[E] is to be further analyzed so that three million passengers comes out as in (71) (section 9.4.2), or its pluralization in (72) (section 9.4.2.0): (71) [∃X : ∃E (∃eμ count[eμ, E, 3,000,000] passengers[E,X])] (72) [∃X : ∃E (∃Eμ count[Eμ, E, 3,000,000] passengers[E,X])] In support of the analysis, there will be arguments that three million is, as first paraphrased above, “now counted to three million,” expressing a relation (section 12.1.1) between an episodic, indexical event eμ of measurement (section 12.1.0), the counting, and the events E (of being a passenger) counted. Recall from section 9.4.2 that an event eμ of counting ζ occurs within a scene in which ζ appear countable: (73) count(eµ, ζ ) → ∃sɶ(∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) & countable[ sɶ, ζ ]) (74) countable[ sɶ,ζ ] ↔ ∃sɶ ℜ[ sɶ, ζ , sɶ ] Such a scene resolves the ζ to be counted under a reticule fixing them within their own addresses: (75) ℜ( sɶ, ζ , sɶ ) ↔df ∃ξ(∀x∀y((ξx & ξy & x ≠ y) → ∃t [ sɶ(t ), x, y, sɶ ]) 4 & ∀x(ζx → ∃y(ξy & overlap(x,y) & ∀z((ζz & overlap(z,y)) → z = x))) & ∀x(ξx → ∃y(ζy & overlap(x,y) & ∀z((ξz & overlap(z,y)) → z = x))) (76) ℜ[ sɶ, ζ , sɶ ] ↔ df ℜ( sɶ, ζ , sɶ ) & ∀sɶ ′((ℜ( sɶ, ζ , sɶ ′) & ∀α sɶ (sɶ α sɶ → sɶ ′α sɶ )) → sɶ ′ = sɶ ) An event of counting n-many ζ fits them under a reticule of n-many addresses, with a standard definition of n-many: (77) For ordered pairs θ, injective[θ] ↔df ∀x∀y∀z(((θ() & θ()) → y=z) & ((θ() & θ()) → y=z)) (78) card[X,Z] ↔df [∃θ: injective[θ]](∀x(Xx → ∃z(Zz & θ())) & ∀z(Zz → ∃x(Xx & θ())) (79) card[X,n] ↔df [℩Z: N(n) & ∀z(Zz → (N(z) & z < n))] card[X,Z]
Measuring Events
615
For visual counting, ‘count[eμ, ζ, n]’ is the defining relation among an event of measurement, the scene or conditions of observation for that measurement, what is measured and the number measured (see section 9.4.2): (80) Visual counting count[eμ, ζ, n] ↔df ∃sɶ∃sɶ (∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) & ℜ[ sɶ, ζ , sɶ ] & card[sɶ, n]) If three million passengers is to be parsed as in (71), we are not quite ready to let loose unsupervised the combination of ‘passengers[E,X]’ with ‘count[eμ,E, 3,000,000]’. The hesitation arises facing a pas de deux, a scene that is unambiguously of one dance and two dancers: (81) [∃X : ∃E (∃eμ count[eμ, E, 2] dancers[E,X])] two dancers (82) [∃X : ∃E (∃eμ count[eμ, E, 1] dance[E,X])] one dance The event that counts as one dance must not, equally an event of being dancers, also count as one dancer. It is straightforward enough to define solo events and to understand that in counting dancers or passengers, one is counting events of being a dancer or being a passenger in which one is alone: (83) solo[E] ↔df ∀e∀x0∀x1(Ee → (participate(e,x0) → (participate(e,x1) → x0=x1))) Such an understanding must not of course intrude on counting dances, pas de deux, that are not solo performances, for which reason I assume it emerges from the meaning of the agentive morphology, dance + er: (84) dancers[E,X] ↔ solo[E] & participate[E,X] & ∃E′(overlap[E,E′] & dances[E′]) If it is granted that passengers, dancers, etc., denote only solo events, it is true to the meaning of three million passengers that such solo events be what is counted to three million and the nominal phrase parsed as in (71) counts exactly them. As first mentioned in section 9.4.2.0 discussing the practical reasoning below, the meaning of its conclusion calls for counting that is the summation of two measurements, each conducted at boarding: (85) Thirty passengers on American Airlines 100 departing JFK at 6:00 p.m. needed special meals. (86) Thirty passengers on American Airlines 104 departing JFK at 8:00 p.m. needed special meals. (87) Therefore, sixty passengers (on AA 100 and 104) needed special meals (on their flights). Such examples prompt the choice of (72), the pluralized version, to represent the canonical logical form for cardinal predicates, with the understanding that further
616
Chapter 12
restriction ‘sg[Eμ]’, singular morpheme, is present if a single event of measurement is intended. The summation of measurements is defined to guard against counting anything twice (see section 9.4.2): (88) a. sum(eμ) = n ↔df ∃ζ count[eμ, ζ, n] b. ¬Eμeμ → (∀e′μ(E′μe′μ ↔ (Eμe′μ ∨ e′μ = eμ)) → sum(E′μ) = sum(Eμ) + sum(eμ)) (89) count[Eμ, ζ, n] ↔df n = sum(Eμ) & [∀eμ : Eμeμ][∃ζμ : [∀v: ζμv] ζv] ∃n count[eμ, ζμ,n] & [∀v: ζv][∃ζμ : ζμv & [∀v: ζμv] ζv][∃eμ : Eμeμ] ∃n count[eμ, ζμ,n] & [∀v: ζv][∀eμ: Eμeμ][∀e′μ: Eμe′μ][∀ζμ : ζμv & [∀v: ζμv] ζv] [∀ζ′μ: ζ′μv & [∀v: ζ′μv]ζv]((∃n count[eμ, ζμ,n] & ∃n count[e′μ, ζ′μ,n]) → eμ = e′μ) 12.1.0
Syntactic evidence that number words denote events of counting
Measurements are dated events. Numerosity as counts are, too. So much is essential to the conceit that number words serve as adverbs—three million as “(now) counted to three million”—and it may survive whatever else has been suggested above about counting in a dense domain. The episodic content of number words finds some unexpected corroboration in the syntax of prenominal modification. In Bolinger 1967 (see also Larson 1998; Larson and Marušič 2004; Larson and Takahashi 2007), it is remarked that there is no contradiction in (90)’s report that the stars listed in an atlas of visible stars were invisible that night: (90) On Mt. Wilson that night, some invisible visible stars hung low on the horizon shrouded in the Los Angeles smog. Sentence (91), permuting the adjectives, cannot make the same report, since the adjective interpreted with respect to the context of utterance must occupy a position peripheral to the one that is not indexical: (91) #On Mt. Wilson that night, some visible invisible stars hung low on the horizon shrouded in the Los Angeles smog. Thus, (91) says that stars visible that night are in principle invisible. Without contradiction threatening, (92) is three-ways ambiguous: (92) On Mt. Wilson that night, the telescopes missed some audible, visible stars. It may be understood to describe stars among those cataloged with radio and visiblespectrum emissions, or those with such emissions detectable that night, or those with radio detectable from Mt. Wilson that night and cataloged in the atlas of visible stars. It cannot be understood to describe stars cataloged for radio emissions and
Measuring Events
617
visible that night. That is, a nonindexical adjective does not occur to the left of an indexical adjective, which (93) corroborates: (93) On Mt. Wilson that night, the telescopes missed some audible, invisible visible stars. The sentence is false unless the stars missed were audible that night. To avoid contradiction from “invisible, visible,” one understands “invisible that night, visible as cataloged,” and then any other adjective further left is also forced to be indexical. The above observations carry over in (94)–(96), replacing some in some … stars with seven. As before, all adjectives left of an indexical adjective are themselves indexical: (94) On Mt. Wilson that night, the telescopes missed seven visible stars. (95) On Mt. Wilson that night, the telescopes missed seven audible visible stars. (96) On Mt. Wilson that night, the telescopes missed seven invisible visible stars. Sentences (97)–(101) prompt an apparently new observation, namely, that adjectives to the left of the cardinal seven are also indexical. Sentence (97), for example, can only be understood to report that the missed stars were visible that night, and (101) cannot escape the contradictory implication that the stars were both visible and invisible then: (97) (98)
On Mt. Wilson that night, the telescopes missed a visible seven stars. On Mt. Wilson that night, the telescopes missed an audible seven visible stars. (99) On Mt. Wilson that night, the telescopes missed an invisible seven visible stars. (100) On Mt. Wilson that night, the telescopes missed an audible visible seven stars. (101) #On Mt. Wilson that night, the telescopes missed an invisible visible seven stars. If seven attributes that number property that is as immutable as the stars themselves, the new observation does not fall under the first, and seven is found to deviate from the attribution of other immutable properties. In contrast to (101), (102) does not force a contradiction: (102) On Mt. Wilson that night, the telescopes missed some invisible visible gaseous stars. On Mt. Wilson that night, the telescopes missed some invisible visible binary stars. On Mt. Wilson that night, the telescopes missed some invisible visible mutually attractive stars. On Mt. Wilson that night, the telescopes missed some invisible visible gravitationally entangled stars.
618
Chapter 12
In contrast and also in accord with the first observation, predicates that must be interpreted indexically with respect to the frame of reference at Mt. Wilson that night again force a contradiction: (103) #On Mt. Wilson that night, the telescopes missed some invisible visible rising stars. #On Mt. Wilson that night, the telescopes missed some invisible visible Eastern stars. #On Mt. Wilson that night, the telescopes missed some invisible visible left-drifting stars. The effect of seven on interpretation to its left can be made to fall under the first observation and fit the pattern in (103) if cardinal attributions are themselves understood to invoke an indexical frame of reference, glossing seven stars along the lines of “stars measuring (then) at 7” or “stars counted (then) at 7.” 12.1.1 Semantic evidence that number words denote a relation between an event of counting and the events counted
Some further confirmation that words denote a relation between an event of counting and the events counted is found in a class of predicates that Schwarzschild (2008, 2009) has identified, stubbornly distributed predicates or stubs, which refuse to apply collectively to what plurals refer to.5 Sentence (104), containing the stub large, applying only to the individual violet, is false if each is small despite a large bunch of 100, which in (105), not containing stubs, is said to weigh 1 pound and cost $100, much more than any single violet: (104) 100 violets are large. (105) 100 violets weigh 1 pound and cost $100. Similarly, if each interview was short, the stub long in (106) does not allow it to be understood that their entire series was nevertheless long: (106) The interviews were long. In a singular plural (see section 12.2), in the position between the article an and a numeral, it has been observed that a modifier denotes a single event in which participate collectively the many NP, which describes a position where stubs should not occur. Yet they do in (107) and true to this position apply collectively, in contrast to their tokens in both predicative position (108) and in attributive position to the right of the numeral (109): (107) a. A long 10,000 popsicle sticks were lined up end to end and called art. b. I sat through a long five skits. c. Many a long five skits are performed without intermission.
Measuring Events
619
(108) a. F Ten thousand popsicle sticks that were lined up end to end were/are long. b. F Five skits I sat through were long. c. F (Many) an uninterrupted five skits are long. (109) a. F A lined-up 10,000 long popsicle sticks were called art. [Note: Not read “10,000-long”] b. F I sat through five long skits. c. F (Many) an uninterrupted five long skits are performed Off-Broadway. The event the stub describes in this position must frame the matrix event, as it does in (107c) and (110), the being long measuring out the performance or filming, in minimal contrast with (111), where it makes no sense that publication is as long: (110) Many a long five skits are filmed for uninterrupted broadcast / without interruption. (111) *Many a long five skits are published as a single play / in a single volume. (112) Cf. Many an unrelated five long skits are published in a single volume. That there is some context where the stub long is not stubbornly distributive, alongside contexts where it is, defeats any suggestion that a monadic predicate ‘long(x)’ is stubbornly distributive over the things x. If that is its meaning, that’s its meaning anywhere and everywhere it is tokened. If a simple cardinal predicate three really denotes a measuring event as if to say “to be counted three,” then the same ought to be said for long, that it is really “to be measured long,” ‘long[eμ,ζ]’. Here it does not matter whether the ζ measured long are sticks or events of being sticks. At issue is reference to an event eμ of measurement in which the length of ζ is taken. A measurement is of course a telic event and so to make a stative out of it—as do the predicative usage, They are long, and the inner attributive usage ascribing an intrinsic property, 10,000 long popsicles—is to say that they more or less always—consistently—measure long. In demonstrative or indexical position, however, only a single telic event is engaged: (113) A now/then measured long 10,000 popsicle sticks are lined up end to end. (114) Ten thousand popsicle sticks that were lined up end to end measure long. [Note: The past-tense were long in (108a) is the past-tense counterpart of stative are long, not a report of an episodic, telic event of measurement but of a persistent past state of always measuring long.] (115) A lined-up 10,000 measuring long popsicle sticks were called art. That is to say that in predicative and inner attributive positions, long has a modal flavor with respect to the protocols of measurement, intending report of a property that is persistent, constant, or at least replicable across various conditions of
620
Chapter 12
measurement within some implicit deviation. But they—in the plural—are still them under various conditions such as scattering the popsicles, where that linear, collective measurement cannot be repeated: there is not a length that is long in that scatter. In contrast, intrinsic individual popsicle lengths (what the stub does apply to) are preserved, no matter how the popsicles might be displaced, and measured the same way no matter the collective arrangement. Correlatively, a nonstub, weigh 1 pound or cost $100, as a measurement of a collective does not measure that collective by a method so fragile and sensitive to the exact position or alignment of the things collectively measured. (116) 100 violets are large. (117) A bunch of 100 violets is large. Again, (116) cannot say that 100 violets are large just in case they are in a bunch, since there would not be a bunch to measure letting the violets scatter. But a bunch of 100 violets remains a bunch just as large, under any displacement of that bunch, as (117) requires. In the outer position between article and numeral, (107), a single event of measurement is denoted with no implication of a similar result under alternative conditions, and thus stubs lose their stubness. As these remarks imply, stubness is not intrinsic to the lexical items, long and large, but a consequence of their meaning and their interaction with the aspect of the phrase in which they occur, for which they must be ‘long[eμ,ζ]’ and ‘large[eμ,ζ]’, not *‘long(x)’ or *‘large(x)’.6 12.2
Singular plurals: [A(n) AP k NP.PL], [A(n) (AP) NP.SG and NP.SG]
The Eventish apparatus—adverbialization and supermonadicity, scenes and frames of reference, and so on—has rescued a naive understanding of and, always the sentential connective, and of the things that nominal descriptions refer to. Despite their effects on the truth conditions of the host sentence, the descriptions the left skater and right skater and the skaters refer to the same two skaters, and the ill-matched pitcher and catcher and the batterymates are, as one might have hoped, merely alternative descriptions of the same two ballplayers. (In baseball, a pitcher and catcher pitching and catching to each other are said to be batterymates). Yet if the ballplayers are more than one, what is there one of in (118)–(121) to warrant the singular article a(n)? (118) a. An (ill-matched) pitcher and catcher are starting for the Red Sox tonight. b. *An (ill-matched) pitcher and catcher is starting for the Red Sox tonight. (119) a. An ill-matched two batterymates are starting for the Red Sox tonight. b. *An ill-matched two batterymates is starting for the Red Sox tonight.
Measuring Events
621
(120) a. An ill-matched seven infielders and outfielders are starting for the Red Sox tonight. b. *An ill-matched seven infielders and outfielders is starting for the Red Sox tonight. (121) a. An ill-matched seven fielders are starting for the Red Sox tonight. b. *An ill-matched seven fielders is starting for the Red Sox tonight. The plural reference of these indefinite descriptions makes number agreement on the verb obligatorily plural, prompting the conclusion that a(n) makes itself felt only within the description, as argued both by Ionin and Matushansky (2004a, 2004b, 2006) in discussing what they call the modified cardinal construction with simple NPs in (119) and (121) and by Heycock and Zamparelli (2003, 2005) discussing the coordinate construction in (118). Despite apparent high position, the indefinite article is here no more a Determiner or quantifier than it is in (122)–(126):7 (122) a. Many a fielder is leaving tonight with a minor injury. b. *Many a fielder are leaving tonight with a minor injury. (123) a. *Many fielders is leaving tonight with a minor injury. b. Many fielders are leaving tonight with a minor injury. (124) a. *Many an ill-matched seven fielders is leaving tonight with minor injuries. b. Many an ill-matched seven fielders are leaving tonight with minor injuries. (125) a. *Many an (ill-matched) infielder and outfielder is leaving tonight with minor injuries. b. Many an (ill-matched) infielder and outfielder are leaving tonight with minor injuries. (126) a. *Many an ill-matched seven infielders and outfielders is leaving tonight with minor injuries. b. Many an ill-matched seven infielders and outfielders are leaving tonight with minor injuries. Safely hidden within ‘[∃X:…a(n)…]’, it may be allowed that the indefinite article occurs at full strength to relate them X, the many, to a single class-as-one of themselves (Ionin and Matushansky 2004a, 2004b, 2006; see Cartwright 1994 on Russell [1903] 1938). Alternatively, the article may be weakened so as not to imply one of anything. For Heycock and Zamparelli (2003, 2005), it falls into the orbit of count/ mass morphology, expressing an anticumulative condition on the denotation of NP. The cumulativity of plurals (and mass terms)—if these are fielders and those are fielders, then they all are fielders—fails both the simple, singular noun and their
622
Chapter 12
coordinations. That is, it is not the case that if he is a fielder and she is a fielder, they are a fielder, and it is not the case that if these are an infielder and outfielder and those are an infielder and outfielder, then they all are an infielder and outfielder. The article a(n), according to Heycock and Zamparelli, demands a complement that fails at cumulativity. Thus, *a veteran fielders contrasts with a veteran seven fielders, and *a veteran infielders and outfielders with both a veteran seven infielders and outfielders and a veteran four infielders and three outfielders.8,9 Indeed if it is assumed that what is a such-and-such is what the indefinite description denotes, then seven ill-matched fielders are a such-and-such and two who are an infielder and an outfielder are a such-and-such. It is then either soon discovered when the many are also one (Ionin and Matushansky 2004a, 2004b, 2006) or there is a retreat (Heycock and Zamparelli 2005) from implying that a such-and-such is one such. Yet to excuse a from counting one undermines the contrast between (127) and (128): (127) a. In two separate trials in the same week, twenty-four sequestered jurors have deadlocked after a fortnight deliberation. b. Two judges in two trials sequestered twenty-four deadlocked jurors. (128) a. F In two separate trials in the same week, a sequestered twenty-four jurors have deadlocked after a fortnight deliberation. b. F Two judges in two trials sequestered a deadlocked twenty-four jurors.10 With an American jury comprising twelve jurors, there must have been two sequestrations and two deadlocks. What falsifies (128) is just the implication that the twentyfour jurors were sequestered together in one sequestration or in one deadlock. It must be the use of the indefinite article a(n), all that tells apart (127) and (128), which implies as much. If a(n) does not retreat from implying one, and it is still held that the jurors referred to must be a one, perhaps it should matter that twelve jurors are one jury and twenty-four jurors are two juries. That is, in this context, there is not one thing that twenty-four jurors are. Imagine, however, that the defendants are members of the same crime family that has met one night to bribe the jurors deciding the family’s fate or has circulated twenty-four names in a single list of home addresses and beloved pets. It suffices to grasp a single conspiracy or a single list, enough to carve what went on as a bribing, a blackmailing, or a deathlisting, to warrant (129) despite the twenty-four jurors remaining the same two juries: (129) a. In two separate trials in the same week, a bribed/blackmailed/deathlisted twenty-four jurors have deadlocked after a fortnight deliberation. b. Two judges in two trials sequestered a bribed/blackmailed/deathlisted twenty-four jurors. What is counted as one are events described by the modifier—sequestrations, deadlocks, bribings, blackmailings, and deathlistings—rather than metaphysical confections of the things the entire description refers to.
Measuring Events
623
Whether there is one or two sequestrations or deadlocks, it is clear enough in these examples what events to count. What is counted should first be countable, which elsewhere can be elusive, as when trying to count states of skirts being long: (130) Twelve long skirts are hanging in the showroom. (131) *A long twelve skirts are hanging in the showroom.11 How is it that there is one state of being long rather than 12 such states or some other number among the events that (130) and (131) describe? The events e such that ∃x long(e,x) resist counting as much as the denotation of any mass term.12 In contrast, states that are the results of actions more readily individuated can be counted as their results: (132) A lengthened twelve skirts are hanging in the showroom.13 A similar contrast is found among (133)–(135). My neuroses include a fear of home invasion and of being stranded without house keys. The report (133) is vague as to whether I have one or as many as six locks to my apartment: (133) Neurotic as I am, I have six duplicate keys for my apartment door. (134) Neurotic as I am, I have a duplicate six keys for my apartment door. (135) Neurotic as I am, I have a duplicated six keys for my apartment door. In contrast, (134) seems to imply six keys for the same one lock, providing grounds for locating a single state of being duplicate. (If one has it in mind, I suppose one could also justify a(n) contrasting being duplicate for this one apartment with being duplicate for others.) For (135), a single visit to the locksmith warrants a single action of duplicating several keys, leaving the number of locks as vague here as in (133). The counting and countability here concern the events or states described by the modifier duplicate(d). The keys are six; events or states satisfy the conditions on a(n), which appear here to be the same as on any other occurrence of a(n).14 In the modified cardinal construction, a(n) signals that among some countable events around, one is being described. What is in the neighborhood of such an event may very well be different from the neighborhood of several. Thus, despite embedding a(n) deep within the construction, its occurrence in the description when that description is adverbialized affects the neighborhood locating the events described by the matrix sentence. Consider the ambiguity of (136)–(137): (136) Seven bloodred sunrises preceded/announced/augured seven days’ bloody battle. (137) Seven bloodred sunrises have (always) preceded/announced/augured seven days’ bloody battle. The speaker may intend either that the seven days’ bloody battle follow the seventh sunrise, or that the sunrises and battles are interleaved—that is, the first battle
624
Chapter 12
following the first sunrise and so on until the seventh sunrise and battle. With a(n), in contrast, the seven days’ bloody battle must follow the seventh sunrise: (138) A bloodred seven sunrises preceded/announced/augured seven days’ bloody battle. (139) A bloodred seven sunrises have (always) preceded/announced/augured seven days’ bloody battle. With some further remarks, the contrast between (140)/(141) and (142)/(143) can also be put down to the difference between an adverbial describing several events that frame those described in the matrix and an adverbial describing only one such event: (140) Twelve green stoplights got me home without stopping. (141) Twelve green stoplights have gotten me home without stopping. (142) #A green twelve stoplights got me home without stopping. (143) #A green twelve stoplights have gotten me home without stopping. For (140)/(141), it suffices that my progress getting home, itself thought of as many or as a continuum of events, is interleaved among twelve distinct and spatiotemporally separated states of being green. For (142)/(143), those same twelve stoplights are parsed as a single state of being green.15 If, for any frame of reference and resolution of events, causes are understood to precede their effects (see Schein 2002, 320ff.), it is hard to picture that there could be a single being green of twelve stoplights and it results in my getting home, whether the latter is itself thought of as one, several, or a continuum of events. If, in recognizing the plural number agreement in (139) and (143), it is granted that these descriptions refer simply to seven sunrises or to twelve stoplights, the contrasts between (136)/(137) and (138)/(139) and between (140)/(141) and (142)/(143) are unexplained unless a(n) counts one state of being bloodred or one of being green and what events or states are counted matters thanks to adverbialization.16 The meaning of a(n) in the modified cardinal construction, as just illustrated in (136)–(143), extends to modified coordinations of NPs. As above, (144) is another example of the simple modified cardinal construction, which contrasts with (145) in the expected way: (144) Last night in different parts of the city, two lightning strikes in heavy traffic injured two nearby motorists. (145) #Last night in different parts of the city, two lightning strikes in heavy traffic injured a nearby two motorists. Sentence (144) reports two states of being nearby, each in a different part of the city; in contrast, (145) makes no such report. Its only felicitous interpretation is the
Measuring Events
625
irrelevant, distributive one that each of the two lightning strikes injured two motorists near it.17 Like (144), absent a(n), the disjunctive interpretation of (146)’s coordination allows motley motorists and pedestrians to have been injured by the one lightning strike or the other: (146) Last night in different parts of the city, two lightning strikes in heavy traffic injured some nearby motorists and pedestrians. (147) #Last night in different parts of the city, two lightning strikes in heavy traffic injured a nearby motorist and pedestrian. But with a(n), (147) excludes that a motorist was injured when nearby the site of one strike and a pedestrian nearby the other. Again, the only felicitous interpretation reports that each lightning strike injured a motorist and pedestrian.18 What is striking about this construction is its determinate syntax. From a position preceding the modifier, a(n) counts nearbys and not anything else. In particular, (147) cannot be used to describe the same one person, a schlemazl struck first while motoring in one part of the city and then again later that night on foot in another. Similarly, (148) cannot be report of one Good Samaritan acting on two different occasions, one while motoring and another while on foot: (148) #Last night in different parts of the city, before emergency services arrived, two accident victims flagged down a nearby motorist and (nearby) pedestrian. What a(n) demands in (147) and (148) is that there be a single nearby answering the description that it is a being nearby to two far-flung incidents, and as so much in Eventish comes down to relations among events, it is not a being nearby of person or thing but of motorizing and being on foot. What (147) and (148) demand is a single nearby that locates motorizing and being on foot near the two incidents. Note that this demand, deriving just from the meaning of a nearby motorist and (nearby) pedestrian, applies with equal force to the distributive interpretations, implying, implausibly of the same person, that the motorizing and being on foot were at the same time and place.19 Contrary to what is urged here, suppose that a(n) does not count nearbys or any events modifiers denote. Instead, the indefinite article counts only what the indefinite description denotes, persons or things. Where before we were left to wonder when the many sequestered jurors are one, we may now also wonder when the one schlemazl or Good Samaritan are many, completing the circle of a metaphysical hell. Respite from it is in a logical syntax that applies adverbialization to a description in which a(n) counts events.
626
Chapter 12
Recall from section 12.1 that counting to three million in the interpretation of three million passengers engages a protocol and events of counting in a description with the structure (149): (149) [∃X : ∃E (∃eμ count[eμ, E, 3,000,000] passengers[E,X])] Thus seven fielders should be ‘[∃X : ∃E (∃eμ count[eμ, E, 7] fielders[E,X])]’, which is to be embedded in the logical form for an ill-matched seven fielders. The collective ill-matchings to be counted are not as numerous as the fielders or the same as their states of being so. Rather, as usual, the participants in some ill-matching and some fielding are said to coincide: (150) [∃X:∃E(∃E1∃X ill-matched[E1,X] [℩E1: pro1] O[E1,E] ∃eμ count[eμ,E,7] fielders[E,X])] ‘… ill-matched seven fielders’ Pace some skepticism (Schein 2006), I will take the singular to express an arithmetic property meaning ‘one’, ‘∃eμ count[eμ, ζ,1]’, introducing sg as in (151) to be variously applied to events ‘sg(E)’ or to other things ‘sg(X)’: (151) sg for λζ. ∃eμ count[eμ, ζ,1] The conclusion of the above discussion is that in an ill-matched seven fielders what is counted to one is an ill-matching: (152) … [∃X:∃E(∃E1 sg(E1) ∃X ill-matched[E1,X] [℩E1: pro1] O[E1,E] ∃eμ count[eμ,E,7] fielders[E,X])] Consonant with Ionin and Matushansky (2004a, 2004b, 2006) and with the syntax in (122)–(126), a(n) qualifies something mentioned within the indefinite description, assuming at least provisionally that a(n) is one way to pronounce sg. The indefinite description itself is to denote fielders. The analysis extends to the coordination of NPs, as in an ill-matched pitcher and catcher, modulo the logical syntax for the disjunctive interpretation: (1) The reporters and superheroes that stalked Metropolis read The Daily Planet. (5) … [The X: ∃E1 W[E1,X] [℩E2: ΦΨ]O[E1,E2] (Φ∃X∃E2 reporters[E2,X]) and (Ψ∃X∃E2superheroes[E2,X])] … The coordination pitcher and catcher is thus as in (153), accepting Heycock and Zamparelli’s (2003, 2005) qualification that it differs from pitchers and catchers in containing NPs of singular denotation: (153) a. … ∃E1 W[E1,X] [℩E2: ΦΨ]O[E1,E2] (Φ∃X∃E2 pitcher[E2,X]) and (Ψ∃X∃E2catcher[E2,X]) …
Measuring Events
627
In a pitcher and catcher, absent an explicit modifier such as ill-matched, the events counted to one must be the presentation or participation events ‘W’ themselves, which meet the conditions for counting, as Heycock and Zamparelli (2003, 2005) hold, just in case they are each of one pitcher and one catcher and not of some pitchers and some catchers: (154) … [∃X: ∃E1 sg(E1) W[E1,X] [℩E2: ΦΨ]O[E1,E2] (Φ∃X∃E2 pitcher[E2,X]) and (Ψ∃X∃E2 catcher[E2,X])] With the logical forms of (152) and (154), an ill-matched seven fielders and a pitcher and catcher retain their ordinary and naive reference to seven fielders and two batterymates, respectively. Yet both indefinite descriptions describe participation in a singular event, which, under adverbialization, frames the events described by the matrix predicate and thus has the effects discussed above on the felicity and truth conditions of the whole sentence. 12.2.0
Generalized singular plurals: [QX : …
SG
…]
Given the plural reference of the indefinite description, as reflected in the obligatory plural number agreement, Ionin and Matushansky (2004a, 2004b, 2006) and Heycock and Zamparelli (2003, 2005) dissociate the article and anything else that may express the singular in an ill-matched seven fielders and a pitcher and catcher from the existential quantification and its variable, which remain resolutely plural, ‘[∃X: … a(n) …]’. The overt expression of many and most any in (122)–(126) and note 7 makes plain that the article is demoted to a position well within the DP. When the things quantified over are not what is counted as one, something else is, the events described by a modifier when one follows the article, eliciting the same effects on meaning observed in the preceding section and reviewed below. Joining many a(n), most any a(n), and any a(n) in eliciting those effects is every. When appearing with simple NPs, every, as much as a(n), occurs only in a singular count nominal: (155) a. Every fielder is/*are leaving the game with minor injuries. b. *Every fielders is/are leaving the game with minor injuries. c. *Every grass on the field is/are wet. (156) a. A fielder is/*are leaving the game with minor injuries. b. *A fielders is/are leaving the game with minor injuries. c. *A grass on the field is/are wet.
628
Chapter 12
Like bare a(n), a cardinal construction and the coordination of NPs excuse the things quantified over from counting as one and result in plural number agreement: (157) Many an (ill-matched) nine fielders are/*is leaving the game with minor injuries. (158) Many a pitcher and catcher are/ *is leaving the game with minor injuries. (159) Every (ill-matched) nine fielders are/*is leaving the game with minor injuries. (160) Every pitcher and catcher are/*is leaving the game with minor injuries. The same morpheme every is tokened in (155) and (159), and thus the quantifier must not of its own demand a singular variable x, *[every x : …]. It seems that any expression of the singular is as dissociated from every as it is from the existential quantifier, ‘[∃X:…a(n)…]’, and from many and most in the translations of an illmatched nine fielders, a pitcher and catcher, many/most any an ill-matched nine fielders, and many/most any a pitcher and catcher. If indeed quantifiers quantify over what NPs denote, for every (ill-matched) nine fielders and every pitcher and catcher, logical form should likewise start out ‘[every X:…]’ replacing ‘∃’ with every in (152) and (154) mutatis mutandis. The expression of the singular within ‘every NP’ so resembles its expression in many a(n) NP, most any a(n) NP, and any a(n) NP as to invite a morphological rule that every incorporates a(n), every < ever + a(n),20 if sg is usually pronounced “a(n)” as assumed above. Also like bare a(n), all singular plurals formed from either simple plural NPs or their coordination require intervention from a cardinal: (161) *Many an (ill-matched) fielders are/is leaving the game with minor injuries. (162) *Every (ill-matched) fielders are/is leaving the game with minor injuries. (163) Many an (ill-matched) nine fielders are/*is leaving the game with minor injuries. (164) Every (ill-matched) nine fielders are/*is leaving the game with minor injuries. (165) *Many an (ill-matched) infielders and outfielders have/has taken the field in the ninth inning in despair. (166) *Every (ill-matched) infielders and outfielders have/has taken the field in the ninth inning in despair. (167) Many an (ill-matched) nine infielders and outfielders have/*has taken the field in the ninth inning in despair. (168) Every (ill-matched) nine infielders and outfielders have/*has taken the field in the ninth inning in despair.
Measuring Events
629
(169) Many an (ill-matched) six infielders and three outfielders have/*has taken the field in the ninth inning in despair. (170) Every (ill-matched) six infielders and three outfielders have/*has taken the field in the ninth inning in despair. In contrast, however, to bare a(n), no modifier need intervene with many + a(n), most any a(n), any a(n), and every between quantifier and cardinal:21, 22 (171) Many a nine fielders have taken the field in the ninth inning in despair. (172) Every nine fielders have at some point or another taken the field in the ninth inning in despair. (173) Many a nine infielders and outfielders have taken the field in the ninth inning in despair. (174) Every nine infielders and outfielders have at some point or another taken the field in the ninth inning in despair. (175) Many a six infielders and three outfielders have taken the field in the ninth inning in despair. (176) Every six infielders and three outfielders have at some point or another taken the field in the ninth inning in despair. (177) *A nine fielders have taken the field in the ninth inning in despair. (178) *A nine infielders and outfielders have taken the field in the ninth inning in despair. (179) *A six infielders and three outfielders have taken the field in the ninth inning in despair. With all singular plurals, when the things quantified over are not what is counted, the events described by an intervening modifier are what is counted as one, with the effects on meaning observed in the preceding section. Thus, in the context described earlier for (127) and (128), (180) is false and (181) vacuous in that no sequestration or deadlock is of twenty-four jurors, each being of twelve: (180) a. #In last month’s trials, many a sequestered twenty-four jurors have ordered lunch in. b. # Last month, judges personally interviewed many a deadlocked twentyfour jurors. (181) a. #In last month’s trials, every sequestered twenty-four jurors have ordered lunch in. b. # Last month, judges personally interviewed every deadlocked twentyfour jurors.
630
Chapter 12
(182) a. In last month’s trials, many a twenty-four (sequestered) jurors have ordered lunch in. b. Last month, judges personally interviewed many a twenty-four (deadlocked) jurors. (183) a. ?In last month’s trials, every twenty-four (sequestered) jurors have ordered lunch in. b. ?Last month, judges personally interviewed every twenty-four (deadlocked) jurors. Without understanding why twenty-four should be critical for lunch or interview, the sentences (182)–(183) are somewhat odd, but not for lack of twenty-four jurors who can be divided among several sequestered juries. As before (see (130)–(132)), if the individual events or states are not easily recognized, counting them is unacceptable (184)–(185) in contrast to counting discrete events (186)–(187) or not counting at all the events described by the modifier (188)–(189): (184) *Many a long twelve skirts are going to be hung in the showroom next to twelve blouses. (185) *Every long twelve skirts are going to be hung in the showroom next to twelve blouses.23 (186) Many a lengthened twelve skirts are going to be hung in the showroom next to twelve blouses. (187) Every lengthened twelve skirts are going to be hung in the showroom next to twelve blouses. (188) Many a twelve long skirts are going to be hung in the showroom next to twelve blouses. (189) Every twelve long skirts are going to be hung in the showroom next to twelve blouses. Again, if, as in (190)–(191) (analogous to (136)–(139)), seven sunrises are not to be counted as a single bloodred event, the sentences may be taken to say that the seven sunrises and seven days of battle are interleaved. In contrast, if, as in (192)–(193), seven sunrises participate in a single event of being bloodred, the seven days’ bloody battle must follow the seventh sunrise: (190) Many a seven bloodred sunrises preceded seven days’ bloody battle. Many a seven bloodred sunrises have (always) preceded seven days’ bloody battle. (191) Every seven bloodred sunrises preceded seven days’ bloody battle. Every seven bloodred sunrises have (always) preceded seven days’ bloody battle.
Measuring Events
631
(192) Many a bloodred seven sunrises preceded seven days’ bloody battle. Many a bloodred seven sunrises have (always) preceded seven days’ bloody battle. (193) Every bloodred seven sunrises preceded seven days’ bloody battle. Every bloodred seven sunrises have (always) preceded seven days’ bloody battle. Direct semantic argument has shown that in an ill-matched nine fielders, many an ill-matched nine fielders, and every ill-matched nine fielders something despite the nine is counted to one, namely, the events, ill-matchings, described by the overt modifier. Complex singular plurals, such as many a nine ill-matched fielders and every nine ill-matched fielders, corroborate in showing that the ill-matchings are no longer counted to one as soon as the modifier is no longer adjacent to a(n). What, in that case, is then counted to one, assuming a(n) always counts something? In a pitcher and catcher, it was further conjectured that W-ings are counted, where no other events are overtly described,24 and the same could perhaps be said here, that a(n) counts as one an event in which nine ill-matched fielders participate. In so counting, which fits the word order well enough, neither a(n) nor any other expression of the singular needs to be weakened for the sake of the singular plural. The lesson above is that to be singular is to be associated with an independent morpheme ‘sg’ expressing a property as in (151) to be variously applied to events or other things. Which it is is a matter of syntax. When ‘sg’ includes within its scope a cardinal or a coordination of NPs, it counts events rather than the things the host DP quantifies over. Recognizing that ‘sg’ sometimes counts events rather than these things, spares us from either denying that a(n) has singular meaning (Heycock and Zamparelli 2003, 2005) or accepting the mystery of a Trinitarian semantics (Ionin and Matushansky 2004a, 2004b, 2006) where three is also one. For the translation to logical form, the above amounts to an argument that ‘sg(E1)’ occur within any logical form for an ill-matched nine fielders or every ill-matched nine fielders as it does in (194) and (195): (194) [∃X : ∃E(∃E1 sg(E1) ∃X ill-matched[E1,X] [℩E1: pro1] O[E1,E] ∃eμ count[eμ,E,9] fielders[E,X])] ‘An ill-matched nine fielders’ (195) [Every X : ∃E(∃E1 sg(E1) ∃X ill-matched[E1,X] [℩E1: pro1] O[E1,E] ∃eμ count[eμ,E,9] fielders[E,X])] ‘Every ill-matched nine fielders’
632
Chapter 12
These translations are fixed exactly as they appear given the assumptions that— (196) i. A quantifier Q quantifies over what NP denotes and binds a variable of the same sort tokened in both restriction and nucleus: a. [QX : NP[X]] (pl[X] & Φ[X]); or, b. [Qx : NP[x]] (sg[x] & Φ[x]); ii. A quantifier phrase in subject position binds a variable in subject position. Given that the quantifier phrases in (157)–(176) and (180)–(193) are all in subject position, they bind variables in subject position according to (196ii). The obligatory plural number agreement in all these cases then settles on (196ia), which (194) and (195) instantiate. Recall however that an implication from plural number agreement to a plural variable has been entirely out of place since chapter 2, where the plural number agreement in (197) is a plural pronoun occurring as a descriptive anaphor to singular antecedents within separate antecedent clauses: (197) Not a single philosopher early morning in New Jersey and not a single linguist later in the day in California have swapped wine futures in online trading. If that usage is not specifically restricted to coordination, then (157), quantifying over singular events rather than fielders, could be assimilated to (198): plural number agreement in (157) occurs as a descriptive anaphor for the fielders just as the plural pronoun in (198) does. (157) Many an (ill-matched) nine fielders are leaving the game with minor injuries. (198) a. Many a time that there are nine fielders ill-matched, they are leaving the game with minor injuries. b. Often, if there are nine fielders ill-matched, they are leaving the game with minor injuries. This alternative is revisited below in discriminating many a one or more fielders from many fielders. 12.3
Singular plurals, distributive plurals, and distributive singulars
In all the singular plurals, the alleged occurrence of ‘sg’ correlates with overt expression of the singular—with a(n), in an ill-matched nine fielders, many an (ill-matched) nine fielders, and most any an (ill-matched) nine fielders and with every, where (155)–(156) show it to be as singular as a(n). Distributive, plural quantifiers—many
Measuring Events
633
infielders, few infielders, most infielders—do not hint at overt singular morphology. Yet in a sense shortly elaborated, they too demand, in recognition of a contrast with plural (in)definites some fielders and the fielders, some latent notion of the singular to characterize their distributivity. Here too the use of the singular must leave room enough for the coordination of NPs in many infielders and outfielders, few infielders and outfielders, and most infielders and outfielders not to restrict the quantifier to domains containing a fielder only if he is himself both infielder and outfielder, *[Qx : infielders[x] and outfielders[x]] and *[QX : sg[X] & infielders[X] and outfielders[X]]. Like in the singular plural, a coordination of NPs deflects what counts as one here from applying to its denotation, deflected as before by syntax and scope to count as singular in some sense something other than a fielder, perhaps an event in which any number of infielders and outfielders participate. Thus the distributivity of many infielders comes to rest on a morpheme separate from both quantifier and noun. With mischief in mind, one might identify it as yet another (tacit) occurrence of ‘sg’ aggravating the extent to which the DPs in (199) are assembled from the same or very similar morphemes: (199) a. many fielders b. many a fielder c. many a one or more fielders In fact, no two of the DPs in (199) occur with synonymous effect as discussed below, and the problem is to discern their logical forms. To appreciate it, note that to begin with the thought that fielders denotes pluralities of one or more fielders, ‘fielders(X)’, and many quantifies over them, ⌜[many X: Φ[X]]⌝, is to forgo daylight between many fielders and many a one or more fielders.25 A complex, singular plural such as many a one or more fielders and a distributive plural many fielders differ in three respects. The former is felicitous wherever the simple, plural indefinite description it embeds, one or more fielders, is.26 In asserting that there are many of these, it looks like distributive, full-blooded collective quantification, counting many of such collections. (200) Many fielders have huddled in the outfield. (201) Many a one or more fielders have huddled in the outfield. Despite combination with collective predicates such as huddle, a distributive plural quantifier, many fielders, always counts the individual fielder, and thus the first difference between them is what is counted many, section 12.3.1. Second, the difference in quantification affects descriptive anaphora, implying a difference in structure, section 12.3.2. A pronoun they whose antecedent is many fielders refers to all the many fielders about whom something true has been said. But a pronoun the antecedent for which is many a one or more fielders makes its reference dependent on
634
Chapter 12
time and place and only to some one or more fielders, never to all the many fielders (who are distributed among the many a one or more), making what has been antecedently said true. Third, despite combination with some collective predicates, what can be said of some many fielders cannot always be said with the use of the distributive plural many fielders, which always remains at least semidistributive, chapter 13, in contrast to many a one or more fielders, which is, as already mentioned, as collective in its reference as the embedded indefinite one or more fielders. In these three respects, counting, descriptive anaphora, and semidistributivity, the result of combining many fielders with a collective predicate in (200) remains more singular than the collective quantification in (201) and thus underlines the contrast between (199a) and (199c). Before proceeding to this contrast, the next section briefly marks the contrast between (199a) and (199b), distributive plurals and distributive singulars. 12.3.0
Distributive plurals and distributive singulars
The next two sections find daylight between (199a) and (199c) in opposing the singular quantification of many fielders to the plural quantification of many a one or more fielders: (199) a. many fielders b. many a fielder c. many a one or more fielders Insisting on singular quantification for many fielders should not, however, deprive it of all plural sense as that remains the basis for discriminating it from many a fielder (199b) and, in general, for telling apart distributive plurals, many fielders, most fielders, any fielders, all fielders, from distributive singulars, many a fielder, most any a fielder, any fielder, every fielder. Number agreement is taken here to be a descriptive anaphor referring to whatever they did in the plural and to whatever it did in the singular. If reciprocity implies a (sub)event with at least two participants and the reciprocal itself embeds a descriptive anaphor referring to others in an event other than them or it (Schein 2003), then it should be that the plural NP in (202)–(204) affords reference (via number agreement) to an event of communication in which mobile phones participate, within which each is related to mobile phones other than it.27 On the other hand, the singular number agreement with singular NP referring to an event of communication in which only one mobile phone participates chokes off reciprocity in (205)–(208):28 (202) Many mobile phones communicate with each other. (203) Any mobile phones communicate with each other. (204) ?Both mobile phones communicate with each other.
Measuring Events
(205) (206) (207) (208)
635
*Many a mobile phone communicates with each other. *Every mobile phone communicates with each other. *Any mobile phone communicates with each other. *Each of the two mobile phones communicates with each other.
Although both sets of quantifier phrases are distributive and thus singular in that sense, it will matter for a descriptive anaphor dependent on the content of those phrases, whether it attempts reference to the mobile phones in an event or to the mobile phone in an event. 12.3.1
Counting many
A singular plural, many (a) one or more mobile phones, and a distributive plural, many mobile phones, differ in what they count. To illustrate, suppose all mobile phones are organized into networks that exclude all communication to the outside and provide total interconnectivity inside. In the city of Mobile, there happens to be a monopoly, and yet (209) is true because many mobile phones belong to it: (209) In Mobile, many mobile phones communicate with (each other and) only each other. (210) In Mobile, many (a) one or more mobile phones communicate with (each other and) only each other. In contrast, the complex, singular plural counterpart in (210) is false: communicating with only each other are only the one one or more mobile phones, the monopoly, to which all the mobile phones of Mobile belong.29 Any fewer mobile phones do not communicate with only each other, but with the other mobile phones of Mobile, too. The same divergence between distributive plural quantifier and complex, singular plural quantifier is illustrated in (211)–(214): (211) Many natural numbers are members of sets of natural numbers. (212) Uncountably many natural numbers are members of sets of natural numbers. (213) Many (a) one or more natural numbers are members of sets of natural numbers. (214) Uncountably many one or more natural numbers are members of sets of natural numbers. If it isn’t many unless it is uncountably so, both (211) and (212) are false, but (213) and (214) are true (Cantor’s Theorem). As there are as many enumerations of one or more natural numbers as there can be one or more natural numbers, sentences (216)–(218) join (213)–(214) in counting the uncountably many, while (215), a variant of (211) with modifier, continues to count only the countable: (215) Many enumerated natural numbers are members of sets of natural numbers.
636
Chapter 12
(216) Many an enumerated one or more natural numbers are members of sets of natural numbers. (217) Many (a) one or more enumerated natural numbers are members of sets of natural numbers. (218) Many an enumeration of one or more natural numbers is/*are of members of a set of natural numbers. Heading this contrast is the minimal pair (211) and (213), which logical form should now take pains not to render equivalent: (211) Many natural numbers are members of sets of natural numbers. (213) Many (a) one or more natural numbers are members of sets of natural numbers. If many is a quantifier ⌜[DP many : Φ[ξ]]⌝ in many natural numbers and many a one or more natural numbers, a view rejected below, and both are first taken to quantify over pluralities of one or more, as in (219), many must be construed to count only the individuals among them as in (220), in effect reducing plural to singular quantification, if the correct interpretation for (211) is to be derived: (219) [Many X : … natural numbers[X]] (pl[X] & Φ[X]) (see (196)) (220) [Many X : Ψ[X]] Φ[X] ↔ [Many y : ∃X(Xy &Ψ[X])] ∃X(Xy &Φ[X]).30 (221) [Many y : [∃X : y = {X}] natural numbers[X])] [∃X : y = {X}](pl[X] & Φ[X])31 But if many is so construed, plural quantification as such must be abandoned in the translation of (213). Counting only individuals forces on (213) quantification over classes as one as in (221) to correctly count its many. But the only warrant for a class as one, a(n), can be occupied elsewhere (e.g., (216)) counting events as one, where even so, the many are still the uncountably many classes. Apart from the implications of this analysis for a(n), which have been rejected in the preceding sections, parsing many natural numbers as (219) may get the count right in (220) but it misses the essential (semi)distributivity over singular natural numbers that many natural numbers requires from the predication, Ψ[X] and Φ[X], in contrast to many a one or more natural numbers. Conceding that many natural numbers is singular in some sense in virtue of its distributivity, nothing is miscounted if (211) and (213) are translated along the lines (222) and (223) respectively suggest, with the same many (thus rejecting (220)): (222) [Many X : sg[X] & natural numbers[X]] (pl[X] & Φ[X])32 (223) [Many X : one or more[X] & natural numbers[X]] (pl[X] & Φ[X]) A good count, counting pluralities of one or more for (213) and individuals for (211), could be achieved counting pluralities in both—pluralities of just one in the case of
Measuring Events
637
(211) as in (222). One way or another, what counts many must differ for many natural numbers and many a one or more natural numbers. A more robust difference in their logical syntax invites itself in if number predicates denote events of measurement, as here supposed. It may then be that many natural numbers counts (events of) natural numbers and many a one or more natural numbers counts counts of one or more of them as in (224), so that (213) is to be read as (226): (224) [Many eμ : sg(eμ) ∃E(∃n(one-or-more(n) count[eμ,E,n]) natural numbers[E,X])] [℩X: there[eμ, X]] Φ[X] (225) ‘[℩X: there[eμ, X]]’ for ‘[℩X: sg(eμ) ∃E(∃n(one-or-more(n) count[eμ,E,n]) natural numbers[E,X])]’ (226) “Many a counting one or more natural numbers, the natural numbers counted therein are members of sets of natural numbers.” Both many natural numbers and many a one or more natural numbers are equally distributive and singular in what they count, singular events: the former—an event (or state) of being a natural number, the latter—a count. But, as counting events are not themselves members of sets of natural numbers, the subject of (213) is a descriptive anaphor to what is counted. As this implies and the next section confirms, there is more that tells apart the logical syntax of (211) and (213) than (222) and (223) allow. But if the distinction in logical syntax is not to be as slight as it is (222) and (223), it must be that one or more is indeed ‘∃n(one-or-more(n) count[eμ,E,n])’ offering something other than natural numbers, events of counting eμ, to be counted many. 12.3.2
Singular plurals and distributive plurals as antecedents for anaphora
To say something about many fielders, using a distributive plural, and then in the next sentence to refer with a pronoun they to them is often to refer to all the many fielders about which something true has been said. To say something about many a one or more fielders, using a singular plural, and in the next sentence to refer back with a pronoun they is just to refer to some one or more, without an intention to refer to all the fielders making what has been said true. To illustrate, imagine a midcentury canvas of many, many line segments. Many of the line segments form triangles, and at least as many do not, forming other figures or nothing at all. No triangles share sides, so that it is true that (227) Many line segments form a triangle. They are three times as many as the triangles.
638
Chapter 12
The pronoun in (227) manages to refer to all and only the line segments forming the canvas’ triangles, comparing their number to that of the triangles. In contrast, the singular plural confines the pronoun’s reference to some three line segments, those at any of the occasions counted among the many. These three are far fewer than the canvas’ triangles, (228), being three times as many as just one triangle, (229). (228) #Many a three line segments form a triangle. They are three times as many as the triangles. (229) Many a three line segments form a triangle. They are three times as many as the triangle. Telescoping in similar fashion down to the individual event is more fickle when the antecedent is many line segments. The distributive plural contrasts with the singular plural in rejecting telescoping in this context (230), while accepting it in others, (232): (230) #Many line segments form a triangle. They are three times as many as the triangle.33 (231) Many a three line segments form a triangle. They meet at three vertices. (232) Many line segments form a triangle. They meet at three vertices.34 Slighting details, the above argues for a pronounced difference in structure setting apart many line segments and many a one or more line segments. As already hinted at in the paraphrases (see (198)), the singular plurals have a modal flavor, quantifying over alternative circumstances where line segments are counted: there are many circumstances where one can count that three line segments form a triangle. The alternative circumstances survey the canvas, where each alternative is nothing more than a region, or frame of reference, within which triangle-forming line segments are counted three. Given that three is explicit in counting to 3 under a frame of reference and protocol for measurement eμ ‘count[eμ,E,3]’, the modal force can be given explicit representation:35 (233) [Many eμ : ∃E(… a … count[eμ,E,3] … ∃X segments[E,X])] Φ[eμ] It should count as further evidence for number to be explicitly related to counting that it now affords a formal distinction between many a one or more line segments and many segments, rescuing them from synonymy. Conforming to restricted quantification, quantification in (233) over measurements eμ in the restriction to many must also quantify over them in its matrix. But in quantifying over measurements, many in many a one or more line segments does not quantify either over line segments or over that which forms a triangle, that which the subject of the sentence is presumed to refer to. It is rather as in (234) paraphrasing (233). A pronoun,
Measuring Events
639
unspoken in (235), occurs in subject position as a descriptive anaphor referring to three line segments: (234) Many times that there are three line segments, they form a triangle. (235) Many a three line segments form a triangle. In reaching for the contrast in anaphora across sentences (227)–(232), it is not far-fetched to assume that quantifiers over measurements and their epistemic conditions, like epistemic adverbials in general, include within their scope the most peripheral elements of a clause: (236) [Many eμ: a three line segments[eμ])] [Φ framed by eμ, they form a triangle] Many a three line segments form a triangle. (237) ∃eμ [Φ framed by eμ, [iMany X: line segments[X])] theyi form a triangle] Many line segments form a triangle. Thus, as in (236) and (237), many a three line segments must move to a higher position than many segments, in order to include within its scope framing by frame of reference. In (227)–(232), a pronoun attempting reference to the line segments faces antecedent sentences with two different structures. When the antecedent is many a three line segments, as in (228), (229), and (231), the sentences conform to the schema in (238): (238) Under many measurements eμ where there are three line segments, they form a triangle thereeμ. … they [i.e., the (three) line segments … thereeμ] … If the intention is to refer with the pronoun back to three line segments, they can only be three with respect to the individual measurement: (239) Under many measurements eμ where there are three line segments, they form a triangle thereeμ. Under any eµ of those measurements … they [i.e., the (three) line segments … thereeμ] … Provided that the descriptive content of the pronoun in copying content from the antecedent clause must include the variable eμ and so refers to the line segments in eμ forming a triangle under eμ, the pronoun cannot refer to all the triangle-forming line segments on the canvas, for that would require replacing without license the singular eμ with the plural variable Eμ: (240) *Under many measurements eμ where there are three line segments, they form a triangle thereeμ. Under those measurements Eµ … they [i.e., the line segments … thereEµ] … As noted above (see (228)), many a three line segments confines the reference of they in the second sentence to the three line segments under a single frame of reference.
640
Chapter 12
Where the antecedent is many line segments, (227), (230), and (232) fit schema (241): (241) ∃eμ Under eμ, many line segments in eμ form a triangle in eμ. … they [i.e., the line segments … thereeμ] … Here closing off the singular variable eμ in the second sentence immediately obtains reference to all the triangle-forming segments on the canvas: (242) ∃eμ Under eμ, many line segments in eμ form a triangle in eμ. Under that measurement eµ … they [i.e., the line segments … thereeµ] … Since there is just the one frame of reference here, for the pronoun to refer to fewer line segments—to just three—requires an adverb quantifying over events of forming a triangle—conventional telescoping as in (243), subject apparently to conditions relating the aspect of the adverbial clause and that of the matrix clause, observed in the contrast between (230) and (232). (243) ∃eμ Under eμ, many line segments in eμ form a triangle in eμ. Under that measurement eµ, whenever-e line segments form a triangle at e … they [i.e., the line segments … at e thereeµ] … This difference in structure ought to be recognized: many a one or more line segments ‘[Many eμ : a one or more line segments[eμ])]’ quantifies over measurements and their varying epistemic conditions, driving it to a higher position than that of many line segments ‘[Many X: line segments[X])]’ quantifying over line segments. Recognizing this difference in structure between many a one or more line segments and many line segments gains some purchase on the difference between them for anaphora that is also consonant with the difference between them in counting, as discussed in section 12.3.1. Thus, many line segments and many a one or more line segments diverge in their meaning, in their internal logical syntax, and in the logical syntax of the clauses that host them, none of which can be achieved without the introduction of explicit reference ‘eμ’ to events of measurement and the construal of number as expressing a relation to them, ‘count[eμ,E,n]’.
13
Antisemidistributivity vs. Conjunction Reduction Redux
The sentential connective and in (1) and (2) has been absolved from collective and divided reference: (1) The infielders sluggishly and the outfielders swiftly covered the field. (2) The nine infielders and outfielders covered the field. Instead, unspoken plural event pronouns (‘[℩E1,2: pro1,2]’) intercede bearing plural reference: (3) [The X: ∃E infielders[E,X]] ∃E1 (W[E1,X] sluggishly[E1]) and [The X: ∃E outfielders[E,X]] ∃E2 (W[E2,X] swiftly[E1]) [℩E1,2: pro1.2][℩E3: pro3]O[E1,2, E3] ∃E3 ∃X Agent[E3,X] [℩E3: pro3][℩E4: pro4] Cause[E3, E4] ∃E4 covered[E4] … (4) [The X : nine[X][℩E1,2: pro1,2]W[E1,2, X] ∃E1∃X infielders[E1,X] and ∃E2∃X outfielders[E2,X]] … In the nine infielders and outfielders, the sentential coordination says only that there be infielder-ing and there be outfielder-ing. The apparatus for collective reference, saying that nine participate (‘W’) in these events, remains outside the phrases and conjoins. Section 12.2 bore witness to the morphological weight and syntactic position of the peripheral material, and section 14.2 will say more about its meaning. Now it happens that to be a fielder is to be an infielder or outfielder, and vice versa (5). And so, if all there were to the logical syntax of nominal coordination is that which can be fashioned from and—infielders and outfielders just as it is embedded in the above apparatus for collective reference—the simple noun fielders and coordinate noun phrase infielders and outfielders ought to denote the same (6): (5) fielder(e,x) ↔ (infielder(e,x) ∨ outfielder(e,x)) (6) fielders[E,X] ↔ infielders and outfielders[E,X]
642
Chapter 13
The same holds of the simple noun musicians and the coordinated instrumentalists and vocalists, if all music is instrumental or vocal:1 (7) musician(e,x) ↔ (instrumentalist(e,x) ∨ vocalist(e,x)) (8) musicians[E,X] ↔ instrumentalists and vocalists[E,X] Then, any context where simple NP and coordinate NP do not substitute salva veritate, as in the alternation between the banal truths of (11)–(12) and the incoherent (9)–(10), presents a puzzle the solution to which must discard the alleged equivalence: (9) #Many fielders are a major league defensive lineup. #Many fielders are major league defensive lineups. (10) #Many musicians are an ensemble. #Many musicians are ensembles. (11) Many infielders and outfielders are a major league defensive lineup. Many infielders and outfielders are major league defensive lineups. (12) Many instrumentalists and vocalists are an ensemble. Many instrumentalists and vocalists are ensembles. If the meanings of fielders, infielders, outfielders, musicians, instrumentalists, and vocalists are too plain to admit alteration, the failure of substitutivity in (9)–(12) suffices rather to demand revision either in the meaning of and itself or in the logical syntax that embeds NP coordination. Holding and innocent, the revision here (chapter 14) will be to reject bare NPs and suppose that all NPs occur within AdrPs, addressing scenes or frames of reference, so that the logical syntax may be paraphrased (including adverbialization) as in (13)–(14): (13) #Many musicians are an ensemble. Many musicians at an address in a frame of reference while musicians at an address in a frame of reference are an ensemble. (14) Many instrumentalists and vocalists are an ensemble. ‘Many instrumentalists at an address in a frame of reference and vocalists at another address at that frame of reference while instrumentalists at an address in a frame of reference and vocalists at another address in that frame of reference are an ensemble.’ Simple nominal phrases and conjoined nominal phrases describe distinct presentations of the same subject matter. The instrumentalists and vocalists are of course the musicians. But in choosing (14), the speaker commits herself to frames of reference that segregate instrumentalists and vocalists. Scenes of instrumentalists here and vocalists there segregate them in a way that scenes of musicians here & there
Antisemidistributivity vs. Conjunction Reduction Redux
643
need not. This escapes the alleged equivalence of simple and coordinate NP without explanation yet for the contrast between (11)–(12) and (9)–(10) that disproves it. For the little that has been said so far, the contrast could just as well have held in reverse, so that the sentences with simple NPs were true and felicitous and those with coordinate NPs, not. In all contexts that reject substitution, it is the simple NP that is rejected, typically from occurring within what the previous sections have called a distributive plural, in contrast to singular plurals and plural (in)definite descriptions: (15) #Many musicians are an ensemble. (16) Many a two or more musicians are an ensemble. (17) The/some/many musicians are an ensemble. A false start succumbs to the temptation in the contrast between (15) and (17) to parse (15) as singular distributive quantification (as in fact was assumed in previous sections). For a syntax thus conforming to (18), all that should have mattered to preserve truth is that the things X that simple NP and coordinate NP denote are the same, as in (8) (whatever passes in the events E): (18) [Many x : ∃X(Xx & ∃E NP[E,X])] are-an-ensemble[x]. The dilemma is aggravated where the baseball players are the fielders, who are also the batters, but they are not the fielders and batters, so it appears from a failed substitution: (19) # Many baseball players are a powerhouse squad. (20) Many fielders and batters are a powerhouse squad. (21) #Many baseball players are nine powerhouses. (22) Many fielders and batters are nine powerhouses. If the baseball players, the fielders, and the batters are all the same, then except for what and might manufacture from them, how else could simple NP and coordinate NP come to denote a difference? Something is indeed to be manufactured if there is a further thought that antisemidistributives,2 the predicates rejecting the substitution, including be nine powerhouses, be a powerhouse squad, be an ensemble, be a defensive lineup above, be a dynamic duo in (23), and be two harmonizing dynamos in (24), are a breed that denotes wholes wholier than other collective predicates, (25)–(28), and so refuse any expression of distributivity, including in particular determiner all and floated all (29)–(30): (23) #Many musicians are a dynamic duo. #Many musicians are dynamic duos. (24) #Many musicians are two harmonizing dynamos.
644
(25) (26) (27) (28)
Chapter 13
Many Many Many Many
musicians musicians musicians musicians
pair up in a dynamic duo / dynamic duos. are paired up in a dynamic duo / dynamic duos. are in a dynamic duo / dynamic duos. are harmonizing duet partners.
(29) #All musicians are a dynamic duo. #All musicians are dynamic duos. (30) #The musicians are all a dynamic duo. #The musicians are all dynamic duos.3 Absent substitution salva veritate, it appears that musicians in (23) does not denote the same things as instrumentalists and vocalists in (31), which denotes instead the wholier ensembles, like in (32), that can be dynamic duos: (31) Many instrumentalists and vocalists are a dynamic duo. Many instrumentalists and vocalists are dynamic duos. (32) Many ensembles of instrumentalist and vocalist are a dynamic duo. Many ensembles of instrumentalist and vocalist are a dynamic duo. (33) Many a two instrumentalist and vocalist are a dynamic duo. If the logical form for these sentences conforms to (18), there is little to conclude other than that and manufactures reference to such ensembles, as Aristotle would have wanted all along, contrary to the eponymous thesis of this book. But, if musicians does not denote what instrumentalists and vocalists does, how is it to be explained that no interpretation of (34) is true (cf. (35))? (34) F The musicians are not the instrumentalists and vocalists. (35) The musicians are not the ensembles (of instrumentalists and vocalists). The bling ontology and alternative meaning for and founder on the false assertion in (34) of the ontological distinction supposed. In looking elsewhere for an explanation of the substitutivity puzzle that spares and, the assumptions that have quickly garden-pathed into (34) must be abandoned: the distributivity of many NP does not warrant its translation as a (first-order) distributive quantifier (contra (18)), and what distinguishes antisemidistributive predicates from other collective predicates does not genuflect to a mystery of wholier wholes. Something else must tell apart musicians from instrumentalists and vocalists, a reference to scenes, as mentioned above, that locate musicians here and there but segregate instrumentalists here and vocalists there, and something else is to explain why as much should matter for substitution into many NP and not into some many NP and for antisemidistributive predicates and not for other collective predicates. Sketching out the explanation (see section 13.1), the first step is to dispel the illusion of distributive quantification, removing the expression of quantity to reveal
Antisemidistributivity vs. Conjunction Reduction Redux
645
a stark contrast between bare plurals (36) and those with articles (40) in their combination with antisemidistributive predicates: (36) (37) (38) (39)
#Unsolved murders were a cluster near the Green River in Washington. Unsolved murders clustered near the Green River in Washington. Unsolved murders were clustered near the Green River in Washington. Unsolved murders were in a cluster near the Green River in Washington.
(40) Some/the unsolved murders were a cluster near the Green River in Washington. Elsewhere, null articles support scanning measurements, imperfective measurement, as it were, whereas the overt article presupposes a single, perfective measurement: (41) Baseball legends by the hundreds were Jews. (42) Bullet holes pockmarked the street by the thousands, traveling from one end of the green line to the other. (43) #Some/the baseball legends by the hundreds were Jews. (44) #Some/the bullet holes pockmarked the street by the thousands, traveling from one end of the green line to the other. If this, that, the, all definite articles, differ merely in the epistemic conditions they impose on reference, the same may be all that distinguishes null and overt article above. The null article presupposes the location of the referents across several scenes or frames of reference, which are essentially plural, whether they cohere as a continuous panoramic scan or as discontinuous, scattered montage. In contrast, the overt article presupposes the location of the referent in a single scene from a fixed point of view for a fixed frame of reference: baseball legends, “some across scenes who are baseball legends,” vs. some baseball legends, “some in a scene who are baseball legends.” Modulo this difference in epistemic conditions as reflected in the embedded reference to plural or singular scenes or frames of reference, baseball legends and some baseball legends are equally plural, indefinite descriptions of baseball legends. Equally slight is therefore the difference between, say, some many musicians and many musicians. Both are plural indefinite descriptions of musicians. The latter is crucially not a distributive quantifier, but it does presuppose that the many musicians are scattered among plural scenes. Already some relief for and is in sight: (45) a. [∃X: ∃E∃σ(there[σ, E] many[E] instrumentalists and vocalists[E,X])] b. [∃X: ∃E∃Σ(there[Σ,E] many[E] instrumentalists and vocalists[E,X])] The logical syntax both for some many instrumentalists and vocalists and for many instrumentalists and vocalists is strictly plural or second-order, as in (45), and is thus relieved of the burden when many instrumentalists and vocalists is mistakenly parsed
646
Chapter 13
of finding for singular x-values that are each an ensemble of instrumentalist and vocalist. But with such a slight difference between some many musicians and many musicians, the apparent difference in their distributivity in combination with antisemidistributive predicates, (15) vs. (17), wants explanation. Consider then the following illusion of formal distributivity that specific content may induce: (46) Atoms scattered among large molecules jiggled only with the other atoms in the molecule. (47) Scattered atoms captured within large molecules jiggled only with the other atoms in the molecule. Given that the atoms are in the first place scattered, it must be that the atoms are distributed among several molecules, since any one molecule is a dense cluster of atoms. Given that there are several molecules, singular reference to any one (the molecule) finds support only via the intervention of a (semi)distributive operator. The sentences appear unambiguous and (semi)distributive. In contrast, sentences (48) and (49) are ambiguous, or perhaps only vague, between a reading that distributes the atoms among several molecules and one that is about a single cluster in a single molecule: (48) Atoms clustered in large molecules jiggled only with the other atoms in the molecule. (49) Clustered atoms captured within large molecules jiggled only with the other atoms in the molecule. Sentences (50) and (51) are a minimal pair attesting to the contrast: (50) Scattered atoms jiggled only with the other atoms in their molecule. (51) Clustered atoms jiggled only with the other atoms in their molecule. As (51) may be true of the solitary cluster and solitary molecule, clustered atoms is not itself a distributive quantifier, but rather a plural indefinite description, and if (semi)distributivity is understood, it is by the intervention of a (semi)distributive operator: (52) Clustered atoms all jiggled only with the other atoms in their molecule. But surely there can be no difference in the logical syntax of (50) and (51), which differ only in the substitution of one lexical item, scattered, for another, clustered. If (50) has only a (semi)distributive reading, it is merely an inferential illusion that rules out as nonsense the reading that would allow combination of a solitary scattering and solitary molecule. We should not contrive for it a logical syntax that formally represents its inferred distributivity.
Antisemidistributivity vs. Conjunction Reduction Redux
647
These remarks now carry over to the crucial contrast between (53) and (54) and between (55) and (56), where the null articles are felt to be distributive:4 (53) #(Many) unsolved murders were a cluster near the Green River in Washington. ‘Some across various scenes (many) unsolved murders were an in a scene cluster …’ (54) Some (many) unsolved murders were a cluster near the Green River in Washington. ‘Some in a scene (many) unsolved murders were an in a scene cluster …’ (55) #Innumerable stars light-years away are that galaxy at the edge of the universe. ‘Some across various scenes innumerable stars are that in a scene galaxy …’ (56) Some innumerable stars light-years away are that galaxy at the edge of the universe. ‘Some in a scene innumerable stars are that in a scene galaxy …’ The scenes or frames of reference for (many) unsolved murders are plural and may yet answer to further conditions on their epistemic scatter—for example, that they are not path-integrated to a single frame of reference. Yet, intrinsic to the concepts of a hexagonal carbon ring of atoms, a cluster of murders, or a galaxy is that a hexagon with its constituent atoms, a cluster of murders, or a galaxy of stars are all found within a single frame of reference or scene for hexagon, cluster, or galaxy. What would it mean for a hexagon or cluster to be scattered across several scenes or frames of reference and recognized as such with a geometry defined only for and within a single frame of reference?5 Whether geometric or arithmetic, determinate measurement often implies measurement under a given singular frame of reference. The antisemidistributive predicates all involve some such measurement. Insofar as reference to a singular frame of reference adheres to the very concept of a cluster, it presumably attaches to the morpheme cluster in all its occurrences, entraining then the question why the collective predicates in (57)–(59) are not also antisemidistributive: (57) (Many) unsolved murders clustered near the Green River in Washington. (58) (Many) unsolved murders were clustered near the Green River in Washington. (59) (Many) unsolved murders were in a cluster near the Green River in Washington. The answer is in the further syntax that distinguishes (57)–(59) from (53)–(54). In all of (53)–(59), there are to be considered the scenes or frames of reference Σ that unsolved murders are scattered across and the scene or frame of reference σ for a
648
Chapter 13
cluster. It must be that the copular construction in (53)–(54) affords no change or discontinuity in perspective from unsolved murders to cluster so that the Σ are the σ, resulting in the contrary conditions being imposed on the same scenes or frames of reference. When instead the morpheme cluster is embedded in a prepositional phrase (59), or the decomposition of the verbal clustered or were clustered reveals it to be a secondary predicate denoting the end state of some other, precedent actions (57)–(58), it is allowed that the Σ frame unsolved murders in their precedent actions and the perspective then shifts without any scene continuity to a σ of a cluster. Those scattered events of unsolved murders meet an end in which (however determined) the participants in those events constitute a cluster in some distant, independent scene σ. Neglected in most discussion of the antisemidistributive predicate is how the same morpheme cluster changes its stripes, denoting wholier wholes in one collective construction but not the next. Here there is nothing in the logical syntax or logical type of an antisemidistributive predicate to distinguish it from any other predicate. It is just that that which is a cluster is so under a single frame of reference, the very same one, it may be, that some unsolved murders implies frames some unsolved murders. In contrast, the frames of reference that unsolved murders scatters them among are not a single frame of reference that frames them in a cluster. This is harmless in those collective constructions with enough room for a shift in perspective from the scattered frames of reference to the one, which is however exactly the respect in which the copular construction is deficient. This then sketches out an explanation for how (many) unsolved murders and some/the (many) unsolved murders diverge in meaning while maintaining a common syntax as plural (in)definite descriptions: (60) a. [∃X: ∃E∃σ(some[E] there[σ, E] (many[E]) NP.pl[E,X])] b. [∃X: ∃E∃Σ( there[Σ,E] (many[E]) NP.pl[E,X])] The bare plural (many) unsolved murders presupposes the murders’ location across plural frames of reference, which precludes the singular frame of reference necessary for the nondistributive, collective interpretation that some (many) unsolved murders supports. The contrast between (61) and (62) demands nothing more for its explanation than the formal difference in (60), eschewing more radical surgery to reconstruct a distributive quantifier ‘[Many x: ∃X∃E(Xx & NP.pl[E,X])]’: (61) a. #Many innumerable stars light-years away are that galaxy at the edge of the universe. b. #Many well-known musicians are that charity orchestra. (62) a. Some many innumerable stars light-years away are that galaxy at the edge of the universe. b. A many well-known musicians are that charity orchestra.
Antisemidistributivity vs. Conjunction Reduction Redux
649
Note that substitution of a coordinate NP does not rescue this nondistributive, collective interpretation: (63) a. #Many innumerable stars and black holes light-years away are that galaxy at the edge of the universe. b. #Many well-known instrumentalists and vocalists are that charity orchestra. As far as the collective interpretation goes, there is no substitution puzzle. And so, given that the formal distributivity of many NP is itself a mirage, there is as yet nothing to confound a syntax and semantics for coordinate NPs that affirms their equivalences such as (64) and their substitution in (60): (64) musicians[E,X] ↔ instrumentalists and vocalists[E,X] Rather, the substitution puzzle presents itself when coordinate NP rescues a semidistributive interpretation that both bare plurals and plurals with overt articles reject equally (65)–(66) (cf. (69)–(70)): (65) a. #Many stars are a galaxy in some corner of the universe (or another). b. #Many musicians are a charity orchestra in one inner city or another. (66) a. #The/some many stars are (all) a galaxy in some corner of the universe (or another). b. #The/some many musicians are (all) a charity orchestra in one inner city or another. (67) a. Many stars and black holes are a galaxy in some corner of the universe (or another). b. Many instrumentalists and vocalists are a charity orchestra in one inner city or another. (68) a. The/some many stars and black holes are (all) a galaxy in some corner of the universe (or another). b. The/a many instrumentalists and vocalists are (all) a charity orchestra in one inner city or another. The semidistributive interpretation of a friendly collective predicate, as in (69)–(70), allows that the many stars or musicians are distributed among several galaxies of fewer stars or among several charity orchestras of fewer musicians: (69) a. Many stars cluster as a galaxy in some corner of the universe (or another). b. Many musicians fill the roster of a charity orchestra in one inner city or another. (70) a. The/some many stars (all) cluster as a galaxy in some corner of the universe (or another). b. The/some many musicians (all) fill the roster of a charity orchestra in one inner city or another.
650
Chapter 13
That is, it may be that they are scattered among several collective events each of which on its own is as the collective predicate describes it. But, with recalcitrant antisemidistributive predicates, no interpretation of (65)–(66) allows the stars to be scattered among several galaxies or the musicians among several orchestras, and now the puzzle is that substitution of a coordinate NP licenses just such an interpretation in (67)–(68). It is a puzzle about the syntax and semantics of semidistributivity per se. When the subject is uncontroversially a plural (in)definite description as in (70), a tacit or overt semidistributive operator (all) intervenes to include within its scope description of the singular event: each cluster is a galaxy in some corner of the universe, and each roster fills a charity orchestra in an inner city. The remarks assimilating many NP.pl and some many NP.pl to the same syntax imply the intervention of a semidistributive operator across the board when semidistributivity is at issue. The puzzle then reduces to consideration of what about the semidistributive operator and antisemidistributive predicates run afoul of each other and how substitution of a coordinate NP manages to repair it. Note that what has already been said about antisemidistributive predicates is of no use here, namely, that they measure under a singular frame of reference. Granting that, the problem is why the antisemidistributive operator can’t scatter stars or musicians among just such several frames of reference, each of which frames the measure of a galaxy or a charity orchestra. Whatever this reason, why does a combination of this operator with a coordinate NP, stars and black holes or instrumentalists and vocalists, do better, and how is this accomplished without stain on the semantic innocence of and? Essential to the syntax and semantics of semidistributivity (section 13.0.1.1) and to its rescue in (72) after an antisemidistributive predicate has rejected it in (71), (71) #Many elms are a cluster in the middle of the forest. (72) Many elms that are clustered are a cluster in the middle of the forest. is a (tacit or overt) singular descriptive anaphor referring to an event in a domain of events countable enough to support such reference: (73) #Many elms are there a cluster in the middle of the forest. #Many elms are in that place a cluster in the middle of the forest. #Many elms are in such a place a cluster in the middle of the forest. (74) Many elms that are clustered are there a cluster in the middle of the forest. Many elms that are clustered are in that place a cluster in the middle of the forest. Many elms that are clustered are in such a place a cluster in the middle of the forest.
Antisemidistributivity vs. Conjunction Reduction Redux
651
According to (72)/(74), for any of the events or states of being clustered, it is a cluster in the middle of the forest, where presumably the clusterings are discrete and equinumerous with the clusters they are said to be. In contrast in (71)/(73), there are not discrete or countable states of being elms, except at best for the states of being an elm coincident with the lifetime of an individual elm, and none of these is a cluster. Whatever the merits of the proposal, the simple contrast between (71) and (72) is the coup de grâce for rival accounts of antisemidistributivity: it is absurd to explain (71) in terms of a logical type for the antisemidistributive predicate be a cluster—the type for wholier wholes—and have it dissolve away in (72) after a well-chosen relative clause modifies the subject. Correlatively, it is absurd that the contrast between (71) and (72) imply much of a difference in what there are many of, whether the NP is modified or not. Nor is there any semantic innovation in what NP denotes when coordinate NP elms and beeches substitutes for a simple NP: (75) #Many elms are a cluster in the middle of the forest. (76) Many elms and beeches are a cluster in the middle of the forest. It will suffice to assimilate (75)–(76) to (77)–(78), supposing that all NPs are AdrPs, as mentioned at the start of this section, and that all AdrPs in the scope of the same null or overt article address the same frame of reference, as the gloss in (78) suggests: (77) #Many trees where there are elms (somewhere/anywhere) are there / in that place / in such a place a cluster in the middle of the forest. #Many trees where there are elms (in some spots) are there / in that place / in such a place a cluster in the middle of the forest. (78) Many trees where there are elms somewhere/anywhere and beeches elsewhere are there / in that place / in such a place a cluster in the middle of the forest. Many trees where there are elms in some spots and beeches in others are there / in that place / in such a place a cluster in the middle of the forest. It seems that a frame of reference so organized as to fix the location of elms at some addresses and beeches at other addresses presupposes a discrimination of frames of reference both discrete enough to support singular reference to such frames and large enough to comprehend the measure of a cluster of trees therein. The many trees, given the null article, are scattered among plural frames of reference, but any one of these is large enough for elms and beeches and a cluster of them. But with nothing more to go on than that the frames of reference frame one or more elms, distributive singular reference to frames of reference that frame a cluster falls short. If so, there is, as is to be shown, nothing in the contrast between (75) and (76) that warrants a novel sense for and, as should already be implicit from the paraphrases in (77)–(78).
652
13.0
Chapter 13
Semidistributivity
The predicates in (79)–(82) are collective in that the solitary fielder is no huddle and the solitary mobile phone, no network. Yet the sentences allow that the many fielders are not one huddle but as many as there are teammates huddled around, and the many mobile phones more than one network: (79) The many fielders have huddled around a teammate. (80) Some many mobile phones are networked all and only with each other. (81) Many fielders have huddled around a teammate. (82) Many mobile phones are networked all and only with each other. This semidistributive interpretation needs some extra help, an intervening operator (or more) to distribute the many that the (in)definite descriptions the many fielders and some many mobile phone refer to among several collective events, each a huddle or a network. The same holds of many fielders and many mobile phones if they too are plural indefinite descriptions as the preceding section promises. Extra help is still needed, if instead many fielders and many mobile phones are, as section 12.3 presumes, first-order distributive quantifiers, ‘[many x : fielders[x]]’ and ‘[many x : mobile phones[x]]’, in order to find for the individual fielder or mobile phone others of the same kind with whom to participate in the collective event:6 (83) [∃X : ∃E(∃eμ∃n(many(n) count[eμ,E,n]) fielders[E,X]))] [∀x : Xx][∃X : Xx & ∃E fielders[E,X]] ∃E have huddled … [E,X] (84) [∃X : ∃E(∃eμ∃n(many(n) count[eμ,E,n]) mobile phones[E,X]))] [∀x : Xx][∃X : Xx & ∃E mobile phones[E,X]] ∃E … networked … [E,X] (85) [Many x : fielders[x]][∃X : Xx & fielders[X]] ∃E have huddled … [E,X] (86) [Many x : mobile phones[x]][∃X : Xx & mobile phones[X]] ∃E … networked … [E,X] As remarked in the preceding section, some predicates, antisemidistributives,7 reject semidistributivity per se: (87), where the intervention of semidistributive operators is needed, fails to express an equivalent of (88), while (89) manages the equivalent of (90) without it, in that each of what is counted many is a cluster, a fully distributive interpretation:8 (87) #Many elms are a cluster in the middle of the forest. (88) a. Many elms cluster in the middle of the forest. b. Many elms are clustered in the middle of the forest. c. Many elms are in a cluster in the middle of the forest.
Antisemidistributivity vs. Conjunction Reduction Redux
653
(89) Many a one or more elms are a cluster in the middle of the forest. (90) a. Many a one or more elms cluster in the middle of the forest. b. Many a one or more elms are clustered in the middle of the forest. c. Many a one or more elms are in a cluster in the middle of the forest.9 The antisemidistributivity of semidistributivity is treated in section 13.1, preliminary to which is a general account of semidistributivity. First, section 13.0.0 rejects translations like (83)–(86) that rely on plural existential quantification over fielders or mobile phones to distribute them among several collective events. Instead, in section 13.0.1, semidistributivity proves to be singular distributive quantification directly over collective events. 13.0.0
Semidistributivity without plural reference to persistent objects
A pair of arguments (Schein 1993, 52ff., 159ff.) show that semidistributivity never quantifies over pluralities and thus reject ‘[∃X : Xx & Φ[X]]’, the toy translation of the semidistributive operator entertained in (83)–(86). First (Schein 1993, 52ff.), evaluating the semidistributive interpretation of (91) in the context depicted in (92), let S be a plenary caucus of many senators and S1, … , Sn be all its subcaucuses, proper subsets of two or more senators: (91) Many senators cosponsored exactly ten amendments. Many senators cosponsored in some session or other exactly ten amendments. Many senators cosponsored exactly ten amendments in some session or other. Many senators have cosponsored exactly ten amendments. Many senators have cosponsored in some session or other exactly ten amendments. Many senators have cosponsored exactly ten amendments in some session or other. (92) S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S, S1, … , Sn, S The events of (92) are a history of Senate sessions at each of which exactly one amendment is sponsored, and no amendment is sponsored more than once. It is recorded for each session which of the subcaucuses or plenary caucus cosponsor its amendment. The plenary caucus S is the cosponsor of exactly ten amendments over the course of the legislative history depicted. Note that every senator in the plenary caucus has, however, been a cosponsor in the course of (92) of quite a few more than ten amendments. Besides participating in the plenary, every senator has been a member of numerous subcaucuses sponsoring amendments.
654
Chapter 13
In (92), sentences (91) are all false no matter what exactly ten amendments is taken to measure in their acceptable interpretations. If it measures each event, each Senate session, the sentences are of course false since at each only one amendment is sponsored. If, instead, exactly ten amendments takes the measure of a senator’s career as in (93), (91) are again false in that each senator has been a cosponsor of more than ten amendments: (93) F [Many x: senators[x]][Exactly 10 y: amendments[y]] ∃e[∃X : Xx & senators[X]] cosponsor[e,X,y] Sentences (91) pointedly omit the interpretation that would be true here if the measure of a Senate caucus could be taken throughout its legislative history: (94) *[Many x: senators[x]][∃X : Xx & senators[X]][Exactly 10 y: amendments[y]] ∃e cosponsor[e,X,y] The senators of the plenary caucus S are such that they are many and they cosponsor exactly ten amendments in the course of its history. While it is coherent to track the individual senator across his career, including occasions in cooperation with others, semidistributivity does not do the same for the senators of the plenary caucus. To whatever extent the semidistributive interpretation feels like quantifying over pluralities, it is not robust enough to think of them as persisting through events. In contrast to many senators, the plural (in)definite descriptions, the senators or some senators, do just that when they refer directly to a plurality and are not themselves interpreted semidistributively, obtaining interpretations for (95) and (96) that are true in (92) in virtue of S’s history (see (97) and (98)): (95) The senators of a plenary caucus cosponsored exactly ten amendments. (96) Some senators cosponsored exactly ten amendments. (97) [The X: senators[X]][Exactly 10 y: amendments[y]]∃e cosponsor[e,X,y] (98) [Some X: senators[X]][Exactly 10 y: amendments[y]]∃e cosponsor[e,X,y] The complex singular quantification, many a one or more senators, in (99) would also be true, along with (95) and (96) while the distributive quantification in (91) remains false, in a context replicating the events of (92) for many different plenary caucuses: (99) Many a one or more senators cosponsored exactly ten amendments. (100) [Many X : one or more[X] & senators[X]][Exactly 10 y: amendments[y]]∃e cosponsor[e,X,y] As expected, many a one or more senators and not many senators quantifies over pluralities exactly as some senators or the senators does. In contrast, the semidistributive operator bridging many senators and the collective predicate in (91) must have a syntax and semantics that, in effect, stipulates for whatever plural quantification
Antisemidistributivity vs. Conjunction Reduction Redux
655
is invoked, ‘[∃X : Xx & senators[X]]’ in (93) and (94), its narrow scope with respect to exactly ten amendments in object position, ruling out (94). Further considerations (Schein 1993, 159ff.) yield the stronger result that semidistributivity must be expressible without any intrusion from a plural quantifier [∃X : Xx & Φ[X]]. It is a puzzle to reconcile in (101) apparent collective reference to the same vaudevillians dancing and singing with the scope of no more than three ballads that they sang together: (101) No vaudevillians danced together to no more than three ballads that they sang together. (102) *[No x : vaudevillians[x]][∃X: Xx & vaudevillians[X]] … together[X] … [no more than three ballads that X sang together[X]] … Sentence (101) need not be the vacuous falsehood in (102) that no vaudevillian is among some who dance and sing together to no more than three ballads. Of course, (102) is falsified by any vaudevillian—for example, Fanny Brice, for whom there is at least one other whom she has not appeared on stage with, Al Jolson. Rather, (101) means that no vaudevillian has no more than three ballads that he sang and danced to together with other vaudevillians, which would be true if vaudevillians were many years on the circuit with the same troupe performing the same routines. The puzzle is how to extract this meaning, in which the two tokens of together and the pronoun they are coreferent and covariant without occurring within the scope of a plural quantifier binding them all, as in (102). To escape the vacuous falsehood, a semidistributive operator is defined (Schein 1993, chap. 12), which in effect replaces the faulty ⌜[∃X : Xx & Φ [X]]⌝ with a term that refers in (103) to whatever events if any vaudevillian x participated in with other vaudevillians: (102) *[No x : vaudevillians[x]][∃X: Xx & vaudevillians[X]] … together[X] … [no more than three ballads that X sang together[X]] … (103) [No x : vaudevillians[x]] … [℩E: [∃X: Xx & vaudevillians[X]]W[E,X]] … together … [no more than three ballads that they sang together] … The remaining fragment of the sentence then says there are no more than three ballads among these events where the vaudevillians sing and dance together, and the sentence as a whole says that no vaudevillian is so inexperienced as to have song-and-danced to no more than three such ballads. The term referring to events embeds within it a plural quantifier and so denies a plural quantification over vaudevillians that includes within its scope together or the pronoun they. Rather than bound variables, together and they contain descriptive anaphora the content of which draws only on thematic relations to events (see Schein 1993, chap. 8).10 Accordingly, direct plural predication of vaudevillians X plays no role in semidistributive predication. Semidistributivity is an illusion of plurally quantifying over
656
Chapter 13
vaudevillians, which turns instead on quantification over events and on how they are described.11 13.0.1
Semidistributivity via singular reference to collective events
The interaction of semidistributivity and nonincreasing quantifiers, exactly ten amendments and no more than three ballads, is of interest in two respects. First, the existence of the target interpretations, in which the nonincreasing quantifier measures the sum of an individual’s cooperation with others across her lifetime, proves that plurally quantifying over others with whom she cooperates cannot be essential to the semidistributive interpretation, as it collapses into the vacuous falsehoods mentioned.12 Some other method, such as the one suggested above and elaborated below, must be called on to derive semidistributivity for sentences like (91) and (101), repeated here: (104) Many senators have cosponsored no more than ten amendments. (105) No vaudevillians danced together to no more than three ballads that they sang together. The suggestion here will combine in a single adverbial clause what more felicitous paraphrase in (106) accomplishes with separate expressions of distributivity and comitativity: (106) Many senators each have with others cosponsored no more than ten amendments. Many senators each have participating with others cosponsored no more than ten amendments. Many senators each have in the events of their participation with others cosponsored no more than ten amendments. Distributivity, it is, because no more than ten amendments measures the individual senator’s collaborative history. A comitative adverbial phrase is needed to refer to that history that subsumes sundry collaborations with ever-changing legislative allies. The logic and logical syntax of the adverbial clause introduced should recall the comitative phrases discussed in section 2.6, especially those in languages such as Slavic and Spanish where they contribute to number agreement: (107) Marta con María levantaron el piano Martha with Maria lifted.pl the piano (Camacho 1997, 206) (108) Marek z Piotrem podniesli pianino M.nom with P.instr lifted.pl piano ‘Mark with Peter lifted the piano.’ (McNally 1993, 376)
[Spanish]
[Polish]
Antisemidistributivity vs. Conjunction Reduction Redux
657
The logic, as Lasersohn (1995, 70) observes, is that a comitative phrase in contrast to instrumental phrases does not drop salva veritate in an inference to (111): (109) Russell, with Whitehead, wrote the whole of Principia Mathematica. (110) Russell, with a Whitehead pencil, wrote the whole of Principia Mathematica. (111) Russell wrote the whole of Principia Mathematica. The plural number agreement in Slavic and Spanish underlines this point that the comitative phrase shifts the action so that the solitary subject has no relation to cause and effect on the piano except via her participation with others. The second respect that commands interest in the interaction between semidistributivity and nonincreasing quantifiers is that granted that some other method is provided such as the one just sketched, the fact that the vacuous interpretations of the preceding section and their approximations are not freely available as alternative interpretations alongside the targeted ones implies that plurally quantifying over others of the same kind, *‘[∃X: Xx & vaudevillians[X]]’ *‘[∃X: Xx & senators[X]]’, is not an operation that logical form makes freely available. It turns out that it does not exist at all. The illusion of quantifying over pluralities is purchased only by quantifying over events with plural participants, in comitative phrases, which is subject to constraints on event quantification to be introduced shortly. Illusion is unmasked in a contrast between few students and no students or not any students. The former distributive like many students needs the semidistributive operator qua comitative phrase to combine with collective predicates. In contrast, no students or not any students, plural indefinite descriptions, quantify in their own right over pluralities, taking some license to decompose them as ‘¬[∃X: ∃E students[E,X]]…’. This outright quantification over pluralities is exempt from the constraints on event quantification that attach to the semidistributive operator that few students must invoke for its illusion of quantifying over pluralities. The illusion is unmasked when it confronts the real thing as in a minimal pair that follows in (113) and (114).13 Suppose that antiwar demonstrations of 1000 each have taken place on three successive Sundays in front of Battell Chapel on the Old Campus. An antiwar organization with a committed campus membership of 1500 had divided itself into three battalions, Alpha, Bravo, and Charlie, of 500 each. Each Sunday a different battalion was held back in reserve in the event that reinforcements might be needed. Participating in 1969 were (112) 4 May 11 May 18 May Alpha & Bravo Alpha & Charlie Bravo & Charlie
658
Chapter 13
Every one of the 1500 attended exactly two demonstrations. It is unequivocally true that (113) a. Few students gathered only once. Fewer than two students gathered only once. Not more than two students gathered only once. b. Many students gathered at least twice. More students gathered at least twice than (did) not. Most students gathered at least twice. But the following are ambiguous: (114) a. No students gathered only once. b. Not any students gathered only once. One interpretation of (114) entails (113), one that itself conceals a distributive operator; but, another just denies that some students gathered only once, which is a false denial since every Sunday saw a different crowd, and thus the students of any given Sunday did gather only once, namely, that Sunday. This last interpretation exploits the plural reference that (114) affords, ‘¬[∃X: ∃E students[E.X]] …’, which is absent when the quantification is obligatorily distributive as in (113). There is no ambiguity in (113), measuring the frequency of what happened either to an individual student in her history or in the history of some particular students. Yet something that could be mistaken for this last reading can be restored to sentences similar to (113) in describing the individual events. Suppose that each demonstration lasted exactly 1 hour. Sentences (115) are now ambiguous: (115) Few students (have) gathered for no more than 90 minutes. Fewer than two students (have) gathered for no more than 90 minutes. Not more than two students (have) gathered for no more than 90 minutes. Like (113), one interpretation comments on the individual student’s career, an interpretation that is true since every one of the 1500 has been 2 hours in demonstrations. But the sentences have acquired a false interpretation too: the many students have all attended demonstrations each of which is a gathering for no more than 90 minutes. That is, it is false that few students were in gatherings each of which is for no more than 90 minutes. The adverb for no more than 90 minutes measures the individual event, a gathering, with plural participants. In this context, there are three such events, and it is false that few students participated. The reading is to be paraphrased this way rather than that few were among some students who gathered for no more than 90 minutes. Returning to (113), although one can always ask whether some particular students gathered once or twice, it makes little sense to ask about a particular event, a gathering, whether that gathering occurred once or twice, hence no ambiguity is recognized.
Antisemidistributivity vs. Conjunction Reduction Redux
659
If what is at issue is a description of events, then whatever may restrict that description stands to alter the available interpretations. In particular, relative clauses have such an effect, via adverbialization. Consider some of my gangster relatives famous in the 1930s for a series of nine jewelry store burglaries, sometimes solo and sometimes with an accomplice: (116) Lieb & Louie Lieb & Lazar Meyer & Moe Meyer & Murray 15 rubies 15 emeralds 15 sapphires 15 diamonds Lieb 10 topaz
Louie 10 opals
Lazar Meyer 10 amethysts 10 garnets
Moe Murray 10 citrines 10 peridots
(117) Few(er than two) thieves (have) gathered (up) / collected / rounded up / reaped no more than five gemstones. (118) No thieves (have) gathered (up) / collected / rounded up / reaped no more than five gemstones. (119) Not any thieves (have) gathered (up) / collected / rounded up / reaped no more than five gemstones. Sentence (117) is unequivocally true of (116) despite that every thief is among some, one of fifty-four of sixty-three combinations, who have not burglarized any jewelry stores. Note again that these fifty-four falsify an interpretation available to (118) and (119), via their plural reference ‘¬[∃X: ∃E thieves[E.X]]… ’. The contrast again illustrates that few and many are obligatorily distributive while (in)definite descriptions and their negated counterparts are not. Consonant with what was observed above, the semidistributive interpretation measures the sum of the individual thief’s collaborations throughout his career. Yet, even with the distributive quantifier in place, the measure of individual careers can be suspended when a relative clause interrupts: (120) Few(er than two) thieves who never planned (their heists) (have) gathered (up) / collected / rounded up / reaped no more than five gemstones. Cf. Few(er than two) thieves who never planned never gathered up a single gemstone. The unwavering truth of (117) becomes falsehood and an insult to my family of professionals, all of whom, if they never planned for it, never stole. The only reason that any of fifty-four combinations did not is that certain relatives were not on speaking terms enough to plan a bar mitzvah let alone a burglary. Note the implications for a logical form so impoverished that it parses (117) and (120) respectively as ‘[Few(er than two)x: Fx] Hx’ and ‘[Few(er than two)x: Gx] Hx’. In this context, for x to be a thief, ‘Fx’, i.e., one of my six gangster relatives), is for x to be a thief in a state with some others of never having planned burglaries, ‘Gx’. That is to say ‘Fx’ and ‘Gx’ are coextensive in this context, and prefixing these quantifiers to the same
660
Chapter 13
matrix ‘Hx’ should make (117) and (120) equivalent here. Rather, although the restrictions do denote the same individuals, (117) and (120) distinguish themselves with different matrix clauses: (121) Few(er than two) thieves while thieves, (have) gathered (up) / collected / rounded up / reaped no more than five gemstones. (122) Few(er than two) thieves who never planned (their heists) having never planned, (have) gathered (up) / collected / rounded up / reaped no more than five gemstones.14, 15 The relative clause in (120) was chosen to illustrate the power of a relative clause to compel consideration of those combinations of thieves that play no role in evaluating (117) or in the events depicted in (116). But these combinations participate in the states of having no plans that the relative clause describes and the rest of the sentence evaluates what might or might not have happened as the aftermath of not planning. A relative clause may equally well restrict evaluation of the individual’s collaborative history to what has happened within the events of (116). Rather than the career, a salient interpretation of (123) and (124) measures only the individual burglary, which becomes the favored interpretation when the event in a jewelry store is described in (124) rather than the nine in jewelry stores as in (123): (123) Few thieves who (have) ransacked / emptied local jewelry stores (have) gathered (up) / collected / rounded up / reaped more than fifteen gemstones. (124) Few thieves who (have) ransacked / emptied a local jewelry store (have) gathered (up) / collected / rounded up / reaped more than fifteen gemstones. What falls short of more than fifteen gemstones is each of the nine burglaries. As (125) and (126) make plain with an explicit adverb, the matrix predicate is interpreted as a comment on the individual burglary: (125) Six thieves who emptied local jewelry stores (at some time or another / at different times / on several occasions) then & there / immediately / soon gathered (up) / collected / rounded up / reaped no more than fifteen gemstones. (126) Six thieves who emptied local a jewelry store (at some time or another / at different times / on several occasions) then & there / immediately / soon gathered (up) / collected / rounded up / reaped no more than fifteen gemstones. The intervening relative clause has a perhaps unexpected power in commandeering the interpretation of the matrix predicate to keep it from applying to entire careers.
Antisemidistributivity vs. Conjunction Reduction Redux
661
If there is an intervening relative clause, its temporal reference must broaden to an entire career as in (127) and (128) in order to recover the interpretation salient in (117) when there is no relative clause: (127) Few persons of interest who are thieves that (have) ransacked/emptied local jewelry stores (have) gathered (up) / collected / rounded up / reaped more than fifteen gemstones. (128) Six persons of interest who are thieves who emptied local jewelry stores (at some time or another / at different times / on several occasions) gathered (up) / collected / rounded up / reaped no more than fifteen gemstones. I assume, as with any sequence of tensed clauses, one is hard pressed to interpret tense in the second except in reference to tense in the first, and so it is when the matrix clause is preceded by a tensed relative clause: tense in the matrix clause is a temporal pronoun that finds its antecedent in the tense of the relative clause. If the matrix tense is taken to be singular, for reasons of aspect or overt morphology, it is a covariant descriptive anaphor (Partee 1973, 1984):16 (129) Few thieves that emptied a local jewelry store immediately / soon / within 3 minutes after entering it gathered up more than fifteen gemstones. (130) Six thieves who emptied a local jewelry store immediately / soon / within 3 minutes after entering it gathered up no more than fifteen gemstones. (131) Few thieves that emptied local jewelry stores immediately / soon / within 3 minutes after entering gathered up more than fifteen gemstones. (132) Six thieves who emptied local jewelry stores immediately / soon / within 3 minutes after entering gathered up no more than fifteen gemstones. The semidistributive operator phrase always refers to the individual’s lifetime collaboration indifferent to whom with. If, as mentioned above, the operator is restricted to events with a solitary participant, a fully distributive operator is derived. Referring to episodes properly within a lifetime history depends on the success of a descriptive anaphor to find an antecedent description that manages to refer to those episodes, which then frame or locate in their neighborhood the events described by the matrix. This anaphor with respect to its antecedent events may itself be singular and hence distributive or not, as (123), (124), and (129)–(132) illustrate, leaving (133) and (134) with two dimensions of variation (see sections 13.0.1.0 and 13.0.1.1): (133) Few thieves that emptied local jewelry stores pocketed more than fifteen gemstones. (134) Few thieves that emptied a local jewelry store pocketed more than fifteen gemstones.
662
Chapter 13
If the semidistributive operator is restricted to be fully distributive and a singular temporal anaphor relates the pocketing to a burglary at the individual jewelry store, more than fifteen gemstones measures the individual thief’s take at the burglary. If, instead, the temporal anaphor makes plural reference to the jewelry store burglaries, more than fifteen gemstones measures the individual thief’s personal career earnings.17 If the semidistributive operator is left as it is to derive reference to an individual thief’s collaborative history, combining it with a singular temporal anaphor obtains that interpretation where more than fifteen gemstones measures an individual jewelry store’s loss from a burglary. Finally, if the semidistributive operator is left properly semidistributive and the temporal anaphora refers plurally to the burglaries, more than fifteen gemstones measures the total number of gemstones that a thief has pocketed in collaboration with others in the course of emptying jewelry stores, the interpretation first introduced to argue that semidistributivity is not plural quantification over thief with other thieves. 13.0.1.0 Reference to collaborative vs. individual histories
A semidistributive operator is defined in (135) and followed with some comments on its technical details:18 (135) Σ satisfy ⌜[ALLCO Ei : Φ] Ψ⌝ ↔df [℩E: ∃Σ′(Σ′ ~i Σ & Σ′(⌜Ei⌝)=E & Σ′ satisfy Φ)] (i) [∀e: Ee] (ii) [℩ECO: ∀eCO(EcoeCO ↔ (iii) [∃E′: E′e & [∀e′: E′e ′] Ee′] (iv) [∃E″ : N[E′,E″] & ∀X(W[E′, X] ↔ W[E″, X])]E″eco)] (v) (vi) ∃Σ′(Σ′ ~i Σ & Σ′(⌜Ei⌝)= ECO & Σ′ satisfy Ψ) First, distributivity, ‘[∀e0: E0e0]’, here and elsewhere (e.g., all in (136)–(137)), applies to the events E0 for which ∃X passengers[E0,X] rather than to the objects X that ∃E0 passengers[E0,X]: (136) Many passengers (all) had an empty seat between them. (137) Some/the many passengers (all) had an empty seat between them. Surely all of us who have flown frequently in the past 10 years have been on a flight with an empty seat beside us. But if due to overbooking and overcrowding, it almost never happens, (136) is false, especially as the advertisement the airline intends. The items counted many are the items to satisfy any distributive or semidistributive condition. Each of whatever is many is what matters: (138) [Many x : ∃X(Xx & ∃E0 passengers[E0,X])]19 [ALLCO E0 : ∃X passengers[E0,X]] [∃E1: N[E0,E1]]W[E1, X] O[E1, E2] ∃XPossessor[E2, X] Overlap[E2,E3]have[E3] …
Antisemidistributivity vs. Conjunction Reduction Redux
663
(139) [Some/The X: ∃E0 (many[E0] passengers[E0,X])] [ALLCO E0 : ∃X passengers[E0,X]] [∃E1: N[E0,E1]]W[E1, X] O[E1, E2] ∃XPossessor[E2, X] Overlap[E2,E3]have[E3] … Suppose (136) and (137) happen to be true of a particular flight with 200 passengers on board seated in 100 pairs with an empty seat in between. For (135), the Ei in this instance number 200, and of course these 200 passenger events each have a single participant. The definite description that follows in (135), lines (iii)–(v), ‘[℩ECO: …],’ refers for each passenger event to whatever (if anything) the participant in that event did in the neighborhood of passenger events jointly or severally with the participants in one or more of these passenger events. As remarked earlier, if the semidistributive operator co-occurs with a singular NP such as every passenger, it will reduce to the fully distributive, given some assumptions about the meaning of singular count nouns such as passenger. As assumed for adverbialization throughout, the content Φ of the semidistributive operator ⌜[ALLCO Ei : Φ]⌝ in (135) is fixed by the NP of its antecedent DP—singular passenger when its antecedent is every passenger, for which two alternative parses are entertained: (140) [Every x : ∃X(Xx & ∃E0 passenger[E0,X])][ALLCO E0 : ∃X passenger[E0,X]] … (141) [Every x : ∃E0 passenger[E0,x])][ALLCO E0 : passenger[E0,x]] … For any given Φ, the first line of (135) denotes those events each of which is to be considered with perhaps some others among them. For Φ in (140), [℩E0: ∃X passenger[E0, X]] are these events, and for (141), they are [℩E0: passenger[E0, x]], with singular count noun in place. Recall from section 12.1 that to count dancers rather than their pas de deux or passengers rather than passages, these count nouns are taken to denote solo events. That is, (142) solo[E] ↔df ∀e∀x0∀x1(Ee → (participate(e,x0) → (participate(e,x1) → x0=x1))) (143) dancers[E,X] ↔ solo[E] & participate[E,X] & ∃E′(-er[E,E′] & dances[E′]) (144) passengers[E,X] ↔ solo[E] & participate[E,X] & ∃E′(-er[E,E′] & passages[E′]) Singular reference to that dancer or this passenger, I do not take to be so severe as to restrict itself to a single event of being a dancer or passenger. It is rather to refer to the events of such for the same one participant. Let encore events be those with all the same participants, and let singly denote encore, solo events: (145) encore[E] ↔df ∀e0∀e1((Ee0 & Ee1) → ∀x(participate(e0,x) ↔ participate(e1,x))) (146) singly[E] ↔df solo[E] & encore[E]
664
Chapter 13
The singular count noun is just the singular of the plural count noun in meaning: (147) passenger.sg[E,X] ↔ passengers[E,X] & singly[E] With every passenger set in the logical form in (141), the semidistributive operator (135) looks for comitativity among the events [℩E0: passenger.sg[E0, x]] for given x. But these are solo, encore events that can therefore amount to no more than a collaboration with the same one x, wherein the semidistributive operator reduces to full distributivity, as it should if it is allowed to occur within the scope of singular every passenger or each passenger.20 Consider next the alternative logical form (140), for which the semidistributive operator looks to the events [℩E0: ∃X passenger.sg[E0, X]]. Here there is simply a failure of reference of the definite description, as there are no events E0 that are solo and encore yet include all passengers among their participants. This too is as it shoud be, as no semidistributive operator should afford to *every passenger had an empty seat between them the otherwise coherent meaning that every passenger with another passenger had an empty seat between them. Recall ((133)–(134)) that one dimension of the ambiguity in using plural DPs (e.g., many passengers or some/the many passengers) alternates semidistributivity and distributivity, which requires supplementing the semidistributive operator to represent that the passenger lifted the overweight bag by herself: (148) Many passengers (all) lifted an overweight bag into the overhead bin. (149) Some/the many passengers (all) lifted an overweight bag into the overhead bin. Perhaps a separate fully distributive operator is to be defined, an each alongside the semidistributive all-with-others and stipulated to be in complementary distribution with it. Alternatively, allow that singular morphology may be introduced into the semidistributive operator itself with further repair (cf. “telescoping,” Poesio and Zucchi 1992) to provide an antecedent for the singular morpheme: (150) [Many x : ∃X(Xx & ∃E0 passengers[E0,X])] [ALLCO E0 : passengers.SG[E0,x]] [∃E1: N[E0,E1]]W[E1, X] O[E1, E2] ∃XAgent[E2, X] Cause[E2,E3]lift[E3] … (151) [Some/The X: ∃E0 (many[E0] passengers[E0,X])] [∀x: Xx][ALLCO E0 : passengers.SG[E0,x]] [∃E1: N[E0,E1]]W[E1, X] O[E1, E2] ∃XAgent[E2, X] Cause[E2,E3]lift[E3] … 13.0.1.1 Singular reference to events
Recall again that the (semi)distributive operator sweeps through a lifetime referring for each of the many to all of whatever she did, if anything, alone or with help from
Antisemidistributivity vs. Conjunction Reduction Redux
665
others. As remarked in the discussion of (133) and (134), referring to an episode properly contained within a lifetime of collaboration depends on the success of a descriptive temporal anaphor to find an antecedent description that manages to refer to that episode (see (123), (124), (129)–(132)): (133) Few thieves that emptied local jewelry stores pocketed more than fifteen gemstones. (134) Few thieves that emptied a local jewelry store pocketed more than fifteen gemstones. (152) [Few x : ∃X(Xx & ∃E0 thieves that emptiedi local jewelry stores[E0,X])] [ALLCO E0 : ∃X thieves that emptiedi local jewelry stores [E0,X]] [∃E1: N[E0,E1]]W[E1, X] [℩E2: then&therei[E2]] O[E1, E2]∃XAgent[E2, X] Cause[E2,E3] pocket[E3] … The interpretation of the (spatio)temporal anaphor, ‘[℩E2: then&therei[E2]]’, singular or plural, codependent or not is the gateway for variation in semidistributive interpretations. If a measure such as more than fifteen gemstones is somehow understood to measure something less than a lifetime of collaboration, it is only because the (spatio)temporal pronoun refers to it. Recall the demonstrations outside Battell Chapel, of which (154) is false since every student attended two: (153) 4 May 11 May 18 May Alpha and Bravo Alpha and Charlie Bravo and Charlie (154) Many students gathered only once. The battalions had their own commanders, but the antiwar organization, anxious to maintain the peace, organized the battalions into brigades under the command of a single officer responsible for the day’s theater of operation. The Alpha and Bravo battalions formed the AlephBeth brigade; the Alpha and Charlie battalions, the AlephGimel brigade; and the Bravo and Charlie battalions, the BethGimel brigade. Every brigade demonstrated only once; thus, it is true to say, at least under one reading, paraphrased in (156), that (155) Many students who belonged to the AlephBeth brigade gathered only once—on May 4. (156) a. Many students who belonged to the AlephBeth brigade then/therein gathered only once. b. Many students who belonged to the AlephBeth brigade, (while) being students who belonged to the AlephBeth brigade, (while) belonging to the AlephBeth brigade gathered only once.
666
Chapter 13
In (156b), the first adverbial phrase corresponds to the obligatory adverbialization of NP, and the second, to an interpretation of the spatiotemporal anaphors, overtly pronounced in (156a). Despite the spatiotemporal vocabulary, then and therein, it should first be acknowledged that the state of belonging to the AlephBeth brigade is of dubious spatiotemporal position, except perhaps for the position of the brigade itself when in the field. Even temporally, there is no suggestion that membership in the battalions or brigades was not constant and simultaneous throughout May 1969. Nevertheless, spatiotemporal vocabulary, including my use of the neighborhood relation N, is recruited to navigate logical space: (157) While belonging to the AlephBeth brigade, they gathered only once. (158) Belonging to the AlephBeth brigade, they gathered only once. Both (157) and (158) convey that they gathered only once qua members of the AlephBeth brigade. A descriptive point often repeated here is that the adverbial phrases derived from the descriptive contents of NPs, show the same range of modification relations attested elsewhere, such as in (157) and (158). Sentence (155) also has an interpretation, which, while making a claim about only 1000 students, is false like (154) in that every one of them has demonstrated twice. The matrix tense need not be anaphoric to the antecedent relative clause, instead referring directly to a past moment, say, spring 1969. Even if there is some anaphoric relation asserting that the time referred to overlaps or coincides with the time described by the relative clause, that reference might in this case serve indeed only to fix the time of the gathering. If so, there is nothing about reference to events or times in the matrix clauses of (154) and (155) to tell them apart. Note that a relative clause describing the same students as the one in (155) nevertheless fails to rescue a (nonvacuous) truth to the extent that a causal or qua relation makes little sense: (159) (160) (161) (162)
Many students who gathered on May 4 gathered only once. (While) gathering on May 4, they gathered only once. Many students who demonstrated on May 4 gathered only once. (While) demonstrating on May 4, they gathered only once.
It is not that qua May 4 demonstrators, they gathered only once, or that gathering May 4 caused them to gather only once, or that in satisfaction of the conditions of gathering or demonstrating May 4, they gathered only once. Considering next purely temporal relations, for the given context and domain of three events of which it is known that only one is on May 4, reference in the matrix back to the time of the May 4 demonstration is about as good and vacuous in (159) and (160) as it is in (163) and (164): (163) Many students who gathered only once on May 4 gathered only once on May 4. (164) Many students who demonstrated only once on May 4 gathered only once on May 4.
Antisemidistributivity vs. Conjunction Reduction Redux
667
In other contexts that leave in play the number of demonstrations on May 4, (159) and (160) of course have the sensible and informative interpretation that many students demonstrating May 4 participated in only one of the demonstrations occurring that day, where once again a spatiotemporal anaphor in the matrix clause manages to restrict what measures as only once to events properly within the lifetime collaboration of the May 4 demonstrators. With three demonstrations and the (maximal) state of belonging to the AlephBeth brigade, the sentences above draw on a domain of countable events. In general, a semidistributive interpretation distributes the subject’s referents among several collective actions each described by the matrix clause only if those actions belong to a domain of countable events each of which has plural participants. Sentence (165) reports on the rolling production of a season of Seinfeld with screenwriters in continuous scene-to-scene rotation and collaboration: (165) Many screenwriters who scripted 484 minutes of sitcom for twenty-two episodes in eight sessions consumed no more than fifteen urns of coffee. The sentence is a ready report either that the screenwriters’ collective consumption was no more than fifteen urns of coffee that season, or that each of them consumed individually no more than fifteen urns of coffee for the season. Looking for semidistributive events—that is, collective events with more than one participant—the salient such events that are countable (i.e., discrete) are just two, the season’s scripting and the season’s consumption. Anything else is an effort. Perhaps the season can be parsed as eight scriptings in eight separate sessions. For the speaker to be understood to intend this domain of eight events depends on details of syntax—the position of in eight sessions—intonation and context. I suppose if the number of sessions is itself in question and (165) is uttered with a rising intonation on “eight,” the speaker can use it to say that the sessions were in fact eight, and in each no more than fifteen urns of coffee were consumed: (166) Many screenwriters who scripted 484 minutes of sitcom for twenty-two episodes in eíght sessions in them each consumed no more than fifteen urns of coffee. Or maybe not. The problem for speaker and hearer, if no more than fifteen urns of coffee is to measure consumption spanning less than the entire season, is to spell out the reference of a plural count pronoun, as in (166), referring to the lesser events to be measured distributively. Moreover, it is necessary but not sufficient for the events to be countable. They must also frame a plausible protocol for measuring consumption. The twenty-two episodes are counted, and it is easy enough to parse the scripting, where, when, and by whom, into twenty-two events according to which script the word contributed at the moment belongs to. But if the episodes were
668
Chapter 13
scripted in fragments, with the scripting of the fragments of any one episode scattered, interleaved, and overlapping the scripting of other fragments, none of these twenty-two events occupies a continuous spatiotemporal region. It is hard to imagine the point or protocol for measuring coffee consumption in those events, and correspondingly hard to imagine that these events are intended. Even if the episodes are scripted serially but some are begun in one session and finished in the next, it is not obvious why something beginning one evening, ending for the night, and resuming the next morning is to be considered as a single event of coffee consumption. With some imagination the question can be answered: coffee consumption measures stress and correlatively work effort, effort on a sitcom is an indicator of its expected quality, and therefore that episode the scripting of which consumed the most coffee is predicted to be the funniest: (167) Many screenwriters who scripted twenty-two episodes consumed no more than fifteen urns of coffee. So, none of those screenwriters are going to win an Emmy for any of those twenty-two episodes they collaborated on. Absent such an understanding between speaker and hearer, it is unlikely that (167) is intended to measure the coffee consumed scripting the single episode. For the Seinfeld sentences, the events that the relative clause denotes do not vary with screenwriters, whether they are the season, eight sessions, or twenty-two episodes. It is not that every screenwriter is involved with twenty-two episodes in eight sessions that possibly differ from those of any other screenwriter. When the interpretation of the relative clause includes such covariation, pronouns referring back to the events described are likewise covariant, and the condition remains in force that their domain be countable. For (168)–(173), champagne is often shared at tables in the first-class dining room on a cruise ship, with first-class passengers sometimes table-hopping during dinners of several courses. Of a sedate evening with no table-hopping, it is easy enough to understand (168a) as measuring a table’s total champagne consumption for the evening’s duration. That is what constitutes a sharing. The sentence is still ambiguous, measuring either the table’s consumption at no more than five gulps or that of each passenger at that table. Sentence (168b) is about the rate at which champagne flows at a table throughout dinner: (168) a. Many first-class passengers who shared champagne drank no more than five gulps. b. Many first-class passengers who shared champagne drank no more than five gulps before more was ordered. With table-hopping, as before, what happens at a single table may be thought of as a sharing, no matter that its participants might not be in it for the duration. But
Antisemidistributivity vs. Conjunction Reduction Redux
669
table-hopping introduces another dimension for ambiguity. Sharings are not the same sharing if their participation differs. For any first-class passengers, what they share with each other in the course of the evening could thus count as a sharing. Because of the table-hopping, these are gerrymandered events no one of which needs to occupy a continuous stretch of time or even of space. Yet, if champagne consumption were a measure of mutual trust and camaraderie, these are the events that would be assessed in evaluating (168a) to determine how much certain groups trust and like each other. Reference to these events becomes more implausible to the extent that they are meant to frame or locate a physical, ongoing process such as the champagne’s flow rate, (168b). Given that (168) describes events of sharing champagne, countability implies a certain temporal condition whether the events are individuated by table or by their participants. If, say, by table, all the champagne sharing at a given table that evening belongs in some event or another in the intended domain of events, and among the events in the intended domain, none of the champagne sharing at a given table overlaps. The champagne that flows at a given table continuously and without interruption all evening participates in a salient single sharing: no subinterval of that event occurs alongside the larger event in any countable domain of events. If the flow should pause for food or replenishment, the speaker and hearer may regard these interruptions as irrelevant for what is still to be considered a single event spanning the evening at a given table, or it may be that what happens between two pauses is to be regarded as indeed one event of sharing champagne. If the latter, the temporal condition implies that no event properly containing a pause can also belong to the intended domain of countable events. The contexts where the many smaller events are intended are special, since the description itself of sharing champagne does nothing to evoke the smaller events intended. But imagine that at the time of this comment, speaker and hearer witness that the champagne at a given table seems to arrive in waves of varying and uncertain amount, then demonstratively, (168) can be used to mean that many champagne-sharing passengers drank no more than five gulps per wave. The physical circumstances, pauses or waves, afford demonstrative reference to countable, smaller events that the description itself does not hint at. Physical separation is not however what is in general important. It is rather that the basis for event measurement be determinate for speaker and hearer. The champagne may in fact flow without interruption, but I assume that prior arrangement between speaker and hearer could make the successive 10-minute intervals the intended domain of events respecting the temporal condition, in which case no more than five gulps applies to the 10-minute interval. Sometimes the relative clause itself describes a domain of discrete countable events. Singular pronouns in the matrix referring back to these events provide classic instances of covariant descriptive anaphora. A magnum is a 1.5-liter bottle. The
670
Chapter 13
description in (169) thus parses the champagne sharing into discrete events, no matter how many magnums flow at a given table. These events are available for plural count reference and distributive quantification. To each of them the measure no more than five gulps applies and the pronoun it refers to the magnum shared in the individual event: (169) Many first-class passengers who shared a magnum of champagne drank no more than five gulps (from it) (before more was ordered). With the bare plural in (170), the events referred to, similar to (168), may each coincide with all the champagne shared at a given table, or it may, like (169), parse this into smaller events each of which consumes a single magnum. (170) Many first-class passengers who shared magnums of champagne drank no more than five gulps (before more was ordered). This latter meaning is not as felicitous expressed by (170) as it is by (169), partly because of the distracting ambiguity, but also, as suggested by the contrast that follows, because of the spatiotemporal and casual factors that go into event segmentation, the conditions that govern the perception of one or many events, independent of its description. It seems to me that (171) (cf. (133)) more readily measures the theft at the individual jewelry store than (170) measures sharing the individual champagne magnum, despite the formal similarity of the sentences: (171) Many thieves that emptied local jewelry stores pocketed no more than fifteen gemstones. What happens from one jewelry store to the next is scattered in time and transports its participants from one location to the next. It is obvious that these are distinct, discrete events both spatiotemporally and in consideration of their local causes and effects, and it is correlatively an effort to think of these as a single event. None of this is true to support the recognition of distinct events as the passengers at a given table share one magnum after another. If this is what is meant, better stick to (169). In mentioning the magnum, a bottle, (169) makes it straightforward how it should apply to those tables at which four magnums are shared. There are four events to be measured there. What is straightforward is lost when the bottle is replaced by an equivalent amount, despite the exactness of that amount: (172) Many first-class passengers who shared 1.5 liters of champagne drank no more than five gulps (before more was ordered).
Antisemidistributivity vs. Conjunction Reduction Redux
671
Here, without further supplement, it is indeterminate how to individuate events of sharing 1.5 liters at a table where 6 liters are shared, and a vague amount exacerbates the problem: (173) Many first-class passengers who shared liters of champagne drank no more than five gulps (before more was ordered). Of course, the thesis that semidistributivity rests entirely on plural reference to events (with multiple participants) finds encouragement in these judgments that quiver in exquisite sensitivity to conditions on event segmentation and the successful description of discrete, countable events. 13.1
A null determiner
The construction ⌜[Many Φ] Ψ⌝, without overt article, and its related comparative quantification, few, most, more … than, etc. (Hackl 2001a, 2001b, 2009), contrast with its counterpart plural (in)definite descriptions with an article present, ⌜[The/ some many Φ] Ψ⌝.21 The former exclude from both nucleus (see (87)) and restriction (see (174)) the antisemidistributive predicates, the so-called genuine collective predicates, which (in)definite descriptions admit (see (176) and (177)): (87) *Many elms are a cluster in the middle of the forest. (88) a. Many elms cluster in the middle of the forest. b. Many elms are clustered in the middle of the forest. c. Many elms are in a cluster in the middle of the forest. (174) *Many elms that are a cluster cluster in the middle of the forest. (175) a. Many elms that cluster cluster in the middle of the forest. b. Many elms that are clustered cluster in the middle of the forest. c. Many elms that are in a cluster cluster in the middle of the forest. (176) a. The many elms destroyed by Dutch elm disease in New York were a cluster in the middle of Central Park. b. Some very many elms that were destroyed by Dutch elm disease were a cluster in the middle of Central Park and the oldest in North America. Many others perished throughout the city. (177) a. The many elms that were a cluster in the middle of Central Park were destroyed by Dutch elm disease. b. Some very many elms that were a cluster in the middle of Central Park were the oldest in North America and perished from Dutch elm disease. I take it as imperative that the same morpheme many occurs throughout many elms, the many elms, some many elms. I have promised to let this morpheme express the
672
Chapter 13
arithmetic concept of being many, aligning it in its syntax and semantics with morphemes such as twelve to be predicative and not intrinsically distributive, so that the above are all plural, (in)definite descriptions:22 (178) a. [∃X: ∃E∃f(some/the[E] there[E, f] (many[E]) elms[E,X])] b. [∃X: ∃E∃F( there[E,F] (many[E]) elms[E,X])] Semidistributivity for both some/the many elms and many elms derives instead from a null semidistributive operator ⌜[ALLCO Ei : Φ] Ψ⌝ (section 13.0.1.0) with wide application elsewhere as in (179), for which it is independently necessary: (179) According to the latest Tonight Show survey, 2480 lovers first met in an elevator—but not the same elevator. If this appeal to a semidistributive operator is to gain any purchase on the problem, there must be some understanding of why the semidistributive operator itself excludes antisemidistributive predicates. It is not after all incoherent to assert either (180) or (181), and still the semidistributive operator tokened in (179) does not reappear in (182) to divide the reference of the subject among several duos. (180) Batman with Robin is/are a dynamic duo. (181) Each with another is/are a dynamic duo of amazing superheroes. (182) *The six superheroes are a dynamic duo. This aspect of the semidistributive operator is discussed in section 13.2 and it has to do with the countability of the events of whatever x did with others that has already been introduced in section 13.0.0. Accepting this promissory note on the semidistributive operator, the exclusion of antisemidistributive predicates from ⌜[Many Φ] Ψ⌝ but not from ⌜[The/some many Φ] Ψ⌝ derives from something further. To the extent that many shares its syntax and semantics with any other cardinal such as twelve in ⌜[Twelve Φ] Ψ⌝, it remains unexplained why it alone suffers from this condition. It can be argued that the full distributivity attaching in this construction to both Φ and Ψ is the effect of some independent operator, independent of many. The contrast between the construction without overt article and its counterparts with articles present ((87)/(174) vs. (176)/(177)) extends to several (several elms vs. the/ some several elms), numerous, various, varied, assorted, scattered, and so on, and also to contrast existential bare plurals as in Elms cover the hillside and some-plurals, Some elms cover the hillside. To escape extensive lexical ambiguity, to conform to a morphology that in several cases cannot be mistaken for determiners (e.g., numerous, various, varied, assorted, scattered), and to subsume the bare plurals, antisemidistributivity in the construction without overt article should be factored out and into an independent operator. None of the lexical items with which it occurs are them-
Antisemidistributivity vs. Conjunction Reduction Redux
673
selves intrinsically distributive, in accordance with the underlying concepts they express. The following illustrate the generalization of the construction to other lexical items. Sentences (183) and (185) exclude antisemidistributive predicates in both restriction and nucleus with numerous, various, and assorted when the article is absent. Its presence in (187) and (188) allows them: (183) *Numerous/various/assorted sports legends honored last night are a Jewish Hall of Fame. (184) a. Numerous/various/assorted sports legends honored last night are in a Jewish Hall of Fame. b. Numerous/various/assorted sports legends honored last night make up a Jewish Hall of Fame. (185) *Numerous/various/assorted sports legends who are a Jewish Hall of Fame were honored last night. (186) a. Numerous/various/assorted sports legends who are in a Jewish Hall of Fame were honored last night. b. Numerous/various/assorted sports legends who make up a Jewish Hall of Fame were honored last night. (187) a. The numerous/various/assorted sports legends honored last night are a Jewish Hall of Fame. b. Some numerous/various/assorted sports legends honored last night are a Jewish Hall of Fame. (188) a. The numerous/various/assorted sports legends who are a Jewish Hall of Fame were honored last night. b. Some numerous/various/assorted sports legends who are a Jewish Hall of Fame were honored last night. The topological contiguous also appears in constructions that contrast the absence and presence of articles. Suppose that four contiguous Great Plains states west of the 100th meridian coincide with four contiguous geological zones. The states are four squares flanking longitude 109o W and latitude 37o N, and a 45-degree rotation of these lines of latitude and longitude defines the boundaries for the four geological zones. Consequently, no state is a geological zone, and no geological zone is a state, although it remains that the four geological zones are the four states, collectively. Sentences (189) and (190) are either false in implying that a zone is a state or unacceptable in attempting to use a truly collective predicate in contexts that are
674
Chapter 13
obligatorily distributive. The targeted interpretations are, however, conveyed in (191) and (192) with articles present: (189) *Contiguous geological zones without surface boundaries are Great Plains states west of the 100th meridian. (190) *Contiguous geological zones that are Great Plains states west of the 100th meridian are without surface boundaries. (191) a. The contiguous geological zones without surface boundaries are Great Plains states west of the 100th meridian. b. Some contiguous geological zones without surface boundaries are Great Plains states west of the 100th meridian. (192) a. The contiguous geological zones that are Great Plains states west of the 100th meridian are without surface boundaries. b. Some contiguous geological zones that are Great Plains states west of the 100th meridian are without surface boundaries. The striking opposition of bare plural existentials and some-plurals is attested in the contrast between (193) and (197b) and between (195) and (198b). Here there is nothing in prenominal position expressing a numeric, topological, or any other distributional property: (193) *Unsolved murders were a cluster near the Green River in Washington. (194) a. Unsolved murders were clustered near the Green River in Washington. b. Unsolved murders were in a cluster near the Green River in Washington. c. Unsolved murders formed a cluster near the Green River in Washington. (195) *Unsolved murders that were a cluster had a common modus operandi. (196) a. Unsolved murders that were clustered had a common modus operandi. b. Unsolved murders that were in a cluster had a common modus operandi. c. Unsolved murders that formed a cluster had a common modus operandi. (197) a. The unsolved murders were a cluster near the Green River in Washington. b. Some unsolved murders were a cluster near the Green River in Washington. (198) a. The unsolved murders that were a cluster had a common modus operandi. b. Some unsolved murders that were a cluster had a common modus operandi. There is little temptation to treat unsolved as a determiner or quantifier, and thus the distributivity that distinguishes the existential bare plural and some-plural must come from elsewhere, from an operator that is presumably available to the construc-
Antisemidistributivity vs. Conjunction Reduction Redux
675
tions also without overt article containing many, numerous, various, assorted, and contiguous.23 All of this is warrant for a null operator that occurs when no overt article does, thus assimilating ⌜[Many Φ] Ψ⌝ to ⌜[Det many Φ] Ψ⌝. The contrast in antisemidistributivity is related to a more basic difference of meaning that separates (in)definite descriptions with overt articles from their articleless counterparts. It is so in the Fantasy Baseball leagues of Brooklyn that (199) Baseball legends in the hundreds were Jews. (200) Baseball legends by the hundreds were Jews. (201) Baseball legends hundreds at a time were Jews. One even loses count: (202) Untallied baseball legends in the hundreds were Jews. (203) Untallied baseball legends by the hundreds were Jews. (204) Untallied baseball legends hundreds at a time were Jews. Sentences (199)–(204) advance no generic claim about the nature of baseball legends but a more modest existential one about the number of those that just happen to have been Jewish. The sentences nevertheless command a certain authority as a summary of measurement extended in time in a complete survey of baseball legends, which when in progress registered them as Jewish hundreds at a time. It is as if a description of the measurement itself occurred under imperfective aspect: while counting baseball legends, hundreds were being counted (until untallied in the end). A premise that the counting was recent enough for what was so to still be so allows an inference from (199)–(204) to their present-tense counterparts (205)–(210), without implying that something is still ongoing hundreds at a time. The adverbial modifies the counting, which can be understood to have occurred in the past despite the present tense of the assertion:24 (205) Baseball legends in the hundreds are Jews. (206) Baseball legends by the hundreds are Jews. (207) Baseball legends hundreds at a time are Jews. (208) Untallied baseball legends in the hundreds are Jews. (209) Untallied baseball legends by the hundreds are Jews. (210) Untallied baseball legends hundreds at a time are Jews. In contrast, the indefinite descriptions in (211)–(222) preclude that the count went hundreds at a time, forcing the adverbial to modify the matrix predicate, be Jews, which is anomalous when the adverbial describes a process in progress. Sentence
676
Chapter 13
(216) is no better than an assertion of some baseball legends that they are Jews hundreds at a time. (211) Some baseball legends in the hundreds were Jews. (212) #Some baseball legends by the hundreds were Jews. (213) #Some baseball legends hundreds at a time were Jews. (214) Some baseball legends in the hundreds are Jews. (215) #Some baseball legends by the hundreds are Jews. (216) #Some baseball legends hundreds at a time are Jews. In (217)–(222), untallied is less a current summary of the speaker’s own experience of measurement than an allusion to some particular failed or missing measurement perhaps by some third party, as in saying that some baseball legends untallied in the official records but whom I know to number in the hundreds were Jews. (217) Some untallied baseball legends in the hundreds were Jews. (218) #Some untallied baseball legends by the hundreds were Jews. (219) #Some untallied baseball legends hundreds at a time were Jews. (220) Some untallied baseball legends in the hundreds are Jews. (221) #Some untallied baseball legends by the hundreds are Jews. (222) #Some untallied baseball legends hundreds at a time are Jews. This understanding of untallied rescues (217) and (220) from the contradiction that there is no tally and it is in the hundreds. The threat of contradiction is a consequence of the structure. In contrast to bare existential plurals, an indefinite description does not allow that untallied describe a state that is the result of measurements proceeding hundreds at a time according to the adverbial phrase. If the adverbial phrase is not to incoherently modify be Jews, both it and untallied apply to the count tout court without any sense of extended process and result. It appears that (untallied) baseball legends and some (untallied) baseball legends express measurement under imperfective and perfective aspect respectively. With respect to this classification, 613 baseball legends joins the perfective (in)definite descriptions, and many baseball legends, the imperfective existential bare plural, conveying that a counting of many proceeding hundreds at a time found baseball legends who were Jews:25 (223) Many baseball legends by the hundreds were/are Jews. (224) Many baseball legends hundreds at a time were/are Jews. (225) #613 baseball legends by the hundreds were/are Jews. (226) #613 baseball legends hundreds at a time were/are Jews. As earnest a tease as this is, I am not about to launch into an analysis of nominal aspect. Rather, the attested phenomenon is invoked to warrant a limited conclusion about the difference in meaning separating (in)definite descriptions from articleless
Antisemidistributivity vs. Conjunction Reduction Redux
677
quantification, one that I think first presents itself on introspection and finds corroboration in the contrastive judgments above. The arithmetic of being many and being 613 is much the same. What tells apart ⌜[Many Φ] Ψ⌝ and existential bare plurals from ⌜[Some Φ] Ψ⌝ and ⌜[613 Φ] Ψ⌝ are the epistemic conditions presupposed for the reported measurement. The former, the imperfective so-called, reports measurement that sweeps multiple frames of reference F (perhaps even a continuum of such) without any promise of their path integration into a single frame of reference. It could at worst be no better than a count of the hand waves in the optic flow through the egocentric frame of reference of a VIP whose motion is too swift or erratic for an accurate fix on the waving admirers swirling around or passing by, except for within a frame of reference local to a snapshot in the optic flow (see section 9.4.1). In contrast, the so-called perfective reports a measurement for a single frame of reference f to which all the measurement scenery Sɶ has been path-integrated so that it constitutes a topographical survey of that allocentric frame of reference (see section 9.4.1, (178)–(180)): ɶ ɶ → topo-survey( sɶ, f , n)) . (227) topo-survey[Sɶ , f , n] ↔ def ∀sɶ(Ss (“n” for narrator/navigator) (228) orient( sɶ, f , n) ↔ def ∀t(LTsɶ ( sɶ)(t ) → orient( sɶ(t ), f , n)) topo-survey( sɶ, f , n) ↔ def orient( sɶ, f , n) f © (229) orient(ṡ,f, n) ↔def ṡ© = projection(t(ṡ©),σf(ṡ©),πf(ṡ©), f, l(ṡ©), No( sɺ ),o(ṡ )) & n sees at t(ṡ) that ∃πf ∃l ∃N of ∃o ṡ© = projection(t(ṡ©),σf(ṡ©),πf, f, l, N of ,o)
Recall that a topographical survey of f affords a perspective-free representation since all that is surveyed can be addressed using only the allocentric coordinates for f, without reference to the observer’s positions in f . The cinematic scenery and its reticulation for what is visually counted may at any given moment subtend only a small subspace. But if it all constitutes a survey of a single frame of reference for a larger space to which the scenery is all orienting, then whatever is counted has a fixed, determinate location in that larger, single frame of reference. In that sense, the count is perfective—perspective-free—an objective measurement for the allocentric frame of reference to which the scenery has been path-integrated. As much as narration is designed for concurrent spatial orientation and navigation, it also has a mission to settle once and for all How-the-World(-of Manhattan)-Is as reflected in the path integration of visual experience to just such a 3-D model, a single frame of reference. It can then be no surprise that there is a vocabulary to which belong some, the, 613, expressing a precision that strives to be objective—allocentric and perspective-free—in contrast to the vocabulary, the null article, many, few—used to convey the proximate thoughts that are guidance toward archival fact.
678
Chapter 13
Let the null determiner express the indifferent epistemic condition that holds between events and frames of reference for them in scenes here, there, or anywhere (i.e., passim), with candidate logical forms for elms and many elms as shown: (230) passim[E, F] ↔def ∃Sɶ ∃F en-scène[E, Sɶ , F , F]26 (231) [∃X: ∃E (∃F passim[E,F] elms[E,X])] (232) [∃X: ∃E (∃F passim[E,F] ∃Eμ [∃n : many(n)] count[Eμ, E, n] elms[E,X])] The epistemic conditions passim are weak. That is, they are consistent with a disorienting sweep through untold frames of reference and with a montage of discrete misplaced memories, as desired, but also with the path integration to a perspectivefree representation that further morphology (e.g., an overt article) will impose. Given the greater mission to represent objective conditions, I will leave it to the pragmatics to implicate that the speaker favors the null determiner only when she can do no better, when she has access only to scenery that cannot be path-integrated. The perfective epistemic condition is some, variously pronounced as some in some elms, or as an in combination with singular morphology, or unpronounced in combination with cardinal predicates, as in (some) 613 elms: (233) some[E, F] ↔def [∃Sɶ : ∃f ∃n topo-survey[Sɶ ,f, n]] ∃F en-scène[E, Sɶ , F , F] (234) [∃X: ∃E (∃F some[E,F] elms[E,X])] (235) [∃X: ∃E (∃F some[E,F] ∃Eμ count[Eμ, E, 613] elms[E,X])]
These logical forms are spare in leaving implicit and therefore to inference a few points. They say that the scenes are path-integrated and orienting, a topographical survey, for some narrator/navigator n. Well, if it is these scenes that are so and the narrator/navigator of this cinerama has warrant to know as much, it must be that she herself is their well-oriented narrator/navigator, without the logical form being explicit in making reference to n de se. Also, if such is her relation to a path-integrated f for the events E, then their addresses F are almost certainly to be coordinates in F that are this same f, without troubling to make all this explicit too.27 Given that all nominal phrases are AdrPs (sections 9.0–9.1), the logical forms for elms, some elms and (some) many elms and (some) 613 elms should first be revised to reflect the constant presence of addressing: (236) [∃X: ∃E (∃F passim[E,F] ∃F (At[F , E, t(E),F] elms[E,X]))] (237) [∃X: ∃E (∃F passim[E,F] ∃F (At[F , E, t(E),F] (∃Eμ [∃n : many(n)] count[Eμ, E, n] elms[E,X]))] (238) [∃X: ∃E (∃F some[E,F] ∃F (At[F , E, t(E),F] elms[E,X])] (239) [∃X: ∃E (∃F some[E,F] ∃F (At[F , E, t(E),F] ∃Eμ count[Eμ, E, 613] elms[E,X])]
Antisemidistributivity vs. Conjunction Reduction Redux
679
Further revision is prompted recalling (section 9.1) and accommodating the reference and definite description of frames of reference that quantify into AdrPs: (240) a. [℩F : Φ] … [∃X: ∃E (passim[E,F] ∃F (At[F , E, t(E),F] … elms[E,X]))] …; or, b. … [∃X: [℩F : Φ] ∃E (passim[E,F] ∃F (At[F , E, t(E),F] … elms[E,X]))] … (241) a. [℩F : Φ] … [∃X: ∃E (some[E,F] ∃F (At[F , E, t(E),F] … elms[E,X]))] …; or, b. … [∃X: [℩F : Φ] ∃E (some[E,F] ∃F (At[F , E, t(E),F] … elms[E,X]))] … It was further suggested (section 11.1.1) that for bare plurals, this reference is anaphoric and restricted to local, antecedently referred-to frames of reference: (242) … [∃X: [℩F : proi] ∃E (passim[E,F] ∃F (At[F , E, t(E),F] … elms[E,X]))] … The logical forms for elms and some elms thus differ in two respects: in the epistemic relation maintained between events E and frames of reference F, ‘passim[E,F]’ and ‘some[E,F]’, and in how the Fs are referred to (see also sections 13.1.0–13.1.1 below). The combined effect is that the bare plural is a semantic default, indiscriminate in its epistemic conditions and continuing with whatever frames of reference are already at issue. 13.1.0
Disoriented counting
Counting behavior, clicking, is typically direct witness of brief moments, a hand raised or a passenger crossing the aircraft’s threshold. To count more than that which is witnessed is an inference from supplemental knowledge—that everyone present is raising exactly one hand, and no hand belongs to more than one person, or that given the frame of reference for, say, a single embarcation on a single aircraft, the same person cannot be in the different scenes for which the counter clicks. That is, only from the certain knowledge that these scenes cannot be scenes from the same lifetime is counting passengers as they cross the threshold counting persons too. For scattered frames of reference, dim in memory, corresponding to various embarcations, one may remain certain of counting passengers qua 7-hour passengers as these frames of reference are scattered enough from each other so as not to be witnesses of the same flights, but counting persons is lost absent knowledge that none of the scenes counted could belong to the same lifetime. The injunction against counting the same thing twice, ruling out counting by this method persons who have flown, is not so strong as to restrict the count to the events under direct observation lasting no longer than the duration of a passenger’s flight. In a study of how many passengers eventually receive refunds for lost baggage, one might very well have in mind
680
Chapter 13
clicking the counter as passengers cross thresholds to count an event that begins with the passenger’s flight and extends to the fate of her baggage and the years until she sees a refund. Such a historical thread launched when a passenger crosses an aircraft’s threshold is not the same as any other when this same person crosses again, and the speaker, even counting across frames of reference, can be confident of it.28 Relying on the memory of various flights, the risk of confusing persons who have flown is palpable: it is plain enough that the epistemic scatter of frames of reference cannot be integrated under a single, orienting one. Yet the antisemidistributivity to be explained also survives frames of reference that are close neighbors, immune from lapses of memory or the extremes of rapid optic flow, and sample a spatiotemporal region that the speaker might very well have subsumed under a single frame of reference from his perch in the stadium stands: (243) #Many cheerleaders are a half-naked squad. Much as he tries to count them, he loses count if his intention is to count cheerleaders cheerleading simultaneously and discriminated only by spatial position. Although the counter clicks for different spatiotemporal scenes of a cheerleader, there is no confidence that two such scenes are not of the same spatial region, or if the cheerleaders are in motion, that they are not scenes from the same life.29 What the speaker has no access to is a frame of reference in which the scenes clicked for are both orienting—with distinct scenes necessarily of distinct spatiotemporal regions—and discerning of only whole cheerleaders, as a count or any other measurement of them requires. All the speaker can report accurately is that in his due diligence to survey all the cheerleaders, the events as numerous as the occasions on which his gaze centers on and tracks a cheerleader were many. I have in this example exploited stadium distances to frustrate counting and make vivid the contrast between single and multiple frames of reference. Of course even under circumstances less arduous for a single frame of reference, the speaker is free to invoke multiple frames, and as this example illustrates those so invoked may yet be close neighbors. The relation many makes essentially plural reference to frames of reference and thus differs from cardinal 613 not merely in a vagueness about number but in the epistemic conditions for measurement.30 In summary, for any count, determinate in what it counts and counting none of it twice, to count across plural frames of reference is to count short. As in event counting so-called, the events counted do not coincide with lifetimes, in contrast to object counting. Thus, bare many, with its plural frames of reference, always approximates event counting, in contrast to 4000 ships, which may event-count or object-count, and so should always be subject to constraints similar to those that attach to event counting.
Antisemidistributivity vs. Conjunction Reduction Redux
681
In event counting, many is subject to sequencing of events (Doetjes and Honcoop 1997), which is manifest especially when past events counted are related to remote events in the present, as in (69): (69) #(The) 4000 ships that passed through the lock are now in the port. (Adapted from Doetjes and Honcoop 1997) (12) (The) 4000 ships that passed through the lock last year are (the) 1000 vessels registered in Panama. As a point of grammar, the clause structure enforcing the sequencing of events is universal, and thus, despite an illusory exemption in (12), recounts too are subject to it (section 11.1), surfacing in a contrast between (212) and (216): (211) (The) 4000 clients that attended speed-dating events last month were 1000 lonelyhearts searching for soulmates in a city of eight million. (212) (The) 4000 clients that attended speed-dating events last month are 1000 lonelyhearts searching for soulmates in a city of eight million. (215) F (The) 4000 clients that attended speed-dating events last month were 1000 honeymooners/newlyweds. (216) #(The) 4000 clients that attended speed-dating events last month are (now) 1000 honeymooners/newlyweds. An event-counting recount such as (212) (and (12)) is an apparent exception to the sequencing of events only as an inference from a conforming sentence (211) that what was still is. Why this should matter and the vocabulary and logical form that explain it were the preoccupation of earlier sections. Here the point is to subsume many’s anti(semi)distributivity under the results and discussion of those sections. Recounts, say of 4000 Fs that they be 1000 Gs, are, after all, as canonical examples of antisemidistributive predication as saying of some cheerleaders that they are one squad (see (243)), and this class has already set itself apart under evaluation with multiple frames of reference. Recall (see sections 11.0.0–11.1.0) that the event counting rejected when when (69) is uttered out-of-the-blue is rescued under various conditions, in one of which the (69)’s speaker gestures to 4000 monitors transmitting images from 4000 cameras planted on board during 4000 passages through the lock and now bearing witness that their ship is in the port: (69) (The) 4000 ships that passed through the lock are now in the port. (244) #(The) 4000 ships that passed through the lock are now 1000 visitors in the port. Although the monitors rescue (69), a recount (244) remains anomalous despite them. As was pointed out, 4000 frames of reference come with no protocol for
682
Chapter 13
counting 1000 across them. It takes a single frame of reference for the present scene, a window on the port in addition to the monitors, to count 1000: (245) (The) 4000 ships that passed through the lock [waving at the monitors] are (now) [pointing out the window] 1000 visitors in port. If the fault with (244) and its correction in (245) is just the absence of a present, single frame of reference necessary for the intended measurement, then what multiple frames of reference can severally bear witness to should be a safer assertion, including collective, semidistributive predications: (246) (The) 4000 ships that passed through the lock are now berthed near each other. (247) Many ships that passed through the lock are now berthed near each other. A monitor projects a scene of its ship including what is nearby. To recognize within such a scene that a nearby ship also passed through the lock either by a marking or by a glimpse of its camera on board is neither to recognize how many ships in total are at berth nor to recognize that the ships near this one are the same or not as the ships near that one, which does demand a single frame of reference that the monitors do not afford and the speaker using many denies herself. An antisemidistributive predicate, unlike distributive and semidistributive predicates, presupposes a single frame of reference, which puts it in conflict with many presupposing many. Notice that if this is the correct account, it applies with equal force to rule out antisemidistributive predicates in both restrictions and nuclei.31 To complete the account, it remains for section 13.2 to explain why given an optional semidistributive operator, the cheerleaders of (243) cannot be scattered among several (small) squads, whatever their number, each of which is within the neighborhood of a cheerleader’s singular frame of reference.32 Counting across frames of reference, many counts short events, and it is thus expected, as remarked above, that it will be subject to sequencing of events and exemplify the effects of tense, aspect, and context discussed in section 11.0. The sequencing of events violated uttering (69) out of the blue is satisfied in the context where 4000 monitors track the effects of those 4000 past events into the present, and similar contexts rescue event counting in (129) while it fails in (130) absent 4000 events in the present scene that carry the past events into the present: (127) (The) 4000 ships that passed through the lock last year are the subjects of (4000) tedious PBS travelogues. (128) (The) 4000 ships that passed through the lock last year fill the new season on PBS. (129) The 400,000 Flying Tigers that passed over the Hump are famous, and their lost cargoes are still guarded/worshipped by Himalayan cargo cults. (130) #The 400,000 Flying Tigers that passed over the Hump are famous and (they are) on display at the Smithsonian’s National Air and Space Museum.
Antisemidistributivity vs. Conjunction Reduction Redux
683
What defeats event-counting in (130) includes the particular aspect under which the corresponding present effects are taken to be presented. Recall the contrast between (69) and (150): (69) # (The) 4000 ships that passed through the lock are now in the port. (150) (The) 4000 ships that passed through the lock have been in the port now for awhile if they don’t/didn’t have to be elsewhere. The present perfect in (150) is not a “shallow” present and does not imply that the scene before the speaker can be parsed into 4000 present effects or continuations of the 4000 passages through the lock. Antisemidistributive predicates show analogous effects, the contrast between (127)/ (128) and (130) repeating in the contrast between (248) and (249): (248) (The) 4000 aircraft that landed at Tempelhof in 1948 are a newsreel at the National Archives. (249) #(The) 4000 aircraft that landed at Tempelhof in 1948 are a squadron mothballed in the Mojave Desert. The aspectual contrast can be illustrated in minimal pairs that play on the ambiguity of the predicate nominals in (250) and (251): (250) (The) 4000 aircraft that landed at Tempelhof in 1948 were an airlift operation. (The) 4000 aircraft that landed at Tempelhof in 1948 were a cold war delivery service. (251) (The) 4000 aircraft that landed at Tempelhof in 1948 are an (ongoing) airlift operation. (The) 4000 aircraft that landed at Tempelhof in 1948 are an (ongoing) cold war delivery service. Describing a process, which implies, extended in time, that parts of it precede and others may follow, (250) and (251) felicitously event-count and say that 4000 landings were parts of a process that in the case of (251) reaches back to 1948 and continues into the present. On the other hand, if operation or delivery service is meant to refer to a company, a standing squadron of aircraft, event-counting fails just like in (249) in that the 4000 (serial) landings do not correspond to any 4000 subevents that are contemporary parts of being such a company, whether it is located in the past (250) or in the present (251). Analogous contrasts occur with many, as expected if its antisemidistributivity is largely a matter of it always event-counting:33 (252) #Many aircraft that landed at Tempelhof in 1948 were a squadron under the command of US Air Forces in Europe. (253) Many aircraft that landed at Tempelhof in 1948 were an operation delivering military hardware disguised as medical supplies.
684
Chapter 13
Similarly, (254) and (255) are not acceptable reports that characters from different scenes coalesce in the person of the same actor playing multiple roles: (254) #Many characters that appeared on stage tonight were an actor in many roles. (255) #Many characters that appeared on stage tonight were actors in many roles. The same is to be said about (256) and (257) if the duos are taken to be famous teams such as Bogart & Bacall or Hepburn & Tracy: (256) Many characters that appeared on stage tonight were a famous duo in costume. (257) Many characters that appeared on stage tonight were famous duos in costume. The sentences become acceptable if instead it is imagined that the evening was a theatrical revue, much of it famous duets: Mimi & Rodolfo, “O soave fanciulla” (La Bohème, act 1), Papageno & Papagena, “Papageno! Papagena!” (Die Zauberflöte, act 2), Don Giovanni & Zerlina, “Là ci darem la mano” (Don Giovanni, act 1), Porgy & Bess, “Bess, you is my woman now” (Porgy and Bess, act 2), and Tony & Maria, “Tonight” (West Side Story, act 1). Here each of many frames of reference fixes a character on stage addressing a partner, sufficient to witness whether it is of a famous duo, which Mimi and Rodolfo, for example, are. Unlike the temporally scattered appearances on stage that are said to be of the same actors, this understanding of (256) and (257) conforms to all conditions on the sequencing of events. Also, constraints on measuring across frames of reference are respected in that the recognition of a duo is complete within a single one. What is new here and anticipates section 13.2 is that semidistributivity with an allegedly antisemidistributive predicate is acceptable in this particular context despite the earlier remark and observation about (243), common in the literature (Hackl 2001a, 2001b, and the references cited there), that it typically is not. The above examples are all in the past tense so that the effects of aspect and measuring only within or across frames of reference can be observed without distraction from a shift in tense from relative clause to matrix. Introducing shifts to the present Tense puts the contrasts in high relief: (258) Some many white men (who were) parading around in white sheets last night are a local KKK chapter. Some numerous white men (who were) parading around in white sheets last night are a local KKK chapter. Some scattered crackers (who were) hiding in the crowd last night are a mock lynch mob. Some baseball legends (who were) honored last night are a Jewish Hall of Fame.
Antisemidistributivity vs. Conjunction Reduction Redux
685
(259) Many white men (who were) parading around in white sheets last night are in a local KKK chapter. Numerous white men (who were) parading around in white sheets last night are in a local KKK chapter. Scattered crackers (who were) hiding in the crowd last night are in a mock lynch mob. Many baseball legends (who were) honored last night are in a Jewish Hall of Fame. Numerous/Various/Assorted baseball legends (who were) honored last night are in a Jewish Hall of Fame. Baseball legends (who were) honored last night are in a Jewish Hall of Fame. (260) #Many white men (who were) parading around in white sheets last night are a local KKK chapter. #Numerous white men (who were) parading around in white sheets last night are a local KKK chapter. #Scattered crackers (who were) hiding in the crowd last night are a mock lynch mob. #Many baseball legends (who were) honored last night are a Jewish Hall of Fame. #Numerous/Various/Assorted baseball legends (who were) honored last night are a Jewish Hall of Fame. #Baseball legends (who were) honored last night are a Jewish Hall of Fame. (261) OK Many white men (who were) parading around in white sheets last night were a local KKK meeting [sic]. OK Numerous white men (who were) parading around in white sheets last night were a local KKK meeting [sic]. OK Scattered crackers (who were) hiding in the crowd last night were a mock lynch mob. OK Many baseball legends (who were) honored last night were a Jewish Hall of Fame. OK Numerous/Various/Assorted baseball legends (who were) honored last night were a Jewish Hall of Fame. OK Baseball legends (who were) honored last night were a Jewish Hall of Fame. Again, the antisemidistributive predication in (260) that is unacceptable in violation of the conditions on sequencing of events, aspect, and measuring across frames of reference improves when these conditions are met in (261).
686
Chapter 13
There is another respect in which many resembles event counting. Recall that most of sections 11.0–11.1 put in relief the contrast between event counting in (75), which uttered out-of-the blue violates conditions on sequencing of events and requires special contexts, and recounts such as (12) which at first blush appear acceptable without appeal to special contexts (although in fact subject to the same conditions, section 11.1): (75) #(The) 4000 ships that passed through the lock last year are registered in Panama. (12) (The) 4000 ships that passed through the lock last year are (the) 1000 vessels registered in Panama. It was left to section 11.1.1 to point out that unmeasured identity statements such as (248), with bare nominals, fall in with (75), in contrast to recounts, in requiring special contexts to rescue out-of-the-blue violations of sequencing of events: (248) #(The) 4000 ships that passed through the lock last year are vessels registered in Panama. Unmeasured identity statements and recounts also differ in their tolerance for sequencing-of-events violations induced by later relative clauses and sentences: (256) #(The) 4000 ships that passed through the lock last year were (merchant) vessels that are registered in Panama. (257) #(The) 4000 ships that passed through the lock last year were (merchant) vessels(, and) they are registered in Panama. (253) (The) 4000 ships that passed through the lock last year were 1000 vessels that are (only now) registered in Panama. (254) (The) 4000 ships that passed through the lock last year were 1000 (merchant) vessels(, and) they are (only now) registered in Panama. The speculation was that the bare nominals in (256) and (257), in contrast to the recounting nominals, do not introduce a new frame of reference for recount and reorientation of the sentence’s continuation. Rather, a zero determiner’s frame of reference is local, so that if the local one coincides with present Tense and the nuclear events, it cannot hark back to the subject’s past frame of reference. The bare nominals continue with the same events under the same frame of reference subject to sequencing of events just as if the sentence did not introduce any new quantification. Unlike full-blooded DPs that quantify and through their adverbialization requantify over events directly, it is as if the bare nominal only comments like other predicative elements, be registered, be in port, etc., on events already introduced. Allowing that nominals with the indefinite article an are the singular counterparts of bare plurals, the sentences (252)–(257) with many are all unmeasured identity
Antisemidistributivity vs. Conjunction Reduction Redux
687
statements. As such, the violations of sequencing of events incurred are repaired in their recounting counterparts: (252) #Many aircraft that landed at Tempelhof in 1948 were a squadron under the command of US Air Forces in Europe. (262) Many aircraft that landed at Tempelhof in 1948 were one squadron under the command of US Air Forces in Europe. (254) #Many characters that appeared on stage tonight were an actor in many roles. (263) Many characters that appeared on stage tonight were one actor in many roles. (255) #Many characters that appeared on stage tonight were actors in many roles. (264) Many characters that appeared on stage tonight were two actors in many roles. (265) #Many white men (who were) parading around in white sheets last night are a local KKK chapter. (266) Many white men (who were) parading around in white sheets last night are one local KKK chapter. (267) #Many baseball legends (who were) honored last night are a Jewish Hall of Fame. (268) Many baseball legends (who were) honored last night are one Jewish Hall of Fame. If these remarks have been on the right track, antisemidistributivity in both nucleus and restriction derives from plural reference to frames of reference that fail path integration to a single frame of reference. Since this antisemidistributivity is observed in all bare plurals, not only those containing many, it is the null determiner that makes plural reference passim to frames of reference in (183)–(185), (189), (190), (193), (195). Antisemidistributivity is thus generalized to all constructions besides many, which can itself remain a predicative, arithmetic expression. The null determiner in all bare plurals also flags, I assume, a null anaphor referring to local, antecedent frames of reference, as when it occurs in the predicate nominals of unmeasured identity statements. When a bare plural is sentence-initial, as in (183)–(185), (189), (190), (193), (195), its anaphoric null determiner, which would otherwise be unbound, is captured by the independent operator referring to plural frames of reference. 13.1.1
Defining the space referred to via the nominal denotation of its landmarks
An illusion—or so I would like to believe it to be—of a certain quantificational force remains to many and scattered in the bare construction with overt article absent, ⌜[Many Φ] Ψ⌝ and ⌜[Scattered Φ] Ψ⌝. The illusion, if not so, would otherwise
688
Chapter 13
suggest that many and scattered take their place in a derived, complex quantifier or determiner. The more interesting case is scattered, and in the following example, the scatter of interest concerns the distribution of a trait among a population: those with the trait are asserted to be few and randomly distributed—that is, scattered—among those without it. The relevant space and distribution is defined on a population, without distraction from the topographical notion of scatter in concrete space (and time). Concerning that population, very few Jews are major league baseball players, and very few major league baseball players are Jews. All the more, very few Jews are baseball legends, and very few baseball legends are Jews.34 But an inspiration to the people is that all Jewish major league baseball players are and have been legends in their time (according to the Jewish press). Given the facts, it can be reported that (269) Scattered Jews are baseball legends or even major league ballplayers. (270) Scattered baseball legends are Jews. (271) Scattered baseball legends are Jewish major league ballplayers. But (272) is false because they all are legends: (272) F Scattered Jewish major league baseball players are baseball legends. (273) Some scattered Jewish major league baseball players are baseball legends. In contrast, (273) may very well be true if, for example, the Jewish ballplayers are scattered in history, among the major league teams, or in different precincts of Brooklyn. In the bare construction ⌜[Scattered Φ] Ψ⌝, the relation that scattered expresses is that of a decreasing quantifier—no more than a scattering of the Φ among the Φ Ψ. Even with all Jewish major league ballplayer legends, some of them are of course scattered among the others, but the legends are more than a scattering among them, falsifying (272). Apart from its decreasing force, it should also be noted that scattered walks like a quantifier in being co-opted to express a relation between the Φ and the Ψ, forgoing any sense that the scatter might be defined with respect to the external backgrounds available to (273). An arithmetic example repeats the point: (274) Scattered natural numbers are primes greater than 9007. (275) F Scattered prime numbers are greater than 9007. (276) Some scattered prime numbers are greater than 9007. The prime numbers greater than 9007 are scattered among the natural numbers, as (274) says and as suffices for (276), but they are not scattered among the prime numbers, falsifying (275). Accepting all this at face value—that distributivity is the effect of an independent operator referring to plural frames of reference, and scattered nevertheless enters into the expression of a decreasing quantifier—urges adoption of an operator that combines with a relation to form a determiner that is a decreasing distributive
Antisemidistributivity vs. Conjunction Reduction Redux
689
quantifier. Let the participle scattered express the relation that X is scattered in some scattering among the Y: (277) ‘scattered[X,Y]’ for ‘∃E(scatter[E] Theme[E,X] Among[E,Y])’ The Downward Distributive Determinerator, acquired at a biennial NATO arms bazaar in Amsterdam and defined in (278), applies in (279) to scattered to derive the target quantifier: (278) Downward Distributive Determinerator DDD = λRλPλQ ∃X (R(P )( X ) & ∀x( Xx ↔ (P ( x) & Q( x)))) (279)
DDD(λYλX
scattered[X,Y]) = λPλQ ∃X (scattered[ X , P ] & ∀x( Xx ↔ (P ( x) & Q( x))))
The Downward Distributive Determinerator can also make a determiner out of a predicative many provided its demand for a relation is accommodated:35 (280) ‘many[X,Y]’ for ‘∃E∃n(many(n) & For[n,Y] & [∃d : d > n] count[E,X,f, d])’ That is, Many NP is, as commonly assumed, “many among the NP” or “many for the NP,” which may be either a proportion of the NP or a cardinal threshold for them (see Herburger 1997, 2000). The Downward Distributive Determinerator, assuming the determiners derived are in complementary distribution with overt articles and other determiners, encodes the collected generalizations: distributivity is an effect of an operator, the vocabulary at issue is predicative, and the phrases derived are determiners expressing nonincreasing quantifiers. Note that the quantifiers derived are conservative, but they are not invariant under permutation (see Larson and Segal 1995, 298ff., 302ff., and the references cited there), assuming that scattered even as a determiner retains the topological content implying “not clustered.” Without this shibboleth, some might not allow scattered to ferry over to the determiners and will thus join me in another account of its impersonation of a decreasing quantifier, wishing it an illusion. The decreasing force of the construction is easy enough to explain away if the target interpretation is one for which scattered is in focus. As Herburger (1997, 2000) discusses and explains, focus in the (a) example renders an interpretation equivalent to the (b) example: (281) a. b. (282) a. b.
Many SCANDINAVIANS have won the Nobel Prize. Many who have won the Nobel Prize are Scandinavian. Few INCOMPETENT cooks applied. Few cooks applying were incompetent.
690
Chapter 13
If so, focus within a bare plural renders the (a) examples below equivalent to the (b) examples: (283) a. SCATTERED baseball legends are Jews. b. Baseball legends who are Jews are scattered. (284) a. SCATTERED natural numbers are primes greater than 9007. b. Prime numbers greater than 9007 are scattered. But the (b) examples retain a decreasing force, despite the plain fact that their scattered is not a determiner, in that to be scattered is to be no more than a scattering as the following examples and context make vivid: (285) Fallen leaves are scattered in the backyard. Fallen leaves are scattered among other fallen leaves. Both sentences are false of a backyard covered over under a single pile of leaves, even though it contains any number of scatterings: (286) Some fallen leaves—those few I had raked—are scattered in the backyard. Some fallen leaves—those few I had raked—are scattered among other fallen leaves. As Carlson (1977) observed of the bare plural, one can never have it both ways: (287) F Fallen leaves are scattered in the backyard, and fallen leaves are not scattered in the backyard. (288) Some fallen leaves are scattered in the backyard, and some fallen leaves are not scattered in the backyard. As much as the effect of focus on interpretation gains a purchase on the decreasing force of ⌜[Scattered Φ] Ψ⌝, which says that no more than a scattering of the Φ Ψ, it leaves aside that the scatter must be of the Φ among the Φ, that is, that no more than a scattering of the Φ among the Φ Ψ: (289) a. F Scattered Jewish major league baseball players are baseball legends. b. Jewish baseball players who are legends are scattered. (290) a. F Scattered prime numbers are greater than 9007. b. Prime numbers greater than 9007 are scattered. In predicative position in (290b), scattered allows for the only sensible and true interpretation that the large prime numbers are scattered among the natural numbers, rather than among the large prime numbers themselves, as (290a) requires. It will not do to appeal to a syntax that contrasts attributive and predicative positions, saying that a parameter for scattered, Y of “among the Y” (see (277)), must be fixed by NP when scattered is in close construction with it, the attributive position. It will not do, since some scattered prime numbers is not restricted to scatterings
Antisemidistributivity vs. Conjunction Reduction Redux
691
among prime numbers and some scattered Jewish major league baseball players is not restricted either to scatterings among the players. Singling out the context of bare nominals, the restricted meaning, no more than a scattering of NP among the NP, was made an effect of applying the Downward Distributive Determinerator. Singling out the bare nominal construction in this way is pervasive, as further illustrated in the following. A meteorite is a meteor that impacts Earth, which few do: (291) Occasional meteors impact Earth. Occasional meteors are meteorites. (292) Intermittent meteors impact Earth. Intermittent meteors are meteorites. (293) Random meteors impact Earth. Random meteors are meteorites. (294) Rare meteors impact Earth. Rare meteors are meteorites. Sentences (295)–(298) are false, by definition, and (299)–(302) are true: (295) (296) (297) (298)
F F F F
Occasional meteorites impact Earth. Intermittent meteorites impact Earth. Random meteorites impact Earth. Rare meteorites impact Earth.
(299) (300) (301) (302)
The occasional meteorites all impact Earth. The intermittent meteorites all impact Earth. The random meteorites all impact Earth. The rare meteorites all impact Earth.
Sentences (291)–(294) are truths about the distribution of meteorites among meteors; (295)–(298), falsehoods about the distribution of meteorites among meteorites; and (299)–(302), truths that comment on the distribution of meteorites relative to a contextually understood background, either among meteors or in absolute time. If the target interpretation of the bare nominal construction requires focus and a logical form in which the focus is extraposed as in (281)–(284), there could be, I suppose, a syntactic stipulation, as brutal as the Determinerator, that as a condition on its movement out of DP the focus “agree” with NP, binding its free parameter “among the Y.” Both the Determinerator and “agreement” share the insight that this curious feature of the meaning of the bare nominal construction is unrelated to anything else about its meaning. My remaining remarks attempt not to be so blind. Presuming that the decreasing or nonincreasing force of the construction derives from the interpretation of focus and the lexical meaning of the focus—that, for example, a density is not many overlapping scatterings or sparsities, that where there is density there is no scatter—it remains to be explained why the field for the
692
Chapter 13
distribution must be the NP themselves in the case of this construction (the (a) examples of (283), (284), (289), and (290)), no more than a scattering of NPs among the NPs, in contrast to the approximate paraphrases (the (b) examples of (283), (284), (289), and (290)). Notice that, even presuming the focus’ explicit extraposition in logical form (see Herburger 1997, 2000), it will not extrapose downward to embed the focus within a matrix predicate in the scope of Tense. In this respect, the logical form of the (a) examples under focus will always differ from what appears in the (b) examples. Moreover, extraposed from a prenominal position, the focus still denotes whatever it denotes in that position, which is to be exploited in explaining the difference in interpretation. Nominal descriptions participate in an imperfective aspect for measurement, already mentioned in the discussion of (199)–(210). Despite proof of the equal cardinality of the natural numbers, the prime numbers, the composite numbers, and the multiples of 5, (303) is true and contrasts for a fleeting moment with (304), which in that moment seems to directly contradict what is proven: (303) a. Most natural numbers are composite (i.e., not prime). b. Most natural numbers are not multiples of 5. (304) a. Most of the natural numbers are composite. b. Most of the natural numbers are not multiples of 5. The truth (303) expresses is more explicit in (305) and also similar to (306) with the difference noted below: (305) a. Natural numbers are mostly composite. b. Natural numbers are mostly not multiples of 5. (306) a. Natural numbers are most of the time composite. b. Natural numbers are most of the time not multiples of 5. That is, as one scans the number line in continuous measurement, the composite numbers outnumber the prime numbers ((303a), (305a), (306a)) and the numbers indivisible by 5 outnumber its multiples ((303a), (305a), (306a)), which is true for all time whether the measurement is cumulative or local (cf. velocity). The definite description in (304) presents the natural numbers whole and entire and gardenpaths to a perfective measure of them whole and entire, which contradicts cardinal arithmetic. But, on second thought, despite the presentation, the continuous measure that comes more naturally to (303) could have been the one intended, and it is salient when an adverb is explicit about the temporal dilation: (307) a. The natural numbers are mostly composite. b. The natural numbers are mostly not multiples of 5. (308) a. The natural numbers are most of the time composite. b. The natural numbers are most of the time not multiples of 5.
Antisemidistributivity vs. Conjunction Reduction Redux
693
In (307), the temporal dimension in which continuous measurement is extended is the ordering of the natural numbers itself. In (308), the temporal dimension clings to real time and thus the natural numbers are to be thought of as objects that can occur in real time and perhaps recur at different times, which invites two differences of meaning. The first is that the natural numbers whole and entire are taken to persist through time and be the object measured every time, which lands (308) back in contradiction of cardinal arithmetic. Instead, (308) should be taken as in (309) (equivalent to the one in which the definite description is sometimes described as referring to a kind): (309) a. The natural numbers are most of the time that any are (finitely) sampled composite. b. The natural numbers are most of the time that any are (finitely) sampled not multiples of 5. So understood, it is allowed that the same natural numbers turn up in different samples, which delivers the second difference of meaning. It makes no more sense in (307) for a natural number to participate in measurement at different times than it does for one point of time to occur at another. In (307), the natural numbers are the measurement space within which a temporal dimension for mostly is defined. The same is to be said about the relation of the NP natural numbers to most in (303), in support of which and in advance of its formal representation is another example. Tincture of iodine is a 7% solution of iodine in ethanol. Because all tincture of iodine contains iodine, (310) is true and (311) is false: (310) Most tincture of iodine contains iodine. (311) F Most tincture of iodine is just ethanol. On the other hand, because so little of it is iodine, (312) is false and (313), true: (312) a. F Most of the tincture of iodine in the beaker contains iodine. b. F Most of this tincture of iodine contains iodine. c. F Most of all the tincture of iodine contains iodine. (313) a. Most of the tincture of iodine in the beaker is just ethanol. b. Most of this tincture of iodine is just ethanol. c. Most of all the tincture of iodine is just ethanol. The contrast between (311) and (313) defeats the suggestion that tincture of iodine in (311) and the tincture of iodine in (313) refer the same and are related by the same relation to whatever most quantifies over. Most of what is quantified over in (313) is not tincture of iodine. So, to relate it to the tincture of iodine that is tincture of iodine, a metaphysics must intervene that for my purpose is best avoided lying down (see Koslicki 2008 and the references cited there for discussion). But of the relation tincture of iodine[E,X], no matter how obscure the X, it is known that there is no event or events E—no matter how small—that ∃X tincture of iodine[E,X]
694
Chapter 13
without iodine in it. These and only these are the events measured in (310) and (311), said truthfully to contain iodine and falsely to be of just ethanol. These three observations—(i) that most tincture of iodine is restricted to measuring while tinctured in a way that most of the tincture of iodine is not, (ii) that The natural numbers … mostly …, in contrast to The natural numbers … most of the time …, cannot revisit numbers, and (iii) that imperfective measurement, the only way (303) is true, extends over events the NP describes—point to, without yet reaching, a conclusion that events define a measure space and it is the events E that ∃X NP[E,X]. Already these three observations follow naturally from the view of quantification defended here, namely, that the act of measurement (and its protocol) is explicitly represented and vividly illustrated in the clicking counter. Recall (sections 9.4.2 and 12.1) that an event eμ of counting ζ clicks the counter for their equinumerous proxies in a scene for eμ in which the proxies appear countable under some reticule ℜ: (314) count(eµ, ζ ) → ∃sɶ(∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) & countable[ sɶ, ζ ]) (315) countable[ sɶ, ζ ] ↔ ∃sɶ ℜ[ sɶ, ζ , sɶ ] (316) ℜ( sɶ, ζ , sɶ ) ↔df ∃ξ(∀x∀y((ξx & ξy & x ≠ y) → ∃t [ sɶ(t ), x, y, sɶ ] )36 & ∀x(ζx → ∃y(ξy & overlap(x,y) & ∀z((ζz & overlap(z,y)) → z = x))) & ∀x(ξx → ∃y(ζy & overlap(x,y) & ∀z((ξz & overlap(z,y)) → z = x))) (317) ℜ[ sɶ, ζ , sɶ ] ↔ df ℜ( sɶ, ζ , sɶ ) & ∀sɶ ′((ℜ( sɶ, ζ , sɶ ′) & ∀α sɶ (sɶ α sɶ → sɶ α sɶ )) → sɶ ′ = sɶ ) (318) Visual counting count[eµ, ζ , n] ↔ df ∃sɶ∃sɶ (∃α sɶ At(α sɶ, eµ , t( sɶ), sɶ) & ℜ[ sɶ, ζ , sɶ ] & card[sɶ, n]) (319) [∃X : ∃E (∃eμ count[eμ, E, n] natural numbers[E,X])] If it is intended as in (303) and in (305) to count eternal states of being a natural number, then despite the brief contact with the eternal that a click on the counter is at any moment of measurement, (318) precludes that there is a click once for being the number 1 and another one later also for being the number 1, as what is counted is not counted twice over. It is not however precluded that what is counted are short states of being a natural number for 50 milliseconds, but these are excluded more generally in that measurement presupposes a determinate protocol for measuring a countable domain when NP is a plural count term. The speaker is unable to say which states of being a natural number for 50 milliseconds are counted and how. In (308), the subject the natural numbers contains an NP describing the countable, eternal states of being a natural number, natural number lifetimes, but most does not directly measure these or any other states of being a natural number. Rather the sentence goes on to consider whether most time while in an eternal state of being some natural number is also time when the number in that state is composite. Time permitting, the same natural state of being a natural number may be revisited.37
Antisemidistributivity vs. Conjunction Reduction Redux
695
Scattered primes numbers describes primes numbers scattered among prime numbers, and occasional meteorites describes meteorites that are occasional among meteorites. The thought has been that prime numbers are the measure space in which a distribution of prime numbers is said to be a scatter, and meteorites are the measure space for judging some meteorites occasional. It must be a scattering of prime numbers among prime numbers and meteorites occasional among meteorites, because when these DPs are interpreted, space is empty except for the prime numbers and meteorites referred to. As just remarked, three observations, along with (318) from which they follow, point to such measure spaces of events E that ∃X NP[E,X], without yet reaching it. It falls short in, for example, a protocol as good as any satisfying (318) and conforming to intuitive notions of counting, which verifies (320) by counting “One, two, three, four, ONE, one two, three, four, TWO.” (320) Two multiples of 5 are natural numbers less than or equal to 10. Surely the measure space for this counting is not only the multiples of 5, omitting the other natural numbers clicked over—at least not if we have any independent purchase on what a measure space should be. All that (318) demands is a space that includes NP-ings, which could very well include other things that measurement might exploit. Eventually, it must. In (290b) and (276), prime numbers are scattered in the space of natural numbers, even as (290a) stumbles in a space of only prime numbers. (290) a. F Scattered prime numbers are greater than 9007. b. Prime numbers greater than 9007 are scattered. (276) Some scattered prime numbers are greater than 9007. The question is why the space of natural numbers present for the interpretation of scattered in (290b) and (276) does not present itself in thought when interpreting (290a) The conditions on counting in (318), as just pointed out, will need some supplement. Suppose that discourse referents are introduced as landmarks in a space yet to be landscaped. Let ℓ be a frame of reference for a space containing landmarks , which are events or spatiotemporal regions or points. Landmarks[ℓ,] just in case there is a measure d of the distance between landmarks that is invariant under any repositioning of theirs that preserves their relative positions in ℓ:38 (321) Let T be the class of linear translations in n-dimensional space, and let Τ* be the class of partial functions τ on n-dimensional spaces such that for any space σ in its domain, ∃t(Tt & τ(σ) = t(σ)). (A function τ is a movement of the spaces in its domain in possibly different directions.) Landmarks[ℓ,] ↔df ∃d [∀τ : Τ*τ] ( [∀l : Ll ]∃α ℓ (At[α ℓ, l, t(l ), ℓ]&[∃Aℓ: Aℓ α ℓ ]ℜ[ℓ, L, Aℓ ] & At[α ℓ, τ(l ), t(τ(l )), ℓ] & [∃Aℓ: Aℓ α ℓ ]ℜ[ℓ, τ(L), Aℓ ] )39 → [∀l0 : l0][∀l1 : l1]d(l0,l1) = d(τ(l0),τ(l1)))
696
Chapter 13
A frame of reference ℓ (or snapshot scene) is said to resolve the with addresses ℓ, ℜ[ℓ, L, Aℓ ] (section 9.4.2).40 As characterized by (321), landmarks preserve their relative position under movement just in case each has an address resolving it from the others that its movement remains within. Thus, four landmarks in four quadrants may move anywhere in their own quadrants, preserving their relative position. are landmarks for ℓ’s space just in case there is a measure of the distances between them that is immune to errors in their exact placement—a measure that gives the same result for any deviation that at least preserves their relative position. Standard Euclidean distance, that one landmark is 100 meters NE of another, will obviously not do, given that these distances vary under movement. But reckoning distance by counting the landmarks themselves is robust—a landmark remains one landmark N and one landmark E of another, provided all the landmarks’ relative positions remain the same. Landmarks are the beacons by which to navigate with an impaired sense of time and distance a landscape obscured by night and fog. As a space becomes more populated with landmarks, deviation that preserves their relative position narrows. In the limit, with all points landmarks, standard Euclidean distance is robust enough for the allowable deviation, which is reduced to zero. Return now to the contrast between (322) and (323): (322) a. Scattered prime numbers are greater than 9007. b. Occasional meteorites impact Earth. (323) a. Prime numbers greater than 9007 are scattered. b. Meteorites that impact Earth are occasional. It is uncontroversial that what is scattered or occasional is so only in virtue of its distribution relative to a given space or background, whether that space is as abstract as the natural numbers (323a) or as bone-crushing as Earth’s history (323b). Moreover, what is scattered or occasional is so only if the things scattered or occasional are far enough apart as determined by their distances from each other in the given space. Absent explicit mention of a space (“among the NP”), it therefore remains that these sentences do not express a determinate thought unless one is supplied. If the speaker and hearer were free to draw on their experience, it is unclear why implicit reference to a space (there) should induce contrast between (324) and (325). In both (324a) and (325a), the only sensible thing to conclude would be that the speaker has in mind all the natural numbers: (324) a. Scattered there prime numbers are greater than 9007. b. Occasional there meteorites impact Earth. (325) a. Prime numbers greater than 9007 are scattered there. b. Meteorites that impact Earth are occasional there. There is, however, no reason to suppose that every implicit anaphor is free to make demonstrative reference to any space, time, or possible world within reach or to
Antisemidistributivity vs. Conjunction Reduction Redux
697
freely quantify over such. That is reserved here for Tense (Partee 1973, 1984; Stone 1999) (and other (ad)verbal morphemes elsewhere). The anaphors that occur implicitly with tokens of scattered and occasional seek local linguistic antecedents. In (325), the implicit anaphor falls within the scope of Tense, its antecedent, and is thus as far-ranging in its reference as Tense itself. In contrast, in (324), the only linguistic antecedent for there is the NP that scattered or occasional modifies, whether the modifier is focused and extraposed (327) or not (326). Mention of prime numbers or meteorites raises them to salience, now not only as discourse referents but as landmarks. The implicit anaphor there refers to the space they landmark, and scattered or occasional is evaluated with respect to the distances measured in these spaces: (326) [DP Prime numbersi therei-scattered] Φ [DP∃X : ∃E0prime numbersi[E0,X] [℩ ℓ :[℩E0: ∃X prime numbersi[E0,X]] Landmarks[ℓ,E0]] [℩E0: proi]scattered[E0,ℓ]] Φ
[= therei]
(327) [DP Prime numbersi __ j] Φ – therei-scatteredj [DP∃X : ∃E0prime numbersi[E0,X]] Φ [℩ ℓ :[℩E0: ∃X prime numbersi[E0,X]] Landmarks[ℓ ,E0]] [℩E0: proi]scattered[E0,ℓ]
[= therei]
But, as the prime numbers greater than 9007 are not a scatter among the prime numbers, and the meteorites impacting Earth are not occasional among the meteorites, sentences (322) are necessarily false and cannot express the truths that (323) express referring to other spaces.41 Recall that the contrasts under discussion oppose not only (322) and (323) but also (322) and (328), where (328), like (323), can describe scatter among the natural numbers and what is occasional among meteors or in elapsed time: (322) a. Scattered prime numbers are greater than 9007. b. Occasional meteorites impact Earth. (328) a. Some scattered prime numbers are greater than 9007. b. The occasional meteorites impact Earth. There is a difference to be stipulated between null determiners and those pronounced. It has already been suggested (section 11.1.1) that null determiners are defective, entirely dependent on a local antecedent for their frame of reference, in contrast to overt determiners, which themselves quantify over and introduce new ones. The facts of (328) then suggest that the frames of reference overt determiners quantify over are a range as broad as that of Tense itself, drawing freely on the
698
Chapter 13
speaker’s and hearer’s experience. If so, any implicit anaphor that occurs within the restriction to an overt determiner refers as broadly. The illusion that scattered, occasional, and the like are quantifiers or determiners, creations of the Determinerator, dissolves in the coincidence of several phenomena. If so, many may, true to the concept of being many, remain as predicative as scattered and occasional and join them in the illusion. The null determiner they all share in the constructions canvassed flags a relation to plural frames of reference that are not path-integrated, and the antisemidistributive predicates excluded from both the restriction and nucleus in all these constructions cannot be interpreted across such unintegrated frames of reference. It remains for the next section to explain why these predicates cannot be interpreted within the singular frames of reference that a token of the (semi)distributive operator would quantify over. In saying that no more than a scattering of Φ Ψ, ⌜[Scattered Φ] Ψ⌝ fakes a decreasing quantifier. But, assuming that scattered occurs here in focus, the meaning is to say, in effect, that the Φ that Ψ are a scattering. That there is nothing more than a scattering follows from the lexical meaning of scatter—that a scattering is not a part of a clumping. In this respect, focus on many, saying that the Φ that Ψ are many, is harmless in that it is no part of the meaning of many to imply that the many are not also all. Meaning that no more than a scattering of Φ among the Φ Ψ, scattered looked as if it were indeed being coerced into the type of determiner that “lives on” the class that Φs (Barwise and Cooper 1981). Here the resemblance to determiners is an illusion of scattered’s implicit spatial anaphor and its syntax. An answer to the question that prompted skepticism about many as a determiner—why should any of this attach to many and not to 613?—cannot escape some stipulation. An analogy to perfective and imperfective aspect was suggested for nominal phrases and descriptions of measurement. Cardinal predicates such as 613 are perfective, as it were, exact measurements under a single frame of reference; many is imperfective, intrinsically vague and lost among several frames of reference. 13.2
Antisemidistributivity and its antidote
The distributivity intrinsic to distributive plurals, many NP, most NP, et al. (sections 12.3.0–12.3.1), amounts to no more than semidistributivity in combination with collective predicates describing events that no one participates in alone, (179), (329)–(332): (179) Many people in the offices of the Empire State Building first met in an elevator—but not the same elevator. (329) Many superheroes have combined their forces in a dynamic duo. (330) Many superheroes are combined in a dynamic duo. (331) Many superheroes formed together a dynamic duo. (332) Many superheroes will form a dynamic duo only with each other.
Antisemidistributivity vs. Conjunction Reduction Redux
699
Even if the many counted are scattered across plural frames of reference that are not path-integrated to a single one, as just urged in section 13.1, a frame of reference for any one of the many cannot be so narrow as to exclude others participating in an event it frames. There are however collective predicates as in (23), antisemidistributives,42 that resist even semidistributivity: (23) #Many superheroes are a dynamic duo. Neither quantifiers said to be strictly singular and first-order (Schein 1993; Hackl 2001a, 2001b, 2002) nor frames of reference as many as the many explain why the license for (179) and (329)–(332), which overcomes all that enforced solitude, fails to do the same for (23) too.43 Semidistributivity, recalling section 13.0.0, derives entirely from plural reference to events (with multiple participants), without plural reference to persistent objects. That plural reference was shown to be sensitive to conditions on event segmentation and the description of discrete, countable events. Alongside these conditions on the events quantified over, the section surveyed how the description of events in a relative clause governed the interpretation of a matrix clause containing a plural count pronoun referring back to them. The results were obtained without benefit from examples of antisemidistributive predication but should apply here too if they apply in full generality to the reference and quantification over events defining semidistributivity. As much was foreshadowed in the previous section in the discussion of (256), rescued in a particular context where there were countable events satisfying the conditions on the sequencing of events: (256) Many characters that appeared on stage tonight were a famous duo in costume. Here the interaction of antisemidistributivity and reference to events will be considered in more detail. The indefinite and definite descriptions referring to events have been assumed here and argued elsewhere (Schein 1993, 107ff.; 2006, 752ff.) to be essentially plural or second-order. There is further argument coming from semidistributivity itself that the plural quantification is count quantification of what must therefore be discrete events. (333) The vegetables are too heavy for the laboratory scale and too light for the bathroom scale. (Schwarzschild 1991, 1996) (334) These vegetables weighed 1 kilogram. As Gillon (1990) and Schwarzschild (1991, 1996) have pointed out in response to Lasersohn (1989), sentences (333) and (334) weigh the vegetables individually in a fully distributive interpretation or as a single collection but in no other configurations unless the context individuates them, as when it becomes understood that the
700
Chapter 13
vegetables have been divided among several trials each of which is to weigh the contents of a basket of vegetables. In such a context, (333) acquires the additional interpretation that the vegetables in each trial, the contents of a basket, are too heavy for one scale and too light for the other, and (334), that the vegetables divide among trials that turn out each to have been a weighing 1 kilogram.44 Schwarzschild (1996, 82f., 92f.) offers a spatial analog to the temporal individuation just illustrated. Speakers do not hesitate to judge (335) true of (336), parsing the scene into a running-parallel that relates the rectangles’ horizontals and another that relates their verticals. (335) The sides of R1 run parallel to the sides of R2. (Scha 1981) (336)
R1. R2.
Yet (337) fails to be true in (338) or (339), where speakers would sooner go blind than parse these scenes into the runnings-parallel necessary to make the sentence true: (337) The double lines run parallel to the single lines. (338)
(339)
Antisemidistributivity vs. Conjunction Reduction Redux
701
The truth of sentences such as (335) and (337) (as well as (333) and (334)) requires events or states that are discrete and recognizable as such. Besides demonstrative reference to discrete events—to certain experimental trials or to particular scenes—modification that describes discrete, collective events also rescues semidistributivity. Here the influence of relative clauses again shows itself: (87)
#Many elms are a cluster in the middle of the forest. #Most elms are a cluster in the middle of the forest. #More elms are a cluster in the middle of the forest than are not. #More elms than not are a cluster in the middle of the forest.
(340) Many elms that cluster are a cluster in the middle of the forest. Most elms that cluster are a cluster in the middle of the forest. More elms that cluster are a cluster in the middle of the forest than not. More elms than not that cluster are a cluster in the middle of the forest. (341) Many elms that are clustered are a cluster in the middle of the forest. Most elms that are clustered are a cluster in the middle of the forest. More elms that are clustered are a cluster in the middle of the forest than not. More elms than not that are clustered are a cluster in the middle of the forest. (342) Many elms that are in a cluster are a cluster in the middle of the forest. Most elms that are in a cluster are a cluster in the middle of the forest. More elms that are in a cluster are a cluster in the middle of the forest than not. More elms than not that are in a cluster are a cluster in the middle of the forest. Events or states of clustering, being clustered, or being in a cluster can be as discrete as the clusters themselves. Sentences (340)–(342) say that many elms finding themselves in such an event or state do so in the middle of the forest. What many, most, and more-than-not count is unchanged by collective modification, and in this respect, it differs from what is counted under cardinal modification in many a one or more elms. The contrast reported in section 12.3.1 between (211) and (213) survives in (343) and (344): (211) Many natural numbers are members of sets of natural numbers. (213) Many (a) one or more natural numbers are members of sets of natural numbers. (343) Many natural numbers that are collected together in a set are members of a set of natural numbers. (344) Many a one or more natural numbers that are collected together in a set are members of a set of natural numbers.
702
Chapter 13
The many that make true (211) and (343) are the countably many natural numbers themselves, but the many of (213) and (344) are as numerous as the uncountably many sets of natural numbers. With what is counted held constant, there is little in the logical form—neither of DP nor of be a cluster in the middle of the forest—to discriminate (87) and (342). Similarly, there is little on which to rest the striking passage from the incoherent (345) to the banal truths in (346) except for the difference as paraphrased in the adverbials derived by adverbialization of the NPs:45 (345) #Many musicians are an ensemble. #Most musicians are an ensemble. #More musicians are an ensemble than are not. #More musicians than not are an ensemble. #Many musicians when (being) musicians are an ensemble. (346) Many musicians who are together are an ensemble. Most musicians who are together are an ensemble. More musicians who are together are an ensemble than are not. More musicians than not who are together are an ensemble. Many musicians who are together when (being musicians) together are an ensemble. Higginbotham and Schein (1989, 168ff.) remark that the semidistributive, collective predicates, gather, meet, cluster, rain down, etc., are all reducible via Davidsonian decomposition to first-order relations between individuals and events and speculate that the truly collective predicates and relations, be numerous, be a cluster, be an ensemble, outnumber, etc., here called antisemidistributive, involve higher-order predicates and relations echoing the Fregean analysis of number (The Apostles are twelve) as a concept of concepts. Distributive quantifiers such as all, every, many, most are first-order quantifiers over individuals (see also Schein 1993) and thus, in contrast to the, never form terms denoting what the truly collective denote. In the first place, so crude and stark a distinction of logical type (and distinctions in a similar vein in later work (e.g., Winter 1998, 2002) embarrasses the intuition underlying it in answering what the morpheme cluster is supposed to mean occurring both in semidistributive predicates, to cluster and to be in a cluster, and in an antisemidistributive, allegedly higher-order predicate, to be a cluster. It would have to be said as in (347) that X is a cluster if and only if there is a cluster that everything X and only things X are in, which reads like a reduction to a first-order relation despite a stipulation to the contrary: (347) be a cluster(X) ↔ ∃e(cluster(e) & ∀x(Xx ↔ in(x,e))) Embarrassment aside, the distinction could not be more lame, foundering on the contrast between (345) and (346), where it appears, laboring under it, that be an ensemble changes logical type under the influence of a relative clause modifying its
Antisemidistributivity vs. Conjunction Reduction Redux
703
subject. Rather, antisemidistributive predicates imply no distinction of logical type and no pleading that genuine collectives are collectives that are only more so. They are formally like all other predicates, decomposed the same way and embedded in a universal clause structure. Their resistance to (semi)distributive quantification derives on the one hand from the fact that plural event quantification is plural count quantification, and on the other hand, from the substantive nature of measurement, given that these are predicates of measurement, which is constrained when multiple frames of reference are in play to apply only to the singular frame of reference (section 13.1).46 At the center of antisemidistributivity is a lexical conflict between the null determiner that scatters what the NP denotes among frames of reference that are not path-integrated and measures that are plain nonsense unless that which is to be measured is so measured under a single frame of reference. What else could it be for a perfect circle to be a perfect circle, a cluster to be a cluster, or a galaxy a galaxy except that it is so under a single frame of reference and coordinate system where it satisfies the topology of a cluster? (348) #Trees are a perfect circle around the rim of a volcano. (349) #Unsolved murders were a cluster near the Green River in Washington. (350) #Innumerable stars light-years away are that galaxy at the edge of the universe. A departure from what is self-evident is to have assimilated all antisemidistributive predicates, so that an arithmetic predicate, fitting what is counted under a reticule, is also a measure under a single frame of reference, intolerant when its count is scattered among several: (351) #Apostles are twelve. If it means anything at all, the above pleads that reference to a singular frame of reference adheres to the very concept of a perfect circle, cluster, galaxy, or twelveness. Presumably then it attaches to the morphemes circle, cluster, galaxy, twelve, in all their occurrences, entraining then the question why the collective predicates in (352)–(357) and the like are not also antisemidistributive: (352) (353) (354) (355) (356)
Trees circle the rim of the volcano. Unsolved murders clustered near the Green River in Washington. Unsolved murders were clustered near the Green River in Washington. Unsolved murders were in a cluster near the Green River in Washington. Innumerable stars light-years away form that galaxy at the edge of the universe. (357) Apostles are at twelve faithful. It must be that further syntax distinguishes (352)–(357) from (348)–(351). In all of (348)–(357), there are to be considered the frames of reference F that trees, unsolved
704
Chapter 13
murders, innumerable stars, or Apostles are scattered across and the frame of reference f for a circle, cluster, galaxy, or dozen. It must be that the copular construction in (348)–(351) affords no change or discontinuity in perspective from, say, unsolved murders to cluster so that the F are the f, resulting in the contrary conditions being imposed on the same frames of reference—that the Fs are not path-integrated to an f. When instead the morpheme cluster is embedded in a prepositional phrase or when the decomposition of verb clustered or participle clustered reveals that the primitive morpheme cluster occurs as a secondary predicate denoting the end state of some other, precedent actions, it is allowed that the F frame unsolved murders in their precedent actions and the perspective then shifts without any scene continuity to a f of a cluster. Those scattered events of unsolved murders meet an end in which (however determined) the participants in those events constitute a cluster in some distant, independent frame of reference f. Except to recall the precedent in previous sections (sections 10.3 and 11.0) where adverbialization is shown to frame only local events that may be at great remove from those described later in the sentence, I will leave open the grammar that in mediating reference to F and to f distinguishes the decomposition in (352)–(357) from the copula constructions in (348)–(351).47 As for the antisemidistributive predicates themselves, recall that the substitution puzzle presents itself when coordinate NPs rescue a semidistributive interpretation: (358) #Many elms are a cluster in the middle of the forest. (359) a. Many clustered elms are a cluster in the middle of the forest. b. Many elms that are clustered are a cluster in the middle of the forest. c. Many elms in a cluster are a cluster in the middle of the forest. (360) Many elms and beeches are a cluster in the middle of the forest. All that remains for the puzzle’s solution is to explain (see section 14.1.1) how a coordinate NP in (360), in contrast to simple NPs, joins (359) (and (346) and the examples in note 45) in licensing the semidistributive interpretation. All NPs are AdrPs, and all AdrPs within the same DP address the same frame of reference, as the gloss below suggests: (361) #Many trees where there are elms (somewhere/anywhere) are there / in that place / in such a place a cluster in the middle of the forest. #Many trees where there are elms (in some spots) are there / in that place / in such a place a cluster in the middle of the forest. (362) Many trees where there are elms somewhere/anywhere and beeches elsewhere are there / in that place / in such a place a cluster in the middle of the forest. Many trees where there are elms in some spots and beeches in others are there / in that place / in such a place a cluster in the middle of the forest.
Antisemidistributivity vs. Conjunction Reduction Redux
705
It seems that a frame of reference so organized as to fix location of elms at some addresses and beeches at other addresses presupposes a discrimination of frames of reference both discrete enough to support singular reference to such frames and large enough to comprehend the measure of a cluster of trees therein. The many trees, given the null article, are scattered among plural frames of reference, but any one of these is large enough for elms and beeches and a cluster of them. But with nothing more to go on than that the frames of reference frame one or more elms, distributive singular reference to frames of reference that frame a cluster falls short. If so, nothing in the contrast between (358) and (360) warrants a novel sense for and.
14
[DP D AdrP and AdrP]
The reporters and the superheroes that stalk Metropolis and verify (1) may in some famous instances be the same, but in general (1)’s subject refers collectively to those who are reporters and those who are superheroes, each of whom is, more often than not, one of the one and not the other: (1) The reporters and superheroes that stalked Metropolis read The Daily Planet. (2) … reporter[Ei,Xi] and superhero[Ej,Xj] … A sentential connective and flanked by sentences, perhaps just the open sentences expressed by the nominal elements in (2), does not on its own take any steps toward collective reference, and for these sentences to share the same plural variable, Xi = Xj, would only lead to the unwelcome implication that the reporters are the superheroes. Rather, internal to DP are located the structures that delivered the disjunctive interpretation so-called in chapter 1 and discussed further in chapter 6, which allow that the 20,000 students of (3) be divided between those who surrounded the Pentagon and those crowded into the Mall: (3) Twenty thousand students surrounded the Pentagon and were crowded into the Mall. (4) Twenty thousand students W[E1,X] … [℩E2: ΦΨ]O[E1,E2] (Φ∃X surround the Pentagon) and (Ψ∃X be crowded into the Mall). Recall that 20,000 students are said to be the participants (‘W[E1,X]’) in that, referring to some surrounding the Pentagon (by some) and some being crowded into the Mall (by some). A backward event pronoun ⌜[℩E2:ΦΨ]⌝, its content fixed by the conjoined phrases, refers to the collective effect of what the students did. Existential closure disposes of the subject positions within the conjuncts, leaving vague who is at the Pentagon and who on the Mall. Similarly, the definite description in (5) refers to those participating in some infielding and some outfielding: (5) … [The X: ∃E1 W[E1,X] [℩E2: ΦΨ]O[E1,E2] (Φ∃X∃E2 infielders[E2,X]) and (Ψ∃X∃E2outfielders[E2,X])] …1
708
Chapter 14
The sentential connective and is no instrument of collective reference. There is no deriving a disjunctive interpretation for any phrasal coordination without the intervention of a descriptive pronoun referring to the events the phrases describe and some relation between these events and their participants. As in (3), the thematic roles, Agent, Patient, and so on, cannot be counted on to be sufficiently general to describe the divided participation (see the murderers and victims) and might not even fit nominal ascriptions,2 and thus (5), as (3) does, recruits ‘W[E,X]’ for the purpose. What is to be said about those respects in which the coordination infielders and outfielders does behave just like the simple plural fielders despite its eccentric syntax is that according to (5), the logical form for the infielders and outfielders provides a Φ such that the fielders are the W-ers who Φ: (6) fielders[E,X] ↔ infielders and outfielders[E,X] ↔ W[E,X] Φ[E] The logical form for the infielders and outfielders supplies, as it were, another simple head noun W, which, when restricted by Φ, ends up denoting what the counterpart simple plural noun fielders denotes (prior to their divergence in relation to scenes and frames of reference that embedding the NPs in AdrPs will impose). Contrary to the letter and spirit of Generalized Conjunction, with its mandate that a conjunction of phrases belong to exactly the same type and syntactic category as its conjuncts, the disjunctive interpretation finds expression only with the intervention of a logical syntax that puts it at some remove from a simple plural NP. There is in the syntax of Eventish no way to refer to the fielders here except by way of the events of infielding and outfielding in which they participate. Sentence (7) (see section 12.2) allows each accident victim to have been attended by exactly one motorist or pedestrian, and, event counting four, it could very well have been the same Good Samaritan at all four accidents. In contrast, sentence (8) is anomalous as a report of four victims and four passersby (i.e., without distributing to each victim his own nearby four passersby). (7) Last night in different parts of the city, before emergency services arrived, four accident victims flagged down four nearby motorists and pedestrians. (8) #Last night in different parts of the city, before emergency services arrived, four accident victims flagged down a nearby four motorists and pedestrians. Such a report asks four events of motoring and walking to constitute a single being nearby near four accidents that are not themselves near each other in space or time. It does not rescue this report to have the same Good Samaritan whose extended action that night has a corner at each accident and could therefore count as near them all. That action could very well be a being nearby all four accidents, but that action is one and not four. It is events counted four that must constitute the being nearby that cannot be. That is, the plural events E1 in (5), in which the passersby
[DP D AdrP and AdrP]
709
participate ‘W’, are said to be the participants in a being nearby (see section 12.2), requiring plural reference to them, as ⌜[℩Ei: ΦΨ]⌝ provides. Moreover, if in a nearby motorist and pedestrian that which there is a one of is a nearby-ing, in a motorist and pedestrian, it must be a W-ing, as the motoring and walking are two events. Thus section 12.2 has borne witness to the morphological weight of ‘W’ and its syntactic position. Section 14.2, betraying evidence of ‘W’ on other than formal grounds, will show that a W-ing is a scene tracking with fixed, uninterrupted gaze the things it is a scene of, glossing ‘W[E,X]’ as already suggested in chapter 2 as something more intentional and perspectival than a bland assertion of participation. What is seen in (5) reflects just the logical syntax of collective reference. If, further, every DP is a [D AdrP], the internal structure of the infielders and outfielders ought to expand accordingly, [the AdrP and AdrP]. Addressing frames of reference, any DP may impose epistemic conditions, such as ‘passim[E,F]’ or ‘some[E,F]’ (section 13.1), on the scenes or frames of reference referred to, and addresses in a given scene or frame of reference may answer to description of their location within it, such as being on the left: (9) [The X: [℩F: ∏] ∃E(some[E,F] ∃F (INFIELD[F , F ] At[F , E, t(E), F ] infielders[E,X]))] …; or, [℩F : ∏]…[The X: ∃E (some[E,F] ∃F (INFIELD[F , F ] At[F , E, t(E),F] infielders[E,X]))] The above defers to the preceding sections but it is more analysis than presently needed. So let ‘infield[F , E]’ abbreviate ‘infield[F ,F] At[F , E, t(E),F]’. Let adr be a schematic variable for such locative relations, and π a schematic variable for the epistemic conditions on frames of reference. A canonical DP looks like (10) [D X: [℩F : ∏] ∃E (π[E,F] ∃F (ADR[F , E] NP[E,X]))] …; or, [℩F : ∏] … [D X: ∃E (π[E,F] ∃F (ADR[F , E] NP[E,X]))] Combining now the apparatus for collective reference and the coordination of AdrPs, the infielders and outfielders is as in (11), suppressing further mention of any epistemic condition π on the frames of reference: (11) … [The X: [℩F : ∏] [℩E2: ΦΨ] W[E2,X] (Φ ∃E2 ∃F (infield[F , E2 ] ∃X infielders[E2, X ])) and (Ψ ∃E2 ∃F (outfield[F , E2 ] ∃X outfielders[E2, X ]))] …; or, [℩F : ∏] … [The X: [℩E2: ΦΨ] W[E2,X] (Φ ∃E2 ∃F (infield[F , E2 ] ∃X infielders[E2, X ])) and (Ψ ∃E2 ∃F (outfield[F , E2 ] ∃X outfielders[E2, X ]))] …
710
Chapter 14
(12) … [D X: [℩F : ∏] [℩E2: ΦΨ] W[E2,X] (Φ ∃E2 ∃F (ADR[F , E2 ] ∃X NPΦ [E2, X ])) and (Ψ ∃E2 ∃F (ADR[F , E2 ] ∃X NPΨ [E2, X ]))] … ; or, [℩F : ∏] … [D X: [℩E2: ΦΨ] W[E2,X] (Φ ∃E2 ∃F (ADR[F , E2 ] ∃X NPΦ [E2, X ])) and (Ψ ∃E2 ∃F (ADR[F , E2 ] ∃X NPΨ [E2, X ]))] … A point of grammar critical to the discussion that follows is that all AdrPs within the same DP address the same frames of reference, whether reference to them, ‘[℩F : ∏]’ above, is itself internal to the DP or quantified in. Thus, a critical contrast between an infielder and outfielder and an infielder and an outfielder is that the former must locate reporter and superhero in the same frame of reference and the latter need not, introducing as it does separate DPs. The above aspects of the grammar of DP and AdrP will resolve the puzzles of extensional substitutivity announced in section 8.1. In the first substitution puzzle, although the musicians are all and only the instrumentalists and vocalists, instrumentalists and vocalists cannot be freely substituted for musicians—not many instrumentalists and vocalists for many musicians (section 14.1.1). It is because the coordinated AdrPs segregate the instrumentalists and vocalists—two addresses for two AdrPs, assuming that the same address is not mentioned twice, and the simple AdrP does not, for any choice of adr: (13) [℩E2: ΦΨ] W[E2,X] (Φ ∃E2 ∃F (left[F , E2 ] ∃X instrumentalists[E2, X ])) and (Ψ ∃E2 ∃F (right[F , E2 ] ∃X vocalists[E2, X ])) ] (14) ∃E2 ∃F (ADR[F , E2 ] musicians[E2, X ])) For any given frame of reference, if the instrumentalists are at one address there and the vocalists at another (13), then of course these very musicians and these same events of musicianship are somewhere there (14) (i.e., for some adr, (13) ⊨ (14)). But their commingling at that venue cannot imply anything about their presentation as instrumentalists and vocalists (i.e., (14) ⊭ (13)). On these grounds, musicians and instrumentalists and vocalists cannot be substituted for one another salva veritate (section 14.1.1), despite denotation of the same events and same participants. For the second substitution puzzle, recall that the same person can sometimes be an instrumentalist and sometimes a vocalist, sometimes a motorist and sometimes a pedestrian, sometimes an infielder and sometimes an outfielder, or sometimes a civilian and sometimes a superhero. This class of puzzles relates a DP containing a coordination of AdrPs to a coordination of DPs containing the coordinated AdrPs: a civilian and superhero opposite a civilian and a superhero, the civilian and superhero opposite the civilian and the superhero. The opposites cannot be substituted
[DP D AdrP and AdrP]
711
salve veritate in contexts of plural reference (e.g., ‘were at the crime scene’), where the same person can be counted two, Clark Kent and Superman, as a civilian and a superhero but not counted two as a civilian and superhero. The coordination of DPs affords independent reference to distinct frames of reference, and even if the same frame of reference is intended, the separate DPs afford reference to distinct scenes therein for civilian and superhero. In contrast, the coordination of AdrPs puts the participation of civilian and superhero in the same W-ing. If the same W-ing is a single event, a W-ing, as it is in a civilian and superhero (see section 12.2), civilian and superhero are fixed under the same gaze, which cannot see them as two. As this remark implies, plural W-ings should allow civilians and superheroes to cloak their identities in several scenes despite the coordination of AdrPs within a single DP, and so it is that some civilians and superheroes may denote civilians Clark Kent, Bruce Wayne and Peter Parker and superheroes Superman, Batman, and Spiderman (section 14.2). ‘W’s fingerprints on the meaning of a civilian and superhero telling it apart from a civilian and a superhero is more than set theory or mereology can conjure from and, which left to its own devices can only render them synonyms. Section 14.2 will examine what exactly it is about the relation of civilian and superhero to the same frame of reference that is tantamount to putting them in the same photograph. In the last substitution puzzle (chapter 15), simple AdrP lovers and coordinate AdrP lover and belovèd again refer to the same things, and yet the assertion of their identity does not license their substitution salva veritate: (15) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned. (16) The lover and belovèd in either love note are the lovers exchanging them. (17) F The lover and belovèd in the first love note are not the lovers exchanging them. The resolution of this puzzle relies on the above grammar of DP and AdrP but it requires supplement from a tacit adverbial or secondary predicate, similarly oriented, so that what (15) truthfully denies given that the notes’ lovers are the same is their arrangement in their scenes the same way: (18) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned similarly oriented. Sections 14.0–14.1.1 investigate in detail the semantics and logical syntax of conjoined AdrPs, [AdrP and AdrP], resolving the first substitution puzzle and concluding the discussion of antisemidistributivity (chapter 13), the context for this puzzle. Section 14.1.2 interrupts, as it is here that the syntax and semantics of [AdrP and
712
Chapter 14
AdrP] are laid out, with an orphan problem for that conjunction, unrelated to puzzles of extensional substitutivity but urgent if the meaning of and is to remain innocent. In some contexts, a definite description containing conjoined AdrPs, the present English nobles and French princes, refers felicitously to the present English nobles, despite the nonexistence of French princes. The corresponding indefinite description, some English nobles and French princes, cannot however escape implication that both exist. Something then is to be said about the existential commitments of conjoined AdrPs. But note already that there is no warrant for and to mean that any AdrP it conjoins must denote, as this is not true when the conjoined AdrPs are embedded in a definite description. As it will advance the argument of section 14.1, section 14.0 precedes it with a discussion of the relevant empirical generalizations and approach to the second substitution problem, which finds its solution worked out in section 14.2, dedicated to the proper meaning of ‘W(e,x)’, said above to imply that a scene tracks with fixed, uninterrupted gaze the things it is a scene of. The last substitution puzzle is then treated in chapter 15, where the tacit adverbial similarly oriented is defined and couched in the logical syntax for identity and conditioned identity statements seen earlier (sections 10.3 and 11.1). 14.0
Within the same DP: The context for conjoined AdrPs
Both Jimmy Olsen and Clark Kent are reporters for The Daily Planet. Olsen down on his luck never seems to be at the right place at the right time—always a near miss for a piece of the action. It seems to the innocent bystander that Jimmy Olsen and Superman are tracking each other into and out of crime scenes (without ever quite meeting). Clark Kent is the savvier reporter, but being Superman, he has different reasons for showing up just before or after Superman does. Jimmy Olsen and Superman can be described indifferently by either (19)–(20) or (21)–(22): (19) A reporter and a superhero are tracking each other into and out of crime scenes. The reporter and the superhero are tracking each other into and out of crime scenes. (20) A reporter and a superhero were tracking each other into and out of City Hall. A reporter and a superhero were at the crime scene interviewing witnesses. A reporter and a superhero have entered City Hall one after the other. A reporter and a superhero were out of the mayor’s office in 18 minutes. A reporter and a superhero were talking to the police commissioner in private.
[DP D AdrP and AdrP]
713
(21) A reporter and superhero are tracking each other into and out of crime scenes. The reporter and superhero are tracking each other into and out of crime scenes. (22) A reporter and superhero were tracking each other into and out of City Hall. A reporter and superhero were at the crime scene interviewing witnesses. A reporter and superhero have entered City Hall one after the other. A reporter and superhero were out of the mayor’s office in 18 minutes. A reporter and superhero were talking to the police commissioner in private. In contrast, to describe Clark Kent and Superman, those who know their identity can use with some ironic detachment (19)–(20) but not (21)–(22). Even if the full DPs conjoined in (19) token a perspectival relation ‘W[E1,X]’ within, as in (23), the sentence affords separate presentations of the reporter-ing and the superhero-ing: (23) [The X: ∃E1 W[E1,X] … reporter[E1,X]] … W[E2,X] … and [The X: ∃E1 W[E1,X] … superhero[E1,X]] … W[E2,X] … are tracking each other into and out of crime scenes. The eye blinks long enough for a scene change from one DP to the next. Note further that adverbialization, whereby the descriptive content of a nominal becomes the descriptive content of an adjacent adverbial quantifier over events, will apply within the conjuncts. Thus, whatever the presentation of reporter-ing, it will frame only the reporter’s participation in tracking into and out of the crime scenes, and similarly for presentation of superhero-ing. In contrast, within the one DP in (21)– (22), a steady gaze fixes a single presentation of some reporter-ing and some superhero-ing, and under adverbialization that single presentation also frames the tracking into and out of crime scenes: (24) [The X: [℩E2: ΦΨ] W[E2,X] (Φ∃X∃E2 reporter[E2,X]) and (Ψ∃X∃E2superhero[E2,X])] [℩E2: W[E2,X] …] … W[E3,X] … are tracking each other into and out of crime scenes. When it is Jimmy Olsen and Superman, there is a single perspective that takes in without loss of continuity (some of) their reporter-ing and superhero-ing and their comings and goings at crime scenes, and it still affords a presentation of them as two, as the plural number agreement demands. For concreteness, assume a single frame of reference and coordinate system with respect to which all of their movements have been defined. When, on the other hand, it is Clark Kent and Superman, although distinct episodes of reporter-ing and superhero-ing are recorded, there is no single coordinate system that tracks them without leaving tracks that they are
714
Chapter 14
one. For both (19)–(20) and (21)–(22) and for Hoeksema’s (1983, 1988) (25), it is indeed necessary for the plural number agreement that there be distinct presentations, that Dr. Jekyll qua Dr. Jekyll is not Mr. Hyde qua Mr. Hyde, that, in my terms, Dr. Jekyll Dr. Jekyll-ing is not Mr. Hyde Mr. Hyde-ing. (25) Dr. Jekyll and Mr. Hyde were the same person. The contrast between (19)–(20) and (21)–(22) is meant to illustrate a further point, namely, that distinct presentations although necessary are not sufficient. In discussion of such examples, in the contexts normally evoked, which do not include, for example, watching Superman undress for a costume change into Clark Kent, there is no continuity between the presentations either with respect to their subject or with respect to the perspective of the observer. These are scenes from different points of view, and nothing in the logical form for (19)–(20) or (25) represent the speaker as knowing how to reconcile these scenes under translation into a single frame of reference for a single point of view. The logical form for (21)–(22), however, presupposes just this—that there is a single frame of reference and point of view on some action, some reporter-ing and superhero-ing. This is as much—or so it should turn out—as asking the reporter and superhero to sit for the same photograph, which Clark Kent and Superman cannot do, and the speaker knows it who knows that Clark Kent and Superman are one. The reporter-ing and superhero-ing are of course themselves displaced in time, and thus their single presentation is cinematic rather than photographic. It must be that the continuity conditions on what counts as a single cinematic presentation from a single point of view and frame of reference require it to interpolate the time and space between the reporter-ing and superheroing while tracking their subjects (according to adverbialization) through their comings and goings into and out of crime scenes, which again cannot be done without exposing that Clark Kent and Superman are the same, one citizen of Metropolis. If they are not two under the single presentation that (21)–(22) presupposes, then they are not two for the plural number agreement of the sentence. If the sentence tolerates singular number agreement, one may with Clark Kent and Superman in mind continue in the singular as in (29). Otherwise, the contrast between (26) and (28) is just the one observed between (19)–(20) and (21)–(22): (26) A reporter and a superhero are in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. The reporter and the superhero are in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. (27) A reporter and a superhero is in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. The reporter and the superhero is in the field for The Daily Planet and at the vanguard for truth, justice, and the American way.
[DP D AdrP and AdrP]
715
(28) #A reporter and superhero are in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. #The reporter and superhero are in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. (29) A reporter and superhero is in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. The reporter and superhero is in the field for The Daily Planet and at the vanguard for truth, justice, and the American way. Recall from preceding sections (e.g., section 10.0.9) and from chapter 1 (cf. (4) and (24) above) that ‘W[Ei,X]’ is also tokened in the matrix clause. The events related by this token are just those framed by, or located in the neighborhood of, whatever events have been introduced by the adverbialization of the subject. The plural number agreement belongs to the matrix clause and its condition is imposed on this last glimpse of the participants, which must support a percept of them as more than one. In (30)–(31), the coordination of DPs (rather than of AdrPs internal to a DP) does not itself impose a single frame of reference on the morning star and the evening star, which are therefore introduced under perhaps discontinuous, separate presentations, and these frame movements as reported in (30) and (31) that preserve a percept of two:3 (30) The morning star and the evening star are tracing the ecliptic from one day to the next. (31) The morning star and the evening star are following each other across the sky like morning follows day. (32) *From the vantage point of Jupiter, (you can see that) the morning star and the evening star are circling the sun. From the vantage point of Jupiter, (you can see that) the morning star and the evening star is circling the sun. In (32), in contrast, the only respect in which morning star and evening star presentations, which exist only on Earth, frame any presentations from Jupiter is in virtue of being presentations of the same object Venus.4 In this context, the presentations introduced by the nominal descriptions converge on just one for the matrix event, undermining the plural number agreement. When the speaker has in mind Clark Kent and Superman (rather than Jimmy Olsen and Superman), the reporter and superhero and a reporter and superhero diverge in felicitous use from the reporter and the superhero and a reporter and a superhero. This divergence reflects the distribution of a perspectival ‘W[Ei,X]’. In particular, a DP that conjoins AdrPs within can aim at plural reference to a reporter and a superhero only by describing them as the participants under some perspective in some reporter-ing and some superhero-ing, which in turn imposes on them and their action an arrangement that comports with a single perspective and frame of
716
Chapter 14
reference. This condition on the context for conjoined AdrPs inside the same DP will matter to the syntax and semantics of [AdrP and AdrP], in particular in an argument on behalf of the solution to the substitution puzzle that attaches to apparently coextensive simple and conjoined AdrPs, fielders vs. infielders and outfielders. 14.1
AdrPs coordinated
A coordination of plural AdrPs substitutes for a simple plural AdrP without much import for the results and conclusions of chapter 12: what was said about nominal semantics prior to consideration of and within DP. The arguments carry over mutatis mutandis with examples slightly more complicated by the substitution. The event counting demonstrated by singular plurals with simple plural AdrPs (section 12.2), (many) an ill-matched nine fielders, (many) a sequestered twenty-four jurors, (many) a lengthened/*long twelve skirts, and (many) a bloodred seven sunsets, carries over to singular plurals with coordinate AdrPs, (many) an ill-matched nine infielders and outfielders, (many) a sequestered twenty-four seated jurors and alternates, (many) a lengthened/*long twelve skirts and dresses, (many) a bloodred seven sunrises and sunsets. A difference in counting and anaphora was found to separate many mobile phones from many a one or more mobile phones and many natural numbers from many a one or more natural numbers (sections 12.3.1–12.3.2). It separates just the same many smartphones and cellphones from many a one or more smartphones and cellphones and many odd numbers and even numbers from many a one or more odd numbers and even numbers. The semidistributive interpretation obtained with distributive plurals many elms and many senators is not derived by the plural quantification of many a one or more elms and many a one or more senators (section 13.0.0). The same can be said to tell apart many elms and beeches and many Democrats and Republicans from many a one or more elms and beeches and many a one or more Democrats and Republicans. 14.1.0
Landmark [AdrP and AdrP]
Another respect in which coordinated AdrPs behave like simple AdrPs is in determining the space for (imperfective) measurement—and more generally viewfinding as introduced later below—but it is also where they first diverge. The contrast between most tincture of iodine and most of the tincture of iodine (see (310)–(313)) carries over to coordinated NPs. Protium, deuterium, and tritium are the isotopes of hydrogen in nature, and protons, deuterons, and tritons are their atoms. Only tritium is radioactive, occurring in seawater in trace amounts: (33) F Most of the protium(, deuterium,) and tritium in seawater is radioactive. (34) F Most of the protons(, deuterons,) and tritons in seawater are radioactive.
[DP D AdrP and AdrP]
717
(35) T Most protium(, deuterium,) and tritium in seawater is radioactive. (36) T Most protons(, deuterons,) and tritons in seawater are radioactive. Anything containing tritium is radioactive and trips a Geiger counter. Any measurement in (35) and (36) is of events (or states) in which all three (or two) isotopes participate, and therefore most of what is measured is radioactive.5 This last observation also becomes the first where coordinated NPs diverge from coextensive simple NPs as in (37)–(38), and it is an affront to the presumed equivalences in (39)–(40):6 (37) F Most hydrogen in seawater is radioactive. (38) F Most hydrogen atoms in seawater are radioactive.7 (39) a. hydrogen[E,X] ↔ protium and deuterium and tritium[E,X] b. protium and deuterium and tritium[E,X] ↔ W[E,X] Φ(35)[E]8 (40) a. hydrogen atoms[E,X] ↔ protons and deuterons and tritons[E,X] b. protons and deuterons and tritons[E,X] ↔ W[E,X] Φ(36)[E] Recall from section 13.1.1 that the measurement in (311), in contrast to (313), is restricted to measurement while tinctured—that is, to measurement in the constant presence of iodine: (311) F Most tincture of iodine is just ethanol. (313) Most of the tincture of iodine is just ethanol. The restriction derives from the fact that there is no event or events E—no matter how small—that ∃X tincture of iodine[E,X] without iodine in them, a fact that matters twice. First, the adverbialization of the nominal refers to events—while tinctured with iodine—of which it is false that there is just ethanol, and second, as all counting or measurement is counting or measurement of events, everything measured is tinctured with iodine, where the simpler (41) has the logical form (42):9 (41) F Much tincture of iodine is just ethanol. (42) [∃X : ∃E0(∃eμ∃n(much(n) & ∃eμ measure[eμ,E0,n]) tincture of iodine[E0,X])] … (Cf. [∃X : ∃E0(∃eμ count[eμ,E03000] passengers[E0,X])] …) Likewise, any sampling in the course of measurement samples and measures tincture, all of which is iodized: (43) Sampling, much then/now measured tincture of iodine was/is just ethanol. As imperfective measurement is understood, the report in (43) is vague as to whether the sampling now viewed in progress coincides with a single continuous and uninterrupted measurement, say, of a flowmeter with continuous readout of
718
Chapter 14
iodine concentration, or with a series of discrete, punctuated samples and measurements. Either way, (43)’s report is also ambiguous: as one goes along (the imperfective temporal dilation), either (i) again and again, much tincture of iodine is just ethanol, or (ii) much tincture of iodine has accumulated that is just ethanol. I will presume that imperfective aspect requires a description of an essentially egocentric point of view, now-en-scène, the gerund in (43), overt there and tacit elsewhere, which occurs as a frame adverbial, displaced to sentence-initial position, from which it restricts a more distant expression of imperfective aspect’s temporal dilation, a temporal distributive operator: (44) [∃Eμ: (i now-en-scène[Eμ] measuring[Eμ])] … [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] … That is, the sampling or measuring now in progress is such that for any moment therein, the measurement of that moment is so-and-so. The measurement of the moment is the instantaneous one if the flowmeter runs continuously with continuous readout or it is the discrete, extended event of measurement that this moment falls within before it reads out. All of this is of course jury-rigging for a proper treatment of imperfective aspect and the progressive. The ambiguity just noted in what measures much tincture of iodine may play out as a scope ambiguity for the DP: (45) [∃Eμ: (i now-en-scène[Eμ] measuring[Eμ])] … [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] … much tincture of iodine … (46) [∃Eμ: (i now-en-scène[Eμ] measuring[Eμ])] … much tincture of iodine … [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] … Withholding less detail: (47) [∃Eμ: (i now-en-scène[Eμ] ∃ζ ∃n measurei[Eμ, ζ, n])], [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] [∃X : ∃E0(∃n(much(n) & measure[Eμ,E0,n]) tincture of iodine[E0,X])] … ‘Measuringi now, at any time’s measurementj thereini, much tincture of iodine thereinj …’ (48) [∃Eμ: (i now-en-scène[Eμ] ∃ζ ∃n measurei[Eμ, ζ, n])], [∃X : ∃E0(∃n(much(n) & measure[Eμ,E0,n]) tincture of iodine[E0,X])] [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] … ‘Measuringi now, much tincture of iodine therewithali at any measurement thereinI …’ The above logical forms also make explicit that I assume the frame adverbial also frames the nominal description of measurement, which contains pronouns bound to it: measuringi, much tincture theni of iodine… . As tentative as this all is, I should
[DP D AdrP and AdrP]
719
not rule out a logical form that suppresses temporal distributivity altogether in order to represent the summary measure: (49) [∃Eμ: (i now-en-scène[Eμ] ∃ζ ∃n measurei[Eμ, ζ, n])], [∃X : ∃E0(∃n(much(n) & measure[Eμ,E0,n]) tincture of iodine[E0,X])] … ‘Measuringi now, much tincture of iodine therewithali …’ In all the candidate logical forms, there is no measurement in progress or completed except in the presence of tincture and therefore of iodine, none of which is pure ethanol, falsifying as desired both (41) and (43). Returning to (37) and (38), that they are false derives from the fact that among the events E, of there being hydrogen (∃X hydrogen[E,X]) or hydrogen atoms (∃X hydrogen atoms[E,X]), those with tritium or tritons present are rare. But if so, the truth of (35) and (36) is unexpected given the alleged equivalences in (39)–(40). Sampling hydrogen rarely samples tritium, but something about the syntax and semantics of NP-coordination per se enforces that the sampling for measurement and assessment reported in (35)–(36) always contain protium(, deuterium,) and tritium.10 Given the equivalences in (39)–(40), something besides the eccentric syntax of disjunctively interpreted coordination must tell apart hydrogen from protium and tritium. Recall that the two conjuncts are rather AdrPs assigning different addresses to protium and tritium for a given frame of reference. In contrast, any address occurring with hydrogen inside its AdrP locates it all at the same address. This difference is exploited to explain how apparently coordinate NPs and coextensive simple NPs diverge in meaning. The vocabulary of addressing, left, right, top, down, north, south, etc., and the coordinate systems themselves for frames of reference are universal: there is always a left and right, and for any triple of altitude, azimuth, and radial distance, no matter the scale or frame of reference. Scanning and measuring with an instrumental frame of reference mobile and perhaps zooming in or out, even if there is always hydrogen there, on the left one moment, on the right, the next, there need not be both isotopes present to view. Scanning, however, with protium always on the left, as it were, and tritium on the right is always to sample radioactive matter no matter how focused and narrow the instrumental frame of reference becomes. Dressing it up so that AdrPs join the syntax of coordination, (50) has the candidate logical form (51) to represent distributed, repeated measurement and (52) to represent summary measure: (50) Sampling, many protons and tritons are radioactive.
720
Chapter 14
(51) [∃Eμ: (i now-en-scène[Eμ] ∃ζ ∃n count[Eμ, ζ, n])], [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ][∃F: FoR[Etμ,F]] [∃X : ∃E0 ( W[E0,X] [℩E1: ΦΨ]O[E0,E1] [℩E1: ΦΨ] (∃Eμ ∃n(many(n) & count[Eμ,E0,n]) (Φ ∃E1 ∃F (ADR[F , E1 ] [℩t: t = tμ]t(E1)t & ∃X protons[E1,X])) and (Ψ ∃E1 ∃F (ADR[F , E1 ] [℩t: t = tμ]t(E1)t & ∃X tritons[E1,X])))] … ‘Measuring now, at any moment’s measurementi, thereini, many protons theni & somewhere therei and tritons theni & somewhere therei …’ (52) [∃Eμ: now-en-scène[Eμ] ∃ζ ∃n count[Eμ, ζ, n])], [∃F: FoR[Eμ,F]] [∃X : ∃E0 (W[E0,X] [℩E1: ΦΨ]O[E0,E1] [℩E1: ΦΨ] (∃Eμ ∃n(many(n) & count[Eμ,E0,n]) ( Φ ∃E1 ∃F (ADR[F , E1 ] [℩t: t = t(Eμ)]t = t(E1)& ∃X protons[E1,X])) and ( Ψ ∃E1 ∃F (ADR[F , E1 ] [℩t: t = t(Eμ)]t = t(E1) & ∃X tritons[E1,X])))]… … [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] … ‘Measuringi now, thereini, therewithali many protons theni & somewhere therei and tritons theni & somewhere therei … (at any moment’s measurement thereini) …’11 Recall (section 9.4.2) that an event of (visual) counting or measuring is a scene (53), so that reference (‘[∃F: FoR[Eμ,F]]’ in (51)–(52)) to frame of reference under which it is conducted is assured (54):12 (53) count(eμ) → ∃sɶ∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) measure(eμ) → ∃sɶ∃α sɶ At(α sɶ, eµ, t( sɶ), sɶ) (54) FOR[E, F ] ↔ def [∃Sɶ : ∃N orient[Sɶ , F , N ]] ∃F en-scène[E, Sɶ , F , F ] The addresses F are in frames of reference to which the measurement is oriented, presumably the egocentric frame of reference of the instrument or counter or the observer bearing it, which may very well be mobile during the course of its progressive scan. So much is true enough. But in emphasizing the protocol for counting or measurement, it neglects what is already given in cineramic semantics, namely, that any utterance may be presumed narration for concurrent cinerama—whether or not a pronounced adverbial invokes it or measurement is in progress ((55)–(56)) and whether or not the cinerama in progress derives from the scenes witnessed by multiple observers ((57)–(58)): (55) (Surveying the damage after the war), bullet holes pockmarked the walls alongside the narrow street. (56) (Surveying the damage after the war), bullet holes pockmarked the walls alongside the narrow street less and less, further from the front line.
[DP D AdrP and AdrP]
721
(57) (Surveying the damage after the war), bullet holes pockmarked the walls alongside the cities’ narrow streets. (58) (Surveying the damage after the war), bullet holes pockmarked the walls alongside the cities’ narrow streets less and less, further from the front line. Viewfinding, tacit reference to the frames of reference of F induced (FoR[E,F]) by the camera’s motion E, is pervasive. Absent demonstration to some determinate, bounded space, recall from section 9.4.0 that a frame of reference per se, an egocentric frame of reference in particular, is a coordinate system for all space. (59) … [∃F: FoR[Eμ,F]] … ∃F (ADR[F , E0 ] … If so, (59) in (51)–(52), in restricting addresses to those in the egocentric frame of reference (at the moment of observation), does not strictly speaking locate them in view, in the space the current scene’s projection subtends. One might think the camera privileges the space in view, which is made explicit in the relation between scenes and frames of reference: (60) viewfinder(ṡ,fṡ) ↔def f © ∃f(ṡ© = projection(t(ṡ©),σf(ṡ©),πf(ṡ©), f, l(ṡ©), No( sɺ ) ,o(ṡ )) & © © ∃αf en-scène(fṡ, ṡ, αf, f) & ∃αfṡ At(αfṡ, σf(ṡ ),t(ṡ ), fṡ)) For a still, instantaneous scene ṡ (originally) projected in frame of reference f, ṡ is a viewfinder for (sub)frame of reference fṡ just in case fṡ (itself a subspace, recall) is in the scene ṡ, and the space σf(ṡ©) that the scene subtends is (addressable) in the (sub)frame of reference fṡ. In effect, ṡ zooms in visual attention so that any addresses indexed to the viewfinder frame of reference fṡ are all within the space subtended. A cinematic scene, if the camera is in motion or zooming in or out, varies the space subtended so that it is a temporal function from its instantaneous scenes to viewfinder frames of reference Fsɶ : (61) viewfinder( sɶ, Fsɶ ) ↔ def ∀t(LTsɶ ( sɶ)(t ) → ∃f (Fsɶ f & viewfinder( sɶ(t ), f ))) & ∀f (Fsɶ f → ∃t(LTsɶ ( sɶ)(t ) & viewfinder( sɶ(t ), f ))) & ∀t∀f0 ∀f1 ((LTsɶ ( sɶ)(t ) & viewfinder( sɶ(t ), f0 ) & viewfinder( sɶ(t ), f1 ) → f0 = f1 ) As reminded above, any cinerama π may derive from scenes Sɶπ from different sources, which a pluralized viewfinder relation accommodates: ɶ ɶ][∃F : [∀f : Ff ] F ɶ f ]] viewfinder( sɶ, F ) & (62) viewfinder[Sɶ , FSɶ ] ↔ def [∀sɶ : Ss S ɶ ɶ] [∃F : Ff & [∀f : Ff ] F ɶ f ]] viewfinder( sɶ, F ) [∀f : FSɶ f ][∃sɶ : Ss S
722
Chapter 14
Events in view, in virtue of the scenes that are scenes of them, and the addresses the camera privileges are those in the viewfinder. So first relate them all as in (63) to define the derivative relations in (64)–(65): (63) (64) (65)
viewfinder[[Sɶ , F ]] & en-scène[E, Sɶ , F , F ]) VIEWFINDER[ E, Sɶ , F , F ] VIEWFINDER[ E, Sɶ , F , F ]
VIEWFINDER[ E, Sɶ , F , F ] ↔ def VF[ E, F ] ↔ def ∃Sɶ ∃F VF[ E, F ] ↔ def ∃Sɶ ∃F
A few examples will chart the interaction in viewfinding between scenery and the events referred to. It will matter whether the cinerama is a solitary scene sɶ or a montage Sɶ of many and whether AdrPs relate to the same frames of reference, coordinated inside a shared DP, [D AdrP and AdrP], or not, on their own in separate DPs, [D AdrP] … and … [D AdrP]. These examples are preliminary to a discussion that returns to scenes of measurement per se under viewfinding. To begin with viewfinding, examples (66)–(67) contain overt AdrPs, left and right, indexed to the egocentric frame of reference and hold this descriptive content constant as the surrounding DP-structure varies: [DP a [stream AdrP and dry creek AdrP]], [DP some [streams AdrP and dry creeks AdrP]], and [DP a [stream AdrP]] … and [DP a [dry creek AdrP]] …:13 (66) Cycling at twilight, a stream on the left and dry creek on the right drew the local fauna into the gloaming. [∃Ec: Cycling at twilight[Ec]][∃E: (i now-en-scène[E] N[Ec,E])],14,15 [∀t: t(E)t][℩Et: proi & t(Et)t][∃F: FoR[Et,F]] [∃X : ∃E0 (an[E0]16 now-en-scène[E0] W[E0,X] [℩E1: ΦΨ]O[E0,E1] (Φ ∃E1[∃ : VF[E0, ]] (left[E1, , F ] 17 [℩t1: t1 = t]t(E1)t1 & ∃X stream.sg[E1,X])) and ( Ψ ∃E1 [∃A: VF[E0, ]] (right[E1, , F ] [℩t1: t1 = t]t(E1)t1 & ∃X dry creek.sg[E1,X])))] … (67) Cycling at twilight, some streams on the left and dry creeks on the right drew the local fauna into the gloaming. [∃Ec: Cycling at twilight[Ec]][∃E: (i now-en-scène[E] N[Ec,E])], [∀t: t(E)t][℩Et: proi & t(Et)t][∃F: FoR[Et,F]] [∃X : ∃E0 (some[E0,F] now-en-scène[E0] W[E0,X] [℩E1: ΦΨ]O[E0,E1] ( Φ ∃E1 [∃ : VF[E0, ]](left[E1, , F ] [℩t1: t1 = t]t(E1)t1 & ∃X streams[E1,X])) and ( Ψ ∃E1 [∃ : VF[E0, ]](right[E1, , F ] [℩t1: t1 = t]t(E1)t1 & ∃X dry creeks[E1,X])))] …
[DP D AdrP and AdrP]
723
(68) Cycling at twilight, a stream on the left and a dry creek on the right drew the local fauna into the gloaming. [∃Ec: Cycling at twilight[Ec]][∃E: (i now-en-scène[E] N[Ec,E])], [∀t: t(E)t][℩Et: proi & t(Et)t][∃F: FoR[Et,F]] [∃X : ∃E0 (an[E0] now-en-scène[E0] [∃A: VF[E0, ]](left[E0, , F ] [℩t1: t1 = t]t(E0)t1 stream.sg[E0,X])))] … and [∃X : ∃E0 (an[E0] now-en-scène[E0] [∃A: VF[E0, ]](left[E0, , F ] [℩t1: t1 = t]t(E0)t1 dry creek.sg[E0,X])))] … Sentence (66) is a report consonant with the experience of a determined cyclist, gaze focused on the road ahead and on a scene at any moment of which there is stream on its left and dry creek on its right. The logical form records that there is a single event of W-ing (‘an[E0]’), in which both stream and dry creek participate. No single event en-scène is ever divided across scenes. The scene for this event is therefore a viewfinder for the addresses that must be found within it for both stream and dry creek, as needed to render the sentence’s meaning.18 In contrast, (67) and strikingly (68) are also felicitous reports, which (66) is not, of a more leisurely, touristic ride along the same route where the cyclist alternates turning her head to the left for long stretches to study the scenery on that side of the road and then to the right to do the same. Looking left, she always sees a stream; looking right, a dry creek. But, lingering as she does on one side or the other, she cannot be certain that it is the same stream or dry creek she returns to when she looks back the other way. Of course, (67) and (68) also fit the report of the determined cyclist’s ride, too.19 For (67), the logical form records that stream and dry creek are divided among several W-ings (perhaps as numerous as the streams and dry creeks themselves) for which now-en-scène are several cinematic scenes Sɶ that fall within the now of the current cycling episode. The addressed regions for stream and dry creek are shared among these as viewfinders for the spaces they subtend. Note that when the cyclist turns her head 90 degrees to the left, the stream occupies her entire field of view without confinement to its left. In the natural understanding of the sentence, left and right are hemispheres for that frame of reference in which her bicycle and direction of travel remain throughout aligned with . Although she may be utterly lost and disoriented to any frame of reference other than the road, scenes projected in the frame of reference aligned with her nose and line of sight are immediately path-integrated to a frame of reference for the ambient environment, the city block or the road in this case, whether or not she is oriented to anything larger. The logical form allows that the regions found within the viewfinder be regions subsequently characterized by their position in a frame of reference outside the space the viewfinder subtends. With two DPs, (68) obviously awards stream and dry creek their
724
Chapter 14
own scene, and thus it too is a felicitous report of either ride. These sentences also all participate in the scope ambiguity noted above for much tincture of iodine to distinguish distributed repeated measurement, the logical forms shown in (66)–(68), from a summary measure, which would allow the plural streams and dry creeks (67) reports to be spread out along the route with perhaps no more than one in view at any moment. The logical forms recruited in translation in (66)–(68) all conform to the following grammar: DPs and only full DPs initiate (existential) event quantification, cueing adverbialization. Their determiners and articles—demonstratives this, that, etc., as is well known, and (in)definite articles an, some (and null and the (section 13.1))—come with epistemic conditions, and these conditions constrain spatial reference in those AdrPs that share the same DP to the same viewfinder. Viewfinding constrains scenes of measurement, as a result of which simple and coordinate AdrPs differ not only in what is measured or sampled, hydrogen vs. protium, deuterium, and tritium, but also in the events of measurement they allow. Babette prepares her feast (69) in 240 minutes, assembling in 40 minutes the ingredients for four courses, which are finished simultaneously in the last 40 minutes. Her preparation is otherwise a rotation among four stations for periods of 10 minutes during which her devotions are exclusive to one course, all as depicted in (70). Both (71), with NP-coordination, and (72), with DP-coordination, are felicitous reports of the feast’s complete preparation: (69) Soup: Appetizer: Main Dish: Dessert:
Potage à la Tortue Blini Demidoff au Caviar Caille en Sarcophage avec Sauce Périgourdine Baba au Rhum avec les Figues
(70) 40 min. 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 40 min. SAMD S A M D S A M D S A M D S A M D SAMD (71) a. A soup, appetizer, main dish, and dessert took 240 minutes to prepare. The soup, appetizer, main dish, and dessert took 240 minutes to prepare. b. A soup, appetizer, main dish, and dessert have taken 240 minutes to prepare. The soup, appetizer, main dish, and dessert have taken 240 minutes to prepare. c. A soup, appetizer, main dish, and dessert has taken 240 minutes to prepare. The soup, appetizer, main dish, and dessert has taken 240 minutes to prepare.
[DP D AdrP and AdrP]
725
(72) a. A soup, an appetizer, a main dish, and a dessert took 240 minutes to prepare. The soup, the appetizer, the main dish, and the dessert took 240 minutes to prepare. b. A soup, an appetizer, a main dish, and a dessert have taken 240 minutes to prepare. The soup, the appetizer, the main dish, and the dessert have taken 240 minutes to prepare. A report of selected courses, say, appetizer and dessert, faces a choice whether to measure only the time of their active preparation, which totals 160 minutes (73), or the 240 minutes they were in preparation despite interruption from other tasks or distractions.20 (73) 40 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 40 AD A D A D A D A D AD The latter measurement may be reported indifferently with either the NP-coordination in (74) or the DP-coordination in (75): (74) a. An appetizer and dessert took 240 minutes to prepare. The appetizer and dessert took 240 minutes to prepare. b. An appetizer and dessert have taken 240 minutes to prepare. The appetizer and dessert have taken 240 minutes to prepare. c. An appetizer and dessert has taken 240 minutes to prepare. The appetizer and dessert has taken 240 minutes to prepare. (75) a. An appetizer and a dessert took 240 minutes to prepare. The appetizer and the dessert took 240 minutes to prepare. b. An appetizer and a dessert have taken 240 minutes to prepare. The appetizer and the dessert have taken 240 minutes to prepare. In contrast, measuring the scattered intervals of active preparation requires the DPcoordination in (77): (76) a. #An appetizer and dessert took 160 minutes to prepare. #The appetizer and dessert took 160 minutes to prepare. b. #An appetizer and dessert have taken 160 minutes to prepare. #The appetizer and dessert have taken 160 minutes to prepare. c. #An appetizer and dessert has taken 160 minutes to prepare. #The appetizer and dessert has taken 160 minutes to prepare.
726
Chapter 14
(77) a. An appetizer and a dessert took 160 minutes to prepare. An appetizer and a dessert took 160 minutes of active preparation. The appetizer and the dessert took 160 minutes to prepare. The appetizer and the dessert took 160 minutes of active preparation. b. An appetizer and a dessert have taken 160 minutes to prepare. An appetizer and a dessert have taken 160 minutes of active preparation. The appetizer and the dessert have taken 160 minutes to prepare. The appetizer and the dessert have taken 160 minutes of active preparation. The grounds for rejecting (76) might seem plain enough. With only the one token of a(n), the sentence is about a one thing—object, event, or state—that is appetizer and dessert, and it must fit into whatever it took part in. Which 160-minute interval would that be? But this had better be a specious explanation for (76) since it would also be grounds to reject the report in (78) of the appetizer’s active preparation time, where of course the appetizer’s total preparation need not fit within an interval only as long as its active preparation: (78) An appetizer took 120 minutes to prepare. The appetizer took 120 minutes to prepare. (79) An appetizer took 240 minutes to prepare. The appetizer took 240 minutes to prepare. Sentences (76) and (78) are parallel in their structure and evaluation except for what derives from AdrP-coordination. A striking, closer contrast for (76) is (80). Suppose that gastronomy dictates that the courses immediately preceding and following the main dish, the accompaniment, achieve a certain balance in juxtaposition to the main dish: (80) An accompaniment took 160 minutes to prepare. The accompaniment took 160 minutes to prepare. In (76) and (80), accompaniment is to appetizer and dessert as hydrogen is to protium, deuterium, and tritium in (35) and (37). The logical form for the simple AdrP (80) is (81), suppressing some details:21 (80) [[DPAn.sg accompaniment.sg] … W … θ … Tense …]22 (81) [∃F: ∏] [∃X: ∃E0 (Δ an[E0] now-en-scène[E0] [∃ : VF[E0, ]](ADR[E0, , F ] ∃X accompaniment.sg[E0,X]))] [℩E0 : Δ] [∃E1 : N[E0,E1]] …W[E1,X] O[E1,E2] ∃X θ[E2,X] … [∃F: ∏](… Tense[Ei,F] … )
[DP D AdrP and AdrP]
727
The AdrP-coordination an appetizer and dessert in (76) (repeated below) has the logical form (82),23 and the DP-coordination (77) an appetizer and a dessert, the logical form (83): (76) #[[DPAn.sg appetizer.sg and dessert.sg] … W … θ … Tense …] (82) … [∃F: ∏] [∃X: ∃E0 (Δ an[E0] now-en-scène[E0]W[E0,X][℩E1: ΦΨ]O[E0,E1] (Φ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X appetizer.SG[E1, X ])) and (Ψ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X dessert.SG[E1, X ]))] [℩E0 : Δ] [∃E2 : N[E0,E2]] … W[E2,X] O[E2,E3] ∃X θ[E3,X] … [∃F: ∏](… Tense[Ei,F] … ) (77) [DPAn.sg appetizer.sg]…W…θ… and [DPa.sg dessert.sg]…W…θ… Tense… (83) … [∃F0: ∏] [∃X: ∃E0 (Δan[E0] now-en-scène[E0] [∃ : VF[E0, ]](ADR[E0, , F0 ] ∃X appetizer.SG[E0, X ]))] [℩E0 : Δ] [∃E3 : N[E,E3]] … W[E3,X] O[E3,E4] ∃X θ[E4,X] and [∃F1: ∏] [∃X: ∃E0 (Δan(E0) ∃Eμ ∃n measure[Eμ,E0,n] [∃ : VF[E0, ]](ADR[E0, , F1 ] ∃X dessert.SG[E0, X ]))] [℩E0 : Δ] [∃E3 : N[E,E3]] … W[E3,X] O[E3,E4] ∃X θ[E4,X] … [∃F: ∏](… Tense[Ei,F] … ) Prior to parsing what fails in (76)/(82), first fix what for all (71)–(80) conveys that the measurement will be of active or of total preparation time. What the subject phrases in (81)–(83) denote starts with a determination of what events in this context are denoted by the nominal phrases, ‘∃X accompaniment[E,X]’, ‘∃X appetizer[E,X]’ and ‘∃X dessert[E,X]’. Much as ‘∃X Superman[E,X]’ is taken to denote only events of superheroism, it could be that ‘∃X appetizer[E,X]’ denotes only events where some stuff is appetizing, which would be during the appetizer course at table. Of course, that cannot be the usage here without a meal in progress. It denotes rather states of stuff with appetizing in its future (cf. the discussion of liar and backstabber, section 10.2). Whatever one thinks about the existence of objects during their creation, if there are any events E within the 240 minutes of (70) that ∃X appetizer[E,X], there are such events throughout, at any moment while the feast is in preparation. It cannot be that the appetizer exists while under active preparation and slips out of existence when not. The same holds of ‘∃X dessert[E,X]’ and the dessert and of ‘∃X accompaniment[E,X]’ and the accompaniment. If so, the adverbialization of the nominals, while an accompaniment, while an appetizer and dessert, and while an appetizer and while a dessert, cannot be what narrows time to 160 minutes when the intention is to report the time of active preparation. That falls to the thematic relation within the scope of derived adverbials and how the event concept take is glossed in this context. If, for example, the understood thematic relation is Theme, implying movement or undergoing transition, it would hold only during active preparation, the 160 minutes for appetizer and dessert. A locative
728
Chapter 14
relation that holds of a state of being an appetizer for as long as and throughout its period as an active site for preparation would extend to the 240 minutes for appetizer and dessert and for accompaniment. To parse what fails in (76), recall from section 13.1 that an antisemidistributive predicate such as take n minutes to prepare itself implies measurement under a single frame of reference. Thus, the one or more frames of reference F introduced with the Tensed Phrase, ‘[∃F: ∏]’ in the above logical forms, must be a single frame of reference, call it f, under which fit the events that appetizer and dessert participate in or those that accompaniment participates in, measuring 160 minutes active participation or 240 minutes passive participation. That is, there is a single frame of reference f, which verifies (81) and (83) assigning it as the value of variables F0, F1, and F and thus f should be good for both tokens of F in (82) too. What is it then that makes (76) ( = (82)) a false report of what is going on under f, in contrast to the true reports (80) ( = (81)) and (77) ( = (83)) under the same single frame of reference? Walk through the evaluation that renders both (79) and (80) true in (73), where accompaniment is appetizer and dessert: (73) 40 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 40 AD A D A D A D A D AD (79) An accompaniment took 240 minutes in total preparation time. The accompaniment took 240 minutes in total preparation time. (80) An accompaniment took 160 minutes of active preparation. The accompaniment took 160 minutes of active preparation. As remarked earlier, since the sentences are true there is an accompaniment while being accompaniment, and since while being accompaniment denotes some events, it denotes events spanning 240 minutes. While being accompaniment, the accompaniment participates either in some events (‘loc’ for θ in (81)) that turn out to be a total preparation of 240 minutes or in some events (‘theme’ for θ in (81)) that turn out to be active preparation for 160 minutes. Straightforwardly, there is in (73) a single (maximal) event while being accompaniment that measures 240 minutes of total preparation. On the other hand, no single event in (73) measures 160 minutes of active preparation. The truth of (80) rests rather on ten events of active preparation and ten measurements, the sum of which is 160 minutes. Any one measurement or measuring event is the measure of a single point-to-point distance of elapsed time—that is, a single start-stop of the stopwatch (see section 9.4.2.0). The matrix predicates to take 240 minutes and to take 160 minutes are themselves indifferent to the number of measurements, asserting only that the measure of the events referred to sum to n minutes. But what happens in (73) to make (80) true are ten events while being accompaniment and ten measurements, and no others
[DP D AdrP and AdrP]
729
will do. The ten events while being accompaniment are ten events ei, 1 ≤ i ≤ 10, that ∃x accompaniment(ei,x), for 1 ≤ i ≤ 10. These conditions on evaluation and measurement now point to what goes wrong in (76): (76) a. #An appetizer and dessert took 160 minutes of active preparation. #The appetizer and dessert took 160 minutes of active preparation. b. #An appetizer and dessert have taken 160 minutes of active preparation. #The appetizer and dessert have taken 160 minutes of active preparation. c. #An appetizer and dessert has taken 160 minutes of active preparation. #The appetizer and dessert has taken 160 minutes of active preparation. The only active preparation in (73) that measures 160 minutes are these same ten events of active preparation, now required according to (76) to be while being an appetizer and dessert (i.e., ∃x appetizer and dessert[ei,x], for 1 ≤ i ≤ 10). Yet only the first and last are each both appetizer and dessert. The second, for example, is appetizer but not appetizer and dessert. The distinction is vivid under the lens of an imagined verification procedure to measure the active preparation time of either accompaniment or appetizer and dessert. As the feast’s preparation is in progress or even in the course of scanning the timeline left to right, it goes: there is an event of actively preparing accompaniment and it measures 40 minutes, and now there is another event of actively preparing accompaniment and it measures 10 minutes for a running total of 50 minutes, and so on. As verification for (76), this procedure balks at the second event of active preparation. There is first an event of actively preparing appetizer and dessert for 40 minutes, but there is then no 10-minute event of actively preparing appetizer and dessert until the final 40 minutes of active preparation. This same procedure will, however, succeed as verification for the DPcoordination in (77): (77) a. An appetizer and a dessert took 160 minutes to prepare. An appetizer and a dessert took 160 minutes of active preparation. The appetizer and the dessert took 160 minutes to prepare. The appetizer and the dessert took 160 minutes of active preparation. b. An appetizer and a dessert have taken 160 minutes to prepare. An appetizer and a dessert have taken 160 minutes of active preparation. The appetizer and the dessert have taken 160 minutes to prepare. The appetizer and the dessert have taken 160 minutes of active preparation. It is the same ten events of active preparation and ten measures that sum to 160 minutes. Each of the ten events however occurs while being appetizer or while being dessert,24 which is all that adverbialization in (77) imposes on the collective events referred to, what appetizer underwent and what dessert underwent, and said to be 160 minutes long.
730
Chapter 14
Given that the verification of (76) balks at the second event (among others), the analysis of AdrP-coordination should accordingly deliver that ∃x accompaniment(e2,x) but ¬∃x appetizer and dessert[e2,x], falsifying the alleged equivalence in (84) of simple and coordinate AdrPs, an effect, it ought to be, of the tacit addressing occurring within the conjuncts in (82). There should be no equivalence between the logical form (86) for the simple AdrP accompaniment and the logical form (87) for the coordinate AdrPs appetizer and dessert: (84) F accompaniment[E0,X] ↔ appetizer and dessert[E0,X] (85) F (86) ↔ (87) (86) [∃ : VF[E0, ]](ADR[E0, , F ] ∃X accompaniment.SG[E0, X ])] 25 (87) W[E0,X][℩E1: ΦΨ]O[E0,E1] (Φ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X appetizer.SG[E1, X ])) and (Ψ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X dessert.SG[E1, X ]))] Indeed, for any events E0 now-en-scène[E0] and viewfinder for the above AdrPs, (86) merely requires finding accompaniment—appetizer or dessert—there, while (87) demands both appetizer and dessert there. Just as in the example of cycling by a stream and dry creek, any singular scene for an accompaniment and dessert must be a scene of both. If appetizer and dessert always have to be in any scene that is of an appetizer and dessert, then any scenes that (76) describes must compose a sequential scan of the events of (73), just as has been imagined for their measurement. Now, if the scenes of measurement, Eμ in (88), those in which the counter clicks, or as here, those for each of which the stopwatch starts and stops once, then a scene of that second event e2 in (73) of only appetizer is not among the scenes Eμ measured, as desired: (88) … [∃F: ∏] [∃X: ∃E0 (Δ an[E0] now-en-scène[E0] W[E0,X][℩E1: ΦΨ]O[E0,E1] (Φ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X appetizer.SG[E1, X ])) and (Ψ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X dessert.SG[E1, X ])∆ )] [℩E0 : Δ] [∃E2 : N[E0,E2]] … W[E2,X] O[E2,E3] ∃X Theme[E3,X] … [∃F: ∏](… Tense[Ei,F] … O[E3,E4] ∃Eμ measure[Eμ,E4,160] … active preparation[E4] …) In contrast to an accompaniment, measuring the active preparation time of an appetizer and dessert is according to (88) a sum of measurements at each of which there is appetizer and dessert, breaking the false equivalence between simple AdrP and coordinate AdrPs. If it is true that (88) is the logical form for (76), the divergence between coordinate AdrPs and their simple AdrP counterparts is explained. That it is so is still in need
[DP D AdrP and AdrP]
731
of some pleading. First, if the conjuncts contain tacit adr, our understanding when locative phrases are overt as in (89) does make it a safe assumption that the scenes and frames of reference viewfinding appetizer and dessert are the same for the scenes for measurement Eμ: (89) #An appetizer (keeping it) to the right and dessert (keeping it) to the left therein took 160 minutes to prepare. #An appetizer (keeping it) to the right and dessert (keeping it) to the left have therein taken 160 minutes to prepare. #An appetizer first and dessert second therein took 160 minutes to prepare. #An appetizer first and dessert second have therein taken 160 minutes to prepare. What a perverse thought it would be if the measurement were elsewhere. The speaker utters (76) or (89) with something in mind for a protocol of how to measure appetizer and dessert and something in mind for how the appetizer is always at the one location and dessert always at the other. If uttered against the background of events in (73), the speaker must have in mind and intend to refer to a sequential scan of those events (as remarked above). Yet the epistemic stance and perspective on these events thereby assumed has nothing to do with the measurement, for which we are asked to suppose that the speaker has something else in mind also part of her communicative intent if the hearer is to have grasped her expression of a determinate thought. Not likely. If the logical form for (76) arrays adr and measure relations as shown in (88), the above reassures that they all relate to the same location, the same scenes under the same frame of reference. An analog to therein in (89) is present in a fully expanded (88). The events measured of appetizer and dessert are those events, ⌜[℩E0 : Δ]⌝ in (88), that, under adverbialization, define the neighborhood for the events E3, ∃X Theme[E3,X], in which appetizer and dessert participate as Themes under active preparation. It is these events, as numerous26 and coincident with the events E0 of appetizer and dessert, to which plural number agreement refers (see chapter 2), ⌜[℩E3 : [℩E0 : Δ] Ω[E3]]⌝ in (90), and that verify the matrix predicate: (90) … [∃F: ∏] [∃X: ∃E0 (Δ an[E0] now-en-scène[E0] W[E0,X][℩E1: ΦΨ]O[E0,E1] (Φ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X appetizer.SG[E1, X ])) and (Ψ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X dessert.SG[E1, X ])∆ )] [℩E0 : Δ] (Ω [∃E2 : N[E0,E2]] … W[E2,X] O[E2,E3] ∃X Theme[E3,X] Ω) … [∃F: ∏] [℩E3 : [℩E0 : Δ] Ω[E3]](… Tense[Ei,F] … O[E3,E4]27 ∃ Eμ measure[Eμ,E4,160] … active preparation[E4] …)
732
Chapter 14
It suffices for the failed interpretation of (76) (repeated below) and its explanation that the antecedent AdrPs for the event pronoun translating number agreement are as in (90). It will then follow that in the context (70), the speaker has in mind a scan each scene of which has appetizer at one address and dessert at another, and the events said to take 160 minutes are those that the scanned scenes are scenes of, or rather, events of active preparation that coincide with the events the scenes are scenes of. Even if a perverse speaker has in mind to evoke this scan of (70) without intending that it also guide the measurement reported as 160 minutes, she must intend instead another protocol under some other frame of reference for the context (70). This provides its own measurements—that is, scenes of measurement Eμ corresponding to the very same events number agreement’s event pronoun refers to—which is still to say that the events measured are each of appetizer and dessert, as desired. Recall that (76) contrasts with (77) as well as with (80): (76) F An appetizer and dessert took 160 minutes of active preparation. F The appetizer and dessert have taken 160 minutes of active preparation. (77) An appetizer and a dessert took 160 minutes of active preparation. The appetizer and the dessert have taken 160 minutes of active preparation. (80) An accompaniment took 160 minutes of active preparation. The accompaniment has taken 160 minutes of active preparation. The ten events of active preparation divide between appetizer and dessert, which is all (77) demands, and each is also active preparation of accompaniment, satsifying (80). Tacit adrs, which the speaker is also free to consider with DP-coordination, share the fate, I suppose, of their overt counterparts: (91) An appetizer (keeping its preparation) on the left and a dessert (keeping its preparation) on the right have taken 160 minutes of active preparation. (92) Babette’s heralded Blini Demidoff au Caviar on the left and her equally celebrated Baba au Rhum avec les Figues on the right took 160 minutes of active preparation. One interpretation, favored in (92), is incoherent in the context (70). Whether scanning a sequence of smaller scenes or viewing a single scene of all (70), it requires that all the preparation spanning 240 minutes of the appetizer whole and entire appear on the left of every scene entertained, and similarly, all preparation of the dessert whole and entire appear on the right of every scene, which they cannot do intermingled as they are. There is, however, another interpretation that while scanning the feast keeping appetizer on the left and dessert on the right, or thinking of appetizer events always as first, preceding dessert events as second, the appetizer
[DP D AdrP and AdrP]
733
and the dessert have taken 160 minutes of active preparation between them. With two DPs there is independent quantification over viewfinders and their scenes (cf. (83)): (77) [DPAn.sg appetizer.sg]…W…θ… and [DPa.sg dessert.sg]…W…θ… Tense … (93) … [∃F0: ∏] [∃X: ∃E0 (Δan[E0] now-en-scène[E0] [∃ : VF[E0, ]](ADR[E0, , F0 ] ∃X appetizer.SG[E0, X ]))] [℩E0 : Δ] [∃E3 : N[E,E3]] … W[E3,X] O[E3,E4] ∃X θ[E4,X] and [∃F1: ∏] [∃X: ∃E0 (Δan(E0) ∃Eμ ∃n measure[Eμ,E0,n] [∃ : VF[E0, ]](ADR[E0, , F1 ] ∃X dessert.SG[E0, X ]))] [℩E0 : Δ] [∃E3 : N[E,E3]] … W[E3,X] O[E3,E4] ∃X θ[E4,X] … [∃F: ∏](… Tense[Ei,F] … ) They are all of (70), but the scan for the first conjunct is scenes with appetizer on their left and nothing on their right, and similarly the scan for the second is scenes of dessert on the right with nothing on the left. The ten events of active participation are divided between the scans, and the sum of their measures is 160 minutes. The interpretations obtained are equivalent with or without locative predicates. As remarked when first introduced, the contrast between (76) and (80) replicates that between (35) and (37) (repeated below): (35) T Most protium(, deuterium,) and tritium in seawater is radioactive. (37) F Most hydrogen in seawater is radioactive. What has been explained is that coordinated AdrPs diverge in meaning from their simple AdrP counterparts if the conjuncts contain unspoken locatives. Now a likely lacuna looms. If unspoken locatives are omitted freely, (35) and (37) should share in (37)’s meaning, and (76) and (80) should share in (80)’s. If not impossible (see note 5), (35) is rather reluctant to express the equivalent of (37). Likewise, it seems harder to understand (76) than (77) as an equivalent for (80). As just remarked, the latter two end up equivalent whether or not locative predicates are inserted into (77). What then explains the hardship on (35) and (76) in conveying those interpretations that they should express naturally absent tacit locative predicates? An answer draws an analogy between DPs and clauses. If, as a point of grammar, clauses always contain some expression of reference to time, context, or, as these notions are enlarged, to spatiotemporal frames of reference, suppose DPs do too. An immediate consequence is that the logical form that would simply omit the locative predicates adr of (90) or (93) is ungrammatical, failing to relate what DP describes to F. After all, what could be the point of interpreting every sentence under a frame of reference except to locate what is talked about there too? It is also to be a point of grammar that there is not room enough in a DP with coordinate
734
Chapter 14
AdrPs for a relation to frames of reference ‘adr[E0, ,F]’ to occur outside the conjuncts and thus these relations occur within the NPs: (94) *… [∃F: ∏] [∃X: ∃E0(an[E0] W[E0,X] [∃ : VF[E0, ]]ADR[E0, , F ] [℩E1: ΦΨ]O[E0,E1] (Φ∃E1 ∃X appetizer.sg[E1,X]) and (Ψ∃E1 ∃X dessert.sg[E1,X]))] Syntax requires that appropriate choices for adr in (90) and (93) locate appetizer and dessert in the given frames of reference. A further stipulation, which I had thought to derive from other considerations, will however remain as such: the choices for multiple adr within DP must be different—appetizer somewhere and dessert somewhere else.28 14.1.1
In the neighborhood of coordinated AdrPs
In a singular plural, a bloodred seven sunrises, an event or state counts as one, and many participate in it. What is in the neighborhood of one event may differ from the neighborhood of several coincident events. (136) Seven bloodred sunrises preceded/announced/augured seven days’ bloody battle. (138) A bloodred seven sunrises preceded/announced/augured seven days’ bloody battle. In this respect the adverbials derived by adverbialization from a bloodred seven sunrises (138) and seven bloodred sunrises (136) denote different events (section 12.2), the one and the many, with different neighborhoods, although the nominals themselves denote the same sunrises. In fact, anything that might be said about the events—their number, the protocol for counting them, the frames of reference or scenes of their observation—characterizes and is preserved by the derived adverbials referring to them, of which section 14.1.0 has provided elaborate examples. Simple nominals and those derived by coordination, such as fielders and infielders and outfielders, coextensive as to who participates, may also vary in the events described in context. What is to count as one event or state when considering any and all in which one or more fielders participate may not be the same as when considering only those in each of which are some infielders and outfielders or only those in which there is an infielder and outfielder. These nominals, adverbialized, have different implications for the interpretation of the sentences in which they occur. The fielders are always the infielders and outfielders, and the events are the same too, in that to field is always to infield or to outfield; yet as in section 14.1.0, simple AdrP and coordinate AdrPs are not equivalent. The coordinate AdrPs seg-
[DP D AdrP and AdrP]
735
regate infielder and outfielder, and the coordinate AdrPs describe only scenes of them both, unlike the simple AdrP: (95) F fielders[E0,X] ↔ infielders and outfielders[E0,X] (96) F (97) ↔ (98) (97) [∃ : VF[E0, ]](ADR[E0, , F ] ∃X fielders[E0, X ])] 29 (98) W[E0,X][℩E1: ΦΨ]O[E0,E1] (Φ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X infielders[E1, X ])) and (Ψ ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X outfielders[E1, X ]))] An antidote to antisemistributivity (section 13.2) in (345) is to provide as in (346) an antecedent description of a collective state or event as in (346): (345) *Many musicians are an ensemble. *Most musicians are an ensemble. *More musicians are an ensemble than are not. *More musicians than not are an ensemble. *Many musicians when (being) musicians are an ensemble. (346) Many musicians who are together are an ensemble. Most musicians who are together are an ensemble. More musicians who are together are an ensemble than are not. More musicians than not who are together are an ensemble. Many musicians who are together when (being musicians) together are an ensemble. Scenes containing both instrumentalists and vocalists are as potent an antidote:30 (99) Many instrumentalists and vocalists are an ensemble. Most instrumentalists and vocalists are an ensemble. More instrumentalists and vocalists are an ensemble than are not. More instrumentalists and vocalists than not are an ensemble. Many instrumentalists and vocalists when (being) are an ensemble are an ensemble. Any scan that answers to the description in (99) is parsed into scenes of both instrumentalist and vocalist. The resulting contrast between (345) and (99) is just that seen earlier between (35) and (37): (35) T Most protium(, deuterium,) and tritium in seawater is radioactive. (37) F Most hydrogen in seawater is radioactive.
736
Chapter 14
The predicate radioactive is hardly antisemidistributive. Yet, because the scenes scanned according to (35) are sampling amounts of seawater with all three isotopes including rare but radioactive tritium, the predicate applies as if it were a stative, collective predicate applying to amounts of all three and not also to any lesser amount at all of hydrogen.31 The disjunctive interpretation of coordinate predicate nominals, infielders and outfielders or Hungarian and Romanian below, is antisemidistributive, like be an ensemble, in resisting distributive plurals in both matrix predicates (100) and relative clauses (102). In contrast to the plural definite and indefinite descriptions in (101) and (104), which easily divide their reference in a disjunctive interpretation between Hungarians and Romanians, the distributive quantification so-called (see section 13.1) in (100) and (102) is about dual nationals who in their person embody ethnic division (cf. be a cluster in (87)–(90) in section 13.0).32 (100) a. Most Transylvanians are Hungarian and Romanian. b. Many Transylvanians are Hungarian and Romanian. c. Few Transylvanians are Hungarian and Romanian. (101) a. The Transylvanians are Hungarian and Romanian. b. Some Transylvanians are Hungarian and Romanian. (102) a. Most Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. b. Many Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. c. Few Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. (103) a. Most Transylvanians such that they are Hungarian and Romanian are rootless cosmopolitans. b. Many Transylvanians such that they are Hungarian and Romanian are rootless cosmopolitans. c. Few Transylvanians such that they are Hungarian and Romanian are rootless cosmopolitans. (104) a. The Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. b. Some Transylvanians who are Hungarian and Romanian are rootless cosmopolitans.33 (105) a. The Transylvanians such that they are Hungarian and Romanian are rootless cosmopolitans. b. Some Transylvanians such that they are Hungarian and Romanian are rootless cosmopolitans.
[DP D AdrP and AdrP]
737
Disjunctively interpreted, be Hungarian(s) and Romanian(s) thus joins be an ensemble, be a cluster, be a perfect circle, be a galaxy and be twelve, be 1000 honeymooners, outnumber my fingers, etc., among the antisemidistributive predicates, without a meaning that is plainly either geometric or arithmetic. Yet, recall from sections 13.1–13.2 that at the center of antisemidistributivity is a lexical conflict between the null determiner (co-occurring with predicative many) that scatters passim what the nominal denotes among frames of reference that are not path-integrated to a single one and a predicate that demands a single frame of reference for what it denotes. For a geometric predicate, to be a cluster, perfect circle, or galaxy is for their points to be so on the same graph paper (i.e., under a single frame of reference and coordinate system). An arithmetic predicate, it was concluded, in fitting what is counted under a single reticule, is also a measure under a single frame of reference intolerant of scatter among several. Although be Hungarian(s) and Romanian(s) implies no measurement, the conjoined AdrPs do imply that to be Hungarians and Romanians is to be so at addresses in the same scene in the same frame of reference—demanding a single frame of reference for them and so conferring on the conjoined AdrPs their standing as an antisemidistributive predicate.34 Recall from section 13.2 that antisemidistributivity is overcome if the AdrP itself denotes discrete, countable events each framing a scene in a single frame of reference large enough for the collective event or state the antisemidistributive predicate denotes. But also recall (section 13.1.0) that for any count, determinate in what it counts and counting none of it twice, to count across plural frames of reference is to count short, like event counting so-called. As in event counting, the events counted cannot be the lifetimes that object counting counts or anything else that endures beyond the scattered frames of reference that are unintegrated into something less transient. DPs with null determiners, many NP and others, were therefore found to be subject to sequencing of events (Doetjes and Honcoop 1997) and the anticonvergence condition, just like event counting. Thus, the minimal pair in (252) and (106) contrasts the antisemidistributivity of (252) with its antidote in (106) in which each landing is of a squadron delivering: (252) #Many aircraft that landed at Tempelhof in 1948 were a squadron under the command of US Air Forces in Europe. (106) Many aircraft that landed at Tempelhof in 1948 after dark were a squadron/ wing/formation/detail delivering military hardware disguised as medical supplies. Turning to disjunctively interpreted AdrP-coordinations, (107) correlatively implies landing and taxiing down the runway in pairs and (108), in various or unknown numbers:
738
Chapter 14
(107) Many Berlin airlift aircraft were a cargo transport and fighter escort. Many aircraft that landed at Tempelhof in 1948 were a cargo transport and fighter escort. (108) Many Berlin airlift aircraft were cargo transports and fighter escorts. Many aircraft that landed at Tempelhof in 1948 were cargo transports and fighter escorts. Deprived however of an antecedent description of discrete, countable events—or, better said, where the only discrete, countable events coincide with the maximal states of being an individual aircraft—antisemidistributivity forces the interpretation asserting that each aircraft is both transport and escort: (109) #Many aircraft from the Berlin airlift were a cargo transport and fighter escort. (110) #Many aircraft from the Berlin airlift were cargo transports and fighter escorts. As expected now, the sequence-of-events effect (Doetjes and Honcoop 1997; chapter 11) that rules out the present Tense with the antisemidistributive predicate be a squadron in (112) (see (258)–(261)) equally rules out the present Tense with the antisemidistributive coordinate AdrPs in (113): (111) Many Berlin airlift aircraft are C-47s (that are) now mothballed in the Mojave Desert. Many aircraft that landed at Tempelhof in 1948 are C-47s (that are) now mothballed in the Mojave Desert. (112) #Many Berlin airlift aircraft are a squadron (that is) now mothballed in the Mojave Desert. Many aircraft that landed at Tempelhof in 1948 are a squadron (that is) now mothballed in the Mojave Desert. (113) #Many Berlin airlift aircraft are C-47s and F-51s (that are) now mothballed in the Mojave Desert. #Many aircraft that landed at Tempelhof in 1948 are C-47s and F-51s (that are) now mothballed in the Mojave Desert. From within a DP, an AdrP-coordination such as instrumentalists and vocalists in (99) is itself antidote to the antisemidistributivity of a matrix predicate in providing an antecedent description of collective events, scenes of both instrumentalist and vocalist affording subsequent reference to such scenes—instrumentalists at one address in a scene and vocalists at another who in that scene … therein did … . As shown below, it is equally effective against the antisemidistributivity of disjunctively
[DP D AdrP and AdrP]
739
interpreted AdrP-coordination in both relative and matrix clauses as against other antisemidistributive predicates as in (99). It is striking that AdrP-coordination heading a DP licenses the disjunctive interpretation of coordinated AdrPs within a relative clause in the same DP. But in this respect, it simply behaves like any other simple AdrP that provides a description of collective events to overcome antisemidistributivity. In contrast to (100) and (102), (114) and (115) need not be about people with dual ethnicity or faith. They rather comment on the tendency when coming upon scenes with Transylvanians and Walachians that there will be different sorts among them: (100) a. Most Transylvanians are Hungarian and Romanian. b. Many Transylvanians are Hungarian and Romanian. c. Few Transylvanians are Hungarian and Romanian. (114) a. Most Transylvanians and Walachians are Hungarian and Romanian. Most Transylvanians and Walachians are Christian and Jewish. b. Many Transylvanians and Walachians are Hungarian and Romanian. Many Transylvanians and Walachians are Christian and Jewish. c. Few Transylvanians and Walachians are Hungarian and Romanian. Few Transylvanians and Walachians are Christian and Jewish. (102) a. Most Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. b. Many Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. c. Few Transylvanians who are Hungarian and Romanian are rootless cosmopolitans. (115) a. Most Transylvanians and Walachians who are Hungarian and Romanian are rootless cosmopolitans. b. Many Transylvanians and Walachians who are Hungarian and Romanian are rootless cosmopolitans. c. Few Transylvanians and Walachians who are Hungarian and Romanian are rootless cosmopolitans. The effect of coordinate AdrPs on the interpretation of another one inside a relative clause is concisely illustrated in the following contrast: (116) #Most people who are Rh-positive and Rh-negative should not have children together. (117) Most men and women who are Rh-positive and Rh-negative should not have children together.
740
Chapter 14
Sentence (116) is about the nonexistent people with a contradictory Rh factor, while (117) is about a risk that presents itself to men and women with different Rh factors.35 The effects observed are not special to AdrP-coordination, recurring with simple AdrPs that manage by other means to describe collective events, which result in the contrasts between (118)–(120) and (121)–(123). Given that one religion is enough of a burden as it is, (118)–(120) make the true observation that rarely does one suffer two, whereas (121)–(123) comment that Christians and Jews are largely endogamous with respect to the other: (118) Few people are Christian and Jewish. (119) Few uncles are Christian and Jewish. (120) Few nephews are Christian and Jewish. (121) Few people that are related (to each other) are Christian and Jewish. (122) Few uncles to the same nephew are Christian and Jewish. (123) Few cousins are Christian and Jewish. Within relative clauses, (124)–(126) seem only to mean that the person is rare who is paying the price of two religions and has shopped around for a third.36 In contrast, (127)–(128) are readily understood as a remark that few people in interfaith families have reached out for even greater ecumenicism: (124) Few people who are Christian and Jewish have visited Mecca. (125) Few uncles who are Christian and Jewish have visited Mecca. (126) Few nephews who are Christian and Jewish have visited Mecca. (127) Few people related to each other who are Christian and Jewish have visited Mecca. (128) Few uncles to the same nephew who are Christian and Jewish have visited Mecca. (129) Few cousins who are Christian and Jewish have visited Mecca. 14.1.2
Nondenoting AdrPs and existential commitments within coordinated AdrPs
Coordinated AdrPs are nondenoting in the contexts where simple AdrPs are nondenoting, and otherwise entail existential commitments in the contexts where simple AdrPs do too. It needs to be shown that this follows just from canonical logical form, (5), the meaning of the AdrPs coordinated therein, and anything that needs to be said about the contexts hosting coordinated AdrPs. (130) F Some English lords and French princes sit on an EU commission. If some English lords and French princes implies the existence of both English lords and French princes, with the nonexistence of the latter thus falsifying (130), it should not fall to the meaning of and to imply the existence of what its conjuncts denote—
[DP D AdrP and AdrP]
741
certainly not, if and is nothing more than the sentential connective. But also on empirical grounds, it should not, as detailed below, since the same conjoined AdrPs (so it appears), embedded elsewhere, as in the English lords and French princes, may occur without existential commitment to either English lords or French princes.37 The meaning of and, always the sentential connective, may be so defended, but then it remains for a syntax and semantics for AdrP-coordination to explain the distribution of existential commitments in tokens of this construction. For nondenoting simple plurals, I follow my 2006 discussion. The logical syntax of (131)–(132) does not diverge from that of (133)–(134) under any plausible parse of the natural language. Insofar as sentences such as (131)–(132) are true, (131), given the definite description ‘[the ξ : non-self-identical custards[ξ]]’, entails (135). (131) The non-self-identical custards are zero in number. The moons of Venus are zero in number. (132) The zero or more solutions to this equation are all unidentified prime numbers. (133) The custards are twelve in number. The moons of Jupiter are more than sixty in number. (134) The three solutions to this equation are all unidentified prime numbers. (135) [∃ξ : non-self-identical[ξ] & custards[ξ]] zero[ξ] Given the truth of (131)–(132), nothing in the meaning of the article the or the plural morpheme proper entails a nonzero measure of what the description refers to, and neither does the existential quantification in (135) or the evaluation of the variable ξ of plural reference. Plural expressions refer fluently to the many and the none, including plural expressions with coordinate AdrPs: (136) The non-self-identical custards and blancmanges are zero in number. The moons of Venus and satellites of Mercury are zero in number. In explicating the truth of (131), (135), and (136), we need not however entertain the treacherous thought that the primitive lexical concepts “non-self-identical” and “custard” denote the none. Rather, adapting to Eventish the proposal in Schein 2006, some primitive relations enter logical form modified by the operator defined in (138). Perhaps a lexical root never enters logical form bare and in becoming a noun, adjective, or thematic relation proper is modified by the nusquam operator (138). (137) θ[E,X] ↔df ∃xXx & ∃eEe & ∀x(Xx ↔ ∃e(Ee & θ(e,x))) (Pietroski 2003, 282)38 (138) ‖θ[E,X] ↔df (∃xXx ∨ ∃eEe) → θ[E,X]39 (139) ADR[E, , F ] ↔ df ∃eEe → ADR[E, , F ]
742
Chapter 14
The (abbreviated) logical forms for the non-self-identical custards and the non-selfidentical custards and blancmanges, including AdrPs, are (140) [The X: ∃E (‖non-self-identical[E,X] [∃ : ∏] (ADR[E, , F ] custards[E, X ]))] (141) [The X: ∃E0 (‖non-self-identical[E0,X] ‖W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0, ]] ( ADR[E1, , F ] ∃X custards[E1, X ])) and ∃E1 (Ψ [∃ : VF[E0, ]] ( ADR[E1, , F ] ∃X blancmanges[E1, X ]))] If, for the sake of (131), existential commitment has been removed from the, from the plural morpheme, and from the relations via the nusquam operator in (138), neither the event pronouns in (141) nor and impose one. Already in section 2.0, it was remarked that event pronouns, glossed “whatever events if any that … ,” must not imply the existence of events where none are implied: (142) No Columbia student and no Harvard student collaborated to overthrow the government. And, of course, and is merely the sentential connective. Given the truth of (136), and must not imply the existence of custards or blancmanges, which it does not in (141), with its ordinary meaning. A different context and a different phrase, twelve custards and blancmanges, commits itself to the existence of both custards and blancmanges by other means (as offered below), since and, given (136), contributes no existential commitment not already inherent in the phrases it conjoins. Indeed, various contexts enjoin and from any such commitment conjoining AdrPs. The conditionals in (143) are falsified despite an absence of pedestrians if the motorists and cyclists that are on the scene do not behave as described, colliding or directing traffic, and thus and cannot imply the existence of pedestrians or of motorists or cyclists. (143) a. If any traffic signals fail, the nearby motorists, cyclists, and pedestrians will certainly all collide. b. If any traffic signals fail, the nearby motorists, cyclists, and pedestrians start to direct traffic. In contrast, the conditionals in (144) imply that every traffic signal is met with a response from some motorists, some cyclists, and some pedestrians, no doubt an effect of some difference between definite and indefinite descriptions but not of and, which is presumed to mean the same throughout. (144) a. If any traffic signals fail, some nearby motorists, cyclists, and pedestrians start to direct traffic. b. If any traffic signals fail, if there are some/ ∅ /any nearby motorists, cyclists, and pedestrians, they will certainly all collide.
[DP D AdrP and AdrP]
743
The peers of France have followed their king into oblivion, and so below English lords and French peers is coextensive with English lords rather than denoting no royalty absent the French. The English lords are 724 and the Irish senators 60, so that (145) T More than 700 English lords and French peers sit together or on their own in a European parliament. (146) F Fewer than 700 English lords and French peers sit together or on their own in a European parliament. (147) T More than 760 English lords, Irish senators, and French peers sit together or on their own in a European parliament. (148) F Fewer than 760 English lords, Irish senators, and French peers sit together or on their own in a European parliament.40 Again, the defunct French peerage does not render (149) or (150) vacuously true, which are falsified by the lords’ annual commemoration in Parliament of their victory over the French: (149) F No English lords and French peers sit in session on St. Crispin’s Day without reviling the others’ perfidy at Agincourt in 1415. (150) F No English lords and French peers sit in a European parliament without the other.41 Taking and to be the connective surrounded by the syntax in (141) derives the disjunctive interpretation without existential commitment, as required here. Some further remarks and stipulation are necessary if this escape from existential commitment is to be consistent with the observation and analysis from section 14.1.0 that in most protium, deuterium, and tritium in seawater, the measurement events scan seawater always sampling all three hydrogen isotopes—that is, if all three exist, all three are always present in what is measured. The application of the nusquam operator to adr relations has different implications for them (see (139)). A nusquam’d non-adr, θ relation may relate nothing to nothing but neither something to nothing nor nothing to something: (151) Second-Order Existence Tripwire for non-ADR relations θ ∀E∀X((‖θ[E,X] & (∃eEe ∨ ∃xXx)) → (∃eEe & ∃xXx)) An underlying non-adr, θ relation itself relates some events only to some things: (θ[E,X] & ∃eEe) → ∃xXx. But the nusquam’d adr relations may relate scenes—that is, the abstracta of addresses and frames of reference—to nothing. For example,
744
Chapter 14
‘‖left[E, ,F]’42 may leave addresses in frames of reference F vacant if there are no events E. Nusquam’d adr relations imply only that (152) Second-Order Existence Tripwire for ADR ∀E∀∀F (( ADR[E, , F ] & ∃eEe) → (∃aa & ∃fFf )) As observed above, the nonexistent French peers do not make most English lords and French peers vacuous, and this is so despite the tacit adrs and the address the first one assigns the English lords: (153) [∃X: ∃E0 (∃Eμ ∃n(many(n) & count[Eμ,E0,n]) ‖W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0, ]] ( ADR[E1, , F ] ∃X custards[E1, X ])) and ∃E1 (Ψ [∃ : VF[E0, ]] ( ADR[E1, , F ] ∃X blancmanges[E1, X ])))] It must not falsify the scenes’ description that there are no French peers in their assigned location. As much is provided for in the nusquam’d adr (139). If existential commitment has been purged from the, from the plural morpheme, and from lexical relations subject to the nusquam operator (138)–(139) and neither event pronouns nor and impose it, it remains to be explained how twelve custards and blancmanges acquires a commitment to both custards and blancmanges when the (non-self-identical) custards and blancmanges is committed to neither (see (141)) and how some/ ∅ /any nearby motorists, cyclists, and pedestrians implies witnesses for each conjunct in (144) while the nearby motorists, cyclists, and pedestrians in (143) does not. Something about an indefinite article or determiner seems to affect all the conjuncts it is in construction with.43 A syntactic distinction to be proposed between definite and indefinite descriptions correlates with an injunction on indefinites proscribing the irrealis:44 (131) The moons of Venus are zero in number. The moons of some planet are zero in number. (132) The zero or more solutions to this equation are all unidentified prime numbers. (154) #Some moons of a planet are zero in number. (155) #Some zero or more solutions to this equation are all unidentified prime numbers. (156) Most French peers that there are if any have been dispossessed of their titles. (157) #Many French peers that there are if any have been dispossessed of their titles. What is then stipulated is that the complement AdrP in an indefinite description is a realis clause containing the morphology expressing as much. The DP some French peers is rather like a mini- there-construction: [DPthere [AdrP-are French peers]]. As
[DP D AdrP and AdrP]
745
with their tensed-clause counterparts, both conjuncts in a coordination satisfy the syntactic conditions on complements, so that some English lords and French peers is, in effect, [DPthere [AdrP-are English lords] and [AdrP-are French peers]]. Omitting the realis morphology from the AdrPs, the French peers and the English lords and French peers are spared any existential commitments; it is the presence of this morphology within the coordinated AdrPs that extends the existential commitment in indefinite descriptions to both conjuncts. A realis relation is defined in (158), and the logical forms for the minimal pair most English lords and French peers and many English lords and French peers are based on (153) and (159) respectively, differing in the occurrence of the realis relation:45 (158) Actual[E, , F ] ↔ df ∃eEe & At[, E, t(E), F ] (153) [∃X: ∃E0 (∃Eμ ∃n(many(n) & count[Eμ,E0,n]) ‖W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0, ]] ( ADR[E1, , F ] ∃X English lords[E1, X ])) and ∃E1 (Ψ [∃ : VF[E0, ]] ( ADR[E1, , F ] ∃X French peers[E1, X ])))] (159) [∃X: ∃E0 (∃Eμ ∃n(many(n) & count[Eμ,E0,n]) ‖W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0 , ]](Actual[E1, , F ] ADR[E1, , F ] ∃X English lords[E1, X ])) and ∃E1 (Ψ [∃ : VF[E0 , ]] (Actual[E1, , F ] ADR[E1, , F ] ∃X French pee rs[E1, X ])))] Just as the complementizer that demands from its complement finite tense morphology, which must be found in every conjunct if the complement is a conjunction of clauses, some and many, in contrast to the and most, demand realis morphology from every AdrP in an AdrP-coordination, from which the existential commitment of both AdrPs follows. The distribution of ‘Actual[E, ,F]’ throughout the grammar is an empirical question, and something in the nature of Eventish invites more of it. All of the underlying event quantification (i.e., the existential quantification and the definite description glossing the subatomic pronouns) is second-order and does not imply the existence of first-order objects. This is as it must be if the glue holding every sentence together is not to imply the existence of events where none are implied: (161) No Columbia student and no Harvard student collaborated to overthrow the government. Yet of course, once in awhile, one means that something happened, for which at least one existential event quantifier should commit to an event— ⌜[∃E: Actual[E, ,F] Φ]⌝ rather than simple ⌜[∃E: Φ]⌝. Some of the event quantification in a sentence ought to be punctuated with ‘Actual[E, ,F]’. Discovery of its distribution can at least begin with a slightly more extensive survey of the existential commitments emanating from coordinate structures, starting with definite
746
Chapter 14
descriptions. Recall that the nearby motorists, cyclists, and pedestrians in (143) does not imply the existence of all three. In this respect, AdrP-coordination under definite description differs from the counterpart DP-coordination of definite descriptions, as the following example takes care to illustrate. Only definite descriptions that are complete in themselves serve the purpose. Incomplete descriptions, in fixing reference, will often resort to means, like demonstrations, that carry their own existential commitments. The self-contained descriptions in (161) and (162) reveal the targeted contrast. (161) a. The British and Canadian veterans of Ypres erected that monument. b. The British veterans and Canadian veterans of Ypres erected that monument. c. The British veterans of Ypres and Canadian veterans of Ypres erected that monument. (162) The British veterans of Ypres and the Canadian veterans of Ypres erected that monument. Suppose that the speaker observes a photograph of the monument and spots the emblem of the Commonwealth forces, which she knows to have consisted only of British and Canadian troops during the relevant phase of the war. Also, it is known that it was left to the veterans of any particular battle to commemorate it as they saw fit. Such a speaker may felicitously utter any of (161), and she will not have spoken falsely if it turns out that there were no Canadians at Ypres, provided that its British veterans put up the monument. The speaker can use (161) to attribute the monument to whatever British and Canadian veterans of Ypres there were. In contrast, the conjoined DPs in (162) cannot be used to refer to whatever British and Canadian veterans there were at Ypres. Not knowing that both British and Canadian forces fought at Ypres, the speaker does not know enough to warrant (162), and in the circumstances imagined, she would have spoken falsely. Recall that existential commitment has been purged from the definite description operator and much else for the benefit of (131): (131) The moons of Venus are zero in number. (163) The Commonwealth veterans of Ypres erected that monument. The implied existence of Commonwealth veterans in the simple case of (163) instead has alternative sources. In this context, holding a photo of the monument, it is a safe assumption that the speaker intends existing, actual events for those to be anchored by Tense and described by the verb erect. That is, the event quantification associated with Tense, the highest in the sentence, does in this context take the form ⌜[∃E: Actual[E, ,F] Φ]⌝. Then, although the subject itself refers noncommittally to whatever Commonwealth veterans there were if any at Ypres, it is inferred that
[DP D AdrP and AdrP]
747
there were: they are the agents of an actual event, and thematic relations do not relate something to nothing. By the same reasoning, the event quantification in (131) must not imply the existence of an actual event or state of being zero in number, lest it also imply the existence of moons of Venus. Thus, whether or not ‘Actual[E, ,F]’ is tokened in the restriction to the event quantifier is a formal ambiguity separating (131) and (163). It echoes the earlier discussion in section 10.3 distinguishing independent Tense, (258)–(259) and (163), where frame of reference, tense, and the event referred to are located independently of the events described by the adverbialization of the subject, and dependent Tense, (164) and (131), where they are located with respect to them: (258) In June, Venus is Hesperus. (259) #In June, Hesperus is Venus. (164) Hesperus is Venus. It could be said that independent temporal reference just is quantification with existential commitment over actual events in the context.46 Turning to the coordinations in (161) and (162), as in (163) an actual event is referred to, implying that at least some veterans antecedently described erected the monument. This is sufficient for (161), which need not include both British and Canadians among them, and nothing further needs to be said about its semantics. As remarked above, the DP-coordination in (162) implies the existence and contribution to the monument of both British and Canadian veterans. Whatever is said to derive it should not encroach on (165), freed of existential commitments: (165) The British veterans of the Trojan War and the Canadian veterans of the Trojan War are zero in number. The DP-coordination is a coordination of clauses each containing the adverbialization of the DP and a thematic relation. Both (162) and (165) share (166), except for the restriction ‘Actual[E, ,F]’ reserved for (162): (166) [The X : ∃E0 (ΦBritish[E0,X]) …] [℩E0 : Φ] [∃E : N[E0,E]] … W[E,X] O[E,E1] ∃X θ[E1,X] and [The X : ∃E0 (ΨCanadians[E0,X]) …] [℩E0 : Ψ] [∃E : N[E0,E]] … W[E,X] O[E,E1] ∃X θ[E1,X] ∃F [℩E0 : ΦΨ] [∃E: Actual[E, ,F]] (… Tense[E] …) [([DPThe British…]…W…θ…) and ([DPThe Canadian…]…W…θ)… Tense…] With just the one token of the realis morpheme, one may infer, as above, from the existence of an event in which the monument is erected the existence of its agents and what they did, to which the event pronoun ‘[℩E0 : ΦΨ]’ refers. This collective reference to whatever the British veterans did and the Canadian veterans did is,
748
Chapter 14
however, indifferent to whether both participated. To guarantee that both participate, as (162) requires, requires that the realis morpheme occur somewhere within both conjuncts: (167) ∃F [The X : ∃E0 (ΦBritish[E0,X]) …] [℩E0 : Φ] [∃E : N[E0,E]] … W[E,X] O[E,E1] Actual[E1, ,F] ∃X θ[E1,X] and ∃F [The X : ∃E0 (ΨCanadians[E0,X]) …] [℩E0 : Ψ] [∃E : N[E0,E]] … W[E,X] O[E,E1] Actual[E1, ,F] ∃X θ[E1,X] ∃F [℩E0 : ΦΨ] [∃E: Actual[E, ,F]] (… Tense[E,S] …) [([DPThe British…]…W…θ…) and ([DPThe Canadian…]…W…θ)…Tense…] This concedes that the formal distinction between (162) and (165) involves even more syntax. Perhaps it could be said that satellite thematic relations agree in their realis morphology—realis concord—with a realis tense or event predicate, the verb in (162). As a result of realis concord, the realis morpheme proliferates into the conjuncts of (167). Realis concord with a lexical predicate of events finds some oblique support in the contrastive behavior of DP-coordinates in partitive constructions and as noun complements. Despite the nonexistence of French peers, the many English lords sitting in Parliament and reviling the French on St. Crispin’s Day suffice to falsify all the coordinations in (168)–(171), as they did in their nonpartitive counterparts (146) and (149). (146) F Fewer than 700 English lords and French peers sit together or on their own in a European parliament. (168) F Fewer than 700 of the English lords and French peers sit together or on their own in a European parliament. (169) F Fewer than 700 of the English lords and the French peers sit together or on their own in a European parliament. (149) F No English lords and French peers sit in session on St. Crispin’s Day without reviling the others’ perfidy at Agincourt in 1415. (170) F None of the English lords and French peers sit in session on St. Crispin’s Day without reviling the others’ perfidy at Agincourt in 1415. (171) F None of the English lords and the French peers sit in session on St. Crispin’s Day without reviling the others’ perfidy at Agincourt in 1415.
[DP D AdrP and AdrP]
749
Absent the intrusion of realis morphemes into the conjuncts, this is as expected. Paraphrasing the DP-coordination (169): (172) The English lords that there are participate (W) in whatever they participate in and the French peers that there are participate in whatever they participate in, and fewer than 700 of whatever participants there are in whatever if anything they all variously participate in sit together or on their own in a European parliament. Paraphrase and logical form capture what the English lords do alone, and their being more than 700 in the House of Lords falsifies (169). In contrast to a DP-coordination in a partitive construction ((169) and (171)), in a noun-complement construction the nonexistence of French princes comes to render the sentences vacuously true: (173) Fewer than 700 peers of the English lords and the French princes sit together or on their own in a European parliament. (174) No peers of the English lords and the French princes sit in session on St. Crispin’s Day without reviling the others’ perfidy at Agincourt in 1415. Since no one is a peer to both an English lord and a French prince—because the latter do not exist—(173) and (174) are vacuously true despite the activities of the English lords and their peers. In a further contrast, sentences with AdrP-coordination within the noun complement remain false if the peers of the English lords sit in Parliament on St. Crispin’s Day reviling the French: (175) Fewer than 700 peers of the English lords and French princes sit together or on their own in a European parliament. (176) No peers of the English lords and French princes sit in session on St. Crispin’s Day without reviling the others’ perfidy at Agincourt in 1415. Again, nothing further needs saying about the semantics of AdrP-coordination under definite description; the English lords and French princes refers to whatever English lords or French princes there are. Of interest is the contrast between the DP-coordination in a partitive construction (169) and (171) and in a noun complement (173) and (174), which are distinguished only by the presence of the lexical head peers. The noun peers occurs with satellite thematic relations relating the English lords and the French princes to the events or states of being a peer. If these thematic relations are subject to realis concord, ‘Actual[E, ,F]’ occurs in both conjuncts when peers itself is for some reason realis. The structure of the partitive construction, with defective of for content, is constructed entirely from the participation relation ‘W’ without thematic relations proper. The canonical configuration for realis concord as supposed here is missing from the partitive construction absent a
750
Chapter 14
lexical head and satellite thematic relations, and thus the partitive construction escapes from dual existential commitments. Much of this is speculative and yet it has gained some purchase on the summary contrast between (177)–(178) and (179)– (180), varying the content of the conjuncts coordinated: (177) F The English lords and the French princes held a reunion on St. Crispin’s Day 2009. (178) F Many peers of the English lords and the French princes held a reunion on St. Crispin’s Day 2009. (179) T Many of the English lords and the French princes held a reunion on St. Crispin’s Day 2009. Cf. T Many from among the English lords and the French princes held a reunion on St. Crispin’s Day 2009. (180) T Many peers of the English lords and French princes held a reunion on St. Crispin’s Day 2009. The contrast seems otherwise beyond reach were it supposed, fault of the surface syntax, that the English lords and the French princes has the same syntax and reference in (177)–(179) and the same reference as the English lords and French princes. 14.2
Kinematic and object-tracking scenes and frames of reference
Both Jimmy Olsen and Clark Kent are reporters for The Daily Planet. Olsen down on his luck never seems to be at the right place at the right time—always a near miss for a piece of the action. It seems to the innocent bystander that Jimmy Olsen and Superman are tracking each other into and out of crime scenes (without ever quite meeting). Clark Kent is the savvier reporter, but being Superman, he has different reasons for showing up just before or after Superman does. Jimmy Olsen and Superman can be described indifferently by either (19)–(20) or (21)–(22): (19) A reporter and a superhero are tracking each other into and out of crime scenes. The reporter and the superhero are tracking each other into and out of crime scenes. (20) A reporter and a superhero were at the crime scene interviewing witnesses. A reporter and a superhero have entered City Hall one after the other. A reporter and a superhero were out of the mayor’s office in 18 minutes. A reporter and a superhero were talking to the police commissioner in private.
[DP D AdrP and AdrP]
751
(21) A reporter and superhero are tracking each other into and out of crime scenes. The reporter and superhero are tracking each other into and out of crime scenes. (22) A reporter and superhero were at the crime scene interviewing witnesses. A reporter and superhero have entered City Hall one after the other. A reporter and superhero were out of the mayor’s office in 18 minutes. A reporter and superhero were talking to the police commissioner in private. In contrast, to describe Clark Kent and Superman, those who know their identity can use with some ironic detachment (19)–(20) but not (21)–(22). The punchline knocking (21)–(22) out as reports of Clark Kent and Superman is supposed to be that they can no more be two in the same frame of reference than they can sit for the same photograph. The contrast between a scene under a single frame of reference for AdrPcoordination in (21)–(22) and multiple scenes under multiple frames of reference for DP-coordination in (19)–(20) finds some confirmation in the behavior of antisemidistributive predicates. Recall (section 13.1) that being dense, being nine, or being an All-Star team insofar as they imply a measurement imply a single frame of reference for a density, a nonet, or an All-Star team, the constituents of which have not been assembled from across multiple frames of reference. Given several frames of reference, however, these predicates can be applied to each frame of reference measuring entirely within it. In (181), AdrP-coordination, the antisemidistributive predicates cannot apply separately to the conjuncts. It is only the trees that are dense and the ballplayers that are nine or an All-Star team. In contrast, DP-coordination in (182) shares in (181)’s interpretation and admits the one it rejects: (181) a. Some elms and beeches that were (*both) dense in the middle of the forest darkened the forest floor. b. Some Angels and Padres that were (*both) nine in number warmed the All-Star bench. c. Some Apostles and Saints who once were (*both) an All-Star team warmed the benches at Madison Square Garden.47 d. Some twelve Apostles and Saints who once were (*both) an All-Star team warmed the benches at Madison Square Garden. e. An ejected twelve Apostles and twelve Saints who once were (*both) an All-Star team warmed the benches at Madison Square Garden.48
752
Chapter 14
(182) a. Some elms and some beeches that were (both) dense in the middle of the forest darkened the forest floor. b. Some Angels and some Padres that were (both) nine in number warmed the All-Star bench. c. Some Apostles and some Saints who once were (both) an All-Star team warmed the benches at Madison Square Garden. d. Some twelve Apostles and some Saints who once were (both) an All-Star team warmed the benches at Madison Square Garden. e. An ejected twelve Apostles and a dejected twelve Saints who once were (both) an All-Star team warmed the benches at Madison Square Garden. Whatever does allow for the distributed interpretation in (182) must be intrinsic to the structure of the DP-coordination itself, without help from the semidistributive operator, which antisemidistributive predicates reject (section 13.2): (183) Some trees that are some elms and some beeches are (*both/*all) dense in the middle of the forest. (184) Some trees that are some elms and some beeches that are (*both/*all) dense in the middle of the forest darken the forest floor. If, at any point in the analysis of DP-coordination, two frames of reference are introduced, one with each DP, given the injunction against application across frames of reference, the sentence would only have the interpretation applying the antisemidistributive predicate within both: some elms in f0 and some beeches in f1 are densities in F0,1. Whatever does allow for the distributed interpretation in DP-coordination must be absent from AdrP-coordination. In particular, as it would, as just pointed out, compel the unacceptable distributed interpretation, no analysis of AdrP-coordination may introduce more than one frame of reference, which converges strikingly with the argument from Clark Kent, Superman, and sentences (19)–(20) and (21)–(22). That Clark Kent and Superman can no more be counted two under the same frame of reference than they can sit for the same photograph will remain the punchline, but it punches through a connection between a single frame of reference and a still photograph that is not self-evident, which this section examines. The action that the single frame of reference frames in (21)–(22) is, after all, more than a still photograph captures since it tracks through several crime scenes (21) or tracks elapsed time at a single location (22). As soon, however, as the presumed single frame of reference becomes cinematic, it is no longer obvious why it cannot host Clark Kent and Superman counted as two. A failed count to two will rest on counting as two some events or spatiotemporal regions that the interpretation of the sentence forces to overlap in violation of the condition that what is counted be discrete. With the temporal dimension of a cinematic frame of reference, countabil-
[DP D AdrP and AdrP]
753
ity is violated here just in case it is implied that there is some interval of time at which there is both reportage and superheroism whose participants are said to be two, a reporter and a superhero. It is not obvious that (21)–(22) need refer to such overlapping reportage and superheroism. Moreover, it must not imply such a moment of overlap occurs at any of the locations mentioned during the events reported there. Otherwise, (21)–(22) would also become false reports of Superman and Jimmy Olsen, which they are not. The sentences do not assert that reporter and superhero did anything at the very same time and place, and thus nothing asserted compels them to sit for the same photograph. Before pinpointing (21)–(22)’s violation of countability counting Clark Kent and Superman, consider first in more detail the successful count to two in (19)–(20). Crucial to the illusion of two in (19)–(20) (and in (21)–(22), were there no other fault) is that as many frames of reference as crime scenes or other venues be posted at the scene, each with a view not much larger than its scene. Within these frames of reference, no reportage and superheroism overlap: Superman enters, saves the day, exits, and Clark Kent enters, misses the story, exits. As before (e.g., in section 10.3, (258)–(259)), the Tensed Phrase may introduce its own frames of reference, independent of those relating to the subject phrases, and here they are plural.49 The pivotal distinction for (19)–(20) and (21)– (22) will be the frames of reference, reportage, and superheroism referred to within the subject phrases. Whether it is one or more frames of reference, the adverbials derived by adverbialization, while being a reporter and while being a superhero in (19)–(20) and while being a reporter and superhero in (21)–(22), denote events that frame all the action, as described by the local thematic relation, that reporter and superhero participate in at the crime scenes and other venues. It is the events they participate in (under whatever frames of reference have been antecedently introduced) that are said to amount to the tracking each other into and out of crime scenes. If, say, the second conjunct in (19) introduces one frame of reference, it must be large enough to circumscribe a spatiotemporal region that includes the superhero’s scattered appearances at the crime scenes, and similarly for the frame of reference of the first conjunct and the reporter’s scattered appearances. If so, already there is an apparent problem counting just one superhero when presented with several discrete events of superheroism. Recall that object counting so-called is just event counting where the events counted happen to coincide with a lifetime. The count relation is a relation among measuring events (i.e., clicks), events counted (what cues a click), and the measure, a number. With several events of superheroism, how does the counter click only once? One superhero is counted just in case there is just one event of being a superhero, which is to say that the superhero’s multiple appearances at these crime scenes all belong to the same extended and scattered being a superhero, all manifestations of Superman.50 Granting that there may be occasion for such mereological sleight of hand, why isn’t there
754
Chapter 14
also occasion without sleight of hand to count as many superheroes as manifestations of Superman at crime scenes? The question is answered below. Here it suffices to resist temptation from the following. Being a superhero, after all, is not merely some events but a vocation and career, as is being a reporter. So make of this lexical meaning. Let the nominal superhero relate a person to one event in his career only if it relates him to them all: (185) superhero[E,X] ↔ ∃xXx & ∃eEe & ∀x(Xx ↔ ∃e(Ee & superhero(e,x))) & ∀x∀e((Xx & superhero(e,x)) → Ee) To do so would, however, deprive superhero of ever event-counting more superheroes than there are careers as a superhero: Three thousand superheroes have foiled 3000 crimes worldwide this year. Better to consider an epistemic condition on the protocol for measurement, namely, that any two discrete events should not be counted two if it is known that their participants are the same, as it is, by hypothesis, in these multiple appearances of Superman. In event-counting ships passing through the lock, the speaker warrants not to have twice recognized the HMS Pinafore or any other vessel. In suggesting a more robust account below, I do not deny that counting may be subject to such pragmatic considerations as well. Whatever the reason, here one superhero is counted, that is, one being a superhero, a state with scattered manifestations at crime scenes, and similarly for the one reporter here. Now all of this fits under a single frame of reference without a violation of countability, since not even the entire careers of reportage and superheroism ever overlap, let alone their small manifestations (suppressing the urge to argue that him in his underwear is both reportage and superheroism). Not even taking on board entire careers manages to induce the failure in counting under the single frame of reference alleged to discriminate (21) from (19). More is to be taken on board to induce failure. Known to the speaker is a metaphysics that any one being a superhero, even if that career comprises only checkered events, is such only if there is a single person whose being a person contains the events of his career and interpolates a continuous spatiotemporal region between them—and similarly for the metaphysics of being a reporter. If somehow reference to the events of reportage and superheroism also compelled reference and count of the events of being a person, then of course in the case of Clark Kent and Superman, no such two discrete persons can be counted and the plural reference would fail. To discriminate (21)–(22) from (19)–(20), the single frame of reference in (21)– (22) should somehow compel counting persons, while the two frames of reference in (19)–(20) separating Clark Kent and Superman exempt the speaker from counting persons rather than one reporter and one superhero or in some other way escape a violation of countability. A frame of reference is so called serving the purpose of spatial orientation and navigation within the space it frames and thus in this robust
[DP D AdrP and AdrP]
755
sense can be presumed to include representation of the bodies within that space and their motions. A frame of reference for a space large enough for all of Superman’s appearances at crime scenes cannot leave him to vanish at one scene and spring up from the mist at another, without interpolating his trajectory through that space.51 Except for early mention of continuity conditions, the discussion of frames of reference and scenes has so far been so general and abstract as to leave room for such fragmentary, disjointed records, but surely any appeal to scenes or frames of reference invokes them as constituents of what is meant to emulate sane perceptual experience. A speaker’s knowledge and witness that Superman was at, say, two crime scenes remote from one another and her ignorance of all else about his movements meanwhile derive from her experience of two frames of reference no one of which embraces both scenes. Yet the frames of reference quantified over—experienced, actual, or possible—are each a complete representation of the motions of the bodies taken to populate its space. Calling such frames of reference kinematic, a single kinematic frame of reference for both Clark Kent and Superman at all the crime scenes is of a single coincident trajectory for them both. The notion kinematic is extended to any frame of reference or scene s that is similarly closed under the motions within it:52 (186) kinematic(s) ↔df ∀θ∀X∀E∀A∀T ((θ[E,X] & At[A,E,T,s] → ∀E∀A∀T∀σ((θ[E,X] & At[A,A,T,s] & At[A,σ,T,s] & At[A,E,T,σ]) →At[A,E,T,s])) Recall that a measurement under a frame of reference is such that the counter clicks for distinct, let alone discrete, regions as addressed according to that frame of reference. This single trajectory of Clark Kent–Superman then cannot count twice for any one measurement under the one frame of reference that (21)–(22) present. No revision is proposed for what is counted. That remains events, not bodies or trajectories. Event-counting 4000 ships is as before, as it must be if it is not to collapse into object counting: (187) [∃X : ∃E0(∃Eμ count[Eμ,E0,4000] ships[E0,X])] … What is proposed is a refinement in the protocol and conditions for measurement. Events are measurable or countable in a kinematic frame of reference only if there is a projection of those events onto a scene in which their images are within discrete, discernible regions of that scene. In the narrow frame of reference in (188), there are two green regions; there is only one in the larger frame of reference.53 This is not a metaphysical claim, for surely the regions counted under the narrow frame of reference exist as parts of the green annulus under the larger frame of reference.
756
Chapter 14
They are just not to be counted there, under conditions inappropriate for their measurement.
(188)
So it is with HMS Pinafore’s passage twice through the lock. Under the kinematic frame of reference that narrows on the lock, there are two regions of Pinafore, and so two ships passing through. Under the kinematic frame of reference of global positioning satellites, there is only one region of Pinafore. So qualified, it can now be stated that under a single kinematic frame of reference as presented in (21) (NB: no mention of (22)), there cannot be counted two events, one of Clark Kent and the other of Superman, because those events do not occur within two distinct trajectories within that frame of reference, and so plural reference in (21) ultimately fails. It can now be explained how two frames of reference in (19)–(20) afford plural reference to Clark Kent and Superman. The two frames of reference are themselves both of the same space, the first large enough for Clark Kent’s comings and goings at the crime scenes and the second large enough for Superman’s comings and goings at the same crime scenes. Being kinematic, they too are both of the one trajectory. Still the occasion for two frames of reference are the two tokens of the count relation (cf. (187)), one per DP, and the quantificational closure of the variable ranging over frames of reference. But that is also occasion for two acts of measurement, values of the variable Eμ in (187). Under the first frame of reference, answering how many reporters are there, the first count counts one there, corresponding to the one trajectory. Under the second frame of reference, answering how many superheroes are there, the second count counts one again. The subsequent plural reference as reflected in number agreement counts across plural frames of reference, already attested in the meaning of many (counting what may be scattered passim; see section 13.1), which are free of any kinematics applying across them. In short, the one or two frames of reference along with one or two acts of measurement matter for successful plural reference in (19)–(20) and (21)–(22) if the frames of reference are assumed to be kinematic, representing the motions of bodies within the spaces framed, which in turn affects the discernibility within the scenes projected of what
[DP D AdrP and AdrP]
757
is to be counted. As an account of all that can go wrong with plural reference under a single frame of reference, kinematics is still incomplete. For reports of action at a single venue, (22), restriction to kinematic frames of reference is too weak to derive that (22) can be about Jimmy Olsen and Superman but not about Clark Kent and Superman. There is nothing in the reporter events at a single location—nothing in their movements—to tell apart Jimmy Olsen and Clark Kent, except for a bobbing pair of black-rimmed eyeglasses. Moreover, in any kinematic frame of reference narrow enough to frame only the venue mentioned in (22), the reporter events whether of Jimmy Olsen or of Clark Kent are discrete from the superhero events. Like the two passages of the HMS Pinafore under the narrow frame of reference in (188), the reporter enters and exits the frame of reference and then Superman does too. Even Clark Kent can do that with Superman. All of it can—absent further conditions—fall within a single continuous and uninterrupted scene under a frame of reference narrow enough so that Clark Kent-ing and Superman-ing are two discrete events. The sense in which (22) compels Clark Kent and Superman to sit for the same photograph, as it were, has not yet been explained. As remarked above, within a cinematic frame of reference, countability is violated and plural reference fails just in case there is some interval of time at which there is both reportage and superheroism and their participant is said to be two, one reporter and one superhero. Again, (22) will fail as a report of Clark Kent and Superman just in case its meaning demands a frame of reference (or scene) that includes such an interval where there is both Clark Kent-ing and Superman-ing. One route to such frames of reference, which proves to be too strong but is nevertheless instructive, is via object tracking. A scene (or frame of reference) is objecttracking just in case any object located within it is within it throughout the scene’s temporal duration. That is, an object-tracking scene is required to be large enough to track the object’s movement throughout the period under observation: (189) object tracking(s) ↔df ∀X(W[s,X] → ∀t(t(s)t → ∃A At[A,X,t,s])) Given that scenes are projections of the events within a space, a scene that fails to be kinematic implies some defect in the perceptual apparatus, such as the loss of visual information experienced during the aura of a classic migraine. Object tracking reflects rather navigational intentions and an attention to landmarks and to the selection of scenes and frames of reference appropriate for navigational guidance. Scenes may fail to be object-tracking without indication of pathology. It would instead be a substantive finding about communicative intentions if all scenes in the domain of quantification proved to be object-tracking as well as kinematic. I will soon retreat from the assertion that all scenes are object-tracking, but its generality makes it worth pursuing until compelled otherwise.
758
Chapter 14
Suppose then that all scenes are both kinematic and object-tracking. Given the structure of AdrP-coordination (section 14.1.0), (22) puts reportage and superheroism in all the same scenes, say, the one at one address and the other at another, and under adverbialization of the AdrP, these scenes frame all the action constituting the events described. The singular article a(n) presupposes that reportage and superheroism were a W-ing of them both (section 12.2). It is a W-ing only if there is a scene and frame of reference within which it counts as one, and therefore there is a single scene of both the reportage and superheroism. But if that scene is objecttracking and endures long enough to frame all the action, it shows the reporter while the superhero is on stage and the superhero while the reporter is on stage. In such a scene, because it is kinematic, Clark Kent’s and Superman’s trajectories fail to be discrete, not to mention that they do not remain within the distinct regions that AdrP-coordination assigns them; as a result, plural reference in (22) fails. The single object-tracking, kinematic scene would be the cinematic counterpart of the still photograph in which Clark Kent and Superman cannot both appear. The DPs in the DP-coordination in (20) come with their own quantification over frame of reference and scene (cf. (83)). Even holding constant a frame of reference large enough to take it all in, the eye does blink between scenes, which need not be the same time or duration. That scene within which reportage is located need only be of the reporter’s action at the mentioned venue and that within which superheroism is located is similarly confined to the superhero on stage. Note that these scenes are, as required, kinematic and object tracking for their duration, brief as they are. Alongside DP-coordination, (190) with a simple plural may also be a report about Clark Kent and Superman: (190) Two famous citizens of Metropolis were tracking each other into and out of City Hall. Two famous citizens of Metropolis were at the crime scene interviewing witnesses. Two famous citizens of Metropolis have entered City Hall one after the other. Two famous citizens of Metropolis were out of the mayor’s office in 18 minutes. Two famous citizens of Metropolis were talking to the police commissioner in private. Although they are in the same frames of reference, nothing precludes plural scenes, each of Clark Kent or Superman in is own, as in the case of DP-coordination. What undermines (22) as a report of Clark Kent and Superman is a(n) in its insistence on a W-ing in a single scene. In sum, to rule out (21)–(22) but not (19)–(20) or (190)
[DP D AdrP and AdrP]
759
as reports of Clark Kent and Superman, it would suffice that all scenes are objecttracking as well as kinematic. The following however sounds a retreat. In the context for (191) and (192), the lock is so narrow that freighters pass through only one at a time, single file: (191) Four thousand freighters under two flags of convenience passed through the lock last year. (192) Four thousand eastbound Bolivian freighters and westbound Bahamian freighters passed through the lock last year. The sentences are equal event counters in this context. In particular, it satisfies both equally that the same freighter has passed through the lock 4000 times, changing its flag of convenience eastbound and westbound. This is another respect in which an AdrP-coordination and its simple plural counterpart behave alike. According to section 14.1.0, however, any scene under consideration in (192) is of both eastbound Bolivian at one address and westbound Bahamian at another. If now any such scene is also object-tracking, the one freighter that is both the eastbound Bolivian and the westbound Bahamian cannot be counted more than one. The descriptive point is that whatever condition applies in a reporter and superhero to demand a scene that locates a time at which reportage and superheroism overlap, it should not apply and demand the same from some reporters and superheroes or from 4000 eastbound Bolivian freighters and westbound Bahamian freighters. Given the facts of (191) and (192), it is a condition confined to the singular plural construction with a(n), to which I now turn. Object tracking, which implies such a condition, does not characterize all reference to scenes or frames of reference. In the singular plural in (138)–(139), a modified cardinal construction, a(n) counts one being bloodred, the event or state described by the modifier in construction with the article (see section 12.2): (136) Seven bloodred sunrises preceded/announced/augured seven days’ bloody battle. (137) Seven bloodred sunrises have (always) preceded/announced/augured seven days’ bloody battle. (138) A bloodred seven sunrises preceded/announced/augured seven days’ bloody battle. (139) A bloodred seven sunrises have (always) preceded/announced/augured seven days’ bloody battle. It is this one event, with its several participants of seven sunrises, that locates the bloody battle near it, the seven days’ bloody battle following the seventh sunrise. In contrast, (136)–(137) also admits an interpretation that interleaves sunrises and
760
Chapter 14
battles. The force of the singular article is similarly felt in the contrast between (193)–(194) and (195)–(196): (193) A sunrise and a sunset preceded/announced/augured two bloody battles. (194) A sunrise and a sunset have preceded/announced/augured two bloody battles. (195) A sunrise and sunset preceded/announced/augured two bloody battles. (196) A sunrise and sunset have preceded/announced/augured two bloody battles. Sentences (193)–(194) may tell of a battle at dawn and another at dusk, but (195)– (196) only of battles the next day.54 Again, only (197)–(198) and not (199)–(200) can report the procession of minuteman, Thomas Jefferson, marine, Sam Adams in that order: (197) A minuteman and a marine preceded/announced two Founding Fathers in the Fourth of July parade. (198) A minuteman and a marine have preceded/announced two Founding Fathers in the Fourth of July parade. (199) A minuteman and marine preceded/announced two Founding Fathers in the Fourth of July parade. (200) A minuteman and marine have preceded/announced two Founding Fathers in the Fourth of July parade. Recall (section 12.2) that absent an explicit modifier such as bloodred in (138)–(139) or ill-matched in an ill-matched pitcher and catcher, the events that a(n) counts as one in a sunrise and sunset, a minuteman and marine, or a pitcher and catcher are W-ings, the presentation or participation events themselves (see (154), the logical form for a pitcher and catcher). Despite the singular article’s power to enforce participation in a single event in (138)–(139), (195)–(196), and (199)–(200), nothing in the meaning of a(n) (or anything else in these sentences) forces a participant’s role in that event to overlap in time that of any other’s—how else could sunrises participate in the one being bloodred in (138)–(139) or sunrise and sunset in the one W-ing in (195)–(196)? If a being bloodred or a W-ing suffers participants who do not share a moment side by side, other single events are less generous. An event concept imposes eccentric conditions on an event it denotes, as witnessed in the following variation: (201) Four thousand {observed/counted/videotaped/logged/chartered/suspicious/ Customs-inspected}freighters passed through the lock last year. (202) An{observed/counted/videotaped/logged/chartered/suspicious/Customsinspected}4000 freighters passed through the lock last year.
[DP D AdrP and AdrP]
761
(203) Four thousand{watch-listed/registered/pictured}freighters passed through the lock last year. (204) #A{watch-listed/registered/pictured55} 4000 freighters passed through the lock last year. A voyage or passage may be observed, counted, videotaped, logged, chartered, under suspicion, or occasion for Customs inspection. A single sustained event of observation, counting, videotaping, or inspection may itself unfold in time, and a single log or charter may be of scattered events. As a result of these considerations, reference to an event in (202) does not disrupt event-counting 4000 freighters. The sentences in (202) do not differ from (201) in allowing that the 4000 passages through the lock may have been of fewer vessels. In contrast, since what is on a watch-list or in a registry are vessels and being pictured is being in a snapshot, to be distinct participants in any one of these is to be distinct bodies, which disrupts event counting in (204). There are 4000 freighters in that picture, on that watch-list, or in that registry only if there are 4000 vessels. Event counting survives in (203) absent reference to a single being watch-listed, registered, or pictured. For each of the 4000 passages, there is a being watch-listed, registered, or pictured in which the vessel of that passage appears. The above both rehearses the effects of being in construction with a(n) and draws attention to the place of lexical content in defining and individuating events. This is to create a climate hospitable to the suggestion that what excludes Clark Kent and Superman from (22) has to do with what it means to be a W-ing and thus specifically with the lexical content of ‘W’. As remarked above, given (196) and the timing of sunrise and sunset, a W-ing is not a still photograph. It cannot be participation in the same one W-ing that forces reporter and superhero in (22) to share a moment side by side, thereby excluding Clark Kent and Superman. Rather, a W-ing is objecttracking. That is, even if not all scenes or frames of reference are object-tracking, any scene s of a W-ing e is (205) (∃X W[e,X] & ∃α∃t∃π∃f∃l∃o s = projection(e,α,t,π,f,l,o)) → object-tracking(s) Note that if the projected scene of a W-ing is embedded within a longer scene, object tracking is mandated only during the briefer projection. The account ruling out (21)–(22) but not (19)–(20) or (190) as reports of Clark Kent and Superman proceeds as before mutatis mutandis. It continues to rule out (21)–(22) that they are about a single W-ing of reporter and superhero that is object-tracking. Although no harm would come to (19)–(20) or (190) if all scenes were object-tracking, in restricting it to a W-ing, (192) is now spared the implication that the same one freighter did not make 4000 passages through the lock. Suppose it was HMS Pinafore. Every scene under consideration is of HMS Pinafore eastbound at one address within it
762
Chapter 14
and westbound at another address. These are (at least) two events, one of a freighter eastbound and another of a freighter westbound. These two events are among the 4000 counted. Each might be a W-ing and the 4000 might be 4000 W-ings, but nothing in (192) joins the eastbound Bolivian freighter and westbound Bahamian freighter in a single W-ing. Each of 4000 W-ings is indeed object-tracking, but since each is only as long as a single passage through the lock, in no scene of any one of them do trajectories of eastbound and westbound traffic collide, leaving the 4000 discrete and well counted as desired. If not a still photograph, a W-ing is a presentation under a fixed gaze, scaled to its subjects and object-tracking them until distracted. The presentation invoked uttering a cousin is not a close-up of a large nose, and it does not cut her head off as she wanders out of its frame like a bad home movie.
15
The Ordered-Pair Illusion
The last substitution puzzle is an illusion that and manufactures reference to ordered pairs: (1) In a mutual declaration of requited love, the lover and belovèd in the first love note are not the lover and belovèd in the note returned. (2) In a mutual declaration of requited love, the lover and the belovèd in the first love note are not the lover and the belovèd in the note returned. As the lovers exchanging the notes are the same Héloïse and Abelard, there is temptation here to let and under a new guise fashion from lovers their ordered pairs and . As these are not identical, (1) and (2) prove to be true if the AdrP-coordinations in (1) and the DP-coordinations in (2) refer to them. But this bloated semantics for and ends in contradiction in (3), which ought to have had a true reading if the subject referred to an ordered pair and the predicative nominal does not: (3) F The lover and (the) belovèd in the first love note are not the lovers exchanging them. Instead, the resolution of the puzzle relies on the grammar of DP and AdrP of the preceding sections and its spatiotemporal apparatus. In addition, it requires supplement from a tacit adverbial or secondary predicate, similarly oriented, so that what (1) and (2) truthfully deny given that the notes’ lovers are the same is their identical arrangement in their scenes: (4) In a mutual declaration of requited love, the lover here in a scene1 and belovèd there in that scene1 in the first love note while lover here and belovèd there in that scene1 are not the lover here in a scene2 and belovèd there in that scene2 in the note returned while lover here and belovèd there in that scene2, that scene1 oriented the same way as that scene2.
764
Chapter 15
15.0
Perspectival relations within nominal conjuncts
In baseball, the pitcher and the catcher, who work either side of the batter’s box to strike the batter out, are the batterymates in a defensive lineup. That is, pitchers and catchers is coextensive with batterymates. Some pitchers and catchers are in fact better suited to working with one another than others (e.g., knuckleballer Tim Wakefield and Doug Mirabelli). In the interest of science and with apology to Major League Baseball, I will further assume that some pitchers are also catchers—that is, that there are some switch players skilled at both positions. Notice next that it suffices for (5) that someone different pitches the second game or someone different catches it: (5) An ill-matched pitcher and catcher started for the Red Sox yesterday, and another equally ill-matched pitcher and catcher started for them tonight. In particular, it would suffice, remarkably, for the first game’s pitcher and catcher to switch positions for the second game—as if to say that , an ordered pair, started yesterday, and the distinct ordered pair , tonight, suppressing any scruples against tossing ordered n-tuples into the mix of individuals, sums, groups, and whatever, that predicates denote (see section 10.0.3 for relevant discussion). It can also be said that certain pitching and catching frames yesterday’s start, and different pitching and catching frames tonight’s. Paraphrasing an adverbialized logical form, a pitcher and catcher, pitching and catching started yesterday, and another pitcher and catcher, pitching and catching started tonight. If it is not understood how the one’s pitching and the other’s catching could frame the events reported, as in (6), where pitching and catching are not different ways of being disabled or excluded from the game, the identity condition shifts: (6) An ill-matched pitcher and catcher were on the disabled list yesterday, and another equally ill-matched pitcher and catcher were on it tonight. An ill-matched pitcher and catcher warmed the Red Sox bench yesterday, and another equally ill-matched pitcher and catcher kept it warm tonight. It can no longer be the same two ballplayers yesterday and tonight. Yet (6) can be coaxed toward (5) if it is further imagined that the disabled list or bench itself designates a pitcher and catcher held in reserve, or that they were ill-matched to be together on the disabled list or bench. As in the previous sections, the temporal framing need not demand coincidence and the neighborhood relation may be rather intentional and perspectival. To my ear, it is enough for (7) to support the finergrained identity condition that yesterday’s misery and consolation concern the
The Ordered-Pair Illusion
765
pitching and catching in yesterday’s game, and today’s misery and consolation concern tonight’s switched pitching and catching: (7) An ill-matched pitcher and catcher consoled themselves after the game yesterday at the Cask ’n Flagon, and another equally ill-matched pitcher and catcher consoled themselves again at the same establishment tonight. The modulated effect on identity in (5)–(7) articulating pitcher and catcher contrasts with (8)–(13). Substitution of the presumably coextensive (two) batterymates undermines the finer-grained identity condition. It never suffices that pitcher and catcher are understood to switch positions from one game to the next. Sentences (8) and (11), crucially, are simply false if the same batterymates play the two games.1 (8)
An ill-matched two batterymates started for the Red Sox yesterday, and another equally ill-matched two batterymates started for them tonight.
(9)
An ill-matched two batterymates were on the disabled list yesterday, and another equally ill-matched two batterymates were on it tonight. An ill-matched two batterymates warmed the Red Sox bench yesterday, and another equally ill-matched two batterymates kept it warm tonight.
(10) An ill-matched two batterymates consoled themselves after the game yesterday at the Cask ’n Flagon, and another equally ill-matched two batterymates consoled themselves again at the same establishment tonight. (11) Two ill-matched batterymates started for the Red Sox yesterday, and another two equally ill-matched batterymates started for them tonight. (12) Two ill-matched batterymates were on the disabled list yesterday, and another two equally ill-matched batterymates were on it tonight. Two ill-matched batterymates warmed the Red Sox bench yesterday, and another two equally ill-matched batterymates kept it warm tonight. (13) Two ill-matched batterymates consoled themselves after the game yesterday at the Cask ’n Flagon, and another two equally ill-matched batterymates consoled themselves again at the same establishment tonight. More examples are forthcoming.2 Unlike shoes, there are no left and right socks that stand as such on their own. My feet, however, are left and right and a full half-size apart. Sentence (14) is a better report than either (15) or (16) of my experiments with a single pair of silks that do not stretch, one size 12 and another 12½, and (15) improves if the effort is made to connect left or right to the sensation of wetness. But no imagination rescues (17) or (18).
766
Chapter 15
(14) A left sock and right sock did not fit, but another left sock and right sock felt snug. (15) A left sock and right sock felt wet, and another left sock and right sock felt wet too. (16) A left sock and right sock were (as light as) silk, and another left sock and right sock were (as light as) silk too. (17) A mixed-size two socks did not fit, but another (mixed-size) two socks felt snug. (18) Two socks did not fit, but another two socks felt snug. Here too reference that pretends to tell apart one arrangement of some objects from another of the same objects succeeds only if described by a nominal coordination, and only if the objects participate in the event reported by the matrix qua as described (i.e., if their arrangement one way rather than the other also qualifies their participation in the matrix event). Section 15.0.0 examines the first condition connecting arrangements and their individuation by nominal coordinations, and section 15.0.1 examines the second condition according to which the arrangements referred to qualify participation. 15.0.0
Scene correspondence3
I have only two socks—a white one with gray toecap, heelcap, and ankle band and a gray one with white toecap, heelcap, and ankle band— so that (19) is false tout court: (19) The socks in photo 1 are not the socks in photo 2. (20)
Photo 1
Photo 2
Yet one can very well understand (21) to be true of (20), saying what (22) says unambiguously: (21) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2. (22) The left sock and (the) right sock in photo 1 are not correspondingly the left sock and (the) right sock in photo 2.
The Ordered-Pair Illusion
767
Even in simple identity statements, a contrast thus emerges between nominal coordination and a simple coextensive nominal, where the former as in the preceding section seems to elicit a finer-grained identity condition. Some further conditions on the elements of (19)–(22) can be noted right away, reserving their explanation for later. First, the coordinations eliciting finer-grained identity do not include those properly contained within a relative clause: (23) The socks that are left and right in photo 1 are not the socks that are left and right in photo 2. Sentence (23) is false of (20) and can only be met with a protest that the socks are indeed the very same, only switched. Second, concerning simple nominals, an overt correspondingly does not rescue sentences such as (24): (24) The socks in photo 1 are not correspondingly the socks in photo 2. The simple nominal remains awkward, unless what is on display is itself rather explicit about the intended correspondence: (25)
L
R
Photo 1
L
R
Photo 2
Note also that despite the explicit nature of the photos in (25), (19), without overt adverb, is as false of (25) as it is of (20). What is held for later discussion is the contribution of the overt adverb and why mention of left and right requires for its effect on identity conditions the particular syntax of nominal coordination. A more pressing question is already posed by the contrast between (19) and (21), repeated below: (19) The socks in photo 1 are not the socks in photo 2. (21) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2.
768
Chapter 15
If (19) can only be understood to be false of (20), then of course (26) is understood with equal confidence to be true: (26) The socks in photo 1 are the socks in photo 2. Now, if, as they do, speakers find salient an interpretation of (21) that is also true, no syntax and semantics for (21) conspiring to give this result should afford an inference from (26) and (21) to (27): (27) The left sock and (the) right sock in photo 1 are not the socks in photo 1, or the left sock and (the) right sock in photo 2 are not the socks in photo 2. As eager as a speaker may be to embrace a truth in (21), none in her right mind thereby commits herself to (27) or recognizes any interpretation of (27) that could also be true. But suppose a toy semantics were to contrive that coordinated nominals sometimes refer to ordered pairs and do so in (21) (i.e., ≠). What then blinds the speaker who has just asserted (21) (and (26)) to a conclusion that takes (27) to express the truth that ≠{s12,s12½} or ≠{s12,s12½}, continuing to use the coordinated nominals in the same way as she has in (21)? If the logical syntax of identity statements (the copula clause) is as hamstrung as this toy semantics implies, it leaves little recourse other than to tinker with nominal reference (see section 10.0.3) and thus little escape from the dilemma just observed. But a closely related paradigm, one that does not dangle nominal coordination as a red herring, implicates the syntax of the identity statement itself in the contrast between (19) and (21). Sentence (28) with the demonstrations indicated is, like (19), simply false of (20): (28) These[the socks in photo 1] are not them[the socks in photo 2]. Yet (29) is true of what is seen there (as is (30)): (29) These are not them put back in the sock drawer in the same order. (30) These are them put back in the sock drawer in a different order. The pronouns all have the same reference throughout (28)–(30), it being too horrible to think otherwise. What, then, is it to be them simpliciter and yet to be them one way and not another? This question has already been asked and answered in section 10.3.0 for sentences such as (31) and (32): (31) At the mayor’s press conference, Superman was Clark Kent. (32) F At the mayor’s press conference, Superman was Clark Kent in cape and tights. Note that further observations fend off the stray thought that the phrase them put back in the sock drawer in the same order refers in (29) not to the socks but to their pairing with a spatiotemporal location, event, or mode of presentation, .
The Ordered-Pair Illusion
769
The sentences in (33) are true despite the fact that their arrangement this way is not their arrangement that way:4 (33) These arranged this way [photo 1] are them arranged that way [photo 2]. These arranged as they are are them arranged in this other way. The sentences are true in that the socks themselves are indeed the same in the two photos. The stray thought to rescue simple identity by tinkering with novel arguments, these arranged this way, and them arranged that way, , founders on (33), whose truth would be undermined if the identity statement were about pairings of socks with spatiotemporal locations. Sentences like (33) do however become false if how the socks are in photo 2 is misdescribed: (34) These arranged this way [as in photo 1] are not them arranged that way too. These arranged as they are are not them arranged the same way. Sentence (35) toggles true and false, depending on whether the second so is anaphoric to the first so or demonstrative to the second photo (with contrastive stress): (35) These arranged here just so are not them arranged there just so. The pronouns in all the above refer innocently to the socks and the modifiers arranged just so, put back in the sock drawer, etc., without becoming devices of novel reference, are restored as simple modifiers when identity statements are articulated as in section 10.3.0: (36) These socks are not those socks. (37) [These X : ∃E socks[E,X]][℩E : socks[E,X]][℩E0 : N[E,E0]][℩E: E0[E] Present[E]] not ([∃E0: E[E0]] W[E0,X]O[E0,E1]∃Y Theme[E1,Y]O[E1,E2]Be[E2] [Those X : ∃E socks[E,X]][℩E : socks[E,X]][℩E0 : N[E,E0]][∃E1: E0[E1]] W[E0,X]O[E0,E3]∃Y Theme[E3,Y]O[E2,E3]) ‘These socks sock-ing are not those socks socking.’ (38) These socks arranged here just so1 are not those socks arranged there just so2. (39) [These X : ∃E socks[E,X]][℩E: socks[E,X]] [℩E′:N[E,E′] arranged[E′] here just so1] [℩E0 : N[E′,E0]][℩E: E0[E] Present[E]] not ([∃E0: E[E0]] W[E0,X]O[E0,E1]∃YTheme[E1,Y]O[E1,E2] Be[E2] [Those X : ∃E socks[E,X]][℩E: socks[E,X]] [℩E′:N[E,E′] arranged[E′] there just so2] [℩E0 : N[E′,E0]][∃E1: E0[E1]] W[E0,X]O[E0,E3]∃Y Theme[E3,Y]O[E2,E3]) ‘These socks sock-ing arranged just so1 are not (those socks sock-ing arranged just so2).’
770
Chapter 15
(40) These socks arranged here just so1 are not those socks arranged there just so1. (41) [These X : ∃E socks[E,X]][℩E: socks[E,X]] [℩E′:N[E,E′] arranged[E′] here just so1] [℩E0 : N[E′,E0]][℩E: E0[E] Present[E]] not ([∃E0: E[E0]] W[E0,X]O[E0,E1]∃YTheme[E1,Y]O[E1,E2] Be[E2] [Those X : ∃E socks[E,X]][℩E: socks[E,X]] [℩E′:N[E,E′] arranged[E′] there just so1] [℩E0 : N[E′,E0]][∃E1: E0[E1]] W[E0,X]O[E0,E3]∃Y Theme[E3,Y]O[E2,E3]) ‘These socks sock-ing arranged just so1 are not (those socks sock-ing arranged just so1).’ Recall that supermonadicity provides each thematic relation with its own events, and thus overt modifiers such as arranged just so and those adverbial modifiers derived from the adverbialization of nominal content all apply to local events. The nominal and copular syntax is identical throughout (36)–(41). Example (36)–(37) merely suppresses the optional adverbial modification explicit in the others. Note that the descriptive content of the predicative nominal, the adverbial modifier (socking) derived from its adverbialization, and its overt adverbial modifier if there is one all fall within the scope of negation. Thus, although (36) is itself false, the sentence becomes true if the more elaborate content happens to misdescribe the local event or presentation of the socks, as it does in (40)–(41): the socks there in photo 2 are not arranged as they are in photo 1. Sentence (21) shares of course the copular syntax of (19) and (36)–(41): (19) The socks in photo 1 are not the socks in photo 2. (21) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2. The analysis of (21) that proves true of (20) in effect locates within the scope of negation some (unspoken) content that happens to misdescribe the socks as photo 2 presents them: (42) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2 corresponding just so. Surely the possibility of modifying the copular syntax as in (42) should also be present for that same syntax in (19), and it is: (43) The socks in photo 1 are not the socks in photo 2 corresponding just so. The syntax is just the scaffold. All the work to tell apart (19) and (21) is in how corresponding just so is eventually construed in section 15.0.1. As is to be seen, the nominal coordination and the mention of left and right set up a correspondence
The Ordered-Pair Illusion
771
that misdescribes the socks in photo 2, but the correspondences that can be construed for (43) happen to all be so weak as to be true of the sock’s presentation in photo 2, with the result that (19) must be false of (20) while (21) can be understood to be true there. The unspoken corresponding just so should not be mistaken for the correspondingly spoken in (22), a distinction to which I will have to return to complete the discussion.5 15.0.1
“Similarly orienting”
Something in how things are with the socks, their arrangement, or their appearance changes from photo 1 to photo 2 so that they are not the same, as (21) reports, a correspondence having been disrupted: (21) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2. (44) The left sock and (the) right sock in photo 1 are not [the left sock and (the) right sock in photo 2 corresponding just so]. It will not do to say that the photos are the same in the relevant respect—that the intended correspondence holds—just in case they are of the same objects, since of course they are of the same socks. It would be equally self-defeating to say that the correspondence fails just in case the events depicted are different, since typically the point of all these sentences, true or false, is to compare events or situations at some remove from one another (see also section 10.3.0): (45) The left sock and (the) right sock brand new in photo 1 are not the left sock and (the) right sock in photo 2 after 3 years of athlete’s feet and 156 launderings. (46) The starting pitcher and catcher for the Red Sox in Game 7 of the 1986 World Series are not the starting pitcher and catcher for the Red Sox in Game 4 of the 2007 World Series. It is rather that something is the same (or not, as is the case for the photos of (20)) in the scenes of these possibly far-flung events. It is whether or not they are similarly orienting. Two scenes are similarly orienting just in case any participant they share is projected to the same place (more or less) in the two scenes. Although the scenes are of different spaces at different times and under different frames of reference, to be at the same place in two scenes is to be at the same address, at the same coordinates , for the local frame of reference from one scene to the next. In (47), similarly orienting is said to hold between events E0 and E1 just in case a similarity of orientation is found in scenes of them, Sɶ0 and Sɶ1 (l. (i)). Recall from section 9.4.2 that scenes, whatever their optical resolution, are further individuated by a native reticule that can be more or less fine-grained in its
772
Chapter 15
resolution and in what fits within and is assigned its addresses. Scenes are compared for similar orientation only if their resolution is the same, assigning the same things addresses that segregate them from the same other things (l. (ii)). If the scenes are comparable, then they are similarly orienting only if their addressing is indeed the same (ll. (iii)–(iv)). As all the examples under discussion are still lifes,6 the conditions apply to the still-life scenes Sɶ0(t ) that at any moment t constitute cinematic scenes, which could prove to be mistaken once the cinematic cases are given more serious consideration: (47) similarly orienting[E0,E1] ↔df ∃Sɶ0 ∃Sɶ1 ( (i) ∃0 ∃F0 en-scène[E0, Sɶ0 , 0, F0 ] & ∃1∃F1 en-scène[E1, Sɶ1, 1, F1 ] & ɶ ɶ ∀t (∀ζ∀ξ (∃ [S0(t ), ζ , ξ, ] ↔ ∃ [S1(t ), ζ , ξ, ]) → (ii)7 ɶ ɶ (iii) ∀ζ∀ξ∀ (([S0(t ), ζ , ξ, ] & ∃F0 en-scène [ζ , S0(t ), , F0 ]) ↔ ( [Sɶ0(t ), ζ , ξ, ] & ∃F0 en-scène [ζ , Sɶ0(t ), , F0 ] )))) (iv) To illustrate the definition in action, assume a visual experience that looks like (48): (48)
t1
t2
This experience could be of similarly orienting scenes s1 and s2, if they have been projected from frames of reference rotated around the line of sight, f2 = 180-rotation(f1). Note in particular that the coordinates for the gray sock are the same in both of these frames of reference. In fact, for scenes projected from these frames of reference, anything, a heel or a thread, under one frame of reference is at the same address in the other, and thus the scenes so projected will correspond under any resolution. In contrast, suppose that (48) is an experience of scenes that are projections from the same frame of reference f. No sock or any part of one is at the same address with respect to that frame of reference at both t1 and t2, and thus scenes so projected do not correspond at any resolutions that resolve even a sock. If, on the other hand, the scenes resolve nothing smaller than a pair of socks, the scenes may correspond even if projected from the same frame of reference: the socks are the occupants of a perimeter that is addressed the same way at t1 and t2. The socks have remained in the same place under a low resolution that does not discern the fact of their rotation (just as the Great Spot’s molecules, clouds, and eddies remain in some sense in the same place, the Great Spot’s). Similar remarks apply to a visual experience like (49):
The Ordered-Pair Illusion
773
(49)
t1
t2
There are frames of reference, where f2 is related by a rotation around f1’s vertical axis and a displacement of the point of view along the line of sight to a point on the other side of the socks, projecting scenes that are similarly orienting at any resolution. Similarly orienting scenes from the same frame of reference, again however, do not resolve the individual socks. If in the context of utterance for (21) it is understood that the spatial vocabulary is to be interpreted with respect to the same frame of reference and that any scenes invoked should at least resolve left sock and right sock, (21) is true of the visual experience in (48) or (49) in denying that the scenes are similarly oriented: (21) The left sock and (the) right sock at t1 are not the left sock and (the) right sock at t2. In contrast to (48) or (49), (21) is false of the visual experiences in (50) even assuming a single frame of reference and resolution of the socks, as there is no understanding of the sentence under which it could be true there: (50)
t1
t2
774
Chapter 15
The evaluation of the sentence in such a context attends only to the relative positions of the socks suppressing all else in the background and excluding them from the relevant reticules (see note 6), as in the scene component in (51): (51)
t1
t2
Despite the richer visual experience of (50), in uttering a sentence (21) about socks, the speaker selectively attends to scenes about them for whatever reason. Perhaps it coincides with her selective attention to a distinct space, on, say, the surface of the cab’s windshield on which the socks lie rather than to the larger space that includes the world beyond the windshield, or perhaps it reflects her attention to certain landmarks chosen to navigate through a space (50) that they are crucially part of. For the scenes (50) (and also (51)), there is, with respect to the same frame of reference, a region that encloses the gray sock and excludes the white sock at both t1 and t2, although the gray one has turned over on its side and shifted its position within that region.8 Similarly, there is a region within which the white sock is enclosed to the exclusion of the gray sock. There is thus no denying that these are similarly orienting scenes, and as the regions that resolve the socks are themselves left and right, (21) is undeniably false of (50). In minimal contrast, note now that (52) is true of (50) in denying the similar orientation of its scenes: no similarly orienting scenes resolving the socks under the same frame of reference resolve them in regions that are top and bottom: (52) The top sock and (the) bottom sock at t1 are not the top sock and (the) bottom sock at t2. A logical form for (21), carrying over from sections 10.3.0 and 11.1 the logical form of the copular construction, is (53). The tacit similarly orienting occurs with a tacit
The Ordered-Pair Illusion
775
anaphora to the scene that the antecedent subject describes—(now) being similarly orienting to the scene of the events then: (21) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2. [DPThe X: ∃E0 (Δleft sock and right sock in photo 1)] [℩E0: Δ]-W-Theme not( W-Present-Theme be[E4] [DPThe X: ∃E0 (Ω left sock and right sock in photo 2)] [℩E0: Ω]-W-Theme [E6, E4] [℩E0: Δ] similarly orienting[E4,E0] ) (53) [∃F1: photo 1] [The X: ∃E0 (Δ now-en-scène[E0] W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0, ]] (left[E1, , F 1 ]9 ∃X sock.sg[E1,X])) and ∃E1 (Ψ [∃ : VF[E0, ]] (right[E1, , F 1 ] ∃X sock.SG[E1, X ])))] [℩E0: Δ][∃E2 : N[E0,E2]]W[E2,X] O[E2,E3] ∃XTheme[E3,X] not( ∃E4 [E3,E4] ∃E5∃X W[E5,X] Present[E4,ER] O[E5,E6] ∃XTheme[E6,X] O[E6,E4] be[E4] [∃F2: photo 2][The X: ∃E0 (Ω now-en-scène[E0] W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0, ]] (left[E1, , F 2 ] ∃X sock.SG[E1, X ])) and ∃E1 (Ψ [∃ : VF[E0, ]] (right[E1, , F 2 ] ∃X sock.SG[E1, X ])))] [℩E0: Ω] [∃E5 : N[E0,E5]] W[E5,X] O[E5,E6] ∃X Theme[E6,X] [E6,E4] [℩E0: Δ] similarly orienting[E4,E0] ) Note that scenes that do not resolve the socks, one from the other, are similarly orienting (by default, not being comparable, (47) l. (ii)), and thus sentences (27) and (54) turn out false if their interpretation should allow the scene invoked at t2 a resolution that assigns the socks a single address enclosing them both: (27) The left sock and (the) right sock at t1 are not the socks at t2. (54) The top sock and (the) bottom sock at t1 are not the socks at t2. But the logical form of the AdrP [socks at t2] does locate them at an enclosing address. Moreover, as reticules partition space, none will have addresses for both the socks together and the socks segregated (see section 9.4.2, (200) and the surrounding discussion): (27) The left sock and (the) right sock in photo 1 are not the socks in photo 2. [DPThe X: ∃E0 (Δleft sock and right sock in photo 1)] [℩E0: Δ]-W-Theme not( W-Present-Theme be[E4] [DPThe X: ∃E0 (Ω socks in photo 2)] [℩E0: Ω]-W-Theme [E6, E4] [℩E0: Δ] similarly orienting[E4,E0] )
776
Chapter 15
(55) [∃F1: photo 1] [The X: ∃E0 (Δ now-en-scène[E0] W[E0,X] [℩E1: ΦΨ]O[E0,E1] ∃E1 (Φ [∃ : VF[E0, ]] (left[E1, , F 1 ]10 ∃X sock.sg[E1,X])) and ∃E1 (Ψ [∃ : VF[E0, ]] (right[E1, , F 1 ] ∃X sock.SG[E1, X ])))] [℩E0: Δ][∃E2 : N[E0,E2]]W[E2,X] O[E2,E3] ∃XTheme[E3,X] not( ∃E4 [E3,E4] ∃E5∃X W[E5,X] Present[E4,ER] O[E5,E6] ∃XTheme[E6,X] O[E6,E4] be[E4] [∃F2: photo 2][The X: ∃E0 (Ω now-en-scène[E0] [∃ : VF[E0, ]] (ADR[E0, , F 2 ] socks[E0, X ]))] [℩E0: Ω] [∃E5 : N[E0,E5]] W[E5,X] O[E5,E6] ∃X Theme[E6,X] [E6,E4] [℩E0: Δ] similarly orienting[E4,E0] ) Alongside the fundamental contrast between (21) and (27), it was remarked at the end of section 15.0.0 that (27) contrasts with a true (56) and (43): (56) The left sock and right sock in photo 1 are not the socks in photo 2 the same way. (43) The socks in photo 1 are not the socks in photo 2 the same way. That is, an unspoken similarly orienting cannot express what the overt the same way does. This is partly just that similarly orienting is stipulated to mean what it means and cannot convey another way the scenes could be, which context can supply to the overt the same way. It is also in part that the silent descriptive anaphora referring to the scenes require descriptive, linguistic antecedent without power on their own, unlike pronounced definites or demonstratives, to refer to what is salient to the observer in the surrounding context (see section 15.0.2 for further discussion). It has been suggested that the sense in which (21) and the like are true demands nothing from the meaning of and, deriving instead from the denial of a correspondence, with this scenel similarly oriented to that scenei. The unspoken adverbial phrase is truly detached from and independent of coordination, and the presumption is that it may also occur in (19) (see (43)). As just sketched, with this scenel similarly oriented to that scenei does just that without ill-effect. As it turns out, its occurrence affects the sentence’s truth only if the sentence presents two coordinate structures as in (21), in contrast to (19). 15.0.2
Silent
ADR
Without encroaching on the meaning of and, the truth of (21) when the photos rearrange the same socks will derive from a correspondence that fails to hold of the frames of reference fixed when adrs such as left and right are tokened.
The Ordered-Pair Illusion
777
(21) The left sock and (the) right sock in photo 1 are not the left sock and (the) right sock in photo 2. (44) The left sock and (the) right sock in photo 1 are not [the left sock and (the) right sock in photo 2 corresponding just so]. The same is to be said of (57), true of two ballplayers who switch positions between games, except that the adrs tokened are unspoken: (57) The pitcher and (the) catcher in Game 1 are not the pitcher and (the) catcher in Game 2. (See (5)–(7).) (58) The pitcher here and (the) catcher there in Game 1 are not [the pitcher here and (the) catcher there in Game 2 there corresponding just so]. Opposite (21) and (57), it has already been observed that (19) remains resolutely false despite rearrangement of the socks: (19) The socks in photo 1 are not the socks in photo 2. Despite the silent adrs that may slip in wherever nominals are modified, there is no understanding of (19) along the lines of (59) that makes it true. (59) The socks on the left in photo 1 are not the socks on the right in photo 2 (corresponding just so). The socks first in photo 1 are not the socks second in photo 2 (corresponding just so). The socks here in photo 1 are not the socks there in photo 2 (corresponding just so). Whatever is to account for the differential effect of silent adrs on (19) and (57) and for any difference between them unspoken in (19) and spoken in (59), it cannot rest on gross distributional facts. In particular, it cannot without harm to my thesis be claimed that silent adrs are special to nominal coordinations such as (57) and do not occur with the simple nominals of (19), nor can it be claimed that a sequence of silent adrs is never contrastive in meaning (ruling out (59) as an analysis of (19) in favor of a reliably false The socks first are not the socks first), since exactly such a contrast will in part account for (57)’s truth.11 Consider first the simple nominals and the differences between silent and spoken adrs that emerge from a closer inspection of (19), (59), and the like. Let (60) be false and therefore to be distinguished despite its silent adrs from such truths as can be expressed by (61) and (62) and the like. Sentences (61) stand for cases where the adr relation is chosen to be the same, and (62), where chosen to be different. (60) They are not them. (61) They first (there) are not them first (here). (62) They first are not them second.
778
Chapter 15
Sentence (61) threatens to be true only if the frames of reference for the adrs are different and they who (61)’s pronouns refer to are not first with respect to the second frame of reference. A shift in frame of reference can be accomplished by overt demonstrations to distinct spaces—distinct photos, to be concrete: (63) They first there are not them first here. They first here are not them first there. Not all alternations of overt demonstratives with unspoken counterparts cue a shift in frame of reference. Thus, although overt demonstratives signal shifts in frame of reference in any direction, silence only consents to the current frame of reference or reverts to the speaker’s here: (64) (65) (66) (67)
T They first are not them first there. F They first are not them first here. F They first here are not them first. T/F They first there are not them first.
Sentence (67) is understood to be false if the second frame of reference is that introduced with the demonstrative there and true if it is the speaker’s here. Notice that if neither space has been privileged to be the speaker’s here, as with photos that are side by side, silence cannot accomplish any shift, leaving (67) simply false in this case and (64) unacceptable. Even with demonstrative gestures accompanying the pronouns, (66) and (67), to my ear, just lapse into unacceptability as well. Those wild gesticulations, one for each pronoun, do however rescue (68) from certain falsehood without them, when reference to the frame of reference is left entirely implicit: (68) They first are not them first. Without the second pronoun to support demonstration, a shift in frame of reference again becomes impossible, leaving (69) unambiguously false: (69) F They first are not first. Note also that it would be entirely unacceptable to pick up and wave one photo and then the other as in (70) to convey what can be truthfully said with (71) if they are first in photo 1 and not first in photo 2: (70) #They first [shaking photo 1] are not first [shaking photo 2]. (71) They first here [shaking photo 1] are not first here [shaking photo 2]. They first here [shaking photo 1] are not first there [pointing to photo 2]. They first there [pointing to photo 1] are not first here [shaking photo 2].
The Ordered-Pair Illusion
779
The remaining examples without a second subject pronoun conform to the use of silence and overt demonstrative seen earlier: (72) T/F They first there are not first. (73) F They first here are not first. (74) They first are not first there. (75) F They first are not first here. Whatever conditions attach to shifts in frame of reference when reference to them is silent as in (68) surely remain in force in (60) where reference to them remains silent and the silence is extended to the tacit adrs themselves. (60) They are not them. (68) They first are not them first. For more conservative speakers, more reluctant than I have been to recognize any way that (68) could be true (where the pronouns are coreferent), the discussion concludes that (60) could very well be the unspoken counterpart to (68). For those like me, an exaggerated demonstration of them in (68) can also be taken to demonstrate a shift in frame of reference that makes the sentence true, which cannot be done for (60). To distinguish the spoken adrs of (68) from those that might occur unspoken as in (76) for (60), (76) They first are not them first, it can be said, taking in all of what has been observed in (60)–(75), that only the parameters of spoken adrs present themselves to be shifted and only spoken demonstratives shift them (to other than the speaker’s point of view). Silence consents to current parameters or defaults to the speaker’s point of view. If so, two unspoken tokens of the same adr, first, do not threaten (60) with an interpretation that could be true despite the coreference of its pronouns. Yet, it has also been allowed that the unspoken adrs could vary so that (60) is taken to be like (62): (60) They are not them. (62) They first are not them second. A shift in the frames of reference has just been ruled out for both sentences; but, it is with respect to the same frame of reference that (62) and the like express truisms— that what is first is not second, what is on the left is not on the right, and so on. Of course (60) says nothing of the sort despite the unspoken adrs that make it resemble (62). Even for a space, frame of reference, and point of view that have been fixed, adrs might not express a determinate thought unless a parameter fixing a comparison
780
Chapter 15
class is also understood. Recall from section 9.4.0 that in addition to these parameters, natural language left proves to be a comparative that means “more leftward,” for which an additional parameter ζ’ for the comparison class is provided in ‘left(e, ζ, f, ζ’)’, which I have more recently suppressed:12 (77) left(e, α, f, ζ′) iff ∃α′(At(α,e,t(e),f) & At(α′,ζ′,t(e),f) & r(α) * cosine(alt(α)) * cosine(az(α)) < r(α′) * cosine(alt(α′)) * cosine(az(α′)) (78) left[E,,F, Ζ] ↔df ∀e(Ee → [∃α:Aα][∃f: Ff][∃ζ: Ζζ](left(e, α,f, ζ) & [∀ζ: Ζζ](∃αAt(α,ζ,t(e),f) → left(e, α,f, ζ))) & ∀α(Aα → [∃e:Ee][∃f: Ff][∃ζ: Ζζ](left(e, α,f, ζ) & [∀ζ: Ζζ](∃αAt(α,ζ,t(e),f) → left(e, α,f, ζ))) & ∀f(Ff → [∃e:Ee][∃α:Aα][∃ζ: Ζζ] (left(e, α,f, ζ) & [∀ζ: Ζζ](∃αAt(α,ζ,t(e),f) → left(e, α,f, ζ))) & ∀ζ(Ζζ → [∃e:Ee][∃α:Aα][∃f: Ff] (left(e, α,f, ζ) & [∀ζ: Ζζ](∃αAt(α,ζ,t(e),f) → left(e, α,f, ζ))) The comparative meaning of left becomes apparent in a featureless landscape absent determinate celestial meridian, and the infelicity that then results when there is nothing as imagined for (79) to be to the left of: (79) M (80) #The left object. #The object on the left. (81) The right-leaning object. The same featureless landscape, given the orientation of the observer, supports reference to the more leftward object even absent an absolute left side: (82) M1 M2 (83) The left object is Messier Number 1. Here to be on the left is for E to be left of Ζ, ‘left[E,,F, Ζ]’ for some determined Ζ. Similarly, to be first, ‘first[E,,F,Ζ]’ is for E to be presented first among Ζ, for some determined Ζ. With demonstration of an intended space and frame of reference and therefore of its contents, the intended comparison class is often accommodated: a sock is first among the other socks there, or to left of the space’s centerline if the space demonstrated has one, as assumed for the photo examples of the preceding section, where the parameter for a comparison class was suppressed. Absent implicit demonstration or specific reference of some other kind, a speaker’s point of view and frame of reference alone make available no default for this parameter. It is obviously so that The sock is first never expresses the thought that the sock is first among (occupants of) regions addressable from the speaker’s point of view and frame of reference, which are too numerous to be coherently
The Ordered-Pair Illusion
781
ordered. On the other hand, it would not be flat-out incoherent for The sock is on the left to mean that it is left of the meridian through GMT in the speaker’s frame of reference, but I take it that the lexical item left expresses a navigational relation to landmarks, that is, to features of the space under observation or to events occurring within it that survive variation in frames of reference for that space, such as the actual physical perimeter of a photo or even a gesture—a wave of the hand—which creates an event as a stable feature of the spatiotemporal landscape with respect to which space can be partitioned.13 Again, absent such a gesture or reference to more permanent landmarks, there is no default understanding of what the socks could be left of. Similarly, across the vocabulary of spatial orientation and navigation, the parameter fixing the comparison class is without a default value and therefore depends on some specific act of reference for a fix. The unspoken and spoken adrs in (60) and (62) now differ—it has been alleged—in that only spoken adrs present their parameters for specific acts of reference. In failing to share an interpretation with (62), (60) runs up against a slight extension to this condition. Notice that the determinate thoughts, the truisms expressed by (62) and (84), are not made explicit until the generic quantification over spatiotemporal locations or contexts in (85) and (86) is supplemented as in (87) and (88): (84) They on the left are not them on the right. (85) Anywhere, they first there are not them second there. (86) Anywhere, they on the left there are not them on the right there. (87) Anywhere, for any things there, they first there among those things there are not them second there among those things there. (88) Anywhere, for any thing there, they there on the left of that thing there are not them there on the right of that thing there. The parameters of unspoken adrs are not accessible to independent quantification or reference of any kind, on account of which (60) with adrs unspoken never acquires the interpretations on display in (87) and (88) and otherwise available to (62) and (84).14 In a coordination of nominals, recall that the unspoken adrs in (57) are as effective as the spoken ones in (21), (89), and (90) in deriving an interpretation that is true of the same two switching positions: (21) The sock on the left and (the) sock on the right in photo 1 are not the sock on the left and (the) sock on the right in photo 2. (89) The sock first and (the) sock second in photo 1 are not the sock first and (the) sock second in photo 2.
782
Chapter 15
(90) The heavenly body on the left and (the) heavenly body on the right tonight are not the heavenly body on the left and (the) heavenly body on the right a fortnight ago. (57) The pitcher and (the) catcher in Game 1 are not the pitcher and (the) catcher in Game 2. The comparison class is fixed here neither by (implicit) demonstrative reference nor by default to some feature of the speaker’s point of view. It is delivered rather by explicit linguistic antecedent:15 (91) The sock on the left and (the) sock on the right of each other in photo 1 are not the sock on the left and (the) sock on the right of each other in photo 2. The sock on the left and (the) sock on the right of a point between them in photo 1 are not the sock on the left and (the) sock on the right of a point between them in photo 2. (92) The sock first and (the) sock second among them in photo 1 are not the sock first and (the) sock second among them in photo 2. (93) The heavenly body on the left and (the) heavenly body on the right of each other tonight are not the heavenly body on the left and (the) heavenly body on the right of each other a fortnight ago. The heavenly body on the left and (the) heavenly body on the right of a point between them tonight are not the heavenly body on the left and (the) heavenly body on the right of a point between them a fortnight ago. Linguistic antecedents thus appear to be proximate enough for unspoken anaphora with unspoken adrs. As a point of grammar, the same should be allowed to adrs spoken or unspoken that occur with simple nominals: (94) Theyi first among themi are not themi second among themi. (95) Theyi on the left of themi are not themi on the right of themi. The resulting sentences are as good and true as any other asserting that they are first among themselves or to the left of themselves. If such analytically false implications can be tortured out of (94) and (95), with their overt reference to the comparison class, they certainly remain out of reach for (96) and (97): (96) Theyi first are not themi second. (97) Theyi on the left are not themi on the right. Since (89) and (90) manage well enough to find local antecedents for their implicit parameters, it is unlikely to be a point of grammar that keeps (96) and (97) from meaning what (94) and (95) might under duress. It must rather be that what is taken up as implicitly understood and unspoken is less tolerant of nonsense than what can be conveyed in words. Whatever else is to be said about why (96) and (97) fall short
The Ordered-Pair Illusion
783
of (94) and (95), it cannot be denied that the adrs first, second, left, and right occur there as pronounced. But then it cannot be a complaint against an analysis of (60) that the adrs unspoken there are no better than (96) or (97) in expressing the incoherent thought that only (94) and (95) might. In summary, it appears that the grammatical distribution of spoken and unspoken adrs proceeds without taking any notice of coordination structures per se, modifying nominals wherever they are found. Independent conditions on the use of silence to fix reference or shift parameters and on the semantic coherence of what is left implicit conspire so that the use of unspoken adrs is effective and their presence felt only, as it happens, when nominals are coordinated. 15.1
Scenes in the neighborhood
The adverbialized singular descriptions, deriving as it were while alone on the left and while alone on the right, interact with Tense exactly as in section 10.3, where the contrasts in (252)–(253) and (256)–(257) illustrate that to be so-and-so at time t locates some so-and-so-ing at t. That is, the adverbial derived from the predicate nominal and not the one derived from the subject nominal qualifies the time of the event indicated by Tense and the temporal adverbial: (252) T At Halloween, Jules Bagwell was (already) Julia Bagwell. (253) F At Halloween, Julia Bagwell was Jules Bagwell. (256) T At the mayor’s press conference, Superman was Clark Kent. (257) F At the mayor’s press conference, Clark Kent was Superman. (317) [∃E: at the mayor’s press conference] (DP [℩E: Superman]-now-W-Theme … -Past[E]- … [DP [℩E: Clark Kent]-W-Theme- Be …) Recall (see (317)) that a time machine intervenes to travel from the events as described by the subject nominal to a time described by temporal adverbial, Tense, and predicate nominal. The same has to be said for (98), which felicitously denies that the skaters remain on the same side between cycles: (98) The left (skater) and (the) right skater in the current semicycle were not the left (skater) and (the) right skater in the previous semicycle. There is a scene, while one skater is alone on the left and the other alone on the right, surveying the current semicycle, from which one travels in time to a scene of the previous semicycle where the same participants were, while again one was alone on the left and the other alone on the right. What (98) denies, when taken to be true, is the similar orientation of these two scenes: the left and right skater in the current semicycle were not the left and right skater in the previous semicycle similarly oriented.
784
Chapter 15
In contrast to (98), (99) in the present tense is either infelicitous or just plain false: (99) #The left (skater) and (the) right skater in the current semicycle are not the left (skater) and (the) right skater in the previous semicycle. Analogous to (98), the predicate nominals should describe a present, given present Tense, while one skater is now alone on the left and another now alone on the right all impossibly of a past semicycle. An alternative analysis to consider construes the predicate nominals as the sometime (if not now) left skater and sometime (if not now) right skater in the previous semicycle, referring, that is, to the one skater who is now wherever and has been on the left at some other time that is the previous semicycle and to the one skater who is now wherever and has been on the right in the previous semicycle. But this describes a current scene where the skaters are exactly where the subject nominal describes them to be, too. The similar orientation of these scenes thus cannot be denied nor can the identity of the participants themselves, falsifying (99) altogether. The examples that launched section 15.0 now look tamer: (5) An ill-matched pitcher and catcher started for the Red Sox yesterday, and another equally ill-matched pitcher and catcher started for them tonight. (6) #An ill-matched pitcher and catcher were on the disabled list yesterday, and another equally ill-matched pitcher and catcher were on it tonight. For (5) to be a true report of the same two ballplayers switching positions, the events yesterday and tonight must corroborate an apparent nonidentity, which is rather a tacit denial of similar orientation: (100) The ill-matched pitcher and catcher yesterday are not the equally illmatched pitcher and catcher tonight. These events yesterday and tonight must be precise enough to support scenes of them pitching and catching that differ in their orientation, and these scenes, due to the adverbialized descriptions, must in turn frame the matrix events: while pitching and catching in one scene yesterday, they started yesterday, and while pitching and catching in the other scene tonight, they started tonight. This last condition fails in (6) absent any sense in which while pitching and catching, they are on the disabled list. Sentence (6) thus fails as report of Tim Wakefield and Doug Mirabelli being on the disabled list for two games. The sentence could however be true if, for example, pitcher Josh Beckett and catcher Jason Varitek were on the disabled list yesterday and pitcher Daisuke Matsuzaka and catcher Jason Varitek were on it tonight. Beckett and Varitek are always other than Matsuzaka and Varitek, without appeal to their participation in any particular scenes, and while being a sometime (if not then) pitcher and catcher, Beckett and Varitek were on the disabled list yesterday, and
The Ordered-Pair Illusion
785
while also being a sometime pitcher and catcher, Matsuzaka and Varitek were on the disabled list tonight. It was observed that (8) also fails, in contrast to (5), as a report of players switching positions from one start to the next: (8) #An ill-matched two batterymates started for the Red Sox yesterday, and another equally ill-matched two batterymates started for them tonight. Both yesterday and tonight, Wakefield and Mirabelli start while being batterymates (i.e., while pitching and catching). As already seen in section 15.0, there is however no sense in which (101) is true of them, whether similar orientation is invoked or not, and thus there is no sense in which these batterymates yesterday are other batterymates tonight: (101) F The ill-matched batterymates yesterday are not the ill-matched batterymates tonight. In the explanation of (5)–(8) just rehearsed, I appeal to the logical form of sentences such as (100) and (101) discussed in earlier sections. It is a leap of faith to recognize that (100) and (101) figure in such an explanation, in particular in the interpretation of other in (5)–(8). First, I trust that whatever has been said about similar orientation and the tacit adverbial expressing it in (100) and (101) could be said with equal merit of (102) and (103), although I offer no analysis of the comparative construction: (102) The ill-matched pitcher and catcher yesterday are other than the equally ill-matched pitcher and catcher tonight similarly oriented. Cf. The left sock and right sock in photo 1 are other than the left sock and right sock in photo 2. (103) F The ill-matched batterymates yesterday are other than the ill-matched batterymates tonight (similarly oriented). Cf. The socks in photo 1 are other than the socks in photo 2. Second, the token of other in (5) is occasion enough, I assume, for an unspoken comparative clause containing adverbial similarly oriented just as in (102).16 While baseball is a robust game and its neighboring events vivid on the field, the neighborhood relation is not confined to the spatiotemporal but ranges as much as the concept of nearness itself does. Suppose for example that Mr. Thing and Mr. Entity own their own homes on adjoining lots:17 (104) A homeowner and brotherly neighbor cleared one home’s yard, and another homeowner and brotherly neighbor cleared another home’s yard. (105) #A homeowner and brotherly neighbor tore down one fence between them, and another homeowner and brotherly neighbor tore down another.
786
Chapter 15
Sentence (104) is a fair report that, say, Mr. Thing and Mr. Entity cleared Mr. Thing’s yard, and then Mr. Entity and Mr. Thing cleared Mr. Entity’s yard. Clearing up a home’s yard is within the neighborhood of that home’s ownership and of neighborliness to that home’s owner and not within the neighborhood centered on the circumstances surrounding another home. In contrast, (105) fails as a report that they first tore down the front-yard fence that stood on the property line between their lots and then tore down the backyard fence that also stood on the property line. In what sense is tearing down a fence equidistant from both homes within the purview of one’s home ownership and not the other’s?18 The reasoning is as before—the homeowner and brotherly neighbor here are other than the homeowner and brotherly neighbor there. Here is where Mr. Thing is rightfully the homeowner and Mr. Entity, the brotherly neighbor; there is where their rights are reversed. According to (104), here is where Mr. Thing’s yard is cleared, and there is where Mr. Entity’s is. The metaphoric usage is clear enough for the yards’ clearings to be so located. But for (105), here is the location for one fence’s being torn down, and there the location for the other, and there is no basis in fact or metaphor to discern a difference in their locations with respect to property rights. All these sentences with left sock, right sock, pitcher, catcher, homeowner, and brotherly neighbor rely on nominals that on their own relate different participants to different events: the pitcher in the first game is not the pitcher in the second game, and so on. Nevertheless, the effects of coordination persist into (106) and (107), even though the first game’s Jimmy Connors is the same as the second game’s Jimmy Connors in (107), and the very same chessmen, one bishop, one rook, and one pawn, may be the three pieces playing all the checkmates and stalemates: (106) A bishop, a rook, and a pawn will always checkmate in ten moves, but a rook, a pawn, and a bishop only stalemate. (107) Jimmy Connors and Ivan Lendl played the first game, but Ivan Lendl and Jimmy Connors played the second. In such cases, the apparent effects of coordination derive from a silent spatial vocabulary appropriate to the space and frame of reference under consideration assigning to each participant its own address. To have said something true given the universal always in (106), my son’s chess tutor must have had three positions in mind fixed by two coordinates, file a ≤ fi ≤ h and rank 1 ≤ ri ≤ 8, and he must have meant in uttering the first clause of (106) something like (108): (108) A bishop at f1r1, a rook at f2r2, and a pawn at f3r3 will always checkmate in ten moves; … Although nothing was spoken aloud about these three positions, I may have seen them demonstrated or believe them to have been demonstrated just at the moment
The Ordered-Pair Illusion
787
before I intruded on the lesson to see and hear the second clause of (106). Without any further gesture or demonstration from the tutor, I then take him to have intended with (106) something like (109) commenting on the result of a rotation among the three pieces: (109) A bishop at f1r1, a rook at f2r2, and a pawn at f3r3 will always checkmate in ten moves, but a rook at f1r1, a pawn at f2r2, and a bishop at f3r3 only stalemate. There is a grammar to such uses of silence. Without further indications, the three separate positions required for the second clause are taken to be anaphoric to those of the first clause in entirely parallel fashion, so that whatever spatial predicate has been understood to tacitly occur in the first conjunct of the first clause recurs in the first conjunct of the second clause, and so on. Thus, without any new gesture or demonstration to redirect the second clause, (110) can only assert incoherently that the same configuration checkmates and stalemates: (110) #A bishop, a rook, and a pawn will always checkmate in ten moves, but a bishop, a rook, and a pawn only stalemate. What is rejected is to parse (110) as (111), which would make it true and equivalent to (109), but would violate the required parallelism: (111) *A bishop at f1r1, a rook at f2r2, and a pawn at f3r3 always checkmate in ten moves, but a bishop at f3r3, a rook at f1r1, and a pawn at f2r2 only stalemate. The parallelism governs anaphora relating the silent spatial predicates of the second clause to those of the first. The chess tutor may instead sweep up the pieces from the board and felicitously utter (110) while he repositions them for the second clause, without intending any anaphoric relation between the silent predicates of the two clauses. Similar remarks rule out (112) when the very same bishop and rook are involved in both endgames: (112) #A well-positioned bishop and rook checkmated in ten moves, and another even better positioned bishop and rook checkmated in five. If no silent adrs are understood, there is no denying the similar orientation of the chess pieces, and there is no denying their identity. Yet, even if it is intended that a bishop here and a rook there checkmated in ten moves, those same positions are intended in the second clause absent gesture otherwise, and there is again no denying their similar orientation or their identity. As chapter 13 introduced, the disjunctive or collective interpretation that allows [DPD NP1 and NP2] in (1) to refer to some things that are NP1 but not NP2 and things
788
Chapter 15
that are NP2 but not NP1 is derived here within DP as it is elsewhere, as in (3) from chapter 2: (1) The reporters and superheroes that stalked Metropolis read The Daily Planet. (3) Twenty thousand students surrounded the Pentagon and were crowded into the Mall. A sentential connective and is no instrument of collective reference, which instead requires the intervention of pronouns referring to events and relations between these events and their participants: (4) Twenty thousand students W[E1,X] … [℩E2: ΦΨ]O[E1,E2] (Φ∃X surround the Pentagon) and (Ψ∃X be crowded into the Mall). (5) … [The X: ∃E1 W[E1,X] [℩E2: ΦΨ]O[E1,E2] (Φ∃X∃E2 reporter[E2,X]) and (Ψ∃X∃E2 superhero[E2,X])] …; or, alternatively, … [The X: [℩E2: ΦΨ] W[E2,X] (Φ∃X∃E2 reporter[E2,X]) and (Ψ∃X∃E2 superhero[E2,X])] … However this collective reference is brought about—as advocated here or by other means—reference to the same two ballplayers throughout (5) does not help much with how they can be other than they were from one start to the next or with how they fail at it in (6) from one disabled list to the next. Moreover, if collective reference is the same throughout, it wants explanation why the same two ballplayers need the coordinate description in (5) rather than the simple one in (8) to be other than they were from one start to the next: (5) An ill-matched pitcher and catcher started for the Red Sox yesterday, and another equally ill-matched pitcher and catcher started for them tonight. (6) #An ill-matched pitcher and catcher were on the disabled list yesterday, and another equally ill-matched pitcher and catcher were on it tonight. (8) #An ill-matched two batterymates started for the Red Sox yesterday, and another equally ill-matched two batterymates started for them tonight. It is this last observation that sorely tempts the conclusion that an ill-matched two batterymates and an ill-matched pitcher and catcher do not refer to the same, the latter referring to something else like an ordered pair as the result of a special meaning for and. The novel meaning afforded (5) indeed rests on the coordinate structure of the sentence, but it is derived without novelty in the meaning of and or in the things the nominals refer to. It is just that a coordination of two conjuncts affords two occasions, one per conjunct, to make separate comments in the form of locatives, silent or overt, about the ballplayers’ locations. The simple description in (8) has a place for only one, and anything said there about the location of the batterymates applies to them both equally. Without a robust and rather particular
The Ordered-Pair Illusion
789
notion of frame of reference and without three silent locatives fixing locations there, a sentence like (106) has hardly a prayer to convey a truth. In the presence of such locatives silent or overt, assertion that the ballplayers here are other than the ballplayers there comes to deny their similar orientation rather than their numeric identity (see (102) /(103)). As in section 15.0.1, their similar orientation can be truthfully denied just in case they are described by separate conjuncts putting them in different locations. Now as the nominal descriptions become ever more elaborate in support of the assertion that things arranged as they are in one scene are not in the same orientation as they are in another scene, the events described, that is, the scenes themselves, become ever more narrow and precise in how finely drawn their neigborhoods are. Adverbializing these descriptions then locates the matrix events within the neighborhoods so described, the sequencing of events (Doetjes and Honcoop 1997, 291ff.) that discriminates (5) from (6).
16
QED
And is an unambiguous, univocal sentential connective such that Σ satisfy ⌜Φ and Ψ⌝ if and only if Σ satisfy Φ and Σ satisfy Ψ.
The rest has been commentary, some grammar for Eventish, and the novelty of Cinerama Semantics, in which all discourse is narration for spatial orientation and visual navigation. Eventish—the language of thought—has supermonadicity, subatomic descriptive event anaphora, adverbialization, and a canonical sentence structure based on an unspoken vocabulary and on a syntax that includes pervasive Right-Node Raising. The Eventish vocabulary includes thematic relations to events (e.g., ‘W[E,X]’), various bridge relations between them (e.g.,‘Cause[E,E′]’), and a vocabulary for spatial orientation and navigation about scenes, resolution, frames of reference, and addresses within them.
Appendix 1 Descriptive Anaphora and Nonmaximal Reference under Selective Perspectives
To derive the nonmaximal reference on display in (1) and denied to (2), it has been said (Heim 1982; Kamp 1981) that the indefinite description in (1), in contrast to the decreasing quantifier confined to its sentence in (2), has a range that includes the entire discourse within its scope as in (3), perhaps favoring (3c) for its more consistent treatment of the pronouns in (1) and (2) as nearly the same definite description:1 (1) Two men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. (2) *Few men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. (3) a. [∃X: Two men[X]] ||: X came to the office today. X tried to sell encyclopedias. Perhaps there were even others than X who did the same. :|| b. [∃X: Two men[X]] ||: X came to the office today. [℩Y: Y=X] Y tried to sell encyclopedias. Perhaps there were even others than X who did the same. :|| c. [∃X: Two men[X]] ||: X came to the office today. [℩Y: Y=X & men who came to the office today[Y]] Y tried to sell encyclopedias. Perhaps there were even others than X who did the same. :|| Such formal license to indefinite descriptions will, however, forfeit explanation of both those contexts where they do antecede pronouns bearing maximal reference (4)–(6) and its suspension when these contexts are provided yet further context (e.g., (7)): 2 (4) John owns some sheep, and Harry vaccinated them. (Evans 1977)
794
Appendix 1
(5) Some children are allergic to cats; cats have an adverse affect on them. (Sommers 1982) (6) Leif has a chair. It is a Regency or Biedermeier. (After Kadmon 1990) (7) Leif has a chair. It is a Regency or Biedermeier. He has another chair. It is a Biedermeier. His other chairs are in various European styles. (Schein 1993, 224) In contrast to (1), which escapes the implication that all the men visiting the office today tried to sell encyclopedias, (4)–(6) seem to imply respectively that all the sheep John owns were vaccinated, all the children allergic to cats suffer an adverse effect, and the unique chair that Leif owns is a Regency or Biedermeier. The suspension of maximal reference is contingent on the accessibility to speaker and hearer (imaginary or real) of a perspective on the events reported that is itself selective in its purview (Schein 1993, 219–237). On the one hand, witness to events at the office is rarely witness to them all. The encyclopedia salesmen may very well have been all the men who came to the office today under that perspective without being all the men who came to the office. On the other hand, it is not obvious what neutral stance affords a perspective on states of sheep ownership, allergy to cats, or chair possession that isolates John’s ownership of some sheep from others, some children’s allergy from others’, or Leif’s having one chair from his having others. Under what conditions for the speaker is the chair Leif has here not all the chairs he has? In a panoramic survey of the salon, (7) may narrate three successive scenes, the first two, each of a solitary chair, and the third of the chairs in various styles: (8) ∃Π ||: Leif has a chair[Π]. [℩x: chair(x) & Leif has a chair[x,Π]] x is Regency or Biedermeier[Π]. [∃Π′: next[Π, Π′]] Leif has another chair[Π′]. [℩x: chair(x) & Leif has another chair[x,Π′]] x is Regency[Π′] …:|| The use of a singular pronoun does not imply Leif’s ownership of a unique chair, but it does imply the uniqueness of the chair in its scene; (7) would not be coherent narration for a survey that collapses the first two into a larger scene of two chairs. That is, the singular pronoun remains a singular definite description relativized to the perspective at hand—the chair, in this scene, that Leif has—and if it is relativized only to this larger scene or perspective, nonmaximal reference fails, as it should here. Example (7) and the following examples are instances of “the problem of indistinguishable participants” (Heim 1990)—in one discourse, multiple singular pronouns referring to multiple participants the description of which, such as a chair of Leif’s, does not appear to distinguish any one of them from the others. The answer is that
Descriptive Anaphora and Nonmaximal Reference
795
successful nonmaximal reference is dependent on a selective perspective.3 It is a small thing to imagine that narrative continuity is not only a temporal progression, tensed sentence to tensed sentence, but a richer spatiotemporal progression—that narrative is cinematic. Even in narratives that are not of the speaker’s own witness or perspective, quantification over distant selective perspectives remains the only route for singular reference to indistinguishable participants: (9)
Once upon a time no one knows when, there was a twin. She had an identical twin. She was really, really identical to her. She loved her as she loved herself. But she did not love her nearly as well. She envied her her beauty, which she would have all to herself.
(10) #Once upon a time no one knows when, there was a tuplet. (*At some time then,) She had an identical tuplet. She was really, really identical to her. She loved her as she loved herself. But she did not love her nearly as well. She envied her her beauty, which she would have all to herself. (11) Once upon a time no one knows when, there was a tuplet. (At some time then,) An idiot prince visiting the castle identified her with an identical tuplet. She was really, really identical to her, and so she forgave that he loved her and not herself, because he was an idiot, after all. None of these stories implies a world lacking in identical n-tuplets, and yet no illustrator can illustrate the first sentence with an image of more than one twin or tuplet. The unknown time and place of this illustration, like any spatiotemporal region, is one at which to have any identical tuplets is to have them all, and thus singular reference in (10) implies just the one, making it odd to have described as a tuplet a tuplet known in fact to be a twin.4 The singular reference will carry this implication if the initial spatiotemporal-frame adverbial and presentational sentence introduces a perspective on just one tuplet, as the illustrator would have it, and the second sentence’s tense is itself a definite pronoun anaphoric to the events antecedently described, without introducing a new perspective, and the same is true of the third sentence: (12) ∃Π0 ||: [∃t0 : Once(t0) upon a time no one knows when] there was a tuplet[t0, Π0]). [℩t0: time(t0) no one knows when & ∃E0 there was a tuplet[E0, t0, Π0]] [℩x0: ∃E0 there was a tuplet(x0) [E0, t0, Π0]] ∃E1 x0 had an identical tuplet[E1, t0]]. [℩t0: time(t0) …] [℩x0: ∃E0 there was a tuplet(x0) [E0, t0, Π0]] (‘she’) [℩x1: ∃E1 x0 had an identical tuplet(x1)[E1,t0]]] (‘her’) ∃E2 x0 was really, really identical to x1 [E2, t0]. …:|| 5
796
Appendix 1
In summary, a generalization: (13) Nonmaximal reference relative to Π0 Given an initial, stationary perspective Π0, reference that is nonmaximal within Π0 is acceptable for a pronoun (or definite description) the antecedent of which is extrasentential only if a felicitous indefinite spatiotemporal-frame adverbial intervenes to describe a spatiotemporal frame within which the reference is maximal. As noted above, another route to singular, nonmaximal reference allows for a successive attribution of states that are not themselves localized in spacetime if the successive attributions narrate a series of scenes with perspectives within which singular reference is maximal: (14) Once upon a time no one knows when, there was a tuplet. She had an identical tuplet. She was really identical to her. She had another identical tuplet. She was really, really identical to her. She even had another identical tuplet, and she was as identical to her as she was to all her identical tuplets. As perspectives are themselves abstract, one can plant one anywhere with as narrow a field of view as needed, as one does with the initial perspective Π0 on a presentation of a tuplet. Moreover, the continuity conditions from one scene to the next are slight as (14) could just as well narrate a slide show. There are therefore at least as many perspectives on solitary tuplets as there are tuplets, and so granting discourse scope to existential quantification over them would again defeat any condition on singular reference: (15) * ∃Π0∃Π1 … ||: [∃t0 : Once(t0) upon a time no one knows when] there was a tuplet[t0, Π0]). [℩t0: time(t0) no one knows when & ∃E0 there was a tuplet[E0, t0, Π0]] [℩x0:∃E0 there was a tuplet(x0)[E0, t0,Π0]]∃E1 x0 had an identical tuplet[E1, t0, Π1]]. [℩t0: time(t0) …] [℩x0: ∃E0 there was a tuplet(x0) [E0, t0, Π0]] (‘she’) [℩x1: ∃E1 x0 had an identical tuplet(x1)[E1,t0, Π1]]] (‘her’) ∃E2 x0 was really, really identical to x1 [E2, t0]. …:|| Rather, after the initialization of a narrative point of view for the discourse, perspective Π0, any succession of scenes is indexical to the speaker’s proximate experience and nonmaximal reference is achieved by definite description relativized to such scenes:
Descriptive Anaphora and Nonmaximal Reference
797
(16) ∃Π0 ||: [∃t0 : Once(t0) upon a time no one knows when] there was a tuplet[t0, Π0]). (∃Π [℩x0: ∃E0 there was a tuplet(x0) [E0, t0, Π0]] here now[x0, Π].) [℩t0: time(t0) no one knows when & ∃E0 there was a tuplet[E0, t0, Π0]] [℩x0: ∃E0 there was a tuplet(x0) [E0, t0, Π0]] ∃E1 x0 had an identical tuplet[E1, t0 ,Π1]]. (∃Π [℩x1: ∃E1 x0 had an identical tuplet(x1)[E1,t0, Π1]] here now[x0, Π].) [℩t0: time(t0) …] [℩x0: ∃E0 there was a tuplet(x0) [E0, t0, Π0] & ∃Π before[x0, Π]] (‘she’) [℩x1: ∃E1 x0 had an identical tuplet(x1)[E1,t0, Π1] & ∃Π later[x1, Π]] (‘her’) ∃E2 x0 was really, really identical to x1 [E2, t0]. …:|| Once upon a time no one knows when, there was a tuplet (seen) here now. She had an identical tuplet (seen) here now. The tuplet (seen) before was really, really identical to the tuplet (seen) later … Singular reference succeeds in (16) only if the first and second tokens of ‘here now’ denote distinct spatiotemporal regions and locate distinct scenes or perspectives, and before and later denote relative position in a series of scenes spanning this discourse. These conditions set in motion some kind of panorama, without which (14) is incoherent. That shifts in perspective are induced only by indexical shifts in perspective follows from two considerations. The first is the formal constraint that excludes from discourse scope multiple quantifiers over perspectives, as in (15). The second is the substantive remark that the perspectives initialized, ‘∃Π0 ||:’, are not themselves in motion but stationary with respect to location, orientation, field of view, and magnification. Thus any scene change from one sentence to the next in the course of narration is done by definite description of the next perspective. But, given any perspective ∃Π0, which is the next perspective Π1? Definite reference to a later perspective fails unless indexical or demonstrative reference supplements it as in (16). In contrast to the state of having an identical tuplet, events of being identified with one are local to time and place, and thus the identical tuplet identified with at this time and place may easily not be the one identified at any other: (17) ∃Π0 ∃t0 ∃t1 …||: [∃t : t = t0 & Once upon a time(t) no one knows when] there was a tuplet[t,Π0]. [Some t′: t′ = t1 & time(t′) & [℩t: then[t, t0]]Within(t′, t)] An idiot prince visiting the castle identified [℩x: tuplet[x, Π0]] with an identical tuplet[t′]. [℩x: tuplet[x,Π0]] was really, really identical to [℩x: [Some t′: t′ = t1 & time(t′) & [℩t: then[t, t0]]Within(t′, t)]. …:|| ‘An idiot prince visiting the castle identified [℩x: tuplet[x, Π0]] with an identical tuplet(x)[t′]]. She was really, really identical to her, and so she forgave that he loved her and not herself, because he was an idiot, after all.’
798
Appendix 1
The temporal-frame adverbial, spoken or unspoken, at some time then, framing the identification of the first tuplet affords a frame for singular reference: She, the tuplet of Π0, was really, really identical to her, the identical tuplet that at some time t1 an idiot prince identified the tuplet of Π0 with. But, strictly speaking, such a relativization to only a time t1 is still too strong. For suppose that at any time an idiot prince identifies her with one identical tuplet, an idiot prince, perhaps the same one, identifies her with another identical tuplet. Singular reference to the identical tuplet so identified at t1 fails. It is rather necessary to relativize to spatiotemporal regions fine-grained enough to resolve one event of identification from another: (18) ∃Π0 [∃σ0 : spacetime(σ0)] [∃σ1 : spacetime(σ1)] …||: [∃t: t = time(σ0) & Once upon a time(t) no one knows …] there was a tuplet[t,σ0,Π0]. [Some t′: t′ = time(σ1) & [℩t: then[t, time(σ0)]]Within(t′, t)] An idiot prince identified [℩x: tuplet[x, Π0]] with an identical tuplet[t′, σ1]. [℩x: tuplet[x,Π0]] was really, really identical to [℩x: [Some t′: t′ = t1 & time(σ1) & [℩t: then[t, time(σ0)]]Within(t′, t)] An idiot prince identified [℩x: tuplet[x, Π0]] with an identical tuplet(x)[t′, σ1]] …:|| ‘An idiot prince identified [℩x: tuplet[x, Π0]] with an identical tuplet(x)[t′, σ1]]. She was really, really identical to her, and so she forgave that he loved her and not herself, because he was an idiot, after all.’ The spatiotemporal regions are no closer to the speaker than the perspective Π0, equally out of reach of demonstrative reference, and are thus also to be introduced by existential quantification. They too must include the entire discourse within their scope, since the identical tuplet is unique only relative to σ1, as is again evident in the following story: (19) Once upon a time no one knows when, there was a tuplet. An idiot prince visiting the castle mistook her for an identical tuplet. She was really, really identical to her. Later, he again mistook her for an identical tuplet. She was really, really identical to her too. Later still, he again mistook her for an identical tuplet. She was alas as identical to her as she was to all her identical tuplets. We do not know how many identical tuplets she has been mistaken for, perhaps as many as three if she and her siblings are quadruplets or better. If so, the pronoun referring to the identical tuplet she is mistaken for in the second mistake, ‘[℩x: identi-
Descriptive Anaphora and Nonmaximal Reference
799
cal tuplet[x, σ3]]’ in (20), should imply uniqueness only for that occasion, as it does with the quantification over spatiotemporal regions at the discourse level: (20) ∃Π0∃σ0∃σ1∃σ2∃σ3∃σ4∃σ5∃σ6 ||: [∃t : t = time(σ0) & Once upon a time(t) no one knows …] there was a tuplet[t,σ0,Π0]. [Some t′: t′ = time(σ1) & [℩t: then[t, time(σ0)]]Within(t′, t)] An idiot prince mistook [℩x: tuplet[x, Π0]] for an identical tuplet[t′, σ1]. [∃t′: t′ = time(σ2) & [℩t: then[t, time(σ1)]]Within(t′, t)] [℩x: tuplet[x,Π0]] was identical to [℩x: identical tuplet … [x, σ1]] [t′, σ2]. [∃t′: t′ = time(σ3) & Later(t′, time(σ2))] he mistook [℩x: tuplet[x, Π0]] for an identical tuplet[t′, σ3]. [∃t′: t′ = time(σ4) & [℩t: then[t, time(σ3)]]Within(t′, t)] [℩x: tuplet[x,Π0]] was identical to [℩x: identical tuplet … [x, σ3]] [t′, σ4]. [∃t′: t′ = time(σ5) & Later(t′, time(σ4))] he mistook [℩x: tuplet[x, Π0]] for an identical tuplet[t′, σ5]. [∃t′: t′ = time(σ6) & [℩t: then[t, time(σ5)]]Within(t′, t)] [℩x: tuplet[x,Π0]] was identical to [℩x: identical tuplet … [x, σ5]] [t′, σ6]. :|| If, instead, the pronoun were translated the identical tuplet for whom there is a time (and place) later where he mistook her for her, it would incorrectly imply that the same one identical tuplet was whom she was mistaken for in both later mistakes. Nonmaximal reference requires that some parameter to which the pronoun’s reference is relativized attain discourse scope, which is the finding of Heim 1982 and Kamp 1981. If nonmaximal reference is however to remain contingent on the accessibility of a selective perspective, then only the initialization of a perspective and quantification over spatiotemporal regions have scope over the entire discourse. There remains however something too loose about this disjunctive restriction to perspective initialization or spatiotemporal quantification. Absent some indication to the contrary, the perspective initialized is the speaker’s—absent contrary indication, the narrator is the speaker. Our monologues ‘||:…:||’ are not true if an arbitrary perspective ‘∃Π0||:…:||’ makes them true and we have neglected to say as much. Of course “once upon a time no one knows when” is just such an indication of a displaced narrator. Rather than discourse quantification over perspectives, I will assume that speakers may help themselves to an indexical predicate of perspectives, ⌜Πinitial(Πi, δ)⌝, Πi is the initial perspective of the narrator for this discourse δ. As some perspective and narrator are always present and presumed not to change in medias res, any shift to a narrator other than the speaker occurs sooner rather than later in a discourse: (21) ∃σ0 …||: [∃t : t = time(σ0) & Once upon a time(t) no one knows when] [℩Π0 : At[σ0, Π0] & Πinitial(Π0, δ)] there was a tuplet[t,Π0] …:||
800
Appendix 1
With the indexical vocabulary so enriched, the formal structure of discourse restricts discourse-level quantification to quantification over spatiotemporal regions as in (21), which is just to say that sequence of tense—or rather sequence of spatiotense— is the only grammatical relation between sentences so treated, pace Heim 1982, Kamp 1981, and subsequent literature. In (9)–(11) and (19), the shifts in perspective occur only with a new sentence. But if the problem of indistinguishable participants as it occurs within a single clause6 is to be subsumed under the same solution, perspectives shift subatomically too: (22) Once upon a time no one knows when, a twin had an identical7 twin. She was really, really identical to her … (9) Once upon a time no one knows when, there was a twin. She had an identical twin. She was really, really identical to her … It hardly matters to the narration or to the reference of the pronouns in the last sentence, whether the story begins as it did in (9) or now as it does in (22). Given Conjunction Reduction and supermonadicity, it hardly matters for the pronouns whether the story begins as in (23) or as in (24), a spoken counterpart.8 Whatever is said for (24) will apply the same mutatis mutandis to (23): (23) A triplet, an identical triplet, and another identical triplet were courted by the same idiot prince and despised what he gave her, what he gave her, and what he gave her—all for one and one for all! (24) A triplet was courted. An identical triplet was also courted, and another identical triplet was courted. They despised what was given her. They despised what was given her, and they despised what was given her. For any scene or perspective that includes both twins of (9) or (22), neither twin is the one in it that has an identical twin. Similarly, none of the triplets of (23) or (24) is the one triplet with one or more identical triplets in any scene or perspective on two or three of them. But, if such an overpopulated perspective came fully equipped as a frame of reference with addresses, relativizing to the latter would again undermine any constraints on perspective, since anything is trivially unique at the address it occupies: (25) ∃Π∃α1∃α2 ||: [Some x: twin(x) & Π[α1,x]] has [an x : identical twin[x] & Π[α2,x]]. [℩x: twin(x) & Π[α1,x]] is really, really identical to [℩x: identical twin[x] & Π[α2,x]]. :|| But the constraints on selective perspective and the failure of nonmaximal reference absent their satisfaction are just as applicable to (22) (or (23)). Whatever went on in your imagination of a context for (22), it did not include a speaker with a photograph of the twins holding hands or with them both presented to the speaker as she utters (22), although it is obvious that one is uniquely on the left and the other
Descriptive Anaphora and Nonmaximal Reference
801
uniquely on the right. In such a scene, there is still a failure of singular reference provoking the question “Which twin is being said to be really, really identical to which?”9,10 There is something incoherent in joining that perceptual experience and intention to report a being-identical-to with the two different perspectives or scenes necessary for the pronouns’ singular reference. A more likely thought behind the speaker’s utterance of (22) goes something like this: (26) Some twin [here at this twins’ party—she’s supposed to be wearing a button that says “Who am I?”] has an identical twin [I wonder if I can find her here too after I ask button twin], and she is really, really identical to her. The contrast above between events and states in licensing selective perspectives also persists among the subatomic perspectives distinguishing otherwise indistinguishable participants: (27) Some tuplet has an identical tuplet, and she is really, really identical to her. (28) Some tuplet was identified with an identical tuplet, and she is really, really identical to her. At a safe distance from family portraits, (27) can be true and felicitous of any n-tuplets, among whom only two are identical (and the rest fraternal). The sentence is however infelicitous of tuplets that include three or more identicals. After a moment’s reflection, it is clear that her, with antecedent an identical tuplet, is where singular reference fails in such circumstances. In contrast, the report (28) of an event of mistaken identity does not preclude that its victim has many identical tuplets each of which she has been mistaken for at some time or another. As before, a perspective or frame of reference needs to be of a time and place only long enough and large enough to frame the events under report, at which time and place there had better be only the two tuplets for the singular pronouns to refer to: (29) At some time and place, some tuplet was identified with an identical tuplet, and she is really, really identical to her. As far as objective conditions go, in contrast to events, spatiotemporal boundaries do not isolate the state of having one identical tuplet from that of having any other: (30) #At some place, some tuplet has an identical tuplet. (31) #At some time and place, some tuplet had an identical tuplet. If an identical tuplet is had there, they all are, offering no route to singular reference, unless again the epistemic conditions of the witness can be invoked: (32) A tuplet, as I learned meeting some of them, had an identical tuplet, and she is really identical to her. As I learned meeting some of them again, she had another identical tuplet, and she is really, really identical to her too.
802
Appendix 1
One might have thought that if anything is a single scene of two at a minimum, it is a single scene of having an identical twin or of having an identical tuplet. But singular reference to indistinguishable participants in such a scene is rejected. Under supermonadicity, there is less than meets the eye to tell apart the first sentence of (22) from the first two sentences of (9). In (22), the subject is solo in an event that could just as well be thought of as a presentational one to be related in some way to the following event of having an identical twin. In fact, it was remarked in section 1.8 that the spatiotemporal-frame adverbials in (33)–(34) like the secondary predicates apply to the events described by the predicate nominals, the being so-and-so, and not to the events that present the subject: (33) For some evenings in 1892, Venus was Hesperus (aligned with a crescent moon). For some evenings in 1892, Venus was the evening star (aligned with a crescent moon). (34) For some evenings in 1892, Hesperus was Venus *(aligned with a crescent moon). For some evenings in 1892, the evening star was Venus *(aligned with a crescent moon).11 This just corroborates the structure that singular reference to indistinguishable participants demands of (22), (27), and (28): There was Venus, and for some (but perhaps not all) evenings then in 1892, she was Hesperus; *There was Hesperus, and for some (but perhaps not all) evenings in 1892, Hesperus was Venus.12,13 The above discussion has for the sake of argument opposed the state of having an identical tuplet and the event of being mistakenly identified with one. The former is both stative and abstract—too abstract for a location distinct from the location of having two identical tuplets, if one has two. A mistaken identification, however, is both eventive and occurs at a certain spatiotemporal location. To discover the limits on spatiotemporal-frame adverbials in creating nonmaximal reference, the confound should be resolved and consideration given to concrete states with definite spatiotemporal location.14 So consider reports from a safety inspector who has inspected a container just received on the receiving dock: (35) This shipment contains glass containers. Some / a few contain lye. They are however shatterproof. (36) This shipment contains glass containers. Some / a few contain lye. If they shatter, 100 kilograms of lye will be released into the cargo bay. (37) This shipment contains glass containers. Some / a few contain lye. They contain no more than 100 kilograms of the caustic compound and occupy no more than 1 cubic meter of cargo space.
Descriptive Anaphora and Nonmaximal Reference
803
The pronouns in (35)–(37) bear maximal reference. Uttering (35), for example, would be infelicitous if the shipment contains any glass containers containing lye that are not shatterproof. In this respect, the stative (35)–(37) contrasts with the eventive (38)–(39): (38) This shipment contains glass containers. (Until they shattered,) some / a few contained lye. They were shatter-resistant. (39) This shipment contains glass containers. (Until they shattered,) some / a few contained lye. They contained no more than 100 kilograms of the caustic compound and released it into the cargo bay. The glass containers that were shatter-resistant and shattered could very well have been few of many glass containers containing lye. Even for simultaneous shatterings with some distance between them, it seems to me that the inspector may utter (38)–(39) intending to comment on just one of them, perhaps to explain next how the shatter-resistant shattered or to track its effects. That the shatterings are to be discrete in order for nonmaximal reference to be felicitous is drawn out in the following example. As in section 1.5.0, 18 barrels of the vintage’s 180 filled 5400 bottles in 75 minutes: (40) Eighteen barrels filled some bottles with Rhône varietals. The flow from all 18 barrels is simultaneous and uninterrupted for the entire 75 minutes, and as the barrels empty, all 5400 bottles fill up at an equal rate (perhaps via an impossible arrangement of hoses from barrel to bottle). There are thus no temporal intervals in which some but not all the bottles are being filled and no temporal intervals in which some but not all the 18 barrels are filling 5400 bottles. For the first spatial configuration, suppose that the bottles are in a tight 90 × 60 array. Gesturing to all this, it is said: (41) Eighteen barrels filled some bottles with Rhône varietals. They were 80% Grenache. The pronoun refers maximally to all 5400 bottles that the 18 barrels filled, and thus (41) is false if only 60 bottles contained a particular blend that was 80% Grenache, despite the eventive activity engaging all 18 barrels that they delimit. What event of filling 60 bottles is there such that the speaker could intend us to understand that its bottles were filled with 80% Grenache? Suppose instead that the 5400 bottles are clustered into cuvées and the cuvées spread out across the floor of the shed. One of the cuvées is a blend of 80% Grenache in 60 bottles. Although the speaker may be less than fully cooperative in neglecting to signal which event is the grounds for her assertion, there is now among many such events an event, at a cuvée, of the 18 barrels filling some barrels with Rhône varietals, and it can be said of its bottles that they were 80% Grenache. Under both spatial configurations, the eventive activity
804
Appendix 1
contained within the 60 bottles is what it is. The difference is that in the former, it does not constitute an event discrete enough to be one event in a domain of countable events—not even in a domain of events of fixed dimension (e.g., those that fill 60 bottles). Under the latter spatial configuration, there are as many events as there are cuvées—some of 60 bottles, others perhaps not; some 80% Grenache, others not.15 Thus, the discourse-level quantification over spatiotemporal regions licenses nonmaximal reference only if what occurs within a region is one (or more) countable events as described. Nonmaximal reference is tied to existential, count quantification over events or states. Nonmaximal reference obtains referring to the participants of one (or more) events when there are other events around that also satisfy the description. Of course, if count quantification is at issue, it is no surprise that nonmaximal reference is inaccessible (except under a panoramic survey) when describing states, as in (9)–(10) and (35)–(37). But the crucial test on which nonmaximal reference rests is not whether the state or event has spatiotemporal boundaries and not whether it is eventive or stative but whether it is one among many.16 In (43), the logical form for (42), the plural morpheme pl flags count terms17 and is tokened twice in the last two lines. It is left open which are effective and necessary. If the first is, its host reads that some events are so-and-so. If the first pl is omitted, the result says instead that some eventery such that it is some fillings is so-and-so: (42) Eighteen barrels filled some bottles with Rhône varietals. They were 80% Grenache. (43) ∃σ0∃σ1 …||: [℩Π0 : At[σ0, Π0] here(σ0) now(σ0) Πinitial(Π0, δ)] (0 [∃X : 18 barrels[X, Π0]] [∃E0 : At[E0, Π0]] (Participate[E0, X] Source[E0, X]) 0) [℩E0 : pro0][℩E1 : pro1] Cause[E0,E1] (1 [∃ l1: l1 = place(σ1) At[l1, Π0]] [∃E1 : At[E1, l1] [∃Eπ0 : At[Eπ0, Π0] pl[Eπ0]][∀e: E1e] Eπ0e ] [∃X : some bottles[X, Π0]] fill.pl with Rhône varietals[E1,X] 1) …:|| ‘There are 18 barrels here (among many others), and these 18 barrels somewhere here filled in one (or more) fillings some bottles with Rhône varietals. They, the bottles filled with Rhône varietals in those fillings, were 80% Grenache.’ The point is that the discourse may still be thought of as navigating from one spatiotemporal region to the next, leaving it to the narrative description within to say how the events should look in the regions encountered. Returning finally to the contrast between the indefinite description in (44) and the decreasing quantifier in (45), they are alike in their confinement to the first
Descriptive Anaphora and Nonmaximal Reference
805
sentence, from where they serve as the antecedents for pronominal definite descriptions: (44) Two men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. (45) *Few men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. With nonmaximal reference contingent on a selective perspective and with only indefinite spatiotemporal-frame adverbials launching them, what discriminates (44) and (45) narrows to whether the indefinite description and the decreasing quantifier play equally nice with such spatiotemporal-frame adverbials:18 (46) At some point, two men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. (47) *At some point, few men came to the office today. They tried to sell encyclopedias. Perhaps there were even others who did the same. For a given perspective Π on a day at the office,19 the preceding remarks suggest (48) for (44): (48) ∃σ0 …||: [∃t0: t = time(σ0)] [℩Π0 : At[σ0, Π0] Πinitial(Π0, δ)]two men came to the office today[t0, Π0]. [℩X: [∃t0 : t = time(σ0)][℩Π0 : At[σ0, Π0] Πinitial(Π0, δ)] two men(X) came to the office today [t0, Π0]] X tried to sell encyclopedias. Perhaps there were even others who did the same. :|| In contrast, given the day at the office, the decreasing quantifier in (45) is somehow barred from within the scope of the spatiotemporal-frame adverbial and the new perspective selective in its attention to events there, which results in maximal reference for the pronoun: (49) *∃σ0 …||: [∃t0 : t = time(σ0)] [℩Π0 : At[σ0, Π0] Πinitial(Π0, δ)]few men came to the office today[t0,Π0]. [℩X: [∃t0 : t = time(σ0)][℩Π0 : At[σ0, Π0] Πinitial(Π0, δ)] few men(X) came to the office today [t0, Π0]] X tried to sell encyclopedias. Perhaps there were even others who did the same. :|| (50) a. ∃σ0 …||: [∃t0 : t = time(σ0)] …, Two men …:|| [=(44)] b. *∃ σ0 …||: [∃t0 : t = time(σ0)] …, Few x: men …:||[=(45)]
806
Appendix 1
The infelicity of a spatiotemporal-frame adverbial qualifying the decreasing quantification in (45) does not preclude the decreasing neither in (51) from falling within the scope of a spatiotemporal quantifier that fixes the initial perspective quantifier for the entire discourse in (52) and derives licit nonmaximal reference: (51) Two men came to the office today. Neither could sell encyclopedias. Perhaps others did better. (52) ∃σ0 …||: Two men came to the office today. Neither could sell encyclopedias. Perhaps others did better. :|| In (51), the first sentence with an indefinite description licenses a selective perspective, to which reference in the second clause is merely anaphoric. The contrast between (44) and (45) surely tells of some difference in their logical syntax. As just sketched, it is distilled as a difference in the presence of a tacit spatiotemporal adverbial, which decreasing quantification does not tolerate as well as indefinite description. In all other respects, (44) and (45) are the same: spatiotemporal parameters constrain the entire discourse while all nominal quantification remains sentence-bound, and the pronouns therefore are descriptive throughout. In support of the empirical thesis that unspoken pronouns are the Eventish device for cross-reference to events, the unspoken pronouns should show the same variability in meaning as spoken pronouns in similar contexts, in particular, varying maximal reference according to the antecedent descriptions. They do as already illustrated in (53)–(56) from section 1.5.0: (53) F Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals, in no more than 50 seconds. (54) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals, in no more than 50 seconds. (55) F Eighteen barrels filled bottles with Rhône varietals, it filling no more than 60 bottles each with 750 mL of at least two Rhône varietals, ((and (it)) (filling them)) in no more than 50 seconds. (56) Eighteen barrels filled bottles with Rhône varietals, it filling 60 bottles each with 750 mL of at least two Rhône varietals, ((and (it)) (filling them)) in no more than 50 seconds. Logical forms for (53) and (54) appear in (57) and (58): (53) F Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. (57) (j[∃X: 18 barrels[X]] ∃E Source[E,X]) [℩Ej: proj][℩Ei: proi]Cause[Ej, Ei] [℩Ej: proj](i[℩Ei: Cause[Ej, Ei]][No X: more than 60[X] bottles[X]] [each x: Xx][∃e′: Eie′] filled with at least two Rhône varietals[e′]))
Descriptive Anaphora and Nonmaximal Reference
807
(54) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. (58) (j[∃X: 18 barrels[X]] ∃E Source[E,X]) [℩Ej: proj][℩Ei: proi]Cause[Ej, Ei] [℩Ej: proj](i[℩Ei: Cause[Ej, Ei]][∃X: 60 bottles[X]] [each x: Xx] [∃e′: Eie′] filled with at least two Rhône varietals[e′])) The first event pronoun, thatj or ‘[℩Ej: proj]’, denotes by description the events of their source-ing, for some 18 barrels. In a vintage of 180 barrels, reference to whatever happened to 18 of them is itself not maximal reference to the source-ing from (any 18) barrels in that vintage. But, by anyone’s reckoning, 18 barrels in subject position includes within its scope the rest of the sentence and thus the event pronoun denotes the events of X source-ing, [℩E: Source(E,X)]. There is in fact an embarrassment of riches for achieving this effect within the sentence. Besides the simple fact about the subject’s scope, the apparatus for cross-sentential nonmaximal reference already seen in (43) (repeated below) for the benefit of pronouns such as those in (59) and (60), initiating a discourse perspective Π0 centered on some 18 barrels, is deployed in (53) and (54) as well: (59) Eighteen barrels filled some bottles with Rhône varietals. They were 80% Grenache. (60) Eighteen barrels filled bottles, and it filled no more than 60 bottles. (43) ∃σ0∃σ1 …||: [℩Π0 : At[σ0, Π0] here(σ0) now(σ0) Πinitial(Π0, δ)] (0 [∃X : 18 barrels[X, Π0]] [∃E0 : At[E0, Π0]] (Source[E0, X]) 0) [℩E0: pro0][℩E1: pro1] Cause[E0,E1] [℩E0: pro0](1 [℩E′1: Cause[E0, E′1]][∃ l1: l1 = place(σ1) At[l1, Π0]] [∃E1 : [∀e: E1e]E′1e At[E1, l1] [∃Eπ0 : At[Eπ0, Π0] pl[Eπ0]][∀e: E1e] Eπ0e ] [∃X : some bottles[X, Π0]] fill.pl with Rhône varietals[E1,X] 1) …:|| ‘There are 18 barrels here (among many others), and these 18 barrels somewhere here filled in one (or more) fillings some bottles with Rhône varietals. They, the bottles filled with Rhône varietals in those fillings, were 80% Grenache.’ If so, the pronoun ‘[℩E0 : pro0]’ could equally well describe the events under Π0 of 18 barrels’ source-ing, indifferent to whether the indefinite description 18 barrels is itself part of the event pronoun’s descriptive content or exported outside it.20 In (53) and (54), a second, backward event anaphor, ‘[℩E1: pro1]’ in (57) and (58), denotes events of bottles each filling with 750 mL of at least two Rhône varietals. With (53)’s assertion that there are no more than 60 such bottles among those that the 18 barrels fill, the event pronoun refers maximally to all events of such bottles so filled. With (54)’s assertion, that there are 60 such bottles, it refers nonmaximally
808
Appendix 1
to the events of 60 bottles so filling, for some 60 among the 5400 that the 18 barrels fill—the same variation in reference that it is subject to in (61) and (62): (61) F It lasting no more than 50 seconds, no more than 60 bottles each filled with 750 mL of at least two Rhône varietals. (62) It lasting no more than 50 seconds, 60 bottles each filled with 750 mL of at least two Rhône varietals. But, outside the scope of 60 bottles in both (54) and (62) and without further supplement, a pronoun with descriptive content something like the filling of bottles (by the 18 barrels) each with 750 mL of at least two Rhône varietals refers to filling 5400 bottles, and a pronoun with the content the filling of 60 bottles (by the 18 barrels) each with 750 mL of at least two Rhône varietals fails to refer at all in that there are as many such fillings as there are ways to choose 60 bottles from among the 5400. The intervention of a spatiotemporal-frame adverbial cannot be assumed. Within the sentence, nonmaximal reference is immune to the conditions on spatiotemporal localization: (63) Eighteen shell corporations share sixty tax shelters each with at least two limited partnerships, for the benefit of no more than three multinational conglomerates. (64) a. Eighteen shell corporations share sixty tax shelters each with at least two limited partnerships, and it is for the benefit of no more than three multinational conglomerates. b. Eighteen shell corporations share sixty tax shelters each with at least two limited partnerships, and they / the tax shelters benefit no more than three multinational conglomerates. (65) Eighteen shell corporations share sixty tax shelters, and each (of them) is shared with at least two limited partnerships, and it is for the benefit of no more than three multinational conglomerates. Unless the speaker is taken to have in mind a particular sixty tax shelters, the pronouns in (64) imply that the eighteen shell corporations share no tax shelters each with at least two limited partnerships other than the sixty shared for the benefit of no more than three multinational corporations. In (65), the anaphora in the second sentence implies that the corporations share none other than the sixty tax shelters altogether. There is no sidestepping these implications of cross-sentential anaphora to the extent it makes little sense to isolate in one location the ownership of some tax shelters from the ownership of any others. Yet none of this seems to matter within sentence (63), where nonmaximal reference is unimpeded and there is no implication that sixty tax shelters so described are the only such (see note 13).
Descriptive Anaphora and Nonmaximal Reference
809
Without metaphysical recourse, the contrast in nonmaximal reference that persists between no more than 60 bottles in (53) and 60 bottles in (54) rather reflects a contrast in epistemic stance. Endorsing in spirit another aspect of Heim 1982 and Kamp 1981, a difference of type (which may perhaps be explained away by the pragmatics; see note 18) distinguishes (in)definite description from other quantification. To introduce an (in)definite description is to introduce concurrently a selective perspective centered on what is described (see note 20): (66) [∃X: ∃E (60 bottles[E, X] At[E, Πi])] (67) *[No X: ∃E (more than 60 bottles[E, X] At[E, Πi])] As remarked earlier, quantification over perspectives cannot have discourse scope, but it may have sentential scope, as in (68), where it closes off the perspectival parameters introduced by 18 barrels and 60 bottles: (54) Eighteen barrels filled 60 bottles each with 750 mL of at least two Rhône varietals, in no more than 50 seconds. (68) …||: ∃Π0 ∃Π1 …(0 [∃X: ∃E(18 barrels[E, X] At[E, Π0])] ∃E0 Source[E0, X]) 0) [℩E0 : pro0][℩E1 : pro1] Cause[E0,E1] [℩E0 : pro0](1[℩E′1: Cause[E0, E′1]][∃X: ∃E (60 bottles[E, X] At[E, Π1])] [each x: Xx][∃E1: [∀e: E1e]E′1e] filled with at least two Rhône varietals[E1,x])1 …:|| All that has been said earlier about discourse, spatiotemporal-frame adverbials, and the initialization of the narrator’s perspective remains as a further precision of (68) and the perspectives introduced (see (43)). As the narrative advances from one spatiotemporal location to the next, within a scene on location, the dramatis personae, the participants, are ushered in, by (in)definite description, with selective attention to their staging, with the intention of course that the several presentations are integrated in a coherent scene, throughout which one may recall the barrels or events presented in Π0 or the bottles or events presented in Π1, and thus refer to the bottles in Π1 filling and only their filling. Absent the introduction of Π1 (see (67)) with no more than 60 bottles, reference in (53) to bottles each filling (from the 18 barrels) with 750 mL of at least two Rhône varietals is maximal reference to all such bottles filling, as desired. The contrast between 60 bottles and no more than 60 bottles thus comes down to the latter’s resistance to perspectival quantification, whether licensed by spatiotemporal-frame adverbials with discourse effects or by (in)definite description itself within the sentence.
Appendix 2 Eventish
The logical language Eventish (second-order monadic predicate calculus with binary, restricted quantification, (in)definite descriptions, and descriptive anaphora) comprises: i. Vocabulary Predicates and relations: a. Open classes of event concepts (stab(e)) and nominals (statue(e,x), clay(e,x)), b. Closed classes: 1. Thematic relations between objects and events, both transitive (W(e,x) “(principal) participant,” Cum(e,x) “accomplice,” Agent(e,x), … , in(e,x)), and ditransitive (with(e,e′,x) (instrumental and comitative)); 2. Temporal, causal, and mereological relations between events (Cause(e,e′) or C(e,e′), Overlap(e,e′) or O(e,e′), e≤e′, …); and aspectual classifiers of events (Prog(e), …). Variables, xi (0 ≤ i ≤ ∞), Xi (0 ≤ i ≤ ∞), ei (0 ≤ i ≤ ∞), Ei (0 ≤ i ≤ ∞). Pronouns, spoken and unspoken [N pro], literally pro noun and not pro DP, appearing with DPs—[DPthe X: pro], [DP ℩X: pro]. Lexical monadicity. An atomic formula from first-order predicates or relations of Eventish contains at least one variable ei and no more than one variable, xj. Note: Lexical monadicity is syntax and not semantics. The domains for these variables are unsorted and unrestricted. Despite the alphabetic variance, anything that there is could be a value for any of them. Sentential connectives, and, or, and not. Sentential complementizers, if, for, and that.
812
Appendix 2
Quantifiers: a. The standard vocabulary of binary, restricted first-order quantification (every, most, few, etc.), including a null, unspoken universal quantifier ∀. b. The standard vocabulary for plural (second-order) (in)definite descriptions (some, the, -s, n(X)) including null, unspoken ∃, and ℩ as defined in (1). (1) [℩X: Φ[X]] Ψ[X] ↔df [∃X: Φ[X] & ∀Y(Φ[X/Y] → ∀y(Yy → Xy)) & ∀Z((∀Y(Φ[X/Y] → ∀y(Yy → Zy))) → ∀x(Xx → Zx))] Ψ[X] (Sharvy 1980,1 omitting existential commitment) ii. Syntax with standard second-order semantics (see Schein 1993, 2006) a. If Φ, Ψ are wffs, ⌜ΦΨ⌝ is a wff. Σ satisfy ⌜ΦΨ⌝ iff Σ satisfy Φ and Σ satisfy Ψ. (See Conjunctivism, Pietroski 2005) b. If Φ is a wff, ⌜not Φ⌝ is a wff. c. If Φ, Ψ are wffs, ⌜(Φ (and Ψ))⌝ is a wff, and ⌜(Φ (or Ψ))⌝ is a wff. d. If Ei, a variable and Φ, Ψ wffs, ⌜[℩Ei : if Φ] Ψ⌝ is a wff. (If-clauses as plural definite descriptions of events, Schein 2003) e. If Q is a 1st-order quantifier, vi, a variable and Φ, Ψ wffs, ⌜[Q vi : Φ] Ψ⌝ is a wff. f. If Q is a 2nd-order quantifier, Vi, a variable and Φ, Ψ wffs, ⌜[Q Vi : Φ] Ψ⌝ is a wff. Translation requires the decomposition of all predicates in the natural language that appear to be polyadic into combinations of elements drawn from the Eventish vocabulary. Supermonadicity. In the general case with n nominal or clausal arguments, there are n events described. For example, (2) (Agent(e1,j) C(e1, e2) (Theme(e2,t) butter(e2) (O(e2, e3) (In(e3,k) (O(e3, e4) (With(e4,b) (O(e4, e5) At(e5,m)))))))) Let θ stand for any of the thematic relations between events and objects (i.b.1), Γ for any of the relations, causal or topological, between events (i.b.2), and let Φ[ei] stand for any complex description of ei , itself composed of thematic relations, causal or topological relations, event concepts, and so on. To say of xi that it is so-and-so is to relate xi to an event that itself relates to another event falling under a complex description Φ (which subsumes most of the content of asserting so-and-so): (3) (θ(ei,xi) (Γ(ei,ei+1) Φ[ ei+1]))
Eventish
813
The phrase Φ is in effect a small-clause complement to the causative/topological verb Γ. In (simple) sentences with multiple arguments, the pattern in (3) iterates within Φ, generating the structure in which small clauses are nested: (4) (θ0(e0,x0) (Γ0(e0,e1) (θ1(e1,x1) (Γ1(e1,e2) (θ2(e2,x2) (Γ2(e2,e3) θ3(e3,x3))))))) The schema in (4) shows only the relations from closed classes (i.b). Perhaps every (or many a) small clause also contains an event concept or other modifiers that apply to events, as in (5): (5) (θ0(e0,x0) (Γ0(e0,e1) (θ1(e1,x1) V1(e1) (Γ1(e1,e2) (θ2(e2,x2) V2(e2) (Γ2(e2,e3) θ3(e3,x3) V3(e3))))))) Either way, (4) and (5), and their variants, instantiate (6): (6) (Ψ0[e0] (Γ0[e0,e1] (Ψ1[e1] (Γ1[e1,e2] (Ψ2[e2] (Γ2[e2,e3] Φ3[e3])))))) The logical structure of complementation. The causal or topological relations between events, the junctural relations, are expressed by small verbs in two variables, and the complement to the verb is a phrase intended to describe some related events via one of the variables: (7) [Γi[Ei,Ej] Ψj[Ej]], e.g., [vP Cause[E0,E1] VP[E1]] (see pluralization below) For a given junctural relation Γi[Ei,Ej], the events Ej that Ψj[Ej] describes are to be restricted to those that bear Γi to Ei. That is, the VP-ing under consideration is restricted to E0’s effects, the junctural relation being a causative in the example in (7): (8) [Γi[Ei,Ej] Ψj[Ej]] ⇒ [Γi[Ei,Ej] [℩E: [℩Ej: Γi[Ei,Ej]]E=Ej]Ψj[Ej]] E.g., [vP Cause[E0,E1] VP[E1]] ⇒ [vP Cause[E0,E1] [℩E: [℩E1: Cause[E0,E1]]E=E1] VP[E1]] Note that ‘Ei’ (‘E0’) and ‘Ej’ (‘E1’) remain free. The prefix to Ψ (VP) that complementation introduces, agreement morphology if you like, is intended only to restrict the domain for ‘Ej’ (‘E1’) in the subsequent articulation of logical form. This prefix has an effect on interpretation only if Ψ (VP) contains a nonincreasing quantifier, and hence is often suppressed. The backbone of any natural language sentence is a structure that dovetails event descriptions in a causal/topological chain. Reference to events by definite description. Braids are structures such as (9) dovetailing event descriptions and knotting them together with cross-referring pronouns: (9)
(Ψ0[e0] (Γ0[pro0,pro1] (Ψ1[e1] (Γ1[pro1,pro2] (Ψ2[e2] (Γ2[pro2,pro3] Φ3[e3]))))))
(10) (∃e1Agent(e1,j) C(pro1, pro2) (∃e2Theme(e2,t) butter(e2) (O(pro2, pro3) (∃e3In(e3,k) (O(pro3, pro4) (∃e4With(e4,b) (O(pro4, pro5) ∃e5At(e5,m))))))))
814
Appendix 2
In translation from the natural language, reference to events is by plural definite description and plural existential quantification (see Pietroski 2005; Schein 1993, 2006, 2012; Schwarzschild 1996 and the references cited there). Hence all citation logical forms are in the plural according to the notational conventions below. The pluralization of (10) is (11) ∃E1Agent[E1,X1] [℩E1: pro1][ιE2: pro2]C[E1, E2] ∃E2Theme[E2,X2] butter[E2] [℩E2: pro2][℩E3: pro3]O[E2, E3] ∃E3In[E3,X3] [℩E3: pro3][℩E4: pro4] O[E3, E4] ∃E4With[E4,X4] [℩E4: pro4][℩E5: pro5]O[E4, E5] ∃E5At[E5,X5] Event pronouns of “laziness” at junctural relations. The event variables ‘Ei’ and ‘Ej’ that remain free in (8) are bound by plural event pronouns like all event variables. The first variable ‘Ei’ in a junctural relation is bound by a “lazy” event pronoun falling within the scope of a perspectival quantifier and referring to events under that perspective, as in (13) (see appendix 1): (12) Eighteen barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. (13) (i[∃X: 18 barrels[X]] ∃E Source(E,X)) ∃Π[℩Ei: proi Π][℩Ej: proj]Cause[Ei, Ej] [℩Ej: proj](i[℩Ei: Cause[Ej, Ei]][No X: 60 bottles[X]] [each x: Xx] [∃e′: Eje′] filled with at least two Rhône varietals[e′])) Adverbialization. To say that [Q NP] are so-and-so is to say that [Q NP] while NP-ing (i.e., in events of NP-ing) are so-and-so: (14) [Qαi : ∃ENP[E, α]](θ[Ej,α] Φ) ⇒ [Qαi : ∃ENP[E, α]][℩E: NP[E, α]][∃Ej: N[E,Ej] (θ[Ej,α] Φ) Translation schemas for Right-Node Raising (chapter 2) (15) [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : proj,k] Ωδ) For Right-Node Raised structure (15), the descriptive content of ‘[℩Ej,k : proj,k]’ with antecedent clauses Φ and Ψ (with abbreviation convention (17) for split antecedency): (16) [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : proj,k] Ωδ) ⇒ [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : ([℩Xm : NPm]Φ,[ ℩Xn : NPn]Ψ)[Ej,k]] Ωδ) ‘[DPm D NP](Φ Δδ), and [DPn D NP](Ψ & the events of NPm-Φ-ing and NPn-Ψ-ing Ω)’ (17) ⌜(Φ,Ψ)[Ej,k]⌝ for ⌜[℩Ej: Φ][∀e: Eje]Ej,ke & [℩Ek: Ψ][∀e: Eke]Ej,ke] & ∀E(([℩Ej: Φ][∀e: Eje]Ee & [℩Ek: Ψ][∀e: Eke]Ee])→ [∀e: Ej,ke]Ee)⌝ events Ej,k are the Φ-ings Ej & the Ψ-ings Ek
Eventish
815
The meaning of deletion Δδ in Right-Node Raising according to the translation in (20) is: (18) plural reference to, and (19) second-order predication of, the events that the Right-Node Raised phrase denotes. (18) … [∀e: Eie][℩Eδ : Δδ]Eδe) and … (19) … [∀e: Eie][℩Eδ : Δδ]Eδe) and … (20) [DPm D NP](Φ Δδ) and [DPn D NP](Ψ [℩Ej,k : proj,k] Ωδ) ⇒ [DPm D NP](Φ [∀e: Eje][℩Ej,k : ([℩Xm : NPm]Φ, [℩Xn : NPn]Ψ)[Ej,k] Ω] Ej,ke) and [DPn D NP](Ψ [℩Ej,k : ([℩Xm : NPm]Φ, [℩Xn : NPn]Ψ)[Ej,k]] Ω) ‘[DPm D NP] Φ in the events of NPm-Φ-ing and NPn-Ψ-ing that Ω, and [DPn D NP](Ψ & the events of NPm-Φ-ing and NPn-Ψ-ing Ω)’ Notational conventions Square brackets, Φ[…], to list variables free in formula Φ and used to indicate that the formula or natural language expression to which they are suffixed is not a primitive of Eventish but an abbreviation for a complex expression. Variables occurring with primitives are in parentheses or bare—for example, , e.g., ‘Agent(e,x)’, ‘Rxixj’. Pluralizing abstraction. ⌜[℩V: Φ[v]]⌝ abbreviates ⌜[℩V: ∀v(Vv → Φ[v])]⌝. Pluralization. The pluralization of event concepts V, of temporal, causal, or mereological relations between events, and of thematic relations θ are defined thus (see Schein 1993, 2006): (21) V[E] ↔df ∃eEe & ∀e(Ee → Ve) (22) [Ei, E j ] ↔ df ∃eEie & ∃eEje & ∀ei(Eiei → ∃ej(Ejej & (ei, e j ))) & ∀ej(Ejej → ∃et(Etet & (ei, e j ))) (23) θ[E,X] ↔df ∃xXx & ∃eEe & ∀x(Xx ↔ ∃e(Ee & θ(e,x))) (thematic exhaustivty) X are the θ-ers in E. (Pietroski 2005) (24) For any relation R, ⌜R[ei,Xj]⌝ for ⌜∃Ei(∀e(Eie ↔ e = ei) & R[Ei,Xj])⌝ ⌜R[ei,Ej]⌝ for ⌜∃Ei(∀e(Eie ↔ e = ei) & R[Ei,Ej])⌝ ⌜R[Ei,xj]⌝ for ⌜∃Xj(∀x(Xjx ↔ x = xj) & R[Ei,Xj])⌝ ⌜R[Ei,ej]⌝ for ⌜∃Ej(∀r(Eje ↔ e = ej) & R[Ei,Ej])⌝ ⌜R[ei,xj]⌝ for ⌜∃Ei∃Xj(∀e(Eie ↔ e = ei) & ∀x(Xjx ↔ x = xj) & R[Ei,Xj])⌝ ⌜R[ei,ej]⌝ for ⌜∃Ei∃Ej(∀e(Eie ↔ e = ei) & ∀e(Eje ↔ e = ej) & R[Ei,Ej])⌝ Note: ‘Agent(e,x)’ means that x is an Agent of e. ‘Agent[e,x]’ means that x is the Agent of e (see (23) thematic exhaustivity).
816
Appendix 2
Phrasal parentheses: (25) ⌜Φ Ψ⌝ for ⌜(Φ (Ψ))⌝ (Default binary, right-branching) (26) ⌜(Φ and Ψ)⌝ for ⌜(Φ (and Ψ))⌝ (27) ⌜[Q ξi : Φ]Ψ Γ⌝ for ⌜(([Q ξi : Φ]Ψ) Γ)⌝ ⌜∀viΨ Γ⌉ for ⌜((∀viΨ) Γ)⌝ ⌜∃ξiΨ Γ⌉ for ⌜((∃viΨ) Γ)⌝ (28) ⌜[Q ξi : Φ]Ψ and Γ⌝ for ⌜(([Q ξi : Φ]Ψ) and Γ)⌝ ⌜∀viΨ and Γ⌝ for ⌜((∀viΨ) and Γ)⌝ ⌜∃ξiΨ and Γ⌝ for ⌜((∃viΨ) and Γ)⌝ Anaphoric descriptions and agreement with subscripts for variables. (29) Every songstress sang solo a single melodic line. It was perfect four-part harmony. With adverbialization suppressed: (30) [Every x1: ∃E S[E,x1]] ∃E1Agent[E1,x1] [℩E1: pro1][℩E2: pro2]C[E1, E2] ∃E2 sing solo[E2] [℩E2: pro2][℩E3: pro3]O[E2, E3] [∃X3: a(n)[X3] ∃E L[E,X3]] Theme[E3,X3]. [℩E2: it2] was perfect four-part harmony[E2]. In translating the descriptive pronoun it in (29), the subscript in ‘[℩E2: it2]’ in (30) indicates the abstraction and description intended in context. This pronoun is to denote antecedently described events of singing solo, namely, the events of the songstresses singing solo single melodic lines. The pronoun ‘[℩E2: pro2]’, on the other hand, also denotes antecedently described events of singing solo, but the antecedent provides for a shorter description of just events of singing solo. As this pronoun and its antecedent both fall within the scope of every songstress, it will denote an individual’s performance, her singing solo. (31) Every songstress sang solo a single melodic line, and every songster sang solo a single melodic line. It was perfect four-part harmony. (32) [Every x1: ∃E S[E,x1]] ∃E1Agent[E1,x1] [℩E1: pro1][℩E2: pro2]C[E1, E2] ∃E2 sing solo[E2] [℩E2: pro2][℩E3: pro3]O[E2, E3] [∃X3: a(n)[X3] ∃E L[E,X3]] Theme[E3,X3], and [Every x1: ∃E S[E,x1]] ∃E1Agent[E1,x1] [℩E1: pro1][℩E4: pro4]C[E1, E4] ∃E4 sing solo[E4] [℩E4: pro4][℩E3: pro3]O[E4, E3] [∃X3: a(n)[X3] ∃E L[E,X3]] Theme[E3,X3] [℩E2,4: it2,4] was perfect four-part harmony[E2,4].
Eventish
817
The pronoun it in (31) has a split antecedent, resulting in its reference to the events of songstresses singing solo single melodic lines and songsters singing solo single melodic lines, as indicated by the subscripts in ‘[℩E2,4: it2,4]’ in (32). Note an important feature of this notation. Nothing compels the usage of distinct variables ‘E2’ and ‘E4’ in the descriptions of the songstresses singing and the songsters singing: (33) [Every x1: ∃E S[E,x1]] ∃E1Agent[E1,x1] [℩E1: pro1][℩E2: pro2]C[E1, E2] ∃E2 sing solo[E2] [℩E2: pro2][℩E3: pro3]O[E2, E3] [∃X3: a(n)[X3] ∃E L[E,X3]] Theme[E3,X3], and [Every x1: ∃E S[E,x1]] ∃E1Agent[E1,x1] [℩E1: pro1][℩E4: pro2]C[E1, E2] ∃E2 sing solo[E2] [ιE2: pro4][ιE3: pro3]O[E2, E3] [∃X3: a(n)[X3] ∃E L[E,X3]] Theme[E3,X3] [℩E2,2: it2,2] was perfect four-part harmony[E2,2]. In (33), the antecedent clauses for it are identical except for a single morpheme, nominal gender (not shown).
Notes
1
Introduction
1. Note however that the displayed logical form is as yet a poorer approximation than its paraphrase, failing to imply that sauce and cheese are the only parties to the perfect marriage (see section 1.5). 2. If the conjunction of DPs continues to be misparsed as such, the two meanings for and, sentential and nominal, prove insufficient when conjoined DPs combine with certain genuine collective predicates so called (Dowty 1987; Higginbotham and Schein 1989; Taub 1989; Brisson 1997, 1998, 2003; Winter 1998, 2002; Hackl 2001a, 2001b, 2002): (i) a. The Apostles and the Saints are (both) twelve, and so there are twenty-four Holy Fathers. b. The Holy Fathers who are Apostles and Saints are (*both) twelve. The Holy Fathers who are Apostles or Saints are (*both) twelve. (ii) a. The oaks, the firs, the birches, the aspens, the alders, the elms, and the poplars are sparse in the forest they populate, but the forest is dense. b. F The trees, which are (all) oaks, firs, birches, aspens alders, elms, and poplars, are sparse in the forest they populate, but the forest is dense. F The trees, which are (all) oaks, firs, birches, aspens alders, elms, or poplars, are sparse in the forest they populate, but the forest is dense. The (a) sentences have interpretations that the (b) sentences cannot share in. Thus, if the conjunction of DPs in an (a) sentence refers to a plurality, it refers to one that the subject of the corresponding (b) sentence does not, suggesting yet another meaning for and to bring this about. It could instead be conceded that the interpretation for the (a) sentences at issue derives by Conjunction Reduction from the repetition in both conjuncts of the entire Predicate Phrase—The Apostles are twelve, and the Saints are twelve—or, equivalently, that the coordination may distribute over the entire scope of the conjoined DPs. Such a derivation is, however, unavailable when the predicate is be equinumerous: (iii) a. The elms and the beeches are equinumerous in this region. b. *The trees are equinumerous in this region. *The trees, which are elms and beeches, are equinumerous in this region. *The trees, which are elms or beeches, are equinumerous in this region.
820
Notes
(iv) a. The odd integers and the even integers are equinumerous. b. *The integers are equinumerous. *The integers, which are odd numbers and even numbers, are equinumerous. *The integers, which are odd numbers or even numbers, are equinumerous. The incoherence of the (b) sentences is proof that the subjects of an (a) sentence and (b) sentence do not refer to the same plurality. If the conjunction of DPs refers to a plurality in the (a) sentences, it refers not to the trees or to the integers but to a plurality of two pluralities, one of the elms and the other of the beeches in (iiia), and the plurality of odd integers and the plurality of even integers in (iva), which it takes another meaning for and to accomplish. Schwarzschild’s (1991, 1996) proposal to resist Landman’s (1989a, 1989b) embrace of further ambiguity for and founders here with predicates such as be equinumerous, which reject altogether the referent derived by the original meaning of nominal and. Parsing the elms and the beeches as a DP thus cannot escape commitment to three rather than two ands. The third and goes on to wreak havoc elsewhere. Sentence (v) is unequivocally false and cannot tell apart the forest from the trees or the pair from the many, as (vi) can: (v) F The elms and the beeches are equinumerous and (they) are not (identical to) the (many) trees that are elms or beeches. (vi) The elm forest and the beech grove are equinumerous and (they) are not (identical to) the (many) trees that are elms or beeches. The group of elms and the group of beeches are equinumerous and (they) are not (identical to) the (many) trees that are elms or beeches. Yet, given the alleged reference of the elms and the beeches in the first clause of (v), (v) should allow a true interpretation in that the plurality of two pluralities referred to is not in fact a plurality of more than two. Just the facts of (iii) and (v) are a dilemma with no obvious way out for any view holding that the elms and the beeches is a referring expression. I would treat equinumerous as a covert reciprocal ‘equinumerous with each other’ so that it may fall under the treatment of concept reciprocity (i.e., elms vs. beeches, in Schein 2003, 352–355), which compares the participants of one event, the one in which all and only the elms participate with the participants in the other, the one in which all and only the beeches participate. 3. Further evidence, contra Aristotle, that and is always a sentential connective. Hopi marks switch reference, which typically expresses a relation between clauses. In some languages, conjunctions are so marked, and in Hopi, as Hale and Jeanne (1976) observe, DP-conjunctions are marked according to the following pattern. A conjunction of subjects is marked as Different Subject (DS) for different subject as in (i) and a conjunction of objects is marked for Same Subject (SS) as in (ii): (i) ‘itana ni-q ‘itanu tumala’yta. our father and=DS our mother are working (Hale and Jeanne 1976) (ii) nu’ ‘it taavot ni-t ‘it sowit niina. I this cottontail and=SS this jackrabbit killed (Hale and Jeanne 1976) Hale and Jeanne point out the apparent evidence for Conjunction Reduction. The switchreference marker marks after all a relation between clauses. So, (i) comes from “our father
Notes
821
and our mother are working” where the conjoined clauses have different subjects, and (ii) from “I killed this cottontail and I killed this jackrabbit” where the subjects are the same. But they reject this interpretation, thinking like Aristotle that Conjunction Reduction is impossible with collective predicates as in (iii) and yet the switch-reference marker still appears. (iii) mi’ tiyo’ya ni-q mi’ manaw’ya naami yori the boy and=DS the girl saw each other The Hopi can, however, be taken at face value as evidence for Conjunction Reduction. The apparent DP-conjunction in (iii) in fact conjoins clauses describing the boy’s and the girl’s participation in events. It is therefore no surprise that the switch-reference marker, which expresses a relation between clauses, should mark the DP-conjunction. But there is a fly in the ointment. Consider a logical form for (ii): (iv) Agent(e,I) & kill(e) & (Theme(e,c) and=SS Theme(e,j)) In what sense can the conjoined clauses in (iv) be said to have the same subject? If my conclusions are correct, it must be that the traditional view according to which the switch-reference marker encodes the sameness or difference of the subjects is mistaken. In particular, it would be a mistake to try, as Finer (1985) does, to reduce switch reference to a binding relation involving subjects. Rather, the switch-reference marker must express some other relation between the clauses, one available even where there are no subjects, as in (iv). It just so happens that more often than not this relation coincides with the sameness or difference of the subject if there are subjects. This coincidence accounts for the traditional identification of switch reference with sameness or difference of the subject. As it turns out, there is extensive discussion in the switch-reference literature that addresses the empirical shortcomings of the traditional view. Much of it is summarized by Stirling (1993). For purposes here, it is interesting to note that the exceptions to the traditional identification go all ways: clauses with obviously different subjects will sometimes get the SS marking, clauses with the same subject will sometimes get the DS marking, and clauses with no subjects (weather verbs, for example) will get one or the other depending on what is to be expressed. Stirling’s view is that the switch-reference marker expresses a relation between the events denoted by the marked clauses. The details are rather complicated and vary from language to language, but if it is a relation between events, then in principle, there is no reason it couldn’t hold in (iv). For further discussion, see Camacho 2003. 4. The toy logical form (36) also implies that the shimmying is the shaking, which may be rock ’n’ roll but is otherwise a defect of this illustration. With a richer decomposition (i), the shimmying need not be the shaking although they are both effects of the same action and thus the same rocker: (i) [A x: rocker(x)] Agent(e1, x) & ∃e2 (Cause(e1, e2) & shimmied(e2)) and [A x: rocker(x)] Agent(e1, x) & ∃e2 (Cause(e1, e2) & shook(e2)) Or the reference to the events antecedently described may enter as an adverbial phrase aspiring to the equivalent of (ii), where it is inferred from the spatiotemporal proximity of the shimmying and the shaking that their rockers are the same: (ii) A rocker shimmiedi, and shimmying therei, a rocker shook. Either way, (32) and (33) are distinguished in that (33) renews existential event quantification in the second clause, whereas (33) makes definite reference to the events described in the first.
822
Notes
5. Realism may demand more species if each is to be sparse in an otherwise dense forest (see (iia) in note 2 above). 6. Drawn from the genuine collective predicates (see note 2) that prohibit it, be sparse preempts a tacit (semi-) distributive operator, …are all sparse in a region near human settlement, which could itself include within its scope in a region near human settlement. In any case, such an operator would defeat the attempt to tell (39) and (40) apart, absent reason for it to occur in (40) but not in (39). 7. These observations extend to a contrast between (i) and (ii), where only (ii) may be understood to advance the reference of now: (i) #Kunstler is sitting now, and is standing now. (ii) Kunstler is sitting now, and Kunstler is standing now. Kunstler is sitting now, and he is standing now. Verbal contortions improve (i): (iii) ??Kunstler is sitting now, and is standing … now. Still, I should rather say: (iv) Kunstler is sitting now, and he’s standing … now. 8. Note that if the meaning at issue were an effect of predicative coordination rather than a question for the interpretation of the null subjects of Tensed clauses, it would be a mystery why (i), conjoining participles, has no interpretation equivalent to (ii), analogous to the equivalence between (51) and (52). (i) #An easy model theory textbook will be needed and written within this decade. (ii) An easy model theory textbook will be needed and an easy model theory textbook written within this decade. An easy model theory textbook will be needed and one written within this decade. 9. See Schein 1993, 219–237, for a precedent for perspectives; see Schein 2002 for scenes and reticules. 10. Still no chapter on coordination with Determiner-Sharing (McCawley 1993; Johnson 2000a; Lin 2000, 2002) (as well as several other contexts): (i)
Too many Irish setters are named Kelly, German shepherds Fritz, and huskies Nanook. (McCawley 1993) (ii) Few dogs eat Whiskas and cats Alpo. (Johnson 2000a) (iii) Most real men drink whiskey straight, and effete snobs chilled chardonnay. (Lin 2000, 2002) 11. ‘tease(e)’ denotes the dynamic condition of suffering, enduring, or experiencing teasing (see embarrass, shame, humiliate, torment). 12. That is, if an alleged coordination of DPs, coordinates clauses based on the participation relation, then any sentence of the forms in (i) threatens to confuse the manner of participation associated with the two tokens of DPi unless they describe different events: (i) … DP1 … DP1 and DP2 … … DP1 and DP2 … DP1 …
… DP2 … DP1 and DP2 … … DP1 and DP2 … DP2 …
Notes
823
Besides pairing Agent and Theme as in (75), other pairings force recourse to a vague participation relation, too: (ii) 613 diverse hedge funds variously sold and sold to 365 equity funds and 248 mutual funds in a single day’s trading. (iii) 365 equity funds and 248 mutual funds variously sold and were sold in the course of a day’s trading. (iv) 365 equity funds and 248 mutual funds variously sold and were sold to in the course of a day’s trading. (v) 365 equity funds and 248 mutual funds were variously sold to and sold in the course of a day’s trading. (vi) The preparation of these Balkan delicacies has variously stuffed and stuffed into other vegetables—30 eggplants, 50 peppers, 80 tomatoes, 130 olives, and 210 pearl onions, weighing 34 kilos in total. (vii) For these Balkan delicacies, 30 eggplants, 50 peppers, 80 tomatoes, 130 olives, and 210 pearl onions, weighing 34 kilos in total, were variously stuffed and stuffed into other vegetables. 13. Sentences (i) and (ii) contrast as (83) and (84) do: (i) Jackie saw Oswald assassinate JFK. (ii) Jackie saw JFK assassinated by Oswald. For (iii) and (iv), it would have to be claimed that -tion never nominalizes the causal event so that both (iii) and (iv) derive from “assassination of JFK by Oswald.” (iii) Jackie saw Oswald’s assassination of JFK. (iv) Jackie saw JFK’s assassination by Oswald. In support of this, it is odd to say any of (v)–(vii) of Oswald’s slow brother Jimmy, who was beside on him on the roof of the Texas School Book Depository and believed that his brother was shooting pigeons in the park. (v) (vi) (vii) (viii)
Jimmy saw JFK assassinated by Oswald. Jimmy saw Oswald’s assassination of JFK. Jimmy saw JFK’s assassination by Oswald. Jimmy saw Oswald assassinate JFK.
On the other hand, one can imagine a context for (viii) where it is enough that Jimmy saw Oswald pull the trigger. It seems that the locus of an assassination is always the victim. Gerunds are once again more verbal. So we have (ix) Jackie saw JFK’s being killed by Oswald. (x) Jimmy saw Oswald’s killing JFK. But neither (xi) Jackie saw Oswald’s killing JFK. (xii) Jimmy saw JFK’s being killed by Oswald. I suspect that the judgments shift again with “Oswald’s killing of JFK” toward the nominal, but it’s squishy.
824
Notes
14. In the face of such wayward uses of the copula as Kate’s praising Harry was her scorning Louis or Reagan’s election was the conservative social agenda’s inauguration, it should not be mistaken for strict identity. For important discussion of actions and their description, see Pietroski 1998, 2000, 2005. 15. See Burton and Grimshaw 1992; Kitagawa 1986; Koopman and Sportiche 1991; Kuroda 1988; McNally 1992. As Terje Lohndal points out to me (p.c., June 2010), labeling all verbal coordinations as VP has not survived the introduction of projections such as vP intermediate between VP and TP, but the motivation for internal subjects remains mutatis mutandis. 16. Examples (i–iv) preempt various attempts to circumvent Heycock’s basic argument and otherwise reinforce her point: (i) The fear of people Gorei insulted years ago, hei has yet to overcome. (ii) *Hisi fear of people Gorei insulted years ago, hei has yet to overcome. (iii) The personal threat to himi of legislation that Newt Gingrichj could sponsor, hej lets no politiciani forget. (iv) *Hisj personal threat to himi of legislation that Newt Gingrichj could sponsor, hej lets no politiciani forget. 17. Many thanks to Alexander Williams for insisting on this clarification and for pointing out to me the special status of ‘participate(e,x)’ as argument for the thematic separation of a fully general thematic role. 18. The argument for supermonadicity is for intereventive relations, ‘Cause(e,e′),’ ‘O(e,e′),’ etc., or the relations sui generis in (123b)–(125b), that are at least dyadic, although only dyadic relations are cited in the text. The argument is that the logical syntax introduces reference to an event e, say, an action whose only participants are agents and reference to another event e′, an effect with only themes as its participants, and therefore the sentence, if it is to cohere, is in need of an intereventive relation ‘Cause(e,e′)’ to relate them. Yet, as Pietroski (1998, 2000, 2005) forcefully argues, an instrumental phrase such as with the lens in (i)–(vi) cannot modify either the cause e or its effect e′: (i) (ii) (iii) (iv) (v) (vi)
Nora melted the chocolate. Nora held the lens. Nora’s holding the lens was her melting the chocolate. Nora melted the chocolate with the lens. *Nora held the lens with the lens. *The chocolate melted with the lens.
Nora melts the chocolate by holding the magnifying glass to focus sunlight on it. Her actions melting the chocolate and holding the lens are the same (iii). If, therefore, ‘With(e, the lens)’ applied to the action in (iv), it should apply to it as well in (v). If, instead, the instrumental phrase ‘With(e′, the lens)’ applies implausibly to the effect in (iv), it should apply as well in (vi). Rather, the lens is instrumental between action and effect, and Pietroski concludes that ‘With(e0, the lens)’ applies in (iv) to a large event that begins with Nora’s action and ends with the chocolate’s change of state. In neither (v) nor (vi) is there mention of a large event of cause and effect for instrument to mediate. One may accept the argument at face value and then to reconcile it to supermonadicity, replace the dyadic intereventive relations throughout with triadic relations ‘Cause(e0,e,e′),’ in which e0 is the large event of which the instrumental phrase is predicated that begins with cause e and ends in effect e′. Alternatively, it may be held that the instrumental phrase is itself ditransitive ‘With(e, e′, the lens)’ (see the
Notes
825
appendix in Schein 2002 titled “Comitative and Instrumental with NP”), allowing the intereventive relations cited in the text to remain dyadic (see also section 2.6 below). 19. As in Schein 1993, I recommend examples uncomplicated by intentional agency and cooperation as a single collective agent—hence the barrels of wine as the subject. There is an uncomplicated and immediate judgment here that any barrels that have not contributed even a drop to some given bottles have not filled or helped fill those particular bottles simply because none of that bottled wine can be traced back to the barrels at issue. It is thus recognized without hesitation that 180 barrels did not fill 5400 bottles; only 18 did. Such judgments are essential to any assessment of, say, the efficiency and cost of production. In such a context, there is little temptation to indulge the perverse, nonsensical thought that the 180 barrels are a team as a whole to be credited for each bottle filled. Despite the elaborate conditions for this thought experiment, the judgments elicited are just about whether so many barrels fill so many bottles. All the rest of it about varietals and the single bottle is to secure that these rather straightforward judgments are judgments about sentences with the structure indicated. If so, they imply truth conditions that can be rendered only if this structure is supplemented with descriptive event pronouns, as argued. Essential separation (Schein 1993) is the argument that such combinations of plural quantification and distributive, first-order quantification (requiring barrels, single bottle, and Rhône varietals) must have the indicated structure. Supermonadicity adds junctural, intereventive relations. In (134), “Cause” is chosen for the decomposition of barrels-filling-bottles to flag the spatiotemporal distance between the source’s participation—a pouring forth—and the narrow confines of the wine level rising within the bottles’ walls—the bottles’ filling—which rules out spatiotemporal coincidence ‘O[E,E′].’ Winemaker Winiarski, at the barrels’ spigots, filled the bottles. The action of which he is the agent caused the barrels’ filling, but it is of no import here whether the junctural relation relating his action to effect is the same as the one relating source to effect (see Pietroski 2000 for discussion). 20. Some of the litter can be cleared out rejecting essential separation (Schein 1993) and with it the logical forms (133) and (134). But even the most conservative Davidsonian must admit event pronouns in the case of collective adverbial phrases, in perfect harmony, dissonantly, etc., the point of (129). 21. Descriptive, plural event pronouns also solve the general problem of telicity and a special problem concerning telicity that distinguishes reports of telic, ballistic causation from those of telic, continuous causation (Schein 2002). 22. For (137), imagine a practice room with several rehearsals in progress one of which is the perfect four-part harmony, of which (137) is a true report. But this harmony is surrounded by attempts at harmony in which many a singer is off-key. It is left as an exercise for the reader to rehearse the argument developing this example. 23. Note, in particular, that according to (147), barrels filling, bottles being filled, and Rhône varietals being the filling are all parts of the same events of barrels filling bottles with Rhône varietals, absent supermonadicity. 24. Instead of a dyadic relation ‘Cause(e1, e2),’ between cause e1 and its effect e2 one may prefer a triadic relation ‘C(e1, e2, e1+2)’ to relate them to a larger event of which they are parts. For relevant discussion, see Pietroski 1998, 2000, and the references cited there as well as Schein 2002. 25. See sections 2.4 and 4.0.
826
Notes
26. Translation in (150b) slips into the official plural idiom; see appendix 2. 27. As in Evans 1977, 1980, the cited examples will oppose indefinite descriptions and decreasing quantifiers. Although (in)definite descriptions are also increasing quantifiers, the contrast in (155) and (156) does not reflect an opposition of increasing and nonincreasing quantifiers, since the pronoun in (i) with an increasing antecedent nevertheless bears maximal reference to the Britannica salesmen who came to the office today. Hence the others cannot be other Britannica salesmen who came to the office today. Since nonmaximal reference is tolerated in (ii), many Britannica salesmen must, in contrast to most Britannica salesmen, count as an indefinite description like two Britannica salesmen: (i) *Most Britannica salesmen came to the office today. They had no luck. Perhaps others did better. (ii) Many Britannica salesmen came to the office today, and they did not have much luck. Perhaps there were some others who did better. 28. An iota-operator as an operator implying maximal reference, in contrast to an existential quantifier or the equivalent. This is not intended to prejudice the treatment of the presuppositions associated with definite descriptions in natural language. 29. To my ear, (i), substituting those for these is less successful at nonmaximal reference than (157), seeming to imply that the two were the only men who came to the office today: (i) ??Two men came to the office today. Those (two) men who came to the office today tried to sell encyclopedias. Perhaps there were even others who did the same. This observation comports well with what is to be said about nonmaximal reference in appendix 1, namely, that it is strictly an effect of selective perspective. The proximate demonstrative these in (157) fixes a perspective that tracks close to the two men, from which perspective it is easy to miss other visitors to the office. The distal demonstrative those in (i) however presupposes a perspective so far away that the two men are far from its point of view. What perspective like that would capture those two and yet miss others coming to the office? Failing such, the reference is felt to be maximal. The nondemonstrative definite description in (ii) strikes me as intermediate between (157) and (i): (ii) ?Two men came to the office today. The (two) men who came to the office today tried to sell encyclopedias. Perhaps there were even others who did the same. Something prompts the speaker to abandon a pronoun for the prolix definite description. Perhaps it is to establish reference from her own current perspective that is remote from the events at the office, from which there is a route to any men who came to the office today only if there is a route to all. When the antecedent is a decreasing quantifier, no more than two men in (158), maximal reference is obligatory for any definite description, these men, those men, or the men. 30. (165) suffers the distraction that the pronoun it might not be a backward anaphor and refer instead to something of prior mention such as a system check or test flow. Keep in mind, however, the context in which the 18 barrels fill 5400 bottles in continuous, uninterrupted flow at a constant rate for 75 minutes. Which 50 seconds is salient enough to have been mentioned?
Notes
827
31. The puzzle also finds overt expression in sentences with definite descriptions of events anchored in subject position: (i)
F The filling (of) bottles with Rhône varietals from/by 180 barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. (ii) F The filling (of) bottles with Rhône varietals from/by 18 barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. (iii) The filling (of) bottles with Rhône varietals from/by 18 barrels filled 60 bottles (each with 750 mL of at least two Rhône varietals). The definite descriptions, whose presence again cannot be denied, also appear here to be unambiguous in referring to the entire pour even if (iii) ends up describing an event that is a fraction of it. Any interesting solution will treat them as such and assimilate them to the definite mass descriptions in (iv)–(vi), which are unambiguous and display the problem in equal measure: (iv) F The wine of 180 barrels filled 60 bottles each with 750 mL of at least two Rhône varietals. (v) F The wine of 18 barrels filled no more than 60 bottles each with 750 mL of at least two Rhône varietals. (vi) The wine of 18 barrels wine filled 60 bottles (each with 750 mL of at least two Rhône varietals). From the plain look of it, the wine of 18 barrels refers to the same stuff, and from the plain look of it, the structures are identical except for the substitutions between 180 barrels and 18 barrels and between no more than 60 bottles and 60 bottles. 32. A definite description operator, essentially as in Sharvy 1980 (see Cartwright 1996) but without the existential commitment; see appendix 2, “Eventish.” 33. The speaker who hedges with possibly must have in mind de re possibility: it is possible that the Harvard students participated in something there. If the Harvard students exist, it is otherwise a metaphysical necessity that there are states or events they participate in. Hence, in (209) ‘Eloc’ occurs de re in the scope of possibly, and in (208), ‘e2’ does. 34. Strictly, it may be false that musician(e,x) → (instrumentalist(e,x) ∨ vocalist(e,x)). In alternating harmonica and vocals, James Cotton creates a musical moment that is neither instrumental nor vocal in its entirety. A fussier description, musician in exclusively vocal or exclusively instrumental performance, restores the biconditional. Even the familiar adults and men and women or people and Jews and Gentiles do not support a biconditional, since the transgendered are adults throughout their adult life but that life is neither an event of being a man nor of being a woman and similarly converts are people throughout a life that is not wholly of one faith. Less familiar vocabulary provides examples without elaborate description—for example, passers and servers and receivers if it is clear that a volley is always multiple passes so that to have served and received once is to have been a passer not less than twice. Perhaps flowers and annuals and perennials is an example for all I know about botany. 35. I assume National League rules, in which pitchers must appear at bat and there are no designated hitters, who bat without fielding. 36. be an ensemble and be a powerhouse squad belong to the genuine collective predicates so-called (Dowty 1987; Higginbotham and Schein 1989; Taub 1989; Brisson 1997, 1998, 2003; Winter 1998, 2002; Hackl 2001a, 2001b), all of which give rise to the substitution puzzle.
828
Notes
Collective predicates outside the class (iii) are immune to the substitution puzzle and can be found in minimal pairs with members of the class (i): (i) ??Many flowers are a cluster in the middle of the garden. (ii) Many annuals and perennials are a cluster in the middle of the garden. (iii) Many flowers cluster in the middle of the garden. Many flowers are clustered in the middle of the garden. Many flowers form a cluster in the middle of the garden. Alexander Williams (p.c.) is doubtful about the contrast between (234) and (235) and notes a preference for singular agreement, where again no contrast is felt: (iv) Many musicians is an ensemble. (v) Many instrumentalists and vocalists is an ensemble. I agree that (iv) and (v) are acceptable and equally so, owing to a confound that allows weak DPs to behave as predicates within implicit small clauses. Compare (vi) (Having/there being) Four musicians is a quartet. (vii) (Having/there being) two instrumentalists and two vocalists is a quartet. Plural agreement is crucial to construing Many NP as distributive quantifier and subject of the sentence. A strong determiner eliminates the confound: (viii) ???Most musicians are/*is an ensemble. (ix) Most instrumentalists and vocalists are/*is an ensemble. Again, (x) ??Most adults would be a train wreck of a marriage. (xi) Most Jews and Gentiles would be a train wreck of a marriage. Sentence (x) misses the intended pronouncement that most adults are incompatible and lands on the infelicity that the solo adult would be a poor marriage. Similarly, (ix)–(x): (xii) ??Many adults are a train wreck of a marriage. (xiii) Many Jews and Gentiles are a train wreck of a marriage. The sentiments at issue in (x)–(xiii) are dubious about marriage or interfaith marriage. In contrast, (xiv)–(xv) counsel only against polygamy, in general or interfaith: (xiv) Many adults is a train wreck of a marriage. (xv) Many Jews and Gentiles is a train wreck of a marriage. 37. This case also includes failures of substitutivity between simple description, the lovers, and a coordination of descriptions containing the relevant NPs, the lover and the belovèd. See (v)–(vi) in note 2. Chapter 15 gives the same analysis for the illusion of referring to ordered pairs here and in (242). 38. There is irrelevant distraction from a reading that says of the kind, the Apostles, and not just those of them martyred, that they are twelve and similarly of the kind, the Saints. 39. See note 34. 40. While for illustration and paraphrase. Expected is the range of relations, temporal, causal, causal explanatory, mediating the interpretation of if-clauses or absolutive clauses including variation therein due to differences between tensed and reduced clauses.
Notes
829
Adverbialization is sometimes without apparent effect framing the events reported, The liar was deep asleep. Compare (i) The liar fails a polygraph. Any liar fails a polygraph. Every liar fails a polygraph.
(ii) The sometimes liar fails a polygraph. Any sometimes liar fails a polygraph. Every sometimes liar fails a polygraph.
As expected from adverbialization, the polygraph failures of (i) are an effect of lying, and any lie that escapes detection is a counterexample to what is asserted. In (ii), it seems that sometimes lying makes one anxious enough to fail a polygraph even when telling the truth. To be a sometimes liar is to always be a sometimes liar, for as long as one lives. So, to do something while a sometimes liar is to do something while oneself. Thus, The liar was deep asleep is The sometimes liar while a sometimes liar was deep asleep, with sometimes unspoken. In the literature on descriptive anaphora, several authors (Bäuerle and Egli 1985; Heim 1990, 162ff.; Reinhart 1987; Root 1986; Rooth 1987) have entertained a precedent for the view adopted here, double quantification, in which the second quantifier quantifies over individuals, as paraphrased below, or even over situations in a yet closer precedent (Heim 1990): (iii) Every parent who was supervising a teenager lent her the car. ‘Every parent who was supervising a teenager, any teenager she was supervising, she lent her car to.’ (iv) No parent who was supervising a teenager lent her the car. ‘No parent who was supervising a teenager, any teenager she was supervising, she lent her car to.’ The double quantification is rejected on the grounds that the second quantifier “any teenager she was supervising” would be dependent for its force on the first quantifier—universal in the scope of every parent in (iii) but existential in the scope of no parent in (iv)—which is unheard of. Sentences with adverbial modification present and spoken out loud are subject to similar variability in its quantificational force: (v) Every parent (yesterday) if (he was) supervising a teenager, lent her the car. Every parent (yesterday) while (he was) supervising a teenager, lent her the car. Every parent (yesterday) whenever (he was) supervising a teenager, lent her the car. Every parent (yesterday) any time (he was) supervising a teenager, lent her the car. Every parent (yesterday), (he) supervising a teenager, lent her the car. (vi) No parent (yesterday) if (he was) supervising a teenager, lent her the car. No parent (yesterday) while (he was) supervising a teenager, lent her the car. No parent (yesterday) whenever (he was) supervising a teenager, lent her the car. No parent (yesterday) any time (he was) supervising a teenager, lent her the car. No parent (yesterday), (he) supervising a teenager, lent her the car. It suffices to falsify all the sentences of (v) and (vi) that one parent on one occasion (yesterday) supervised one teenager and lent her the car, for which it must be that the adverbial is universal in (v) but existential in (vi). As interesting an analytic problem as it is, such variability is proof against the tacit adverbialization proposed here only if it is also proof that the adverbial modification that is plain in (v) and (vi) isn’t there either. One should hope that this variability answers to a common analysis that takes in if-, while-, whenever-, any time-, and absolutive adverbial clauses, to which I would join the adverbials derived by
830
Notes
adverbialization. For discussion and directions, see Higginbotham 2003; Gajewski 2005; Leslie 2009; Klinedinst 2011; Herburger 2015. 41. These logical forms suppress supermonadicity as irrelevant to the discussion. 42. Slighted is the variation from speakers like myself for whom (269) is unambiguously macabre to those for whom the contrast between (268) and (269) is rather weak. I thank Alexander Williams for discussion of this point. 43. As paraphrased rather than while being nations or while in alliances since the adverbial modification here is not temporal. There is no reason to expect that the unspoken adverbials derived by adverbialization will be less polysemous than spoken counterparts, absolutives and conditional clauses. 44. This does not preclude a felicitous use of (282) in which salience in context, prior mention, or a speaker’s gesture cues some further implicit restriction: (i) Venus Pond [when first sampled] was colder than Venus Pond [when next sampled]. 45. Also, (i) Superman does what needs to be done while Clark Kent reports it. (ii) #Superman does what needs to be done while Superman reports it. #Superman does what needs to be done while he reports it. #Superman does what needs to be done while reporting it. 46. The details beyond these broad strokes are deferred until chapters 11–12, along with what is to be said about recounts and identity statements: The morning star and the evening star are one planet, The morning star and the evening star are the same, The two stars are one planet, etc. 47. Alexander Williams (p.c.) cautions that some speakers accept a shift in the understanding of the two tokens of the left in (316) and remarks that whatever counts against it in (316) is much weaker than the grammatical constraint that forces Mo and Lee to hate the same jerk: (i) Sonny and Frank are both jerks. Mo hates him, and Lee does too. I agree and would rather compare (316) to (ii): (ii) Sonny and Frank are both jerks. Mo hates him [the first guy], and Lee hates him. Speakers left to their devices to intone different utterances of (ii) will, I imagine, find it acceptable if unlikely for the second token of him to sometimes refer to Frank. But surely it is a point of grammar that the pronoun destressed must refer the same as the first pronoun and stressed contrastively, not: (iii) Sonny and Frank are both jerks. Mo hates him [the first guy], and Lee hates ’m. (iv) Sonny and Frank are both jerks. Mo hates HIM [the first guy], and Lee hates HIM. Whether or not to switch frames of reference in (316) is subject to a grammar that relates such switches to the intonation pattern on the two tokens of the left. 48. Think of two triangulating agents formulating and exchanging course corrections based on outputs and inputs in the language of a visual guidance system. 49. See Schein 1993, 219–237, for discourse-level quantification over perspectives Π to resolve problems of nonmaximal reference in the use of definite descriptive anaphora.
Notes
831
50. With the secondary predicate (and not without), (336) becomes acceptable if for some evenings in 1892 and not others, Hesperus, the evening star, was aligned with a crescent moon, much as one says Hesperus is Venus in the evening or The evening star is Venus in the evening. 51. It may be doctrine that sentences (335)–(336) denote the same semantic proposition, and it may be argued that the contrary indications of speakers’ judgments answer to a pragmatics of word choice distinguishing the cognitive significance, say, of choosing Venus and Hesperus from twice-choosing Venus, while holding the semantics harmless. This special pleading slights the asymmetry between (335)–(336) and thus the grammar of the thoughts they express. If that be the semantic doctrine, so be it. The interest of cognitive science follows the structure of the thought joining some semantics and some pragmatics, however that goes. If Eventish ℐdentity and while Venus and while Hesperus are the terms in which pragmatic strategies are conceived and communicated, one might wonder why it all goes out the window when the semantics comes up. Still, a speaker whose thoughts (335)–(336) conform in structure to their Eventish logical forms can be trained to respond accurately to queries about the semantic propositions her thoughts express. But to start with thoughts structured around ⌜x = y⌝ forfeits an explanation of the speaker’s differential behavior on display in (335)–(336). 52. Lebanese Arabic distinguishes full and partial agreement: (i) kenet alya w marwaan ʕam yelʕabo was.3fs Alia and Marwan prog playing. ‘Alia and Marwan were playing.’ (ii) kenou alya w marwaan ʕam yelʕabo was.3mpl Alia and Marwan prog playing. ‘Alia and Marwan were playing.’ Where number agreement is semantic, as in Lebanese, agreement is overt pronunciation of one of the descriptive event pronouns holding the sentence together, and it copies relational content from that with which it agrees, “the participants’ events” as paraphrased in (iv). (iii) The sauce and the cheese are a perfect marriage of rivals. ‘The sauce participates and the cheese participates, and all that is a perfect marriage of rivals.’ (iv) The sauce participates and the cheese participates, and the participants’ events are a perfect marriage of rivals. If it is singular as in (i), there must be some relation to events that holds of Alia and not of Marwan, else singular agreement faces a failure of reference, *the participant’s events, *the agent’s events, where there are more than one participant and agent playing. What Alia does and Marwan does not cannot be play, it must be something else such as being the primary focus as opposed to mere accomplice. As the referent of number agreement, this must be some events other than the playing itself. One way or another that does not disturb their play, if number agreement is semantic, (i) and (ii) must not be synonyms, with merely a superficial variation in grammatical agreement. There is instead a close analogy between (i) and (344)/ (346), where the asymmetry intrinsic to the comitative construction has much in common with the asymmetric statements of identity among (337)–(343). Chapter 2 studies the quixotic distribution and variation in the semantic conditions on full and partial number agreement, quantifier scope, and descriptive event anaphora. 53. This section owes its existence to Alexander Williams for his trenchant commentary and discussion.
832
Notes
54. Variously, Bach 1994, 2011; Bezuidenhout 2002; Cappelen and Lepore 2004; Carston 1988; Ludlow 2008, 2014; Recanati 1993, 2007. 55. The “binding argument” so called. For discussion, see Elbourne 2008, Martí 2006, Recanati 2007, and Sennet 2008. 56. Ludlow (2008, 2014) points out that there are characteristic behaviors to flag code switching, intonations, or gestures, such as are in use to flag that this token of bank is not to be taken as a second token of the bank tokened previously or that this Bruce is not the same Bruce as that Bruce. Ludlow embraces the open texture (Searle 1978, 1980) or occasion sensitivity (Bezuidenhout 2002) of lexical meaning and recognizes that throughout the occasion, speaker and hearer entrain each other or negotiate on the fly the meanings of the lexical items for the microlanguage then in use between them. Since this is a dynamic process always in progress, it may be that what appears as a second token of the same lexical item earlier tokened has in the meantime had its meaning qualified. Thus, apparently sound inferences such as There’s a baseball game. Therefore, there’s a baseball game (see Bezuidenhout 2002) are so only in the presence of an enthymeme, ∀x(baseball game1(x) ↔ baseball game2(x)), that the two tokens denote the same. 2
DP and DP
1. See Schein 1993, 275–293, for further skeptical remarks about the place of branching quantification in natural language. 2. Also, (i) No Columbia student participated in such a way and no Harvard student participated in such a way that the Columbia student led and the Harvard student followed. 3. I have in mind to set aside the “telescoping” interpretation (see (100) and section 2.1.2) according to which the sentences assert merely that for any such Columbia student, no more than thirty-five Harvard students were a strike committee with her, which is true of the context. 4. Compare (i) Not many a king in legends with the characters herein, not many a knight in such, not many a succubus in the same, and not many a true lady joining them ever tell a story very different from that of Arthur, the Redcrosse Knight, Duessa and Una in Book I of The Faerie Queene. 5. As in Pietroski 2005, 2006, clausal concatenation itself means conjunction: (i) If Φ, Ψ are wffs; then, ΦΨ is a wff, and Σ satisfy ΦΨ iff Σ satisfy Φ and Σ satisfy Ψ. The ‘&’ conjoining the terms of a familiar Davidsonian decomposition are absent from official Eventish, which observes (i) (appendix 2), so that indeed the decomposition of (47) reads more like a subatomic paragraph like (46) rather than a complex sentence.
Notes
833
6. ‘Δ’ is used throughout to mark the deletion site or gap that Right-Node Raising leaves behind in the first conjunct. 7. Assimilating (6), containing increasing quantifiers, to (7), (9), (11)–(13), with decreasing quanfiers, so that they all deploy descriptive anaphora subjects the analysis to the problem of nonmaximal reference (see section 1.5.0 and appendix 1, “Descriptive Anaphora and Nonmaximal Reference under Selective Perspectives”). 8. I ignore the difference between a mass term and a plural count term. The plural event pronouns so-called are, when made explicit, in fact neutral to the contrast, if a mass term, the water, is assumed, as I do (see Burge 1972, 1975), also to divide its reference, denoting severally anything that is water rather than its singular totality. 9. Note that the tacit adverbial in (120)/(122) copies short (Ludlow 1994), as paraphrased rather than whenever there was linguistics professor endorsing operant conditioning. In contrast, (121) copies long to accommodate the presuppositions of helped him. 10. The ‘telescoping’ in (126) improves on (123). I expect that the independently tensed clauses of (126) have greater resources for referring to the events described by antecedent clauses. I return to this point below with more to say about the logical form of (123). 11. Progovac (1998a, 1998b) appeals to syntactic properties of the coordination to conclude that the coordinated DPs are in construction with some other phrase, which is consistent with either the structure advocated here (i) or a structure that in some way preserves the coordination of DPs as deriving a DP, as in (ii): (i) [S[DP1 Φ][and [DP2 Ψ]]] (ii) [DP[DP1 Φ][and [… DP2 …]]] Ψ She notes that in languages with possessive reflexives DP1 cannot serve as an antecedent to a reflexive contained within DP2: (iii) *Jovani i svojai žena su stigli. John and self’s wife are arrived
[Serbo-Croatian]
Icelandic (Gunnar Olafur Hansson, p.c., December 1999) (iv) Jón elskar kettina sína og hundana sína. John loves cats-the self’s and dogs-the self’s (v) Forstjórarnir og aðstoðarmönnum Þeirra sátu fundinn. CEOs-the and assistants their attended meeting-the (vi) *Forstjórarnir og aðstoðarmönnum sínum sátu fundinn. CEOs-the and assistants self’s attended meeting-the She also observes that an epithet’s disjointness from a c-commanding antecedent, (vii) vs. (viii) (Hornstein and Weinberg 1990), suggests that the first conjunct of (ix) is a phrase that properly contains the DP, as would be the case in either (i) or (ii). (vii) *Every senatori thinks that the bastardi will win. (viii) Every senator’si chief-of-staff thinks that the bastardi will win. (ix) Every senatori and the bastardi’s chief-of-staff think that he will win.
834
Notes
I find these arguments sound as far as they go. Note however that if, as argued, there must be some Φ, it is difficult to imagine what it could mean if the coordination composes a DP. In entertaining (ii) and its variants, the syntax literature (see the references cited in Progovac 1998a, 1998b) typically takes Φ to be a copy or discontinuous part of the conjunction and, with no independent interpretation in situ but with the syntactic effects observed in (iii)– (ix). 12. Alongside the examples coordinating nonincreasing quantifiers, which plainly show scopal independence and therefore “telescoping”, there are also examples that can only be analyzed by assignment of wide scope to the first quantifier: (i) No soldier and his rifle have ever been separated. (ii) No owner and her prize pooch entered the judging ring tugging at opposite ends of the leash. (iii) No wolf mother and any of her cubs remain close for very long. (iv) No Park Avenue therapist and any of her patients invested in Wall Street together. These examples fall in with the asymmetric coordinations that appear to allow violations of the Coordinate Structure Constraint, under special conditions (see Postal 1998, chap. 3 and the references cited there). “Telescoping” in (i) would deliver a reading that asserts that no soldier is separated from his rifle, and whenever there is a soldier, he and his rifle are so separated. Notwithstanding the contradiction, it is unclear to me why such an interpretation is unavailable, except to note that the semantic restrictions on “telescoping” are not particularly well understood. 13. In section 2.2, number agreement comes to be identified with this plural event pronoun and is obviously placed at the left edge of the Right-Node Raised phrase in No Columbia student secretly and no Harvard student openly have.pl conspired. Note that ‘[℩Ei,i : proi,i]’ under pain of infinite regress will not have copied from its antecedent clauses content that includes the deletion site itself Δ, as is explicit in (143). See Ludlow 1994 for discussion of copying short. 14. In a variant of (144), replacing the Right-Node Raised phrase with a decreasing quantifier, it is plain in the meaning that the Right-Node Raised phrase does not occur within the scope of either of the decreasing quantifiers in subject position: (i) Not many a Columbia student proposed to the strike committee to adopt a resolution to encourage violent resistance to—and not many a Harvard student accepted to endorse to the strike committee a call for illegal action against—not many an attempt by police to prevent them from chaining each other to the White House fence. Even so and as above, what is said to involve not many a Columbia student and not many a Harvard student relates to violent resistance to and illegal action against attempts by police to prevent them from chaining each other to the White House, exempting the number concerned with other violent resistance or illegal action. I suppose that there is a further extraposition of not many an attempt… outside the scope of not many a Harvard student (see Schein 1993, 211–214), leaving behind a deletion site subject to an interpretation similar to other deletion sites, the details of which I leave as an open problem. 15. See Dresner 2001.
Notes
835
16. The solution to the semantics problem does not lighten the burden on syntax. The semantics delivers a collective interpretation of (i) that allows the thirty talent shows to be divided between John and Mary: (i) John whistled and Mary hummed in thirty talent shows (between them). Unconstrained, a stray event pronoun deployed in the second conjunct of (ii) as it is in (i) will derive a mistaken interpretation true where John whistled in twenty talent shows and Mary only ten. (ii) John whistled in twenty talent shows, and Mary hummed in thirty talent shows. So it appears that an event pronoun referring collectively to John’s whistling and Mary’s humming can occur in the second conjunct, ‘… [℩E: John’s whistling & Mary’s humming] [∃Y: … 30(Y) …]in[E,Y] …,’ only with concurrent deletion in the first. In a more conventional account, (ii) provides no occasion for the semantics to deviate. Rather Right-Node Raising stands apart as an exotic construction in need of a semantics. The question posed is confined to the meaning of phrases related as in (iii) to two positions, presupposing a gap in the first conjunct: (iii) John whistled ti, and Mary hummed ti in thirty talent shows i. If we can do no better, we can at least offer a comparable stipulation: If Φ is an antecedent for a (null) event pronoun, then a token of that event pronoun must complement Φ. It suffices here for a phrase to complement Φ if it is included within the scope of every quantifier that has Φ in its scope. In (i) and (ii), both “… whistle …” and “… hum …” are antecedents for the event pronoun and provide its content when it is understood collectively. It is thus stipulated that tokens of this pronoun occur as complements within both conjuncts, which condition (i) but not (ii) satisfies. If there is later reason to regiment the order in which event variables are introduced into logical form (cf. Dresner 2001), ‘e0, … , ei … ,’ it may be possible to find a more principled derivation for the syntax of Right-Node Raising. 17. Plural morphology does not really mean “more than one,” and singular morphology attaches to mass terms without implying that a singular object is denoted. Definitions (165) and (166) are provisional in using common coin. See Schein 2006 and the references cited there for discussion of the meaning of number morphology. 18. Adverbialization is elsewhere suppressed in this chapter. 19. And, also from some further event quantification, as section 2.2.0 concludes, that intrudes in the matrix clauses, contained either within the Right-Node Raised phrase (i) or within the remnants (ii): (i) No Columbia student anywhere anywhere a Columbia student is does she participate (‘W’) (secretly) Δ, and no Harvard student anywhere anywhere a Harvard student is does she participate (openly): &, somewhere (near), the sg.event of her participating there was being on a strike committee in which the Columbia student there led and the Harvard student there followed.
836
Notes
(ii) No Columbia student anywhere anywhere a Columbia student is does she participate (secretly) & somewhere (near) Δ, and no Harvard student anywhere anywhere a Harvard student is does she participate (openly) & somewhere (near): the sg.event of her participating there was being on a strike committee in which the Columbia student there led and the Harvard student there followed. 20. Prostrate before the Golden Calf “Compositionality,” these authors see in (172) evidence of obligatory group formation in a conjunction of (in)definite descriptions. Unremarked is that the same restriction against singular agreement recurs in (173) and in (180), with no group formation in sight. 21. That is, the parallel structure in demanding the two antecedents demands that content be copied from both, which could be so short as to be the same, the student there and student there or the participant there and participant there. 22. Or even as in (i), leaving the (in)definite descriptions, along with their adverbialization, inside their conjuncts: (i) [[A(n) NP][℩E : NP][∃E0 : there[E,E0]]Φ Δδ] and [[A(n) NP][℩E : NP][∃E0 : there[E,E0]] Ψ Ωδ] (In)definite descriptions are well known to be unbounded in their scope, exempt from the island conditions that constrain distributive quantifiers. The import of this observation in the present setting amounts to the following. The adverbialization of (in)definite descriptions always has the lower in situ scope indicated for ‘[℩E : NP]’ in (184) or (i), lower in scope than the adverbialization of distributive quantifiers shown in (183). The (in)definite description itself (ii) or its restriction (iii) may take a hike alone unbounded: (ii) [A(n) X0 : ∃E0 NP[E0,X0]] [A(n) X1 : ∃E1 NP[E1,X1]] [… [[℩E0 : NP[E0,X0]] … Φ Δδ] and [[℩E1 : NP[E1,X1]] … Ψ Ωδ]] (iii) [∃X0 : ∃E0 NP[E0,X0]] [A(n) X1 : ∃E1 NP[E1,X1]] [… [A(n) X : [∀x : Xx]X0x] [[℩E0 : NP[E0,X]] … Φ Δδ] and [A(n) X : [∀x : Xx]X1x] [[℩E1 : NP[E1,X]] … Ψ Ωδ]] 23. The analogy between number agreement and other pronouns breaks down in (i): (i) *A Columbia student was and a Harvard student were (too) on a strike committee to divest from the war. Why can’t at least the second token of number agreement be plural if it so resembes a plural pronoun that could refer to a Columbia student and a Harvard student? In the end, a subatomic event pronoun is subatomic after all. I assume that number agreement cannot look beyond its own tensed clause for antecedent events to denote. 24. Under the letter of supermonadicity, (204) expands to (i): (i) … ‘[℩Ei: proi][℩Ej: prok]O[Ei,Ej] [A(n) x: ∃E funky disco beat[E,x]] ∃Ej To[Ej,x]’ Supermonadicity is suppressed in this chapter when there is no harm in it. 25. Let alone the discomfort that this plural-object forming con might not be the same as the con in PPs modifying predicates—con going the way of Aristotle’s and. 26. I slight the typological diversity of the phenomenon, even within Arabic dialects. The variation seems to demand a departure from the simpleminded view of number agreement
Notes
837
and number morphology as a direct reflection of reference and logical form. It is doubtful that all the reported variation could equal variation in reference. I will also discount morphological agreement and so-called feature resolution rules. In many languages, it could very well be that such rules of grammar overlay and obscure the semantic effects discussed in the text, and to the extent they do, such languages do not engage the questions raised here. See the authors cited, especially Munn 1999 and Camacho 1997, and the references there, for discussion of the typological variation. 27. Lina Choueiri and her sisters, Nada and Leila Choueiri, for Lebanese and Ibtissam Kortobi for Moroccan served as my generous and patient consultants between December 1999 and November 2000. The data recorded in morpheme-by-morpheme translation was restored to Lebanese in the winter of 2009–2010 by Sarah Ouwayda, to whom I am grateful for her hard labor and for the data’s corroboration a decade after its original elicitation. The data below from Aoun, Benmamoun, and Sportiche 1994, 1999, as well as Munn 1999, has been modified so that agreement with the first conjunct is always feminine, ruling out the possibility of a default, nonagreeing 3m.sg. form. 28. To the extent that much is made of the internal structure of coordinated DPs to explain how to get to the first DP, these accounts do not offer much toward the analysis of number agreement in languages like Basque (163), which suggest that its dependence on phrases properly contained within a coordination is a more general problem. 29. Van Oirsouw (1987, 232) cites Mohammad Mohammad for the Palestinian (i): (i) Gatal ʔelwalad we-l-banaat ʔel-bisse. Killed.3ms the-boy and-the-girl the-cat 30. Camacho (1997, 109ff.; 2003) observes an analogous contrast between estar and ser in Brazilian Portuguese and Spanish. One could well imagine trading in this contrast in meaning for a difference in structure. Suppose there is a notion of “aboutness” and a sentence is “about” its subject or topic. Eventive sentences or what Brentano ([1874] 1973) calls thetic judgments report changes in the environment and are thus directly “about” events. Stative sentences, or Brentano’s categorical judgments, are “about” the objects to which they ascribe some stative property. In (250), the event is the higher argument, and thought of as a single event, determines number agreement. In contrast, the higher argument in (251) is the plurality of the butter and the flour. I am not unsympathetic to this attack on the problem, but without further amendment, it suggests wrongly that the single event that (i) is “about” should also determine singular number agreement: (i) *became.3s the ingredients dough Some further difference in structure between the conjoined and simple nominals in (250) and (i) bears on number agreement. (See von Fintel 1989; Kuroda 1972, 1988, 1992; Ladusaw 1994; McNally 1998; Sasse 1987.) 31. If number agreement is semantic, and singular refers to either a single participant or an event or events of a single participant, then there is a problem in the syntax of (258), or even more simply of (i), in that there is no such singular reference except from within the scope of the universal quantifiers: (i) DeHket kell bent laughed.3fs every girl ‘Every girl laughed.’
838
Notes
Thus, number agreement is not to be interpreted in situ in (258) or in (i), perhaps to be reconstructed to a position from which the verb has moved by Verb Raising. If singular number agreement implicates Verb Raising, it would have to be decided for the coordination in (258) whether it is extracted across the board from both conjuncts or asymmetrically from just the first. The problem recurs in English in (ii) and (iii): (ii) a. So was every girl laughing? (Cf. *So was the girls laughing?) b. So was every girl and every boy laughing? (Cf. *So was Eve and Adam laughing?) (iii) a. Was every girl laughing? (Cf. *Was any girls laughing?) b. Was every girl and every boy laughing? (Cf. *Was Eve and Adam laughing?) (iv) a. b. (v) a. b.
Every girl was laughing so. Every girl and every boy was laughing so. Every girl was laughing? Every girl and every boy was laughing?
Since it is encountered as soon as number agreement is taken to be semantic and is special to neither coordination nor its analysis, I will not offer much to address it. In this chapter, I will assume that there is little to tell apart the logical form of (ii) and (iii) from that of their counterparts in (iv) and (v), however syntax contrives to make it so—and similarly, for the Lebanese Arabic. Section 3.2.1.4 entertains that number agreement is itself a pronominal clitic subject to clitic climbing to higher position, while chapter 3 more generally examines the syntax of constituents that prima facie coordinated PredPs share to answer how so. 32. Aoun, Benmamoun, and Sportiche’s remarks do not discriminate between spatiotemporal proximity and causal proximity. That (259) evokes a stronger condition than (255) is from the Choueiri sisters. 33. Aoun, Benmamoun, and Sportiche (1999, 676) remark that in English the pronoun in (i) cannot covary with the antecedent quantifier, and they imply the same holds for Arabic. A related example in (ii), however, establishes “telescoping” between tensed clauses in English: (i) *Each woman read a story and her child read a story. (ii) Each woman read a story and then her child read it after her. The facts in Lebanese are analogous, (iii) and (iv), with the further qualification that subject inversion in the first clause precludes “telescoping” in the second, (v) and (vi): (iii) *kell mara aryet essa w ebn-a aryou essa *Each woman read.3fs story and child-her read.3ms story. (iv) kell mara aryet essa w baʕdein ebn-a ara-ha baʕd-a Each woman read.3fs story and then child-her read.3ms–it after-her (v) *’aryet kell mara essa w baʕdein ’aryouw-a ebna baʕd-a * Read.3fs each woman story and then read.3ms–it child-her after-her (vi) *’aryet kell mara essa w baʕdein ’aryouw-a saami baʕd-a * Read.3fs each woman story and then read.3ms-it Sami after-her The implications for a scopal analysis of these facts are discussed below. 34. The “sloppy” interpretation of the Right-Node Raised constituent in the first conjunct is fully general, whether introduced there as the content of a pronoun denoting events, or copied
Notes
839
in situ directly. I suppress irrelevant details for the simpler logical form in (115), with the proviso that it should not be mistaken as “official”; see the following note. 35. The logical forms in this chapter have various reformulations in Eventish (chapter 1), which I will suppress in the belief that little except as noted is at issue here that cannot be expressed with simpler and more readable logical forms. Still suppressing adverbialization and the expansion of ‘Agent’ and ‘At’ to include the participation relation ‘W’, (115) in proper Eventish might be (i) ⌜sg.Φ[Ei]⌝ and ⌜pl.Φ[Ei]⌝ (as in (165)–(166)): (i) ∃E0(sg.laughed.3f[E0] [℩E0: pro0] [℩E1: pro1]O[E0,E1] [℩X : ∃E Alia[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]), and ∃E0(sg.laughed.3m[E0] [℩E0: pro0] [℩E1: pro1]O[E0,E1] [℩X : ∃E Marwan[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]) But in truth, the position of ‘[℩E0: pro0] [℩E1: pro1]O[E0,E1]’ and its constituents is underdetermined, at least until further discussion later in this chapter and in chapter 4. The Eventish for (115) could just as well be (ii) or (iii) among others: (ii) ∃E0(sg.laughed.3f[E0] [℩X : ∃E Alia[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0][℩E1: pro1]O[E0,E1] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]), and ∃E0(sg.laughed.3m[E0] [℩X : ∃E Marwan[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0][ ℩E1: pro1]O[E0,E1] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]) (iii) ∃E0(sg.laughed.3f[E0] [℩E0: pro0][℩X : ∃E Alia[E,X]] ∃E1 Agent[E1,X] [℩E1: pro1]O[E0,E1] [℩E0: pro0][℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]), and ∃E0(sg.laughed.3m[E0] [℩E0: pro0] [℩X : ∃E Marwan[E,X]] ∃E1 Agent[E1,X] [℩E1: pro1]O[E0,E1] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]) All (i)–(iii) parse as in (iv), where in the first conjunct laughing E0 is asserted followed by a clause describing Alia’s participation in it (i.e., E0), and then a clause about a linguisti professor’s participation in it: (iv) ∃E0[laughed.3fs[E0] [[Alia [… Agent …][E0]] [A ling prof[… At …][E0]]] and ∃E0[laughed.3ms[E0] [[Marwan [… Agent …][E0]] [A ling prof[… At …][E0]]] The second conjunct about Marwan parses the same way. Squinting only slightly, (iv) is (264), and so I will continue in the text with the likes of (115) leaving aside the “official” language (i)–(iii), which can be reached reversing my steps via (iv). Similar remarks attach to (272), exemplifying collective predication as indicated by plural number agreement, which Eventish could represent as in (v) turning to a plural event pronoun for collective reference: (v) ∃E0(pl.laughed.3m[E0] [℩E0: pro0] [℩E1,1: pro1,1] O[E0,E1,1] [℩X: ∃E Alia[E,X]] ∃E1 Agent[E1,X] and [℩X: ∃E Marwan[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X])
840
Notes
But (v), although true to its meaning, does not look much like (272) since neither of its conjuncts says anything about the laughing E0. Various Eventish logical forms are however available that bear the resemblance, under various assumptions about Lebanese Arabic syntax. If, for example, there is both Verb Raising and Right-Node Raising, the sentence parses as in (vi) and the corresponding Eventish is (vii), in which the gap Δδ is left unexpanded: (vi) [Laughed.mp [[Alia-Agent-Δδ] and [Marwan-Agent- [δ V Φ]]]] (vii) ∃E0(pl.laughed.3m[E0] [℩X: ∃E Alia[E,X]] ∃E1 Agent[E1,X] Δδ and [℩X: ∃E Marwan[E,X]] ∃E1 Agent[E1,X] (δ [℩E1,1: pro1,1][℩E0: pro0] O[E1,1,E0] laugh[E0] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X])) A collectivized event pronoun ‘[℩E1,1: pro1,1]’ prefixes the Right-Node Raised phrase, as discussed in section 2.1.3. Interpreting the gap, recall, introduces a second event pronoun the antecedent of which is the Right-Node Raised phrase, effectively copying it into the first conjunct. This results in a logical form that may be parsed as in (viii), in which there are clauses, one describing Alia’s participation and the other describing Marwan’s participation in the same laughing E0, resembling in this respect (272): (viii) ∃E0[Laughed.mp[E0] [[Alia-Agent-Δδ[E0]] and [Marwan-Agent- [δ V Φ[E0]]]]] Verb Raising is hardly necessary for Eventish or for the formal resemblance to (272). Simply strike ‘laugh[E0]’ from (vii) for a logical form assuming that the verb originates in initial position. For that matter, Right-Node Raising is also not necessary for the resemblance if there is some other excuse for a token of ‘E0’ to be introduced into the first conjunct as in this minimal variant on (v): (ix) ∃E0(pl.laughed.3m[E0] [℩E1,1: pro1,1] [℩X: ∃E Alia[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0]O[E0,E1,1] and [℩X: ∃E Marwan[E,X]] ∃E1 Agent[E1,X] [℩E0: pro0]O[E0,E1,1] [℩E0: pro0] [℩E2: pro2]O[E0,E2] [∃X: a(n)[X] ∃E ling prof[E,X]] ∃E2At[E2,X]) But if it is to be maintained that each conjunct describes participation in the collective event, as (272) presumes, then, given section 2.1, it seems that Right-Node Raising is the route to it with or without Verb Raising. This presumption of (272) is however just that, an artifact of a logical form that restricts itself to a single event variable to tie the conjuncts together, and one that Eventish is under no constraint to prop up, as (v) does not. The arguments to follow about the distribution of existential event quantifiers and the phrases that prompt their introduction do not, I believe, depend on this artifact of the simpler logical forms in which the discussion is couched. Even so, carrying on with (272) rather than (vii) comes with a caution. Commenting on the same event, ‘Agent(e, Alia)’ and ‘Agent(e, Marwan)’ must be read to mean that Alia is an Agent in e and Marwan is an Agent in e. That they are the only Agents in the laughing goes unrepresented, omitting the plural event pronouns—that is, omitting the very instruments for collective, definite reference (chapter 1). Despite the inaccuracy, the uncomplicated (264), (272), and the like are more helpful adjuncts to the argument and discussion of empirical questions in this chapter. 36. Example (276) flags a problem along with the point it illustrates. Alongside obligatory plural number agreement and plural pronoun, singular a strike and singular noun room are
Notes
841
acceptable and interpreted distributively, the effect I assume of a tacit each, which is equally selective in the singulars licensed: (i) They were each on a strike committee that met in their dorm room. (ii) *They were each on a strike committee that met in her dorm room. Lebanese Arabic restricts the distribution of tacit each (see section 2.3.2.3). It is excluded here from intervening to give each a linguistics professor or Woody Allen joke of their own to laugh at, and thus the Lebanese Arabic reflects more transparently the distribution of event quantifiers. I leave aside explaining the behavior of each ruling out (ii). 37. See (vii) in note 35 (and chapter 1) for expanded Eventish. Instead of collective reference to the boy’s and girl’s actions, a pronoun could just as well refer to the laughing itself (see (ix) in note 35). (The event pronoun that supermonadicity introduces with an (in)definite description (and all DPs) includes only the thematic relation within its scope, therefore excluding at a linguistics professor. See section 2.2.) 38. The division also coincides with the distinction in Karttunen 1977, Heim 1982, and Kamp 1981 between DPs that introduce discourse referents and those that are purely quantificational. 39. And perhaps also within the scope of distinct tokens of ‘[∃E0 : there[E,E0]]’ if, as entertained in note 22, distributive quantification involves higher position for more event quantification than just adverbialization: (i) [Every NP] [℩E : NP] [∃E0 : there[E,E0]] [[∃E1: O[E0,E1]]W1 (secretly) Δδ] and [Every NP] [℩E : NP] [∃E0 : there[E,E0]] [[∃E1: O[E0,E1]]W1 (openly) [℩E1 : sg.pro1[E1]].Ωδ] For (i), the content of the descriptive anaphor would be ‘[℩E1 : sg.( there[E,E0] O[E0,E1] W1)[E1]]’. 40. ⌜sg.Φ[E]⌝ as defined in (165). 41. Suppose that the syntax of distributive quantification is movement to a projection ΠP the head of which is the relation there (which might be thought to narrow perspective or frame of reference) (see Szabolcsi 1997, 2009, for discussion of the “cartography” of quantification): (i) [ΠP [D NP] [℩E : NP] ∃E0 [[Πthere[E,E0]] XP]] ΠPs are licensed only by distributive quantification or to meet the syntactic constraint that coordination coordinates only phrases of like category—so that, in particular, ΠPs coordinate only with other ΠPs: (i) … [ΠP [Every teacher][℩E : NP] ∃E0 [[Πthere[E,E0]] [XP[∃E1: O[E0,E1]] W1 Δδ]]] and [ΠP ∃E ∃E0 [[Πthere[E,E0]] [XP [℩Alia][℩E : NP At[E0,E]] [∃E1: N[E,E1]] W1 [℩E1 : pro1[E1]] in roomδ]]] The logical form in (i) assumes the more extreme alternative in note 22 for the contrast between the structure of distributive quantification and (in)definite description. Note that (in)definite description in (i) manages to remain in situ in XP. Existential closure, as shown, or context supplies the argument E to the relation there in the highest projection of the second ΠP. The two tokens of ‘∃E0’, there only because these are both ΠPs, license the “sloppy” interpretation of the event pronouns ‘[℩E1 : pro1[E1]]’, the one shown in the second clause and
842
Notes
the one deleted under Δδ. (Except to reflect usage elsewhere, the near synonyms there, N (for ‘Neighborhood’), At, and O (for ‘Overlap’) need not all be in use.) 42. See the remarks in note 31 on Verb Raising and Verb-Raised number agreement. 43. Number agreement with preverbal DPs provides a closer comparison between English and Lebanese. Lebanese agrees with English on (i): (i) Every Columbia student and every Harvard student was on a strike committee. Moreover, a preverbal conjunction of (in)definite descriptions does not tolerate singular, partial agreement under any circumstances (Aoun, Benmamoun, and Sportiche 1994) ((214) is full agreement and reference to the same object). Also, as in English and contrasting with the postverbal case in (iii), Alia and Marwan can be taken in (ii) to laugh at different professors: (ii) alia w marwaan DeHkou ʕa esteez linguistics Alia and Marwan laughed.3mp at professor linguistics (iii) DeHkou alia w marwaan ʕa esteez linguistics laughed.3mp Alia and Marwan at professor linguistics To complete the paradigm, it should be noted that a mixed coordination also requires plural agreement in preverbal position while allowing each of Alia and the students different targets for their laughter: (iv) alia w kell telmiiz DeHkou ʕa esteez linguistics Alia and every student laughed.3mp at professor linguistics (v) kell telmiiz w alia DeHkou ʕa esteez linguistics every student and alia laughed.3mp at professor linguistics With (ii) and (iv)–(v), the possibility of singular number agreement and the accessibility of a distributive or “sloppy” interpretation part ways. I return to the syntax of (ii) and (iii) later (section 2.3.2.3), where the location of tacit occurrences of each will derive the difference (see note 36). The obligatory plural agreement in (iv) and (v) is unexpected under the current analysis, which accords with the option in English for singular agreement: (vi) Every student and Alia is laughing at a linguistics professor. (vii) Alia and every student is laughing at a linguistics professor. The terms that here translate number agreement themselves denote events, with number being reflected in whether they are the events of the many, whatever some did, or the events of the one, whatever one did. Hoeksema (1983, 1988) makes some further observations that are congenial to the spirit of this approach in suggesting that number counts events, or scenes, individuated by objects rather than the objects themselves: (viii) a. b. (ix) a. b. (x) a. b. (xi) a. b.
Dr. Jekyll and Mr. Hyde *was/*were hung this morning. Dr. Jekyll and Mr. Hyde ?was/were the same person. The doctor and the killer was/*were hung this morning. The doctor and the killer ?was/were the same person. A doctor and a killer was/*were hung this morning. A doctor and a killer ?was/?were the same person. Dr. Jekyll and the killer *was/*were hung this morning. Dr. Jekyll and the killer ?was/were the same person.
Notes
843
To Hoeksema’s examples, one may add: (xii) The morning star and the evening star are/*is following each other season to season like morning follows evening. (xiii) From the vantage point of Jupiter, you can see that the morning star and the evening star is/*are circling the sun. See section 1.6.3 and chapters 11–12 for scene-counting number agreement. 44. A spatiotemporal relation often suffices for “telescoping” between fully tensed clauses, which typically need not be related causally. At the subatomic level depicted in (153), I assume that modifier and modified do not yet include expressions of tense and temporal relations, in contrast to causal relations, which have a way of showing up even in the analysis or decomposition of uninflected verbal phrases. 45. In note 12, a distributive quantifier is extracted from the first clause to include within its scope the entire coordination. If (298)–(300) were treated similarly, the distributive quantifier would include within its scope the subject of the second clause and it could directly bind any pronoun that happened to occur there. Such violations of the Coordinate Structure Constraint, those that are not across the board, are themselves well known (see Postal 1998, chap. 3 and the references cited there) to require a causal or spatiotemporal relation between the events that the conjoined clauses describe. If there is a difference, I have not teased apart the spatiotemporal and causal relations that “telescoping” invokes from those invoked by asymmetric coordination. But, either way (298)–(300) is analyzed, the semantic condition indicates a coordination of clauses. To obtain covariance, either there is adverbial modification of the second clause, or the subject of the first clause is asymmetrically extracted out of that clause under the special and rather similar semantic conditions of that construction. 46. Thanks to Sarah Ouwayda (July 2010), the following examples are much improved from an earlier draft. 47. It is ‘N[E,E′]’ elsewhere in this book. 48. In Eventish, it looks like (i) (suppressing some supermonadicity): (i) ∃El [Laughed.3fs[El] [℩El : prol[El]][℩E0 : pro0[E0]] O[El,E0] [[Every x: ∃E woman[E,x]][℩E: wm[E,x]] [[∃E0: there[E,E0]] W0[E0,x] Δδ and [∀E′: [∀e: E′e]Ele [∃x: ∃E wm[E,x]][℩E: wm[E,x]] there[E,E′] W0[E′,x] Δδ] [∃E″: C[E′,E″]] [℩x: ∃E child[E,x]] [℩E: child[E,x]] [[∃E0: [∀e: E0e]E″e there[E,E0]] W0[E0,x] (δ [℩E0 : pro0[E0]][∃Y: a(n)[Y] ∃E ling prof[E,Y]] [℩E: ling prof[E,Y]] at[E0, Y] δ) ]] 49. The descriptive anaphor her expands as ‘[℩y : woman(y) & … W(e′,y) …]’, and her child as ‘[℩x : child(x) & ∃e(Possess(e) & Possessed(e,x) & [℩x : woman(x) & … W(e′,x) …] Possessor(e,x))]’
844
Notes
50. Recall the content (287) of the event pronoun in (286), where different event quantifiers bind ‘E’: (137) [Every NP] [℩E : NP] [[∃E0 : there[E,E0]] W0 (secretly) Δδ] and [Every NP] [℩E : NP] [[∃E0 : there[E,E0]] W0 (openly) [℩E0 : sg.pro0[E0]].Ωδ] (138) [℩E0 : sg.(there[E,E0] W0)[E0]] Event pronouns elsewhere simply omit the number morphology sg. 51. Or, alternatively, if it would help (which it won’t) to whatever she does there: (i) [℩e″ : C(e′,e″) & [℩y: y = x] W(e″, y)]. Omitting relativization to e′—that is, omitting ‘C(e′,e″)’ for a shorter descriptive content— puts singular reference at risk and does not make it any easier for the conjuncts to describe different events of laughing at a professor with different professors in them. So this alternative is not considered further. 52. ‘[℩e′ : proi]’ may expand as ‘[℩e′ : e′ ≤ e … Wi(e′,x)]’ or as ‘[℩e′ : e′ ≤ e … ∃xWi(e′,x)]’, etc. Crucially, it abstracts on the same variable, e′, as its antecedent description. 53. Returning to (183), note that it violates the constraint in that the first remnant exposes e′ and the second, e″′. “Backtracking” in the use of event variables manages to elude the proposed constraint in that the remnants of both conjuncts in (i) are in e′: (i) ∃e(Laughed(e) & [every x : woman(x)][∃e′ : e′ ≤ e] … Wi(e′,x) & [℩e′: proi][∀e: e ≤ e′] [℩e′: ∃e″(C(e′,e″) & e′ ≤ e″) & [℩y: y = x] W(e′, y) & at(e′, a linguistics professor)] e ≤ e′, and [∀e′ : e′ ≤ e & [∃x : woman(x)] … W(e′,x) …][∃e″ : C(e′,e″)] [[every x : child of hers(x)] [∃e′ : e′ ≤ e″] (… W(e′,x) … & [℩e′: ∃e″(C(e′,e″) & e′ ≤ e″) & [℩y: y = x] W(e′, y)] at(e′, a linguistics professor))) Undermining the comitativity condition, the result is a pathological meaning according to which the woman laughs at a linguistics professor, the same one no matter which if any of her children she laughs at him with, and the child is said to do something with her mother but whatever she does, with or without her mother, is at a linguistics professor. “Backtracking” the event variable apparently does no better to rescue the meaning that combines comitativity with a “sloppy” interpretation of at a linguistics professor. But, to avoid such pathological interpretations, there must also be a constraint against “backtracking” regimenting the use of event variables in the successive restriction of one domain of event quantification by reference to events from a higher domain (cf. the “familiarity” condition of Heim 1982; Dresner 2001): (ii) [∃e1: R1e1e0](… [∃e2: R2e2e1](… [∃e3: R3e3e2](…))) (iii) *[∃e1: R1e1e0](… [∃e2: R2e2e1](… [∃e1: R3e1e2](…))) 54. The formal difference may underlie a more substantive comparison between (197) and (198) according to which (197) expresses a thetic judgment, directly locating some events in the ambient environment, and (198), a categorical judgment, which locates the verifying events relative to each subject’s biography. (See note 30 for references.)
Notes
845
55. Molecular in that they include lexical items and larger phrases in their scope. The subatomic event quantifiers of the next section include only thematic relations in their scope. 56. Singular agreement in (283) and (370), only under the comitative conditions imposed by a different structure, ‘… Cum …’ (see section 2.5.0), and the condition of eventivity makes this structure unacceptable in (283). 57. As before, ‘C(e′,e″)’ is the causal, comitative, or accessibility relation by which adverbs locate events. Note the delicate arrangement between the adverbs, today and yesterday, and the thematic relation ‘W(e″,x)’. In a sentence such as (i), we don’t want to lose the participation of either of the two girls in today’s event or of either of the two boys in yesterday’s event: (i) šerbou benteen lyom w sabyeen mbeereH anninet pepsi drank.pl girl.dual today and boy.dual yesterday bottle Pepsi Thus it would be incorrect to paraphrase (i) as (ii): (ii) There was some drinking in which two girls participated and some of which took place today, and … Rather, it should be as in (iii): (iii) There was some drinking and two girls in some of it today participated, and … The occurrence of ‘W(e″,x)’ within the scope of the adverbs in (378) guarantees that all those that the subject denotes participate at the time mentioned. To fix this scope arrangement, I must appeal to a particular syntactic structure, namely, that a subject occurring to the left of an adverb (e.g., DP-today-Φ) leaves behind within Φ, at least a copy or the original itself of the thematic relation ‘W(e″,x)’. 58. The logical forms (378) and (379) assign drank.pl scope over both conjuncts. An equally plausible alternative leaves it within the first conjunct and puts a coreferring event pronoun in the second. 59. There is here an argument for locating the pronoun cross-referring to events at Δ rather than at a position higher in the first conjunct. It is that its location at Δ allows for the adverbs to induce a distributive, “sloppy” interpretation. 60. But there is dialect variation and object number agreement that I neglect. (See McCawley 1988, 532ff.; Camacho 1997, 184ff., 2003.) 61. Event concepts are first-order—for example, sparse[E] ↔ ∃eEe & ∀e(Ee → sparse(e)) (Schein 1993, 119ff.; 2002, sec. 1.2). Note that if the pines are to delineate a sparsity of only the pines and the oaks, a different one of only the oaks, the thematic relation ‘W[e′,X]’ must express that X exhausts the W-ers in e′: ‘W[e′,X] ↔ ∃xXx & ∀x(Xx ↔ W(e′,x))’ (Pietroski 2005). (See Schein 2002 for a general discussion of collectivized thematic relations.) The exhaustivity that ‘[℩E : pl.pro]’ enforces is only that these events of being sparse involve all and only the pines and the oaks, without specifying more precisely the participants in the individual events. Chapter 1 argued for the separated assertion of exhaustivity that ‘[℩E : pl.pro]’ accomplishes, pointing out that such an assertion is required outside the scope of the modal adverb in The Columbia students and possibly the Harvard students. That argument still holds and is consistent with the present contention that the thematic relations introduce event quantifiers and also express exhaustivity.
846
Notes
62. Pietroski (1998), arguing from rather different considerations, also reaches the conclusion that thematic relations themselves occasion the introduction of an event quantifier. 63. In light of the locality of event quantifiers, the logical form for (163) sketched out with some convenient shortcuts in (237) is found to be misleading: (163) Lindak ardau eta Anderek esnea edaten dabez. Linda wine and Ander milk drink be (3PL) ‘Linda will drink wine and Ander milk.’ (Sjoblom 1980)
[Basque]
(237) … ∃e([℩x : Lx]Agent(e,x) & ∃eC(e,e′) & [℩x : Wx]Patient(e′,x)) and ∃e([℩x : Ax]Agent(e,x) & ∃e′C(e,e′) & [℩x : Mx]Patient(e′,x)) … It should rather be as in (237′), (237′) … ∃e″( [℩x : Lx][∃e : e ≤ e″]Agenti(e,x) & C(proi, proj) & [℩x : Wx][∃e : e ≤ e″]Patientj(e,x)) and [℩x : Ax][∃e : e ≤ e″]Agenti(e,x) & C(proi, proj) & [℩x : Mx][∃e : e ≤ e″]Patientj(e,x)) …) where the event quantifiers licensed by thematic relations have narrowest scope and descriptive pronouns relate what the agent did to what happened to the patient. This logical form is in any case more faithful to the conclusions of chapter 1. 64. Or, perhaps by a tacit both, in the case of the English counterparts to (394) and (395): (i) (ii)
The pines and the oaks were both sparse in a region near human settlement. The pines and the oaks both covered a region near human settlement.
(iii)
The oaks, firs, birches, aspens, alders, elms, and poplars were all sparse in a region near human settlement.
(iv) (v) (vi)
?The pines and the oaks were each sparse in a region near human settlement. ?The pines and the oaks each covered a region near human settlement. ?The oaks, firs, birches, aspens, alders, elms, and poplars were each sparse in a region near human settlement.
(vii) *The pines and the oaks were all sparse in a region near human settlement. (viii) *The pines and the oaks all covered a region near human settlement. [unless nondistributive] Compare (ix) *The trees (which are (the) pines and (the) oaks) are {both/each/all}sparse in a region near human settlement. (x) *The trees (which are (the) oaks, (the) firs, (the) birches, (the) aspens, (the) alders, (the) elms, and (the) poplars) are {both/each/all}sparse in a region near human settlement. With a class of predicates that includes sparse (Dowty 1987; Taub 1989), distributive operators can distribute only over conjoined subjects, (i)–(ii) vs. (xi)–(x). It is sometimes concluded from the evidence of (i) and (ii) that the conjoined subject names an object {the pines, the
Notes
847
oaks} with two parts, and similarly that the subject of (iii) refers to an object {the oaks, the firs, the birches, the aspens, the alders, the elms, the poplars}, any part of which is the trees of a particular variety. If it is given that the conjoined subject refers, then I would have to agree that the meaning of these sentences and the contrast between (i)–(ii) and (ix)–(x) argue that the conjoined subject names the object with the parts as enumerated. There is, however, nothing in the behavior of distributive operators to argue that a conjoined subject refers, since they behave just the same when their antecedents occur in overtly separate clauses: (xi)
The pines suffered from blight, the oaks were infested with borers, and both were now sparse in a region where they had once thrived. (xii) ?The pines suffered from blight, the oaks were infested with borers, and each was now sparse in a region where they had once thrived. (xiii) *The pines suffered from blight, the oaks were infested with borers, and all were now sparse in a region where they had once thrived. (xiv) The oaks were infested with borers, the firs lost their needles, the birches were shivering, the aspens were quaking, the alders were diseased, and the elms were confused with beeches. All were now sparse in a region where they had once thrived. (xv) ?The oaks were infested with borers, the firs lost their needles, the birches were shivering, the aspens were quaking, the alders were diseased, and the elms were confused with beeches. Each was now sparse in a region where they had once thrived. For discussion of the interaction of distributive operators and predicates like sparse, see Brisson 1997, 1998, 2003; Dowty 1987; Gillon 1990; Lasersohn 1989, 1995; Schein 1993, 302; chapter 1, note 1, and chapter 12 of the present book; Schwarzschild 1991, 1996; Taub 1989. 65. See Hinterhölzl 1999, 2006, and the references there for discussion of the West Germanic VP. 66. One could entertain that ‘[℩E : sg.pro],’ the translation of singular number agreement, take the place of the plural ‘[℩E : pl.pro]’ and that its tokens, the copy within Δ and the overt morpheme in the second conjunct, refer respectively to what Boris did and what Natasha did. But such a construal ends up incoherent. Since these tokens of singular agreement are evaluated with respect to the same event, such a construal would say that Boris is its only participant and so is Natasha. 67. Without Δ, (417) would come to resemble (418) too much for any plausible account of reflexivization and we would expect the reflexivization in (413b) and (415b) to be acceptable, just as it is in (414b) and (416b). 68. Or, as Pietroski (2005, 2006) argues, phrasal concatenation—that is, subordination—itself means conjunction (rather than, say, function application): σ satisfies X YP iff σ satisfies X & σ satisfies YP. Thus and but not ‘&’ appears in logical form, retaining in a more perspicuous fashion the contrast between coordination and subordination. (See also Larson and Segal 1995.) 69. Given the opacity of and-phrases, (411) should be emended so that past tense occurs in the second conjunct and is copied along with the rest of Δ into the first:
848
Notes
(411) (∃E)([℩x : Bx] [∃E′ : [∀e: E′e]Ee](Wi(E′,x) & Δ) and [℩x : Nx](Past(E) & [℩E′ : [∀e: E′e]Ee & pl.proi,i] win(E′) & [∃E′ : [∀e: E′e]Ee] (Wi(E′,x) & [℩E′ : sg.proi] [∃Y: $100(Y)] Theme(E′,Y))) 70. English can also represent this interpretation via tacit each. 71. See note 70 above. 72. Molecular in that they include lexical items and larger phrases in their scope. Recall from section 2.3.2.2 that there are subatomic existential event quantifiers, suppressed in (411) and (429), that include only thematic relations in their scope. If the distribution of partitive event quantifiers is exactly as just stated, then the analysis of Basque (163) must claim that both conjucts are tensed. Otherwise, the causal and topological relations bridging thematic relations can provide a further locus for partitive event quantification as suggested earlier. 73. See Schein 1993, 150ff. 74. See sections 2.3.0 and 2.3.2 as well as Schein 1993, 119ff. 75. Given that the sentence contains only one existential, partitive event quantifier, alternative permutations of whatever Boris did, whatever Natasha did, and some of whatever α did suffer from defects analogous to (284)’s. 76. Event domain restrictions as here and in Plurals and Events are sometimes implicit (“modalized”). Event quantifiers appear under two licenses: those that are original with phrases, lexical items such as adverbs, tense, and so on, and those—all definite descriptions glossing pro—that are introduced via the system cross-referencing the events described in antecedent clauses. I assume that original event quantifiers are introduced with original event variables. Hence, the existential event quantifier in (434)/(435) binds a new variable, e′, that has not been seen before. Since the domain of e′ must now be restricted by some relation to higher events e, it does so explicitly, as in (434)/(435). In the case of pro, we come across the pattern ‘∃e(Φi[e] & Ψ[proi])’—that is, ‘∃e(Φ[e] & [℩E : Φ[e]]Ψ[e])’—where the intent is for pro to refer to whatever Φ-ing there is within e, where the definite description is evaluated with respect to the received value of the variable e, without explicitly representing it as such. The official Eventish is, however, to represent explicitly the domain restriction. 77. Recall that the worrisome logical form is specifically one where the Right-Node Raised constituent contains the existential event quantifier, occasioning the distributive (i.e., “sloppy”) interpretation when it is copied into Δ. Hence I cite in (424), (436), and (437), logical forms and paraphrases that locate the existential event quantifier within the scope of both the subject and the comitative phrase. As with (426) and (428), locating it elsewhere will not, however, make a stray Δ any more coherent, even if one is willing to concede that a “sloppy” interpretation of ‘won $100’ is beyond reach. 78. In the text, I treat reflexives as pro-nouns, context-dependent on their antecedents. In the larger scheme of things, I would favor a univocal content, ‘[℩x : self[e,x]]’ (see Peacocke 1981), but I am eager to forestall discussion of modes of presentation and what it means to be the self in an event. Either way, the content of a reflexive is always a relation to events e, and that formal point is all I require to tell apart coordination and subordination. (See section 2.6 for more on reflexives.)
Notes
849
79. As we have just seen, the locality conditions on binding reflexives affect their content as well as their reference. Reciprocals and distributive adverbs show similar effects. Dalrymple et al. (1998b) consider dual readings for the Russian reciprocal drug-druga (men vs. women, and person vs. person) and the distributor po in (i) and (ii): (i) mužčiny s ženščinami brosili drug v druga po basketbol’nomu mjaču. men.nom with women.inst threw other at other po basketballball mužčiny i ženščiny brosili drug v druga po basketbol’nomu mjaču. men.nom and women.nom threw other at other po basketball ball (ii) mužčiny s ženščinami brosili drug v druga po snežku. men.nom with women.inst threw other at other po snowball mužčiny i ženščiny brosili drug v druga po snežku. men.nom and women.nom threw other at other po snowball They note that of the four possible combinations only one is proscribed, namely, that the men and women threw, each person at another, the men using one ball and the women, another. The same restriction applies in English and extends beyond the reciprocal phrase each other to the use of one, (an)other, each, both, as one would expect if the reciprocal phrase is to be analyzed in terms of the latter (see Schein 2003). (iii) The English chafed. The French seethed. (The) one insulted the other and the other answered, both (of them) lobbing a barb at the other’s privates. (iv) The English chafed. The French seethed. (The) one insulted the other and the other answered, each (of them) lobbing a barb at another’s privates. (v) The English chafed. The French seethed. (*The) one insulted another and the other answered, both (of them) lobbing a barb at the other’s privates. (vi) The English chafed. The French seethed. (*The) one insulted another and the other answered, each (of them) lobbing a barb at the other’s privates. The English chafed. The French seethed. (*The) one insulted another and the other answered, each (of them) lobbing a barb at another’s privates. Suppose that one … other is one … other than that one, where both tokens of one indicate the same content. Thus if the one is taken to refer to the English, then the other must refer to the other than the English in this context, namely, to the French, as in (iii) and (iv). In contrast, if again the one refers to the English, another denotes someone in the context other than the English—that is, a French seether. But, if the speaker intends to report what became of all that chafing and seething, which engaged all the English and all the French, she either neglects some of the French in saying that the English insulted some French or has infelicitously withheld that she knows that the English insulted all the non-English—that is, the French. If the point is a national rivalry, it is one nation against the other ((iii), (iv)). On the other hand, if the speaker uses one to denote an individual in the context or the one to denote the individual in each of some (sub-)events in the context, another denotes anyone other than that one, English or French, and the other fails singular reference (unless one makes the effort to accommodate subevents with pairs of participants). Thus, if the point is to describe an Anglo-French melée, it is one against another ((v), (vi)). While the choice between the other and another fixes the reciprocity, it also restricts the distributive modification expressed by
850
Notes
the final gerund (and the reference of them). Once the melée confounds the English and French ((v), (vi)), the barbs they use do not conform to two national standards. In (v), both quantifies only over the two persons, insulter and insulted, within each event. It cannot mean that the English were lobbing a barb (their favorite) and the French, lobbing another. See my “Appendix: Comitative and Instrumental with DP” (Schein 2002) for further discussion of the semantics of comitative phrases. For discussion of their syntax, semantics, and effects on number agreement (to which I return below), see McNally 1993; Camacho 1995, 1997, 2000, 2003; and Dalrymple, Hayrapetian, and King 1998a. 80. The solution to the semantic problem of Right-Node Raising and its application to Lebanese (445)–(465) threatens to overgeneralize, since it seems to be a fully general capacity of the system to make collective remarks about individuals who have been introduced singly in separate antecedent clauses. The question then is why it should ever fail, as it does in Aoun, Benmamoun, and Sportiche’s (1994, 1999) accounts and Aoun and Benmamoun’s (1996) report of certain cases in Arabic: (i)
*ltaʔa Kariim w Marwaan. met.3ms Kareem and Marwaan
(ii)
*tqasəm Omar w Karim l-ɣalla. divided.3ms Omar and Kareem the-harvest (Aoun and Benmamoun 1996)
[Moroccan]
(iii)
*tʕanəq Omar w Nadia. embraced.3ms Omar and Nadia (Aoun and Benmamoun 1996)
[Moroccan]
(iv)
*biħibb Kariim w Marwaan ħaalun. love.3ms Kareem and Marwaan themselves
(v)
*biħibb Kariim w Marwaan baʕḍun. love.3ms Kareem and Marwaan each.other
(vi)
*ʔaʕad Kariim w Marwaan ħad baʕḍun. sat.3ms Kareem and Marwaan near each.other
(vii) *ʔəža Kariim w Marwaan maʕ baʕḍun. came.3ms Kareem and Marwaan with each.other (viii) *raaħ Kariim w Marwaan sawa. left.3ms Kareem and Marwaan together (ix)
*raaħ Kariim w Marwaan tnayneetun. left.3ms Kareem and Marwaan both
(x)
*ʔakal Kariim w Marwaan təffeeħ kill waaħad. ate.3ms Kareem and Marwaan apple each one
(xi)
*raaħ l-walad w l-muʕallim yalli ltaʔo ʕala-l-madrase. went.3ms the-boy and the-teacher who met.p at-the-school
(xii) *rfəD Omar w Karim yəmŝiw. refused.3ms Omar and Kareem go (Aoun and Benmamoun 1996)
[Moroccan]
Notes
851
The sentences (i)–(x) all contain reflexive or reciprocal morphemes (including the reflexive verbal affix -t- in (i)–(iii), as Lina Choueiri points out), or distributive or collectivized items bound to local antecedents, (viii)–(x). The crucial observation, borrowing a page from Camacho 1997, 2003, is that such elements, unlike other pronouns, do not tolerate antecedents split between two positions: (xiii) *Echo introduced Narcissus to themselves. (xiv) *Echo introduced Narcissus to each other’s reflections in the water. (xv)
Echo introduced Narcissus to their reflections in the water.
(xvi) *Echo watched Narcissus admire themselves in the water. (xvii) *Echo watched Narcissus admire each other’s reflections in the water. (xviii) Echo watched Narcissus admire their reflections in the water. Recall now that my account of the distribution of possessive reflexives (see section 2.4) conjectured that reflexives and the like do not refer directly to their antecedents but describe them via the antecedent’s proximate thematic relation: The barber shaves himself means that the barber shaves the shaver. If so, the further restriction against split antecedency can be stated as the requirement that the anaphor copy for its content a single thematic relation. For a sentence with partial agreement, such as (iv) above, the conclusion has been that Kareem W-s and Marwan does not W but Cum-s instead. The condition against split antecedency keeps the reflexive from referring to the W-er and the Cum-er, and thus plural reference fails. As we have seen, with full, plural number agreement, both Kareem W-s and Marwan W-s and the reflexive refers simply to the W-ers: (xix) biħibbo Kariim w Marwaan ħaalun. love.3p Kareem and Marwaan themselves In the case of the extraposed relative clause in (xi), the same could be said for the relative pronoun, namely, that it too is a descriptive anaphor subject to the condition against split antecedency. Its antecedents must both be W-ers, which by itself derives the fact that relativeclause extraposition may apply to DPs that are coordinated but not to DPs that stand in any other relation to one another. Aoun and Benmamoun (1996) also remark that control (xii) cannot occur with partial agreement. Here the present suggestion may run out of steam, but I would first like to check whether PRO tolerates split antecedents in Lebanese as it does in English. If not, then my suggestion extends to this case as well. Otherwise, I should like to verify that the verb translating ‘refuse’ does indeed express an action that supports the eventivity and comitativity the use of partial agreement independently requires. 81. Thanks again to Sarah Ouwayda for improvements to the examples. 82. There appears to be no difference, and I do not understand why, between (i) and (ii) even if it is given that the swimmer dragged down the lifeguard, and similarly no difference between (iii) and (iv), although the lead ball pulls on the chain. (i)
ghere’ l-sebbeeH drowned.ms the-swimmer (ii) ghere’ l-maitrenajeur drowned.sg the-lifeguard
w l-maitrenajeur and the-lifeguard w l-sebbeeH and the-swimmer
852
Notes
(iii) gher’et T-Te”aleh w l-selsleh sank.sg. the-lead-ball and the-chain (iv) gher’et s-selsleh w T-Te”aleh sank.sg the-chain and the-lead-ball 83. The asymmetric meaning of partial agreement in Lebanese can be reproduced in English provided that the comitative phrase occurs in an analogous position, as in (i)–(iv). (i) (ii)
The water, with the infection, (has) spread throughout the city. #The infection, with the water, (has) spread throughout the city.
(iii) The water has, with the infection, spread throughout the city. (iv) #The infection has, with the water, spread throughout the city. (v) (vi)
#The water (has) spread with the infection throughout the city. #The water (has) spread throughout the city with the infection.
(vii) The infection (has) spread with the water throughout the city. (viii) The infection (has) spread throughout the city with the water. Interestingly, VP-internal modification, (v)–(viii), seems to reverse the perspective. If the meaning of with is invariant throughout (i)–(viii), it suggests that it may modify different events from different syntactic positions, as the analysis of transitivity in chapter 1 invites. 84. Comitative phrases do not always agree: (i) Alia and Marwan are light together. So they can ride the same swing. (ii) #Alia is light with Marwan. So they can ride the same swing. #Alia with Marwan is light. So they can ride the same swing. Lebanese partial agreement patterns more like with—at least with respect to the above contrast. 85. Without benefit of experimental evidence, I argue in Schein 2002 that problems of event identity and a prior commitment to Davidsonian logical form lead to a like conclusion that syntactic position induces an asymmetry in meaning, independent from but consistent with lexical content that nevertheless describes symmetric events. 86. For further discussion see chapter 9; appendix 1; Schein 1993, chap. 10; and Schein 2002. 87. I suppress here chapter 1’s conclusion that embedded thematic relations are closed off by existential quantification. The fact that the participant in the higher event is, say, the Agent of the lower event was argued to be an inference from the overlap of these two events. 88. Recall that note 84 reports a reversal in perspective from (i)–(iv) to (v)–(viii), which seems in Talmy’s terms to be a reversal of figure and ground, where in this case the figure’s position is fixed relative to a ground that is itself in motion. (i) (ii)
The water, with the infection, (has) spread throughout the city. #The infection, with the water, (has) spread throughout the city.
(iii) The water has, with the infection, spread throughout the city. (iv) #The infection has, with the water, spread throughout the city. (v) (vi)
#The water (has) spread with the infection throughout the city. #The water (has) spread throughout the city with the infection.
Notes
853
(vii) The infection (has) spread with the water throughout the city. (viii) The infection (has) spread throughout the city with the water. The assignment of figure and ground in (v)–(vii) accords with Talmy’s expectation; (i)–(iv) fall outside the paradigms that Talmy and Gleitman et al. investigate. 89. Langendoen (1978) shows that ‘…each other’ fails equivalence with ‘each…(some of) the other(s)’, of which (562)–(563) and (568)–(569) are further instances: (i) The plates are stacked on top of each other. (ii) #Each of the plates is stacked on (some) of the other(s). (iii) The fish ate each other up. (iv) #Each of the fish ate (some) of the other(s) up. These examples illustrate that each other has an interpretation distinct from, but still perhaps in addition to, the interpretation that a dislocated each conveys. Examples (562)–(563) and (568)–(569) make a different point, namely, that each other does not share the interpretation expressed by a dislocated each. This is unexpected if, as in Heim, Lasnik, and May 1991, the logical form of what surfaces as each other optionally moves each to higher positions within the sentence. For my argument, it suffices that (562)–(563) and (568)–(569) elicit some degree of contrast in the direction indicated. Although a narrow construal of the reciprocal with lexically asymmetric relations, (559) and (565), straightforwardly supports the temporal comparison, a narrow construal with symmetric relations, (562) and (568), does not. Unlike (559) and (565), the only coherent reading of (562) and (568) forces each to be moved out in logical form, which speakers judge to be at least a more remote possibility. The contrast is also quite clear in the following contexts: (v) (vi)
To hit each other at different times is possible. To be against each other at different times is possible.
(vii) #To collide (with each other) at different times is possible. (viii) #To be with each other at different times is possible. As in the text, (vii) and (viii) little tolerate comparison of the time when one is with another with the time when the other is with one. (See Schein 2003 for further discussion.) 90. Sarah Ouwayda rejects plural agreement in the VSO order, the (b) examples of (572)– (573) and (ib) and (iib) below, as implying that the subjects were sentient agents. The implication is absent from SVO and from VSO with singular, partial agreement, the (c) examples of (572)–(573). (i) Beneath Dr. No’s laboratory, a. l-HiiTaan maʕasou James Bond the walls crushed/squashed.3mp 007 b. maʕasou l-HiiTaan James Bond crushed/squashed.3mp the walls 007 (ii) Beneath Dr. No’s laboratory, a. l-HiiTaan lta’ou the walls met.3mp, b. lta’ou l-HiiTaan met.3mp the walls, but 007 had escaped unharmed.
854
Notes
91. 007 may be genuinely confused about which of the floor and ceiling are moving if his only cue is the diminishing wall space. 92. Volitionality, intentionality, and a possible conspiracy confound the comparison of these examples. Comparing the following, Sarah Ouwayda accepts (i)–(iii) but points out that (i) may be irrelevant because the feminine singular is also the form for agreement with inanimate plurals, and so (i) is not clearly first-conjunct agreement: (i) lamma betkoun mghattayeh mniH, btdall l-'adseh w l-mreeeyeh ndaaf when is.3fs coated properly, remain.3fs the-lens and the-mirror clear.3mp (la sniin) (for years) lamma beykoun mghatta mniH, bydall l-kastak w ’zeez when is.3ms coated properly, remain.3ms the-watchband and glass s-see’a ndaaf (la sniin) the-watch clean.3mp (for years) ‘When coated properly, the watchband and glass remain clean for years.’ (ii) lamma ykoun t-ta’s neeshef, bydall l-telvezion w l-ghesseeleh when be.3ms the-weather dry, remain.3ms the-TV and the-washer sheghghaaliin (la sniin) functioning.3mp (for years) ‘When the weather is dry, the TV and the washer remain functioning (for years).’ (iii) lamma btousal ishaaret l-muruur, bydall D-Daw l-awwal w D-Daw when arrives.fs signal the-transit, remain.3ms the-light the-first and the-light t-teeni b-sheere’na xeDer the-second in-street-our green.3mp ‘When the transit signal is received, remain.3s the first traffic light and the second traffic light on our street green.3mp.’ 93. I confess to some sleight of hand here. I have so far given no reason to discount the possibility that full number agreement delivers (i) and partial agreement (ii), for some Z that is not W. (i) … W(e,Alia) and W(e, Marwan) … (ii) … Z(e,Alia) and Cum(e,Marwan) … If so, the same event could be such that both Alia and Marwan W there while Alia Z-s alone there as Marwan Cum-s. In my defense, observe first that there is no respect, perspectival or otherwise, in which plural and singular agreement should force a difference in meaning other than in number (whatever (s)he did vs. whatever they did). (iii) The children have arrived. (iv) Alia has arrived. There is little reason to suppose that the children W according to (iii) but Alia Z-s according to (iv). Whether or not one intends the children to be together in some sense may justify ambiguity somewhere in (iii), but rather than decide where, consider just those utterances of (iii) absent such an intention. There is nothing that (iii) says about the children that (iv) does not also say about Alia, and so there isn’t anything to peg as a difference in meaning between
Notes
855
singular and plural agreement, at least for those utterances of (iii) that do not intend the children to be together. Often enough (iii) and (iv) say the same thing about the children and Alia, say, that they W. Surely (v) joins in this too, and thus (iii)–(v) share the highest thematic relation. (v) Alia and Marwan have arrived. It has been my assumption that verbal number agreement is agreement with the W-ers, with those nominals that have been projected in phrase structure as arguments (the subjects) of relation W. Turning to (vi) and its translation into Lebanese, I have flatfootedly assumed (like Aoun, Benmamoun, and Sportiche 1994, 1999) that partial agreement is like singular agreement everywhere else, agreeing with the argument of the highest projection, Alia, the one and only W-er. (vi) Alia with Marwan has arrived. Enough has been said about the peculiar meaning of partial agreement. Consonant with the above, I have assumed that perspectives and the whole shebang are the analysis of how with Marwan in (vi) modifies a sentence where the syntax and semantics of the position that Alia occupies is just as it would be without modification by with Marwan. Under these assumptions, it is then correct in the text that Z in (ii) could not be anything other than W. If these assumptions are mistaken, I lose the metaphysical argument that puts distance between the “higher” and “lower” events, which then rests on chapter 1’s argument from multiple conjunction and the empirical evidence offered in this chapter to distinguish the surrounding conditions affording an asymmetric perspective from the sometimes symmetric event reported. 94. See the appendix to Schein 2002 for a unification of comitative and instrumental with. 95. See (after examples from Parsons 1990): After the Marseillaise, they saluted. After the Marseillaise’s being performed, they saluted. After the Marseillaise was performed, they saluted. After Ilsa’s departure for Lisbon, Rick was unhappy. After Ilsa’s departing for Lisbon, Rick was unhappy. After Ilsa departed for Lisbon, Rick was unhappy. Despite structural differences internal to the adverbial phrases, we expect them to behave much the same way in modifying their host sentences. 96. Montagnier and Gallo were from rival laboratories in Paris and Bethesda, MD, respectively. 97. Maria Bittner is also one of the Polish consultants for the judgments that McNally (1993) reports. 98. As marginally in English, too: (i) ????After sorting the subjects into two clusters, Anna with Peter have brown eyes, and everyone else lumped into the green-eyed cluster. (ii) *After sorting the subjects into two clusters, Anna with Peter has brown eyes, and everyone else lumped into the green-eyed group.
856
Notes
Collins (1988a, 1988b) observes plural agreement in examples similar to (iii): (iii) Robin (together) with Hillary have run 20 miles (between them). Suppose that Robin and Hillary are unacquainted and run different courses at different times. Nevertheless (iii) may report their accumulated mileage, granted some motivation to tabulate their results together. In contrast, (iv) cannot be understood to report cumulative mileage. With singular agreement, Robin and Hillary must have each traversed 20 miles. (iv) #Robin (together) with Hillary has run 20 miles (*between them). 99. Schein (2002, appendix) says that comitative phrases and instrumental phrases are the same, differing only in their position within the sentence and the antecedents for their free event variables. 100. Examples modeled after those in Lasersohn 1995, 70. 101. It is equally unexpected under the view of subordination endorsed here: (i) … [∃e′ : e′ ≤ e & With(e′, Whitehead)][∃e″ : e″ ≤ e′ …] … (ii) … [∃E′: ∀e′(E′e′→ Ee′ & With(e′, Whitehead))][∃E″: ∀e″(E″e″→ E′e″ & …)] … If (i) or (ii) is all there is to modification by ‘With Whitehead’, it should be possible to drop the modifier salva veritate (assuming for (i) that ≤ is a transitive relation). 102. After some examples discussed in Parsons 1990 and Pietroski 1998. 103. As remarked in note 99, Schein (2002) identifies instrumental and comitative phrases, which is neglected here. I also overlook the respect in which the position of the instrumental phrase in (615)–(616) fixes its event variable. 104. This account of comitative phrases anticipates that they can modify other thematic relations: (i) (ii) (iii) (iv) (v)
Willy Wonka melted the chocolate with the butterscotch. The chocolate melted with the butterscotch. Willy Wonka gave Nora the chocolate with the butterscotch. The chocolate was given to Nora with the butterscotch. ?The chocolate smeared the light side of the toast with the dark side.
And it says nothing about where they cannot: (vi) (vii) (viii) (ix)
*The chocolate smeared on the light side of the toast with the dark side. ?Willy gave Nora with Amy marzipan pigs. *Willy gave Nora marzipan pigs with Amy. ??Willy gave marzipan pigs to Nora with Amy.
Perhaps a comitative phrase is more complex and more elliptical and therefore subject to further syntactic conditions: (x) ?The chocolate smeared the light side of the toast (along) with smearing the dark side. (xi) *The chocolate smeared on the light side of the toast (along) with smearing on the dark side. 105. I import here some handiwork from Camacho 1997, 2003. 106. It is immaterial for the present point whether the event quantifier restricted by an ifclause is universal, existential, a definite description, or first-order or second-order.
Notes
857
107. McNally (1993) proposes that Anna s Petej ‘Anna with Peter’, when and only when it licenses plural number agreement, is a phrase that refers to an object. It can be neither the object that the plural phrases ‘Anna and Peter’ and ‘the children’ refer to, as she and Camacho (2000) show, nor the object that the singular ‘the group of Anna and Peter’ refers to, as Dalrymple et al. (1998a) show. Ontological scruples aside, the proposal divorces the occurrence of ‘with Peter’ in this construction from its occurrences elsewhere and, even so, gains little purchase on its syntactic distribution. It must still be stipulated that this PP, unlike PPs that can be extraposed from DP, remains adjacent to the DP it is in construction with. 108. Note that the placement of number agreement is somewhat arbitrary since I offer no particular account of the verb movement that will bring verb and number morpheme in contact. If an accessibilty relation tags along with every adverb, including the nonclausal yesterday, then the number morpheme may fall within the scope of yesterday as in (i) and nevertheless use an exact copy of the thematic relation to denote only those events of Nora W-ing that took place yesterday: (i) ∃e∃e″ ∃e″″ ([℩x: Nx]Wi(e, x) [∃E′ : (i e′ ≤ e & yesterday(e′))][℩E: O(e,e′)] [SG℩E : proi][℩E″ : proj](O(e,e″) & (j ∃xAgentk(e″,x) & [℩E”’ : [℩E″ : prok](O(e″,e”’) & with Willy)][℩E″: O(e”’,e″)] (∃xAgentk(e″,x) & Cause(e″,e″″) & melt(e″″) …))))
Nora yesterday -sg with Willy melted …
Given the constraints assumed, I do not see number agreement originating in any lower position in logical form than the one shown in (i). My use of the term ‘[℩E: O(e,e′)]’ to effectively restore use of the variable e while continuing successive domain restriction looks equivalent to the modal convention in Plurals and Events according to which none of the quantifiers occurs vacuously in ‘[∃E:Φ][ ∃E:Ψ]Π’, but rather the domain of the second quantifier is understood to be restricted to some events that are Φ. Earlier in this chapter, I chose to represent domain restriction explicitly and to derive certain syntactic effects from the use of variables (see section 2.3.1). I have yet to verify whether a careless appeal to ‘[℩E: O(e,e′)]’ will not undermine the earlier treatments. 3
PredP and PredP: Of Subjects and Ancient Grievance
1. Imagine a narrative convention for heroic sagas that uses an abbreviated name for the hero to hasten the telling, and introduces the full name with all the hero’s titles to signal a new adventure. 2. If it is a freestanding morpheme, it violates Pesetsky’s (1995) generalization that all zero morphology is bound morphology. It may however comport with the generalization, if it is bound, nominal morphology (perhaps like Case), closer in spirit to Lohndal’s (2014) treatment of thematic relations. 3. The slogan in the text is misleading for the discussion in section 1.9, “Contextualism Logicized.” I do not mean here to endorse Contextualism (Stanley 2000), where every truthconditional hiccup is represented in a complete representation of a sentence or utterance’s truth conditions. I am here rejecting a view in linguistic semantics that allows meanings— however much of truth-conditional content is therein made explicit—to be scattered, some with representation in syntax or logical form, others located only “in the semantics,” whatever that is, in an effort to avoid abstract syntax and unspoken structure (cf. Stanley 2000, 428ff.).
858
Notes
4. Among others, Burton and Grimshaw 1992; Kitagawa 1986; Koopman and Sportiche 1991; Kuroda 1988; McNally 1992. Terje Lohndal (p.c., June 2010) points out that labeling all verbal coordinations as VP has not survived the introduction of projections such as vP intermediate between VP and TP, but the motivation for internal subjects remains mutatis mutandis. 5. The reader should therefore remain vigilant that in the course of the chapter, a satisfactory account of subjects and subject positions emerges, whatever other objections are entertained, quelled, or otherwise distract. 6. In anticipation of later discussion (e.g., section 3.4), note that it is necessary for a “sloppy” interpretation but not sufficient that copies of the same phrase fall within the scope of distinct event quantifiers. In (i), for example, the first token of a bottle of Pepsi falls within the scope of an existential event quantifier, and the second, within the scope of a definite description of events. (i) ∃E(I … a bottle of Pepsi …) and [℩E : (I …)]( … a bottle of Pepsi …) But if the definite description, here anaphoric, describes the event of the antecedent clause, both tokens of a bottle of Pepsi end up evaluated with respect to the same events. 7. A sophisticated reply, an ersatz version of Partee and Rooth’s (1983, 369) treatment of (i) and (ii), goes as follows (see section 3.4.1.0 and also chapter 1): (i)
An easy model theory textbook is badly needed and will surely be written within this decade. (ii) A tropical storm was expected to form off the coast of Florida and did form there within a few days of the forecast. As arguments against Conjunction Reduction have held, the lack of ambiguity in (63) indicates that and coordinates predicates. The problem presented by apparent reconstruction in both (68) and (i)–(ii) is addressed by revising what the predicates involved denote. Let the nominals, a rocker, an easy model theory textbook, and a tropical storm, be expressions that (at least sometimes) refer to themselves. For any (one-place) predicate Φ, let substo–Φ be a substitutional predicate such that substo–Φ is true of (an expression) x just in case the sentence x⁀Φ, concatenating x and Φ, is true. Sentences (68) and (i)–(ii) are now all ambiguous according to whether predicates or their substitutional counterparts occur there. The semantics of and applied to predicates is unchanged. When substitutional predicates are coordinated, the predicate derived by coordination is true of x just in case each constituent predicate is true of x. But since the relevant values of x are themselves expressions, a rocker, an easy model theory textbook, a tropical storm, and so on, the interpretation mistaken for reconstruction can be derived without that particular transgression. To be complete, the reply requires a taste that can savor the difference between reconstruction and substitutional predication, and then one needs to supply some explanation for why (63) is never translated with a substitutional predicate (see Partee and Rooth’s (1983) translation when, in contrast to (i) and (ii), extensional predicates are coordinated). The arguments for unpronounced subjects developed in later sections, 3.3 and 3.4, are immune to the sophisticated reply (ersatz or genuine) even granting its completion. Parallelism effects (Fox 1995, 2000) between the conjuncts of (68) governing the scope relations within them present further proof against a coordination of predicates (see sections 3.4–3.6 for discussion). 8. An irrelevant interpretation saying that the rollers have shaken tout court is marginally possible.
Notes
859
9. From which it follows—if conjoined phrases are assumed to be syntactically identical—that the subject of the first conjunct alone is raised outside the coordination in (90), (92), and (93), as championed by Johnson ([1996] 2003, 2000a, 2000b, 2002): (90) Hei has [[ti stirred] and [she shaken]] (92) The rockersi have [[ti shimmied] and [the rollers shaken]] (93) *Diegoi has [[simultaneously ti stirred a mojito] and [Frida shaken a margarita]] 10. If one lingers over these sentences long enough, other ways out come to mind. One might for example imagine that in uttering (94), the speaker really intends to abbreviate (i) or the like: (i) Kunstler is sitting at every odd moment and is standing at every even moment. I think this only confirms the crucial point, namely, that in repeating the tensed auxiliary verb, the speaker intends that the first conjunct describe what is in progress at the present moment and that the second conjunct also describe what is in progress at the very same moment. Taking the speaker to be coherent, one then grasps at descriptions that would reconcile the apparent contradiction. 11. It is coincidence rather than identity, and there is no anomaly if coincidence is possible as in (i), although one might hesitate to identify the glaring with the formulating. (i) Kunstler is glaring at the judge and is formulating his rebuttal. If the coincidence is strong enough to support the inference that one participates in the one event if and only if one participates in the other, then the unpronounced subject of the second clause need not itself be anaphoric referring to Kunstler. Even if it were someone, one would draw the inference to Kunstler based on the coincidence. If, on the other hand, the understood coincidence is too weak for such an inference, the second subject must be taken to be a tacit pronoun referring to Kunstler or alternatively describing whomever is the participant in the event described by the first conjunct. 12. With special thanks to Kyle Johnson (p.c., 2003) for his endurance under elicitation. 13. The examples cite only decreasing quantifiers. I haven’t yet elicited judgments for nonmonotonic quantifiers. 14. The formulation has to be vague in light of examples such as No man can be friends with a woman he finds attractive. He always wants to have sex with her. (Billy Crystal in When Harry Met Sally, cited in Poesio and Zucchi 1992) 15. “Jeroen Groenendijk (p.c.) has pointed out an interesting contrast between anaphora to an ordinary universal and anaphora to an any phrase. While distinctions in modal subordination have not been formally analyzed, the paradigm below appears suggestive of the difference in the domains of quantification of ordinary universals and any phrases being claimed here: (i) Every philosopher is sometimes wrong. He usually doesn’t admit it. (ii) Any philosopher is sometimes wrong. He usually won’t/*doesn’t admit it.” (Dayal 1998, 447n20)
860
Notes
16. Note that the coordination of participial phrases in (158), (160), and (162) falls entirely within the scope of ever. Thus, any pronoun within the second conjunct referring to events described by the first does not find a negative-polarity item in its antecedent. Likewise in (174) and (175), even if the subject of the second participial phrase is a definite pronoun, its antecedent is the subject of the first participial phrase, the coordination falling entirely within the scope of the negative-polarity item. 17. I have shied from commenting on mixed coordinations such as (115): (115) Only a few students have read this book and remember it poorly. Either such cases are intrinsically asymmetric—Only a few students have read this book and having read it, remember it poorly. (Cf. *No rocker has shimmied and (now) shakes to that funky disco music. *No rocker has shimmied and (also) shook to that funky disco music. Every rocker has shimmied and (now) shakes to that funky disco music. Every rocker has shimmied and (also) shook to that funky disco music.) Or I should take up sooner the question that is richly posed in the literature on Germanic syntax of the apparent coordination of clauses of mixed syntactic type, asymmetric coordination as it is called (Höhle 1990, 1991; Heycock and Kroch 1993, 1994; Johannessen 1998; Kathol 1992, 1993; Thiersch 1994; Büring and Hartmann 1998; Schwarz 1998a, 2000; among others). 18. Rather than fix syntactic positions for simultaneously, a friend of predicative coordination might entertain that the contrast between (87) and (88) shows only that the adverb modifies conjoined predicates and not conjoined clauses, to whom the following examples should be pointed out: (i) (ii) (iii) (iv) (v) (vi)
Simultaneously, a rocker has shimmied and a roller shaken. A rocker has shimmied and a roller shaken simultaneously. Has simultaneously a rocker shimmied and a roller shaken? Has a rocker shimmied and a roller shaken simultaneously? Have simultaneously a rocker shimmied and a roller shaken? Have a rocker shimmied and a roller shaken simultaneously?
19. Schwarz (2000, sec. 4.4.5) acknowledges the following evidence of asymmetric extraction from the first conjunct of a coordination of smaller clauses: (53) a. Den Hund hat sie nicht gefüttert und ihn the dog has she not fed and it ‘She did not feed the dog and beat it.’ b. Den Hund hat sie nicht gefüttert und hat the dog has she not fed and has ‘She did not feed the dog and beat it.’ c. Den Hund hat sie nicht gefüttert und sie the dog has she not fed and she ‘She did not feed the dog and she beat it.’
geschlagen. beaten ihn geschlagen. it beaten hat ihn geschlagen. has it beaten
In a coordination of tensed clauses, (53c), the sentence is unambiguous in confining negation to the first conjunct, (¬p & q). In addition to this meaning, (53a–b) allow another in which negation includes within its scope both conjuncts ¬(p & q), and then plainly the direct object den Hund has been extracted from the first conjunct.
Notes
861
Schwarz however resists this conclusion, since he assigns to the strict observance of the Coordinate Structure Constraint (CSC), which forbids such an extraction, a role in explaining the ungrammaticality of (15): (6)
Äpfel ißt der Hans drei und zwei Bananen. apples eats the Hans three and two bananas ‘Hans eats three apples and two bananas.’
(15) *Äpfel wird der Hans drei und zwei Bananen essen. apples will the Hans three and two bananas eat ‘Hans will eat three apples and two bananas.’ (60) Äpfel wird der Hans drei und Bananen zwei essen. apples will the Hans three and bananas two eat ‘Hans will eat three apples and two bananas.’ In confining the topicalized direct object to the first clause, the CSC forces the coordination and subsequent reduction of full clauses so that (15) must be parsed as in (15′), where there is something defective about the gapping and Right-Node Raising corresponding to the indicated deletions: (15′) *Äpfeli wird der Hans drei ti essen und der Hans wird zwei Bananen essen. But unless that defect attaches specifically to the deletion of wird in the second conjunct, it should be as much a blemish in (15″), which allows for the asymmetric extraction in evidence in (53): (15″) *Äpfeli wird [der Hans drei ti essen und der Hans zwei Bananen essen]. There seems in any case some kind of parallelism constraint distinguishing (15) from (60), which should rule out (15″) just the same. What the CSC does for Schwarz is force (15) to contain a verb-phrasal coordination, ruling out a parse that coordinates only DPs as in (15‴): (15‴) *Äpfeli wird der Hans [ [drei ti] und zwei Bananen ] essen. Here however und is always sentential, excluding (15‴), the conditions on grammatical complementation compel Right-Node Raising, and the only parse for (15) is (15″), which is ungrammatical on the same grounds as (15′). It suffices to force (15) into a clausal coordination without it being the largest such coordination. The coordination of smaller clauses is, in turn, consistent with the evidence for the asymmetric extractions of subjects and objects. (Thanks to Kyle Johnson for correspondence about this note, February 2011.) 20. An alternative (θ-criterion, Principles and Parameters) stipulates that a DP must relate to a θ-role—that is, a thematic relation associated with the stem of a main verb to the exclusion of those (e.g., ‘W[E,X]’) that may occur elsewhere, for example, with auxiliary verbs. Accordingly, sentences (195)–(197) are ruled ungrammatical on the grounds that the subject of the auxiliary verb has been deprived of a θ-role by the participial subject that is Agent for shake-. If this is favored, its application to the disjunctive interpretation of (187) requires special pleading. There the θ-roles, the tokens of ‘Agent[E,X]’ inside the conjuncts, are closed off by existential quantifiers. In what sense do they relate to twenty rockers except to retrace a particular derivational history?
862
Notes
21. These simplifying assumptions are not harmless when extended to the progressive. Neither (i) nor (ii), for example, both being extensional, can be maintained without an ontology that includes possibilia—possible, nonactual events of building a house. (i) … Be[Ei] R[Ei,Ej] -ing[Ej] T[Ej,Ek] build[Ek] … (ii) … Be[Ei,Ej] -ing[Ej,Ek] build[Ek] … 22. A fortiori if in fact there is no sense at all in participating in a having when have is used as an auxiliary verb. I assume from the position it occupies that the subject of the auxiliary verb induces lifetime effects on Tense (see section 10.0), which does not of its own require participation in having (‘W[E0,X] … Have[E0,E1]’). It could be that the first mention of participation concerns only the events described by the embedded participle, as a strict raising analysis of the auxiliary verb would imply. 23. To preserve a structure strictly parallel to the conjunct with pronounced subject, the first conjuncts in (214) and (215) articulate the descriptive content of the definite description taken to correspond to the trace left behind by the subject of the tensed auxiliary verb. If correct, adverbialization would have to extend at least as an option to the traces of movement as found here, in contrast to the null pronominal subject of (94). Strict parallelism is not necessary unless the meanings of (189) and (210) insist. The relevant facts would concern the lifetime effects discussed in chapter 10. These are expected between the pronounced subject of the tensed verb and Tense in (189) and (210). But given that this subject and the subject of the first participial phrase refer to the same object, I am at a loss how to construct examples that would reveal whether an independent lifetime effect governs the interpretation of the participial phrase as the logical forms in (214) and (215) assume. 24. The negative-polarity idiom is unacceptable, in contrast to a literal meaning (see (246)): With the lift of a finger, the don had his enemies sleeping with the fishes. 25. Note that the high position assumed for ever is consistent with its position relative to simultaneously (see section 3.2.1.0): (i) No one has ever simultaneously stirred a mojito and shaken a margarita. (ii) ??No one has simultaneously ever stirred a mojito and (*ever) shaken a margarita. 26. A negative-polarity item within the antecedent participial phrase is necessary for referential failure but not sufficient. If the negative-polarity item is subject to Quantifier Raising, as in (i), it appears that raising it outside the participial phrase may rescue the structure. (i) No one has ever confessed to any infraction and his confession gotten him off. In (ii), there is no risk that the negative-polarity item in the least escapes the participial phrase. Here abbreviating the description as in (iii) improves the outcome. Note that to accommodate the second conjunct in both (ii) and (iii), there must now be at least some hesitation on the child’s part: (ii) ?No child has ever hesitated in the least and her parents been unconcerned. (iii) ?No child has ever hesitated in the least and hesitating, her parents been unconcerned. The crucial examples in the text confine the negative-polarity item to the participial phrase and preclude abbreviation. The direct objects cannot be dropped—there are no budgings and there are liftings only in a different sense.
Notes
863
27. Note however the difference in Case assigned—With him not standing, … but She is standing and he not standing. 28. Asymmetric coordination cannot be summarily dismissed (see also notes 17, 46). Freely translating some German examples (Höhle 1990, 1991), we have also in English (i) and (ii): (i) Nobody I know ever got home and the cops were waiting for him or (*he) went to work and the feds had the building surrounded. (ii) Nobody I know has ever gotten home and the cops were waiting for him or ((*he) has) gone to work and the feds had the building surrounded. As (ii) makes vivid, fully Tensed clauses occur as conjuncts to participial phrases within a disjunction of participial phrases. The German examples (iii) and (iv) show a further asymmetry. The fully tensed clauses show the V2 syntax of matrix clauses rather than the verb-final syntax of if-clauses, which appears only in the first conjuncts: (iii) Wenn jemand nach Hause kommt und da steht der Gerichtsvollzieher vor der Tür, … ‘If someone comes home and the bailiff is waiting outside the door, …’ (iv) Wenn jemandi [nach Hause kommt und da steht der Gerichtsvollzieher vor der Tür] oder (*eri) [zum Arbeitsplatz fährt und dort ist das Gebäude umringt durch die Polizei], … ‘If someone comes home and the bailiff is waiting outside the door or (*he) goes to the office and the building is surrounded by the police, …’ Perhaps then it is not so outlandish to imagine that a coordination with participial subject pronounced is a reduct from the likes of (i)–(iv). Note that the tensed clauses in (i)–(iv) must be understood to comment on events antecedently described. 29. Alternatively, a symmetric structure could locate ‘X[E]’ in the first conjunct as well, provided that its event variable comes to be bound within it. The conjuncts in (i) are entirely parallel in structure (in this respect the distribution of ‘X[E]’ more closely resembles that of ‘Tense[E]’ when it appears on the left edge in both conjuncts): (i) DPi Tense Aux [[ti therewith[E] ti W O -en V] and [therewith[E] DP W O -en V]] In a move familiar from syntax, raising to subject of Tense first passes through a position to the left of ‘therewith[E]’ as if this phrase were indeed a complementizer. From this position, adverbialization of the DP trace of movement closes off the event variable (see note 24). 30. To address “the problem of indistinguishable participants” (Heim 1990), as much is necessary to allow the definite descriptions in (i) and (ii) to refer to different rockers: (i) If true rockers play opposite true rockers, the ones graciously introduce the others. If some true rockers play opposite some true rockers, the ones graciously introduce the others. (ii) At the Garden tonight, true rockers played opposite true rockers, and so the ones graciously introduced the others. At the Garden tonight, some true rockers played opposite some true rockers, and so the ones graciously introduced the others.
864
Notes
If the descriptive content of ‘the ones’ in (i) or (ii) is nothing more than the true rockers who play(ed) opposite (some) true rockers, it fails to refer to the rockers who play opposite apart from those rockers being played opposite, and thus fails the meaning of these sentences. As Heim (1990), Ludlow (1994), and others have suggested, such examples recommend that the descriptive content be supplemented by reference to an event, situation, or perspective appropriated in context, as if to say sotto voce something more like (ii) or (iii) (see appendix 1): (iii) If true rockers on the one hand play opposite true rockers on the other, the ones on the one hand graciously introduce the others on the other. If some true rockers on the one hand play opposite some true rockers on the other, the ones on the one hand graciously introduce the others on the other. (iv) At the Garden tonight, true rockers on the one hand played opposite true rockers on the other, and so the ones graciously introduced the others on the other. At the Garden tonight, some true rockers on the one hand played opposite some true rockers on the other, and so the ones on the one hand graciously introduced the others on the other. Heim’s (1982) familiarity condition requires that an indefinite be introduced with the understanding that it is not to be taken as referring to the same things as any preceding term. As a condition on pronounced indefinites, one must therefore always understand (189) this way or else fall into a logical form equivalent to (187)/(258) that intends both tokens to refer to the same twenty rockers. 31. Recall that the subject of the first conjunct has been raised to the subject of the Tensed auxiliary verb phrase and so is deleted from the first conjunct of the conjoined participial phrases. I assume that the first subject is adverbialized where it is pronounced. 32. Recall from section 2.3.2.1 that the agreement in (i) is plural, Alia and Marwan being two, but to each, their own Pepsi as the result of Right-Node Raising from within the scope of distinct adverbs. In this respect, it was observed that (i) contrasts with (ii), where they share the bottle. (i) šerbou drank.pl (ii) šerbou drank.pl
alia Alia alia Alia
mbeereH w marwaan lyom anninet pepsi. yesterday and Marwan today bottle Pepsi w marwaan anninet pepsi. and Marwan bottle Pepsi
The facts are presumably no different for (iii) and (iv): (iii) šerbou drank.pl (iv) šerbou drank.pl
clark Clark clark Clark
mbeereH w superman lyom anninet pepsi. yesterday and Superman today bottle Pepsi w superman anninet pepsi. and Superman bottle Pepsi
Sentence (iii) would be especially felicitous if the reporter in his sack suit drank at the soda fountain yesterday, and the superhero in leotards did today. But since Clark and Superman are the same, the expectation is that one could just as well say (v) and (vi) and preserve the contrast that depends on the interaction of Right-Node Raising and the adverbs, namely, that (v) reports that he drank one Pepsi in his sack suit and another one in his leotards while (vi) that he drank just the one bottle.
Notes
(v) šereb drank.ms (vi) šereb drank.ms
865
clark Clark clark Clark
mbeereH yesterday w and
w and superman Superman
superman Superman anninet bottle
lyom anninet pepsi. today bottle Pepsi pepsi. Pepsi
In English, absent the intrusive adverbs, the “sloppy” interpretation of his date becomes impossible. It fails in (vii) to refer to either the man at the bar or the woman on the phone: (vii) #Clark Kent and Superman has been chatting up his date. #A mild-mannered reporter and a superhero is chatting up his date. 33. As noted, singular agreement excludes collective interpretation of any part of the RightNode Raised constituent, as if a token of the entire tensed auxiliary verb phrase was interpreted independently within each of the antecedent conjuncts. 34. The adverbs may invite but are not necessary for intonation to rescue singular number agreement: (i) *Robin and Hillary was drinking bordeaux last night. (ii) Robin—and Hillary—was drinking bordeaux last night. 35. Note also the following contrast: (i) ?*Clark fecklessly was and Superman adroitly was chatting up his date. (ii) ?Clark just now was and Superman earlier was chatting up his date. 36. Analogous effects under embedding although the if-clause does not seem to require a special intonation for singular agreement even when they be not one. Suppose it is the law that Superman’s invitations are always accepted. The following can be true and only so if the if-clause imposes a single, collective condition where p & q holds, if p & q, then r. An alternative parse as if p,r & if q,r, would be falsified by the second conjunct, Superman never being refused. (Remember not to be alarmed that ‘if p and if q, r’, pronouncing two ‘ifs’, need not be parsed as if p,r & if q,r. Cf. ‘everywhere that p and that q, r’.) (i) If Clark clumsily and (if) Superman deftly says anything suggestive, not everyone in earshot will accept the invitation. If a nerd clumsily and (if) a superhero deftly says anything suggestive, not everyone in earshot will accept the invitation. (ii) If Lex Luthor now and (if) Superman tomorrow says anything suggestive, not everyone in earshot will accept the invitation. If a supervillain now and (if) a superhero tomorrow says anything suggestive, not everyone in earshot will accept the invitation. Whether they be one matters for the collective interpretation of the direct object: (iii) If Clark clumsily and (if) Superman deftly chats up any two women, two women will not accept invitations. If a nerd clumsily and (if) a superhero deftly chats up any two women, two women will not accept invitations. (iv) If Lex Luthor now and (if) Superman tomorrow chats up any two women, two women will not accept invitations. If a supervillain now and (if) a superhero tomorrow chats up any two women, two women will not accept invitations.
866
Notes
The conditional in (iii) may describe situations like at the Good Luck Bar, with Clark trying with one woman and Superman another. In contrast, the conditional in (iv) describes only those situations where Luthor himself tries with two women and so does Superman. Describing a collective effort of theirs as directed toward two women requires plural agreement: (v) If Lex Luthor now and (if) Superman tomorrow chat up any two women, two women will not accept invitations. If a supervillain now and (if) a superhero tomorrow chat up any two women, two women will not accept invitations. 37. Sauerland and Elbourne (2002) show that the British plural, which occurs with a singular term referring to a group to say something about its members, (30b) and (32d), itself neither lowers ((30b) vs. (30a)) nor raises (30d), and it obstructs its subject from lowering ((30b) vs. (30a), and (32b) vs. (32a)). (30) a. b. c. d.
A northern team is likely to be in the final. (likely >> ∃) A northern team are likely to be in the final. (*likely >> ∃) There is likely to be a northern team in the final. *There are likely to be a northern team in the final.
(32) a. b. c. d.
A people with any tradition of trepanning has never been discovered. *A people with any tradition of trepanning have never been discovered. A people with a tradition of trepanning has never been discovered. A people with a tradition of trepanning have never been discovered.
To reconcile the British plural to my text, I should speculate either that i. The British plural is not simple, flat, featural number agreement as the authors suppose, but rather more like a (tacit) floated quantifier, “all (of that)” or “they (of that).” As they point out, strong quantifiers, floated or not, do not lower, for whatever reason; or ii. The British plural, whatever its syntactic structure, does in fact contain a demonstrative, indexical, or specific element, which even if it is lowered manages to convey the near equivalent of a wide-scope interpretation anyway. (31) a. A Germanic people is trying to settle in this land. b. A Germanic people are trying to settle in this land. While acknowledging that (31b) favors having specific people in mind, Sauerland and Elbourne have already rejected (ii), citing the generic contexts in (34) where the British plural is used without specific reference: (34) a. Any committee worth their salt are going to have to look into that. b. A rugby team are likely to break up the furniture. c. If a team have a good coach, they do what he says. But if the indexical or specific reference demanded by (ii) is specific relative to a context, and if the generic quantification in (34) and similar examples subsumes quantification over contexts, the specificity condition of the British plural could be met within and relative to each such context. Without departing from the episodic contexts used in (31) and elsewhere in Sauerland and Elbourne 2002, can one nevertheless find nonspecific uses of the British plural to dispel the suggestion in (ii)?
Notes
867
As in (31), Paul Elbourne (p.c.) and Daniel Harbour (p.c.) find that the British plural in (iii) is rather about some specific hit team in contrast to (iv), which may serve as a report of an inference to the existence of as yet unknown and unidentified Sicilian muscle: (iii) Given the marinara sauce found on the victim, crime investigators believe that a Sicilian hit team have clipped Tony Baritone. (iv) Given the marinara sauce found on the victim, crime investigators believe that a Sicilian hit team has clipped Tony Baritone. Elbourne (p.c.) then remarks, however, that the British plural again escapes specific reference in (v): (v) Given the poisoned marinara sauce delivered to him, crime investigators believe that a Sicilian hit team are trying to clip Tony Baritone. But notice a suggestive shift. To start, in (iii), there is a victim and so there is a fact of the matter. It is about a past event that, as fact, is independent of the investigators’ present beliefs. Since the investigators find victim and marinara sauce, the sentence also relates that the time of the finding is in direct rapport with the formation of present belief. The speaker intends the embedded clause factively, and if so, it must therefore accord with her own account of the past event and of when the investigators have come to have some present belief about it, and that puts on her the responsibility for the British plural (in what looks like a de re occurrence). In (v), on the other hand, nothing has happened and nothing might, except for the delivery (which enters the content of their beliefs, but its time and the delivery itself stand in no salient causal relation to the attributed belief. (Cf. Because they found marinara sauce on the victim, crime investigators believe …, vs. Because poisoned marinara sauce was delivered to him, crime investigators believe … .) That there is no fact of the matter, along with the concord in tense between the present belief and the present trying, allows that the embedded clause be understood to represent only the content of belief (de dicto), therefore leaving with the investigators and what they believe all responsibility for the British plural. I have no basis to prefer (i) or (ii), but the variation in its usage, (31), (34), (iii)–(v), hints of greater complexity in either the structure or meaning of the British plural than would be expected if it were nothing other than number agreement. 38. This section has benefited from correspondence with Kyle Johnson, January 18–29, 2002. 39. See especially Johnson 2002, [1996] 2003. 40. Note that (315) is a “subject-gapped” structure. See section 3.3. 41. These remarks extend to the German (314) and (315) only if the pre-V2 constituent has only carelessly been called topic rather than focus. Subj-Aux inversion in English at least fits the diagnosis. The pre-Aux phrase is a focus: (i) (ii) (iii) (iv)
A vaccine to save the world did Jonas Salk invent. The polio vaccine did Salk invent. The polio vaccine did a Jew invent. A vaccine to save the world did a Jew invent.
Sentence (i), for example, can only be used to say, in effect, that what Jonas Salk invented was a vaccine to save the world. It cannot mean “as for a vaccine to save the world, Salk invented it/one.”
868
Notes
42. Given that the antecedent for distributive each in (i) must itself be plural and that English is free with tacit distributors, an interpretation of (342) that allows each his or her own really big hand fails to demonstrate that tacit singular number agreement introduces the RightNode Raised phrase. (i) Ed Sullivan gave no rocker and (any of) his mama(s) each a really big hand. Both Lebanese and Slavic, however, exclude tacit distributors from such a position (section 2.3.2.3), and thus the translations of (342) are predicted to imply that rocker and mama share a really big hand and that (ii) is therefore rather awkward: (ii) Ed Sullivan gave no rocker and (any of) his mama(s) a handshake / a slap on the back. The translation of (342) and (ii) may, however, be thwarted unless these languages resemble English in their use of at least some decreasing quantifiers. 43. In section 3.2.1.2, participial phrases with pronounced subjects were found to have a leftperipheral projection, as if this small clause had a complementizer. Such a structure would naturally induce opacity and a coordination within it could be large enough to leave no room for the extraction of the first quantifier. 44. Null coordinative pronouns also known as “Subject Gapping” (Höhle 1990, 1991; Schwarz 1998a, 1999, 2000; Zwart 1991). 45. Consider the derivation of (i) or (ii) under the VP-internal Subject Hypothesis (Burton and Grimshaw 1992; Kitagawa 1986; Koopman and Sportiche 1991; Kuroda 1988; McNally 1992; among others): (i) James Brown has shimmied and shaken. (ii) James Brown shimmied and shook. 46. Wilder (1994) vigorously takes up this argument. If the second subject in (i)–(iv) is a null pronoun, the initial topicalization may remain within the first conjunct, violating no known conditions on movement, as shown. What is especially compelling here is the meaning of (iv) indicating that only John contains only the first conjunct within its scope: (i) (ii) (iii) (iv)
The bag hei dropped and proi ran to the exit. Kiss her, Ii didn’t, and proi will probably regret it. In came Johni and proi sat down. Only John did wei lend money to, and proi never expected it back.
Across-the-board movement of the subject would subsequently force the topicalization to be an asymmetric extraction from a coordinate structure, and it would mistake (iv)’s meaning: (i′) (ii′) (iii′) (iv′)
The bagj [hei [[ti [dropped tj]] and [ti [ran to the exit]]]] [Kiss her] j [Ii [[ti didn’t tj], and [ti will probably regret it]]] [In came]j [Johni [[ti tj] and [ti sat down]]] [Only John]j [didk [wei [[ti tk lend money to tj] and [ti never expected it back]]]]
Wilder’s reluctance to countenance such extractions finds encouragement in (i″). When the syntax and semantics of questions make such an extraction the only analysis, the result is ungrammatical: (i″) *What did he drop and ran to the exit? On the other hand, the same considerations imply that an asymmetric extraction from a coordinate structure is the only derivation for the apparently acceptable (ii″)–(iv″), which are
Notes
869
only a few of some well-known exceptions to the Coordinate Structure Constraint (Ross 1967; Goldsmith 1985; Lakoff 1986; Postal 1998, chap. 3 and the references cited there): (ii″) Who didn’t you kiss and will probably regret it? (iii″) ?Whose office did John come into and sat down? (iv″) Who did you lend money to and never expected it back? There is extensive discussion of asymmetric extraction in the Germanic syntax literature (Höhle 1990, 1991; Heycock and Kroch 1993, 1994; Johannessen 1998; Kathol 1992, 1993; Thiersch 1994; Büring and Hartmann 1998; Schwarz 1998a, 2000; among others). Johnson (2009, [1996] 2003) defends asymmetric extraction in other classes of cases. In fact, if acrossthe-board movement is rejected in favor of a null pronoun in the second conjunct, an asymmetric extraction exactly of the kind Johnson envisages is necessary to derive (384) from (385), (393) from (394), and (397) from (398). It may then be that the derivation for Icelandic depicted in (388) should not be dismissed out of hand even though (389) looks like less trouble. Still, whether or not an argument from constraints on movement survives, the plain meaning of (iv) confirms the existence of a null pronoun that does not fall within the scope of its antecedent, as desired. 47. The facts are the same coordinating one quirky predicate to another, whether the quirky Cases they select are the same (dat-dat: i–iv) or not (dat-acc: v–viii) (Hansson, p.c.): (i)
(ii) (iii) (iv)
tuttugu krökkum mun leiðast í skólanum og vera illt í maganum. 20 kids.dat will.sg bore.inf in school-the and be.inf sore in stomach-the ‘Twenty kids will be bored in school and have a tummy ache.’ Það mun tuttugu krökkum leiðast í skólanum og vera illt í maganum. *Það mun leiðast í skólanum tuttugu krökkum og vera illt í maganum. *Það mun leiðast í skólanum og vera illt í maganum tuttugu krökkum.
(v)
tuttugu krökkum mun leiðast í skólanum og langa heim. 20 kids.dat will.sg bore.inf in school-the and long-for.inf home ‘Twenty kids will be bored in school and want to go home.’ (vi) Það mun tuttugu krökkum leiðast í skólanum og langa heim. (vii) *Það mun leiðast í skólanum tuttugu krökkum og langa heim. (viii) *Það mun leiðast í skólanum og langa heim tuttugu krökkum. 48. Wilder (1994, 1995a, 1995b, 1997, 1999) discusses systematic typological differences between Forwards Deletion (or forward null anaphora in my terms) and Backwards Deletion (backward anaphora), which, he argues, a symmetric across-the-board movement does not predict. 49. “Woke up, fell out of bed, Dragged a comb across my head Found my way downstairs and drank a cup, And looking up I noticed I was late. Found my coat and grabbed my hat Made the bus in seconds flat Found my way upstairs and had a smoke, Somebody spoke and I went into a dream.” (Lennon and McCartney, A Day in the Life, 1967)
870
Notes
50. (94) #Kunstler is sitting and is standing. (95) Kunstler is sitting, and Kunstler is standing. Kunstler is sitting, and he is standing. 51. Haiman (1983, 112ff.) notes that in languages such as Turkish, clauses with a high degree of cohesion or continuity with the following clause (including not only identical subject, but also identical tense, mood, and polarity) will appear with verbs in which the suffixes marking these categories are replaced by an invariable suffix: (y)A for simultaneous activity and (y)Ip for sequential activity. Such suffixes are confined to the apparatus of coordination. 52. The contrast between (430) and (432) is that between (63) and (64) and subject to the same treatment. (63) All night long, a rocker has shimmied and shaken. (64) All night long, a rocker has shimmied and a rocker shaken. 53. In proposing that the intentional verb need in (423) prompts reconstruction in order to acquire an intensional argument rather than one referring to an object (see note 7), Partee and Rooth 1983 mistakenly assimilates (430) to (432). It spoils the generalization in generalized conjunction. 54. Modal subordination is at work here too, so that stripping it all away ends infelicitously in (iii), as in (430): (i)
Many new tons of rice will be needed and will surely be farmed and distributed to those who pay. (ii) ?Many new tons of rice will surely be needed and be farmed and distributed to those who pay. (iii) *Many new tons of rice will surely be needed (and) farmed and distributed to those who pay. 55. Although in in every shaking, the rocker there …, there is no prior indication that every shaking has one and only one rocker there, I am assuming a glib accommodation that they do. The coordinative pronoun could just be a naked Russellian definite description formed with the iota-operator with little pragmatic force other than what it literally asserts. Determiners the, this, that, and the null determiner associated with names (see Higginbotham 1988) combine the semantics of definite description with further pragmatic conditions on familiarity, proximity, acquaintance, and so on. 56. See section 1.1, “The Slippery Slope to Conjunction Reduction.” Also, recall from chapter 2 the likes of (i) and (ii): (i) Marvin this afternoon from Great Neck and Bernice this evening from Syosset are arriving at Leonard’s with each other’s spouses in rented Mercedes. (ii) Marvin made a grand entrance and Bernice swept in at the gala with each other’s spouses on their arms. 57. Fox (1995; 2000, 61–62n48) notes that speakers accept the scope inversion in (i), reject it in (vi), and waver in between. Such breaches in parallelism have the same effect throughout although I have not listed the relevant examples. The parallelism is itself discussed in section 3.6.
Notes
871
(i) A guard is standing in front of every church and sitting at the side of every mosque. (ii) A guard is standing in front of every church and sitting at the side of almost every mosque. (iii) A guard is standing in front of every church and sitting at the side of most of the mosques. (iv) A guard is standing in front of every church and sitting at the side of two of the mosques. (v) # A guard is standing in front of every church and sitting at the side of one of the mosques. (vi) #A guard is standing in front of every church and sitting at the side of this mosque. 58. More examples: According to Officer Krupke— (i)
A Shark recklessly and a Jet blindly squared off on every street corner and mixed it up in every schoolyard.
(ii)
A Shark recklessly and a Jet blindly were squaring off on every street corner and mixing it up in every schoolyard. *A Shark recklessly and a Jet blindly were squaring off on every street corner and were mixing it up in every schoolyard.
(iii) (iv) (v) (vi) (vii)
A Shark recklessly and a Jet blindly have squared off on every street corner and mixed it up in every schoolyard. *A Shark recklessly and a Jet blindly have squared off on every street corner and have mixed it up in every schoolyard. A court officer sternly and a court psychologist compassionately cosupervised every inmate and cocounseled every parolee. A court officer sternly and a court psychologist compassionately were cosupervising every inmate and cocounseling every parolee.
(viii)
*A court officer sternly and a court psychologist compassionately were cosupervising every inmate and were cocounseling every parolee.
(ix)
A court officer sternly and a court psychologist compassionately have cosupervised every inmate and cocounseled every parolee. *A court officer sternly and a court psychologist compassionately have cosupervised every inmate and have cocounseled every parolee.
(x) (xi) (xii)
A family trust privately and a pension fund publicly bought out every IPO and refinanced every startup. A family trust privately and a pension fund publicly were buying out every IPO and refinancing every startup.
(xiii)
*A family trust privately and a pension fund publicly were buying out every IPO and were refinancing every startup.
(xiv)
A family trust privately and a pension fund publicly have bought out every IPO and refinanced every startup. *A family trust privately and a pension fund publicly have bought out every IPO and have refinanced every startup.
(xv)
872
(xvi) (xvii)
Notes
A merchant carefully and a broker thoughtfully tasted together every bordeaux and savored together every burgundy. A merchant carefully and a broker thoughtfully were tasting together every bordeaux and savoring together every burgundy.
(xviii)
*A merchant carefully and a broker thoughtfully were tasting together every bordeaux and were savoring together every burgundy.
(xix)
A merchant carefully and a broker thoughtfully have tasted together every bordeaux and savored together every burgundy. *A merchant carefully and a broker thoughtfully have tasted together every bordeaux and have savored together every burgundy.
(xx) (xxi) (xxii)
A merchant yesterday and a broker today tasted from each other’s stock every bordeaux and savored in each other’s cellars every burgundy. A merchant yesterday and a broker today were tasting from each other’s stock every bordeaux and savoring in each other’s cellars every burgundy.
(xxiii)
*A merchant yesterday and a broker today were tasting from each other’s stock every bordeaux and were savoring in each other’s cellars every burgundy.
(xxiv)
A merchant yesterday and a broker today have tasted from each other’s stock every bordeaux and savored in each other’s cellars every burgundy. *A merchant yesterday and a broker today have tasted from each other’s stock every bordeaux and have savored in each other’s cellars every burgundy.
(xxv) (xxvi)
A soldier on Sabbaths and a plainclothesman on weekdays split the guard duty between (the two of) them at every church and partner on patrol around every mosque. (xxvii) A soldier on Sabbaths and a plainclothesman on weekdays are splitting the guard duty between (the two of) them at every church and partnering on patrol around every mosque. (xxviii) *A soldier on Sabbaths and a plainclothesman on weekdays are splitting the guard duty between (the two of) them at every church and are partnering on patrol around every mosque. (xxix)
(xxx)
(xxxi)
A soldier on Sabbaths and a plainclothesman on weekdays have split the guard duty between (the two of) them at every church and partnered on patrol around every mosque. *A soldier on Sabbaths and a plainclothesman on weekdays have split the guard duty between (the two of) them at every church and have partnered on patrol around every mosque. Leo, toward the front of the building, with an FBI agent, and Murray, toward the rear, with a NYPD officer, share surveillance of every church and partner on patrol around every mosque. Leo, from the front of the building, with an FBI agent, and Murray, from the rear, with a NYPD officer, cover every church and secure every mosque.
Notes
873
(xxxii) Leo, toward the front of the building, with an FBI agent, and Murray, toward the rear, with a NYPD officer, are sharing surveillance of every church and partnering on patrol around every mosque. Leo, from the front of the building, with an FBI agent, and Murray, from the rear, with a NYPD officer, are covering every church and securing every mosque. (xxxiii) *Leo, toward the front of the building, with an FBI agent, and Murray, toward the rear, with a NYPD officer, are sharing surveillance of every church and are partnering on patrol around every mosque. *Leo, from the front of the building, with an FBI agent, and Murray, from the rear, with a NYPD officer, are covering every church and are securing every mosque. (xxxiv) Leo, toward the front of the building, with an FBI agent, and Murray, toward the rear, with a NYPD officer, have shared surveillance of every church and partnered on patrol around every mosque. Leo, from the front of the building, with an FBI agent, and Murray, from the rear, with a NYPD officer, have covered every church and secured every mosque. *Leo, toward the front of the building, with an FBI agent, and Murray, toward the rear, with a NYPD officer, have shared surveillance of every church and have partnered on patrol around every mosque. *Leo, from the front of the building, with an FBI agent, and Murray, from the rear, with a NYPD officer, have covered every church and have secured every mosque. 59. A further iteration provides more challenging examples: (i) A silk screen all around and a canopy of vines from above secluded every lover happily and a beloved tentatively in a secret bower of bliss. (ii) A rubber band horizontally and a length of twine vertically bundled tight every framed portrait face up on the bottom and a framed copy facedown on top. Neither silk screen nor vine secludes without the other, and no lover is secluded without a beloved. Similarly, neither rubber band nor twine bundles tight without the other, and a framed portrait is a tight bundle only with its copy. That is, (i) and (ii) are doubly collective, with respect both to the subjects’ participation and the objects’ participation. Yet Quantifier Lowering persists to allow every lover and beloved their own silk screen and canopy and every framed portrait and its copy their own rubber band and length of twine. There are some further obstacles I leave for the future. Recall that the analysis in chapter 2 relies on a couple of plural event pronouns. The first is prefixed to the Right-Node Raised phrase and describes for (474) what the one backhoe does from below and the one crane does from above. If the subjects that are antecedents for this description are lowered into the Right-Node Raised phrase and its unspoken counterpart at the deletion site, the two tokens of this first event pronoun must be lowered along with the subjects in the manner already established in section 3.2.1.4 for lowering number agreement. In section 2.1.3, the relation between Right-Node Raised phrase and deletion site is that of antecedent to a backward descriptive anaphor, the second event pronoun of the analysis, in which the deleted phrase Δδ is taken up as its content: “… [∀e: Eie][℩Eδ : Δδ]Eδe) and … .” But then the lowering of the first subject (and the first event pronoun) into Δδ is lowering into the content of a pronoun,
874
Notes
which seems rather obnoxious. I suspect that casting deletion site and Right-Node Raised phrase as descriptive pronoun and its antecedent is not strictly necessary. In light of the demands of Quantifier Lowering, some other kind of ellipsis should be considered. Or else I let loose ellipsis and reconstruction, where speakers hearing (474) jump to (iii) (iii) One backhoe from below and one crane from above have uprooted every elm and one backhoe from below and one crane from above planted every beech, from which the quantifiers every elm and every beech skip out over their tokens of the coordinated clausal remnant. The Eventish will be good, the translation from English ad hoc, but the meaning must out. 60. Suppose, for example, that the logical forms of (74) and (76) were something like (i) and (ii) respectively: (i) [A y : Pepsi(y)] (Drank Alia y and Marwan y ) (ii) [A y : Pepsi(y)] (Drank every girl y and every boy y ) Economy, as will be seen below, succeeds in allowing the quantifier to be lowered into the conjuncts in (ii) and excluding it from (i). But Economy would in this case apply with equal force to incorrectly exclude lowering in (75). Economy considerations will not distinguish (i) from (iii), the logical form for (75): (iii) [A y : Pepsi(y)] (Drank Alia yesterday y and Marwan today y ) 61. Fox (1995; 2000, 61–62n48) notes variation in the extent parallelism is enforced. Speakers accept the scope inversion in (i), reject it in (vi), and waver in between: (i) A guard is standing in front of every church and sitting at the side of every mosque. (ii) A guard is standing in front of every church and sitting at the side of almost every mosque. (iii) A guard is standing in front of every church and sitting at the side of most of the mosques. (iv) A guard is standing in front of every church and sitting at the side of two of the mosques. (v) #A guard is standing in front of every church and sitting at the side of one of the mosques. (vi) #A guard is standing in front of every church and sitting at the side of this mosque. 62. I do not hear the alleged contrast between (490) and (491)–(492). Does the following improve (490)? The results trickled in slowly, but it quickly became clear that at least one of the film critics admired every movie, and in the end, that every critic would. 63. Long Playing album (not Les Paul), from when vinyl was king. 64. State courts have jurisdiction over offenses against state statutes, and federal courts have jurisdiction over federal infractions. Kosher dietary laws strictly separate dairy and meat. 65. The contrast between definite descriptions and universal quantifiers may, however, be grist for the mill that sizes coordinations according to the distribution of event quantifiers. As distributive quantifiers in a Davidsonian setting, the universal quantifiers must include within their scope an event quantifier that (in)definite descriptions do not, a difference
Notes
875
exploited throughout. This, however, does not hold out much hope for the contrast between Mary and other definite descriptions in disambiguating the first sentence. 66. In light of (i)–(iii), (534) and (535) are probably irrelevant: (i) Having an open wound killed many of my patients, and CHF did too. (ii) Applying a schmear of sunblock alone between naked skin and direct sunlight protected many noses from a serious burn, and the big cabana did too. (iii) Having a schmear of sunblock alone between naked skin and direct sunlight protected many noses from a serious burn, and the big cabana did too. Scope inversion in the first clause may just be an illusion of singular reference to a property, activity, or state as in (i)–(iii). 67. Here the parenthetical remark falsifies the “telescoping” interpretation in that the second internal hard disk and every mainframe cannot all provide the second set of files to the same house server. I think it fairly clear even without the parenthetical that the mainframes must be backing up the house servers already backed up by internal hard disks. 68. Here are minimal pairs. Nuclear fusion illuminates stars. Their rotation and illumination are spatiotemporal coincidences, causally unrelated, and so it seems that the coordination of tensed auxiliary phrases in (i) is unacceptable: (i) #No star was rotating and was glowing hotter than a brown dwarf. (ii) No star was rotating and glowing hotter than a brown dwarf. Changing the circumstances to supply a causal connection seems to redeem the coordination: (iii) Despite unusual friction, no iron ball was rotating fast enough and was (therefore) glowing hot enough to melt. (iv) Despite unusual friction, no iron ball was rotating fast enough and (therefore) glowing hot enough to melt. Although my judgments are less secure, it may not be necessary for the rotation to cause the illumination but enough that they are engaged in some causal connection perhaps conspiring toward some third effect: (v) ?No star was rotating fast enough and was (#therefore) glowing hot enough to generate detectable pulses. (vi) No star was rotating fast enough and (#therefore) glowing hot enough to generate detectable pulses. Logical syntax glosses the coordination of tensed auxiliary phrases as follows, requiring the reference to antecedent events. There may be a variety of relations, as with any absolutive, justifying the speaker’s use of the adverbial modifier. Only when the logical syntax compels there to be such a modifier must the hearer scramble to figure out what the speaker meant by it. (i′) #No star was rotating and rotating, was glowing hotter than a brown dwarf. (iii′) Despite unusual friction, no iron ball was rotating fast enough and rotating, was (therefore) glowing hot enough to melt. (v′) ?No star was rotating fast enough and rotating, was (#therefore) glowing hot enough to generate detectable pulses.
876
4
Notes
PredP and PredP: Coordination vs. Subordination
1. According to Sportiche (1997, 2005), the eccentricity is only superficial with nominal quantification ‘(D NP) Y’ also deriving from ‘D (NP,Y)’. See Conjunctivism (Pietroski 2005, 2006). 2. Thus, note in particular that the presence of an overt, nominative subject in (20b) and (i) does not indicate gapping. (i) He has shimmied and she shaken. Instead, there is a coordination of participial phrases and the subject of the first is raised asymmetrically outside (see Johnson, [1996] 2003, 2002). 3. Accordingly, ‘W’ is not a bound morpheme, as the disjunctive interpretation of the coordination in (i) demands its occurrence outside the coordination, where an adverb can in turn intervene between ‘W’ and any verb within the coordination: (i) The Columbia students swiftly surrounded the Pentagon and crowded into the Mall. 4. Kayne (1994) and Zoerner (1995) hold that the apparent coordination of verbs masks a coordination of VPs. But this observation is not sufficient on its own to account for (26)–(29), since the complements to many bound morphemes are maximal projections (subject to Head Movement), Tense being an example par excellence. Given that the likes of re-VP, co-VP, and Past-VP have well-formed outcomes, the coordination is what goes wrong in re-(VP and VP), co-(VP and VP), and Past-(VP and VP). 5. So, it is specifically rejected that (31) yields to an across-the-board movement of Tense to derive something like (30), which would be masquerading as (31) merely to satisfy morphosyntactic conditions. 6. See section 13.0. 7. See chapter 13. Taub (1989) offers the most interesting characterization. She observes that only those Aktionsarten that include an activity component (i.e., activities and accomplishments) admit semidistributive interpretations. Thus the predicates in (i), which are either statives or achievements, according to the usual tests for Aktionsarten, block semidistributivity. But the activities and accomplishments in (ii) do not. (i) *Few koalas are numerous. *The koalas are all numerous. *Few trees are dense in the middle of the forest. *The trees are all dense in the middle of the forest. *Few senators passed the pay raise. *The senators all passed the pay raise. *Few students decided unanimously to skip class. *The students all decided unanimously to skip class.
Notes
877
(ii) Few students coauthored a novel. The students all coauthored a novel. Few combatants walked in single file. The combatants all walked in single file. Few combatants huddled in the center. The combatants all huddled in the center. 8. Note that it is crucial for the argument that distributive quantifiers such as few combatants, all, each, and so on supply singular (i.e., first-order) variables. No matter what mechanism derives the disjunctive interpretation in (i), its application in (ii), which emerges unambiguous, collapses into a conjunctive condition on the singular subject. That is, all and only those that are among the Bechstein carriers and the Bösendorfer hoisters are that Deathwish Piano mover. (i) The Deathwish Piano movers carried the Bechstein and hoisted the Bösendorfer. (ii) The Deathwish Piano mover carried the Bechstein and hoisted the Bösendorfer. 5
PredP and PredP: (Tense+) Aux Sharing
1. For discussion of analogous sentences, see McCawley 1993; Lin 2000, 2002; Johnson 2000a. 2. The logical form for (10a), for example, must contrive to express that a relevant rocker’s mama shimmies and does so in the past, and so if one’s thinking about logical form is suitably impoverished, logical forms for (10a) can hardly be discerned from those for (10b). Without room to tamper further in logical form, it will then be conceded that the process that allows a second token of shimmied to go unpronounced in (10a) manages somehow to stretch the scope of no rocker. With some enrichment of logical form, the difference in nominal scope between (10a) and (10b) can be seen to correlate with a difference in clause structure. The logical form for (10b) would be something like (i), with independent Tense morphemes in both conjuncts: (i)
([No x : rocker(x)]∃e(Past(e) shimmy(e) Agent(e,x) slow(e)) and ∃e([ιy : hisx mama(y)] Past(e) shimmy(e) Agent(e,y) fast(e))
(ii) [No x : rocker(x)](∃e(Past(e) shimmy(e) Agent(e,x) slow(e)) and ([ιy : hisx mama(y)] shimmy(e) Agent(e,y) fast(e)) (iii) [No x : rocker(x)]∃e(Past(e) shimmy(e) (Agent(e,x) slow(e)) and ([ιy : hisx mama(y)] Agent(e,y) fast(e)) In contrast, the logical form for (10a) would omit Tense morphology in the second conjunct, along the lines of (ii) or (iii). Elaborated further, the second conjunct would come to assert (ii) that his mama is also a shimmier in the event described by the first conjunct, or (iii) that she is also an agent in that event. If (iii) is correct for (10a), where it is especially clear that the coordination is complement to the verb, the contrast in nominal scope between (10a) and (10b) will also fall under the generalization that a decreasing quantifier simply never escapes from a tensed clause. Surely that prospect itself argues for so enriching logical form, but in the text I am favoring for now those arguments for (Tense−) Aux sharing that do not ask the reader to take on my commitments.
878
Notes
3. The contrast confirms some difference in the syntax of logical form that affords the decreasing quantifier wide scope in (3)–(4) and denies it in (6)–(7). Some other candidates for the difference should be mentioned before moving on. Perhaps (3), for example, is parsed “… Past(be shimmying and be shaking) …” in contrast to (6)’s “… Past be shimmying and Past be shaking … ,” where the auxiliary verb is always repeated and a difference in logical form made out only from tokens of Tense. This parse of (3) relies, however, on a treatment of the bound tense morpheme explicitly ruled out in the next section. Also—ad hominem—I can think of no one wishing to repeat the auxiliary verb in the second conjunct who would take comfort to learn that they can, provided Tense is not repeated along with it. Usually, the motivation to repeat the one applies equally to the other. Another possible way to draw out different logical forms, which should be considered in light of the proposals to be made in this chapter, is that (3)–(4) and (6)–(7) are the same in their phrasing, tokening Tense and the auxiliary verb in both conjuncts, but differ only in the distribution and content of their event pronouns and quantifiers. This variation in referring to events that later sections lay out proves not to be very helpful in the present case, where it does appear in the end that the required difference of logical form is that (6)–(7) contain two tokens of the auxiliary verb and (3)–(4), only the one. Note that replacing no rocker with every rocker or Sid Vicious vacates the contrast. 4. As Johnson and Lin discuss, (11)–(13) contrast with (i) and (ii), which, ambiguous, make possible a reading in which negation and the modal are tokened again in the second conjunct: (i) Kim didn’t play bingo and Sandy, canasta. a. Kim didn’t (play bingo, and Sandy play canasta) b. Kim (didn’t play bingo, and Sandy didn’t play canasta) (ii) Ward can’t eat caviar and his guests beans. a. Ward can’t (eat caviar and his guests eat beans) b. Ward (can’t eat caviar and his guests can’t eat beans) Thus, as closely matched a pair as (iii) and (iv) are subject to a basic contrast between auxiliary verbs and main verbs: (iii) Kim did play bingo and Sandy play canasta. a. Kimi did (ti play bingo and Sandy play canasta) b. *Kimi (did ti play bingo and Sandy did play canasta) (iv) Kim played bingo and Sandy canasta. a. *Kimi played (ti bingo and Sandy canasta) b. Kimi (ti played bingo and Sandy played canasta) The structure shown in (iiia) meets all of the auxiliary verb’s requirements and any other structural requirements—so it appears—and must be declared the only analysis of the sentence if (11)–(13) are to remain unambiguous. One may suppose economy considerations compel (iiia) to preempt (iiib). Since a similar analysis should not be allowed to preempt ambiguity in (i) and (ii), there must be something defective in similarly analyzing a main verb as in (iva). Perhaps main verbs cannot originate from a position that includes within its scope a position for a pronounced subject. This would suffice to locate an unpronounced main verb in the second conjuncts of (i), (ii), and (iv). What is interesting and the subject of much dis-
Notes
879
cussion is that this recognition of a gapped main verb then becomes license for gapping more structure, the auxiliary verb and negation in (ib) and (iib). 5. Note that (14) and (18) are not to be confounded either. Accelerate Kunstler’s manic movement between sitting and standing to the point where (14) continues to be a true and natural report, but the observer is unable to make sequential observations in a natural way that would warrant (18). 6. I allow both (i) and (ii), with however a difference of meaning between (iii) and (iv): (i)
There is a bench and a picnic table in the garden. In the garden is a bench and a picnic table.
(ii) There are a bench and a picnic table in the garden. In the garden are a bench and a picnic table. (iii) There is a left tire skidding and a left tire slipping. On the Citroën DS is a left tire skidding and a left tire slipping. On the Citroën DS is the left front tire skidding and the left rear tire slipping. (iv) ?There are a left tire skidding and a left tire slipping. ?On the Citroën DS are a left tire skidding and a left tire slipping. ?On the Citroën DS are the left front tire skidding and the left rear tire slipping. The sentences in (iii) can only describe the catastrophic condition but (iv) allows that the time of the one left tire skidding may not coincide with that of the other slipping, an effect of singular vs. plural reference to events. 7. Lacking the logical form of pronounced subjects in sub-Tense Phrases, recall that sentence (40) presents a question left open for the moment: with only the one token of the tensed auxiliary verb and thinking of its logical form as akin to there is the left front tire skidding and the left rear tire slipping (see notes 2 and 6), we might, by way of (38), expect contrary to fact that (40) mean that at a given moment what goes on is in a condition that if persistent coincides with the left front tire skidding and left rear tire slipping. Yet it seems only to describe catastrophe on the slalom: at a given moment what goes on if persistent coincides with the left front tire skidding and coincides with the left rear tire slipping. 8. I owe Roumyana Pancheva for extensive discussion of this section and for bringing to my attention the present perfect puzzle, as well as the relevant examples and references. No analysis of perfect aspect in general or of the crucial observations in particular is on offer. I am arguing from the facts. So watch your wallet. 9. The following adapts for present purposes the discussion in Giorgi and Pianesi 1997, 84ff. 10. See Pietroski 2000, chap. 4, for discussion of ceteris paribus laws. 11. The discussion has been slanted toward the pragmatics of reasoning, slighting a crucial point of grammar that Giorgi and Pianesi (1997) emphasize, namely, that in English Bulimia has eaten herself full is construed as “Bulimia has eaten herself full now” rather than as *“Bulimia has eaten herself full then.” If I have understood their discussion of the present perfect puzzle well enough to extend it here, speakers of Norwegian, Danish, and Swedish will find translations of (48) as anomalous as English speakers find (48), but speakers of Italian, German, Dutch, and Icelandic will judge them true. Of course such crosslinguistic variation underscores a difference of grammar.
880
Notes
12. More examples: (i)
#Manny sedated himself at lunch and is artificially stimulating himself at dinner. a. He has calmed himself down and (he) has shaken himself up. b. He has (been) calmed down and (he) has (been) shaken up. (ii) Manny sedated himself at lunch and is artificially stimulating himself at dinner. a. He had calmed himself down and (he) has shaken himself up. b. He had (been) calmed down and (he) has (been) shaken up. (iii) Manny sedated himself at lunch and is artificially stimulating himself at dinner. a. He has calmed himself down and shaken himself up. b. He has (been) calmed down and (been) shaken up. (iv)
(v)
(vi)
#Manny sedated himself at lunch and is artificially stimulating himself at dinner. a. Meds have calmed him down and (meds) have shaken him up. b. Downers have calmed him down and uppers have shaken him up. Manny sedated himself at lunch and is artificially stimulating himself at dinner. a. Meds had calmed him down and (meds) have shaken him up. b. Downers had calmed him down and uppers have shaken him up. Manny sedated himself at lunch and is artificially stimulating himself at dinner. a. Meds have calmed him down and (meds) shaken him up. b. Downers have calmed him down and uppers shaken him up.
(vii) #Leon played brilliantly at the competition and died backstage (to everyone’s dismay). He has charmed Carnegie Hall and Carnegie Hall has worn him out. (viii) Leon played brilliantly at the competition and died backstage (to everyone’s dismay). He had charmed Carnegie Hall and Carnegie Hall has worn him out. (ix) Leon played brilliantly at the competition and died backstage (to everyone’s dismay). He has charmed Carnegie Hall and Carnegie Hall worn him out. (x)
#Leon played brilliantly at the competition and died backstage (to everyone’s dismay). He has charmed Carnegie Hall and (he) has worn himself out. (xi) Leon played brilliantly at the competition and died backstage (to everyone’s dismay). He had charmed Carnegie Hall and (he) has worn himself out. (xii) Leon played brilliantly at the competition and died backstage (to everyone’s dismay). He has charmed Carnegie Hall and worn himself out. 13. Nor does the condition apply if the complement to have is an unmixed coordination of passive participial phrases: (i) The fabric had been tested (for 20 minutes) and been torn (through) (a bit) (in 20 seconds). (ii) ?The fabric had been tested (for 20 minutes) and a hole been torn through (it(s surface)) (in 20 seconds). ?The fabric had been tested for 20 minutes and its surface been torn (through). 14. As expected, the scope of decreasing quantifiers licensing bound pronouns and negativepolarity items conforms to the pattern established in (3)–(7): (i)
No fabric of Klein’s has been tested and fibers-schmibers proven it worth bubkes. No fabric of Klein’s has been tested and fibers-schmibers proven hís schmattes worth bubkes. No fabric of Klein’s has been tested and fibers-schmibers proven thát gonif’s goods worth bubkes.
Notes
881
(ii) *No fabric of Klein’s has been tested and fibers-schmibers have proven it worth bubkes. *No fabric of Klein’s has been tested and fibers-schmibers have proven hís schmattes worth bubkes. *No fabric of Klein’s has been tested and fibers-schmibers have proven thát gonif’s goods worth bubkes. (iii) But no fabric from mý factory has been tested for two húndred minutes and a hole (been) torn through mý fabric. (iv) *But no fabric from mý factory has been tested for two húndred minutes and a hole has been torn through mý fabric. 15. Examples are not hard to come by: Prophylaxis has broken down and a pox broken out. The Earth has trembled for a few minutes and hot lava broken through the crust. The Earth has trembled for a few seconds and grown men fallen to their knees in prayer. The Lord has spoken and the Evil One been rebuked. The father has outstretched his arm and the son come home. The leviathan has opened its maw and the vessel been engulfed. Goodwill has flown and malice shown its stripe. Stocks have gone up and bonds gone down. 16. Correlatively, as Danny Fox (p.c.) points out, (i), which gets the causal flow right, is acceptable: (i) The sun has come out and the flood withdrawn. 17. See note 2. 18. I have observed a further condition on this construction, which I record without comment. In (i) and (ii), there is a felt implication that the subject of the second conjunct is lying in wait at the scene and thus in existence prior to the result being reported. Thus, the sentences in (ii) suggest oddly that the holes and the errors were there from the start. The coordination of tensed clauses in (iii) escapes this suggestion and reports straightforwardly the creation of holes and errors. (i)
The vessels have sat in the harbor for 40 years and corrosive chemicals eaten through their hulls. Apple’s vigilance has slackened and Microsoft subroutines shown up in its code.
(ii) #The vessels have sat in the harbor for 40 years and hundreds of holes eaten through their hulls. #Apple’s vigilance has slackened and errors shown up in its code. (iii) The vessels have sat in the harbor for 40 years, and hundreds of holes have eaten through their hulls. Apple’s vigilance has slackened, and errors have shown up in its code. 19. Among others, see Reichenbach 1947; McCoard 1978; Dowty 1979; Giorgi and Pianesi 1997.
882
6
Notes
PredP and PredP: Complementation as a Condition on Subatomic Event Anaphora
1. As displayed, the logical forms (5)–(9) suppress adverbialization (chapters 1, 8–15), which enlarges the structure further. 2. See note 3. 3. Note that (12) (and (7) and (9) earlier) place a token of the Overlap relation outside the coordination, in violation of the condition on bound morphemes (section 4.2). Provided that the content of the pronoun is fixed as shown, the relation could just as well be tokened within the conjuncts as in (i), allowing all the bound morphemes to occur with what they are bound to, in compliance with section 4.2. Also see section 6.1. (i) [The X : … Rx …] ∃E0W[E0,X] [℩E0:pro0]Present[E0] [℩E0:pro0]Will[E0] [℩E0:pro0][℩E1:pro1]O[E0,E1] ∃E1∃XW[E1,X] [℩E1:pro1]Have[E1] [℩E1:pro1] [℩E2:ΦΨ] (O[E1,E2] ∃E2(Φ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (O[E1,E2] ∃E2(Ψ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4])) 4. Sentence (19) is itself a coordination of larger, predicative phrases, which does not in this case block the disjunctive interpretation. See section 6.3. Translations of (20)–(22) and (23)–(25) in French (Dominique Sportiche, p.c.), Italian (Rita Manzini, Luigi Rizzi, p.c.), and Spanish (Juan Uriagereka, p.c.) show the same contrast. 5. Existential closure in the disjunctive interpretation is confirmed, recall from chapter 1, by the interaction with a modal adverb: (i) Three Harvard students cooked up the Harvard beets and possibly garnished them with pearl onions. If (i) is understood with the so-called full Conjunction Reduction reading, then like the paraphrase in (ii), the grounds for using possibly remain fairly open: (ii) Three Harvard students cooked up the Harvard beets, and possibly they garnished them with pearl onions. It may be that the speaker is uncertain whether the beets were ever garnished, although if they were, she’s certain it was the three Harvard students who did it. Or it may be plain that the beets were garnished, and certain that three Harvard students cooked them, but she nevertheless suspects that the three she later saw with the garnish were impostors. In contrast, with the disjunctive interpretation of (i), allowing the Harvard students to be divided between the cooks and the garnishers, the possibility conveyed renders the sentence equivalent to (iii): (iii) Three Harvard students cooked up the Harvard beets, or three Harvard students cooked up the Harvard beets and garnished them with pearl onions. Something went on. It was at least a cooking up of the Harvard beets and maybe also a garnishing, and three Harvard students did it. If there was no garnish, then three Harvard students did what was left to do, the cooking. No possibility that they may have been impos-
Notes
883
tors is entertained here. Sentence (i) would not be felicitous if it were known that the one Harvard student Biff cooked up the beets by himself and suspected that Yalies Eli and Nathan posing as Harvard students Miff and Tiff garnished the beets with pearl onions. On the other hand, the full Conjunction Reduction interpretation, paraphrased in (ii), is appropriate on analogous conditions: it is known that Biff, Miff, and Tiff cook up the Harvard beets and suspected that Eli, Nathan, and Bull posing as them did the garnish. This divergence of the disjunctive interpretation follows directly from its logical form. Existential closure simply removes reference to the Harvard students and any concern for their true identity from within the scope of the modal adverb: (iv) ∃Ei[∃X : 3(X) & … Hx …]W[Ei,X] O(proi,proj) ∃E′j(∃XAgent[E′j,X] C(pro′j,pro″k) ∃E″k (cook[E″k] …)) and possibly ∃E′j(∃XAgent[E′j,X] C(pro′j,pro″k) ∃E″k (garnish[E″k] …)) With (iv), the speaker only hedges the possibility that pearl onions were garnished then. The modal within the conjunct thus provides an assay for the content of the subject within its scope. The coordination is conjunctive if and only if the subjects of both conjuncts refer directly or descriptively to the matrix subject, and then the modal can hedge the reference that falls within its scope. The coordination is disjunctive if and only if neither conjunct contains a reference to the matrix subject, as it must be if the sentence is to remain vague about who participated in the events described by which conjunct. In that case, when the coordination is understood disjunctively, the modal cannot hedge a reference to the matrix subject, since no such reference falls within its scope. 6. Anaphora within is harmless to the disjunctive interpretation, as in (i), where a single existential quantifier within each conjunct binds all the subject positions and closes them off from outside: (i) ∃Ei[∃X : 20,000(X) & …]W[Ei,X] will[Ei] O(proi,proj) ∃E′j∃X(W[E′j,X] … have[E′j] … ∃E″(Agent[E″,X] … surround[E″] …)) and ∃E′j∃X(W[E′j,X] … have[E′j] … ∃E″(W[E″,X] … been[E″] … ∃E‴(Theme[E‴,X] … crowd[E‴] …))) 7. As in section 3.4.0, a coordination of tensed auxiliary phrases is catastrophic for the disjunctive interpretation: (i) The rockersl (([℩E: NPl] Wk[E,X] prok have proj shimmiedj) and (prok … have proj shakenj)) All the event pronouns are properly contained within the conjuncts, and in particular, one at the left edge of the second refers to what the first says the rockers were doing. If indeed the first says they were shimmying, the disjunctive interpretation will die right there. But, as remarked in section 6.1, there is no guarantee that the first conjunct commits to shimmying rather than a collective shimmying and shaking unless there are the further conditions on the second event pronoun proj introduced there and elaborated below. 8. Also, like (25), a disjunctive interpretation is inaccessible to (23), as noted earlier: (23) b. Twenty rockers will have shimmied and will have shaken.
884
Notes
9. The same is said for (i), complying with section 4.2.1. The observation is that there can be no projections of O above the tensed auxiliary phrase. (i) *[The X : … Rx …][℩E1:Φ′Ψ′] (∃E0W[E0,X][℩E0:pro0] O[E0, E1] ∃E1(Φ′ ∃XW[E1,X][℩E1:pro1]Present[E1] [℩E1:pro1]Have[E1][℩E1:pro1] [℩E2:pro2]O[E1,E2] ∃E2∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shimmied[E4])) and (∃E0W[E0,X] [℩E0:pro0] O[E0, E1] ∃E1(Ψ′ ∃XW[E1,X][℩E1:pro1]Present[E1] [℩E1:pro1]Have[E1][℩E1:pro1] [℩E2:pro2]O[E1,E2] ∃E2 ∃XW[E2,X] [℩E2:pro2][℩E3:pro3]O[E2,E3] ∃E3∃XAgent[E3,X] [℩E3:pro3][℩E4:pro4]C[E3,E4] ∃E4shaken[E4])) 10. Since chapter 3 and until this moment, logical forms have suppressed displaying adverbialization. 11. The clitic placement is taken (Kayne 1993; Tortora 2010) to indicate that the auxiliary verb + participle construction is biclausal opposite the monoclausal simple tensed verb construction. 12. Observe that (19) contrasts with (i), where have is the auxiliary verb that enters into a mixed coordination and blocks a disjunctive interpretation: (i) Twenty thousand students surrounded the Pentagon and have crowded into the Mall. The syntax of clitic climbing will not distinguish (i) from (19), but something may have gone awry with the collective reference to events. Compare the nominalizations—the surrounding the Pentagon and being crowded into the Mall, their surrounding the Pentagon and being crowded into the Mall, ?the having surrounded the Pentagon and having crowded into the Mall, their having surrounded the Pentagon and having crowded into the Mall, but *the surrounding the Pentagon and having crowded into the Mall, *their surrounding the Pentagon and having crowded into the Mall. It is as if collective reference under a single description must be uniformly to either some states or some episodic events. In a verbal passive, the copula occurs in an episodic report and thus the events described by the coordination in (19) are uniformly episodic. The disjunctive interpretation of (i) would require mixed reference to episodic surrounding and stative having crowded-into and thus fails. Alternatively, the contrast between (19) and (i) may be taken to support the treatment that assimilates (18) to (58), to the extent that the disjunctive interpretation and the covert do that this analysis requires correlate well with overt counterparts. That is, (19) vs. (i) reflects (iii) vs. (iv). (18) Twenty rockers shimmied and shook. (58) Twenty rockers did shimmy and shake. (iii) ????They do be crowded in there. (iv) *They do have crowded in there. 13. I am indebted here to Kyle Johnson, Richard Larson, and Liz Spelke for discussion and judgments concerning the English data. 14. John D. (1839–1937) founded Standard Oil in 1870, Nelson A. (1908–1979) governed New York from 1959 to 1973, and David Sr. (1915–) chaired the Council on Foreign Relations from 1970 to 1985 and remains an honorary chair. A fourth Rockefeller, David Jr. (1941–), is a
Notes
885
member of the Council; I have made him a cochair, since my informants otherwise protested the plural number agreement. 15. A single token of ‘W[E0,X]’ is ensured when representing the disjunctive interpretation, in order to express the vagueness of who did what. Here I leave open several possibilities (i–iii) that the conjuncts are larger, each containing a token of ‘W[E0,X]’, when representing the conjunctive, distributive interpretation: (i)
[∃X : 4(X) ∃E Rockefeller[E,X]][℩E : Rockefeller[E,X]] [∃E0 : N[E,E0]]W[E0,X]founded Standard Oil … and [∃E0 : N[E,E0]]W[E0,X]governed New York State … and [∃E0 : N[E,E0]]W[E0,X]chair the Council on Foreign Relations …
(ii) [∃X : 4(X) ∃E Rockefeller[E,X]] [℩E : Rockefeller[E,X]][∃E0 : N[E,E0]]W[E0,X]founded Esso … and [℩E : Rockefeller[E,X]][∃E0 : N[E,E0]]W[E0,X]governed NYS … and [℩E : Rockefeller[E,X]] [∃E0 : N[E,E0]]W[E0,X]chair the CFR … (iii) [∃X : 4(X) ∃E Rockefeller[E,X]]i [℩X: proi][ιE : Rockefeller[E,X]][∃E0 : N[E,E0]]W[E0,X]founded Esso … and [℩X: proi][ιE : Rockefeller[E,X]][∃E0 : N[E,E0]]W[E0,X]governed NYS … and [℩X: proi][ιE : Rockefeller[E,X]] [∃E0 : N[E,E0]]W[E0,X]chair the CFR … If so, each conjunct of a conjunctive interpretation may be the occasion for both independent temporal reference and a fresh perspective on it. (For further discussion of scenes and perspectives, see Schein 1993, chap. 10; 2002; chapter 9 below.) 7
PredP and PredP: Conclusion
1. Thus, (20) and (21) are felicitous without semantic condition. (22) ( = (39)) is felicitous just in case semantic conditions on descriptive event anaphora are satisfied, which they are not in (23) and (33c). (34c) and (40) are altogether unacceptable (see sections 3.2.0 and 3.7). 8
Introducing Adverbialization and Cinerama
1. Ruled out is a box of noix and a box of noisettes and no box for all the pralines. The sentence is again true and felicitous if the box of noix and the box of noisettes are inside a third box containing them all. 2. The quantifiers are chosen to defeat the temporal quantification in (i), which is weaker than (17)’s assertion that the motorists injured in collisions under 10 mph were few in total. A descriptive, temporal anaphor such as then in (ii) will do, but its descriptive content is indeed while motorists. See Larson 2003 on quantifying over events rather than times in the logical form for temporal phrases. (i) ∀t few motorists at t were injured at t in collisions under 10 mph. (ii) Few motorists at some time were injured then in collisions under 10 mph. 3. Strictly, it may be false that musician(e,x) → (instrumentalist(e,x) ∨ vocalist(e,x)). In alternating harmonica and vocals, Little Walter creates a musical moment, “My Babe” (1955), that is neither instrumental nor vocal in its entirety. A fussier description, musician in
886
Notes
exclusively vocal or exclusively instrumental performance, restores the biconditional. Even the familiar adults and men and women do not support a biconditional, since the transgendered are adults throughout their adult life but that life is neither an event of being a man or of being a woman. Less familiar vocabulary provides examples without elaborate description (e.g., passers and servers and receivers) if it is clear that a volley is always multiple passes so that to have served and received once is to have been a passer not less than twice. 4. I assume National League rules, in which pitchers must appear at bat and there are no Designated Hitters, who bat without fielding. 5. This case also includes failures of substitutivity between the simple description the lovers and a coordination of descriptions containing the relevant NPs, the lover and the belovèd. 6. There is irrelevant distraction from a reading that says of the kind, the Apostles, and not just the some of them martyred, that they are twelve and similarly of the kind, the Saints. 7. Notice that judgments veer elsewhere in (i): (i) Few musicians abuse their instruments. 8. For a sharper focus, consider a world in which the above facts are the same and it also happens that the one and only acquired instrument is the kazoo, so that all music down to the semihemidemisemiquaver is instrumental and vocal. Sentences (59)–(61) are judged the same, with all kazoos in perfect condition and all vocal folds scarred. So it must be that adverbial modification in (i)–(iii) expresses a relation between the events described by adverbialized NP and events of instrument abuse that is not merely temporal or even spatiotemporal unless the speaker has in mind location in the performer’s vocal tract and airstream. My use of while in paraphrase abbreviates the range of relations spatiotemporal, causal, or causal-explanatory that absolutive modification as in (i)–(iii) in particular and conditionals more broadly invoke: (i) Few instrumentalists instrumentalizing, abuse their instrument. (ii) Few vocalists vocalizing, abuse their instrument. (iii) Few musicians, making music, abuse their instrument. That range requires requires their classification as relations between events proper rather than spatiotemporal locations (see notes 2 and 12; Larson 2003). The confusion in judging the truth of (61) or (iii) is that of violations of the Homogeneity Condition (Fodor 1970; Löbner 1985; Lønning 1987; Roeper 1983), which confound sentences that say so-and-so of a plural or mass definite description whenever what is described is neither all so-and-so nor all not so-and-so. The effect is expected and found (Gajewski 2005) wherever plural or mass definite descriptions are alleged, such as the plural definite description of events that adverbialized NPs and adverbial phrases in general provide (Schein 2003). 9. The validity of (i), which is to be preserved, shows that the plural morpheme should not be tinkered with in breaching (63): (i) The musicians are gifted is true iff The instrumentalists are gifted is true & The vocalists are gifted is true. 10. Sequence of spatiotemporal address (cf. sequence of tense) implies that here1 and here2 in (70) are different addresses.
Notes
887
11. As written, (101) assumes for DPs the canonical logical form in (5), which (100) does not instantiate. For (101) to apply directly to [D AdrP1-and-AdrP2], its logical form should be expanded as in any of (6)–(8). 12. Formalities should not obscure the empirical claim about adverbialization. ‘N[Ei, Ej]’ is intended to be whatever neighborhood or accessibilty relation mediates what adverbial phrases describe and what is described by the phrases modified: (i) (ii) (iii) (iv) (v) (vi)
The infection spreading rapidly, the patient is crashing. Infected, the patient is crashing. The sun setting, time passes. A triangle, this polygon has angles that add up to two right angles. Being a triangle, this polygon has angles that add up to two right angles. If a triangle, this polygon has angles that add up to two right angles.
Whatever the formal analysis of modification in (i)–(vi) and of its variability in being causal, spatiotemporal, or analytic, it should work the same way with nominal adverbialization. I do not expect to find daylight between adverbial modification by a triangle in (iv) and its counterpart adverbialization unspoken in (vii): (vii) A triangle has angles that add up to two right angles. 9
Cinerama Semantics
1. I thank Jessica Rett (p.c., September–October 2012) for extensive discussion of this example and the points it raises. 2. Note that in varying the frame of reference, here … here … here in (8) and (9) does what is denied to left … left … left in (5), presumably because the expressions referring in (5) to frame of reference f are zero pronouns or anaphors, less free than here to introduce new referents. Correlatively, for the intended reading of (8) and (9), each token of here must be read with emphasis, to signal that it is a demonstrative introduced de novo. Reduced stress indicating that the token so pronounced is anaphoric to an antecedent here would signal that the anaphor is to be understood exactly as its antecedent to refer to the same frame of reference f. 3. It follows from these remarks that utterances of “I am here now” are ambiguous according to the frame of reference intended for here. Whichever is intended, an utterance of “I am here now” that turns out to be a (contingent) a priori truth (Kripke 1980; Kaplan 1989) for one frame of reference remains so for any other that might have been intended then and there and successfully demonstrated or referred to. This last remark is meant to flag without prejudice for their semantics or pragmatics, utterances with failed demonstrations as when the mad king suffering a hallucination utters “This is my flagship, and I am here now and will remain at its helm until she is sunk.” The thought the king expresses with “I am here now” does not seem to be an a priori truth. My coy allusion to utterances that turn out a priori is done so with deference to those true utterances of “I am not here now” on answering machines the negation of which cannot be a priori (Predelli 2005) (see section 9.3 for discussion of displaced narrators). 4. Loomis et al. (1999, 128–129) remark that “path integration is the inclusive term referring to the updating of position on the basis of velocity and acceleration information, i.e., without
888
Notes
position-fixing. Authors broaden path integration to include navigation with restricted ‘viewpoint’ as when navigating medieval alleys. … To generalize, path integration is the process of navigation by which the traveler’s local translations and rotations, whether continuous or discrete, are integrated to provide a current estimate of position and orientation within a larger spatial framework. … Gallistel (1990) noted … path integration … allows the traveler to gradually integrate the isolated perspective views encountered into an internal representation (cognitive map) suitable for subsequent piloting.” I of course join those authors who broaden path integration to include visual navigation, my focus. 5. In fact, Yes and No are responses reserved for sound assertion issued with epistemic warrant from a sound mind. Suppose Ray, who believes that Soul will never die, is in dispute with Ringo, who variously envisages its death in relation to January 20, 1964, the release of Meet the Beatles and the launch of the British Invasion: (i)
Ringo: Soul died January 20, 1964. Ray: No, it didn’t. Soul will never die.
(ii) Ringo: Soul will die. Ray: No, Soul will never die. (iii) Ringo: Soul will die January 20, 2064. a. Ray: #No, Soul will never die. b. Ray: Huh, how the hell do you know that?! (iv) Ringo: Soul died January 20, 2064. a. Ray: #No, Soul will never die. b. Ray: #Huh, how the hell do you know that?! c. Ray: Hey man, what’s in your Kool-Aid?! In (iii) and (iv), Ray cannot felicitously respond as in (a) to what he simply believes is false. In (iii), Ray’s response is that of someone who believes that it is epistemically impossible to know the asserted metaphysical possibility, responding thus to an assertion issued without apparent epistemic warrant from a mind otherwise presumed sound. In (iv), Ray can respond neither as in (i)–(ii) nor as in (iii). He responds as someone who knows that it is epistemically impossible to know a metaphysical impossibility, namely, that a future date is past, responding thus to an assertion he judges to have been issued from an unsound mind. Unsurprisingly, the responses between introspective, self-monitoring instruments are designed to flag those discrepancies of fact indicative of suspected system corruption. Only instruments that are well calibrated with each other and trusted to be accurate exchange simple Yes and No. As (iii) and (iv) show, the range of other appropriate responses varies with the nature of what else is believed wrong. A similar point is vivid in discussions of mathematical conjectures: (v) Pierre de Fermat, 1637: “No three positive integers a, b, and c satisfy the equation an + bn = cn for any integer n greater than 2.” a. Colleague: “No, (it’s true) they do not.” b. Colleague: “But yes (it’s false), there are those that do.” c. Colleague: “Huh, mon cher Pierre, how the hell do you know that?!”
Notes
889
(vi) Andrew Wiles, 1995: “No three positive integers a, b, and c satisfy the equation an + bn = cn for any integer n greater than 2.” a. Colleague: “No, (it’s true) they don’t.” b. Colleague: “But yes (it’s false), there are those that do.” c. Colleague: “Really, Andy, lemme see.” The conjecture that Fermat and Wiles assert is certainly either true or false, without room among the colleagues for it to be truth-valueless. Yet Fermat’s colleague’s Yes or No is felicitous just in case her truth claim is either based, like Fermat’s, on a hunch—either in agreement with his or to the contrary—or made with some other dispositive evidence in hand. Absent epistemic privilege equal to Fermat’s, (vc) is her only felicitous response. (Well, also, Amen.) Similarly, Wiles’s colleague’s dissent means that Wiles’s proof is flawed or that she has herself proven the contrary; her assent, that his proof is correct or the conjecture independently proven by her. Yes and No and yes-ing and no-ing are lexica and speech with a full-throated meaning and pragmatics of their own beyond T and F. Their infelicity above is no indication that the semantics of Fermat’s Last Theorem contains a presupposition that has failed. 6. Here philosophers may part company, preferring to wait out the psychological “noise” in armchair comfort until the subject’s commitments to How-the-World-Is are revealed at t + Δt. 7. In articulatory phonology (Browman and Goldstein 1986, 1992), the plan, in the sense of motor control, of a speech utterance is a gestural score (see orchestral score) that coordinates the actions of independent articulators (lips, tongue tip, tongue body, velum, glottis, etc.). Speech perception parses the acoustic signal into the gestures that produced it and the gestural score that planned them, from which is inferred the speaker’s intended diction. Of course all of this is subdoxastic to what is experienced as direct perception of a sentence uttered. Yet the best theory of speech production and perception characterizes it as an exchange of gestural scores, plans, that no one is consciously aware of. It may be a helpful analogy to characterize utterance understanding as the apprehension of a score for the speaker’s thought, details of which are not presented to conscious awareness. 8. What is written above is congenial to a radical pragmatics that launches hyperquotational logical forms directly into a fog of global reasoning, bypassing altogether semantic objects such as propositions. But the hyperquotational logical forms only represent mentation at the moment t0 of linguistic utterance, and I do not mean to preclude that further encapsulated mental processes might arrive at t0 + Δt at other representations. 9. See Abusch 2012 and other papers from “Visual Narrative: An Interdisciplinary Workshop at UCLA,” June 20–22, 2012. 10. But see Chapter 11 (54)–(66) below for examples with very small numbers. 11. Some I thought better to introduce along with their linguistic application: Landmarks[ℓ,ℒ], section 13.1.1; FOR[E,F], viewfinder(ṡ,fṡ), section 14.1.0; kinematic(s), object tracking(s), section 14.2; similarly orienting[E0,E1], section 15.0.1. 12. Often with little regard for anything that might distinguish perspectives, frames of reference, or scene. In this and the following chapters, frames of reference belong to spatial orientation in 3-D space, where a frame of reference includes a coordinate system for space such as altitude, azimuth, and radial distance from an observer’s location at point of view. Scenes are the projection of 3-D space under a given frame of reference onto a 2(.5)-D manifold. The perspective is the n-tuple of parameters sufficient to determine projection and resultant
890
Notes
scene, such as the observer’s location at point of view, line of sight, field of view, frame of reference, resolution, and other optical properties. Answering different considerations, scenes and reticules (i.e., resolution screens) are imported in Schein 2002 to address problems of event identity. 13. Thanks to Elena Herburger (p.c., May 2011) for discussion of this point. 14. For further examples of scene-dependent nonmaximal reference, see (332) and (341) in section 10.3.1 as well as note 93 in chapter 10. 15. If we stand motionless back to back throughout T, then everything that is on the left with respect to my frames of reference F that ∏[f,T,I] is on the right with respect to your frames of reference F that ∏[f,T,you]. If we stand shoulder to shoulder, you to my left, facing the same way with parallel lines of sight, anything between these lines threatens to be on my left and your right. Yet as the parallax between our observations of such objects disappears, the resolutions that we can intend in actual report fail to resolve any positions at great distances that are left for me but right for you. 16. No doubt in part because the tokens are embedded in definite descriptions that occur in a context, the entire discourse, where definite reference succeeds only if they refer to the same left. But that just is the observation that a continuity in frame of reference is imposed throughout the sentence. But, see chapter 1, note 47. 17. Sentences (57) and (58) exemplify an extensional substitution puzzle introduced in section 8.1 (see (48)), the analysis of which is promised for chapter 15. 18. Here I just parrot common treatment of “sloppy” identity in reference among otherwise identical anaphors. 19. Consequently, one leapfrogs the other when colliding at the axis. 20. An endorsement of adverbialized logical form, (62)–(63), should not be mistaken for what it is not. I have no reason a priori to take (i)–(iii) to be synonymous, despite the heuristic of using them and their kind interchangeably in paraphrases: (i) Skating, Peggy pirouettes. (ii) While skating, Peggy pirouettes. (iii) If skating, Peggy pirouettes. Looking at (63), one might entertain that ∅ in (i), while in (ii), and if in (iii) correspond to different choices for ‘N,’ but if (i)–(iii) are not synonymous, I have not looked closely enough at the adverbial derived by (62)–(63) from a nominal description to choose among them or a fourth as yet unseen. There is also no expectation of synonymy among constructions that replace fully tensed clauses with various reduced clauses (see Schwarz 1998b): (iv) (While/If) on the left, Peggy pirouettes. (v) (While/If) skating on the left, Peggy pirouettes. (vi) (While/If) she is skating on the left, Peggy pirouettes. Yet whatever variation in (iv)–(vi) is discovered in the spatiotemporal relations between skating on the left and pirouetting, (vii)–(ix) are expected to duplicate it: (vii) The skater on the left pirouettes. (viii) The skater skating on the left pirouettes. (ix) The skater who is skating on the left pirouettes.
Notes
891
21. For the sake of concreteness, I idealize all coordinate systems as from a reference point for the given frame of reference, surely a mistaken idealization. 22. Patty and Cathy Lane are from The Patty Duke Show (1963–1966), a situation comedy. 23. In The Patty Duke Show, Brooklyn teenager Patty Lane and London debutante Cathy Lane are “identical cousins,” whom the actor Patty Duke realizes with changes in costume, coiffure, posture, diction and accent. The cousins often impersonate each other to perplex hapless dates and to escape sticky situations. 24. Note that there is no presupposition failure when (i) and (ii) are not taken to make any claim about orientation but only a true or false assertion that the earrings seen at Tiffany’s were or were not still the earrings seen at Van Cleef & Arpels. (i) The left earring and right earring that Patty Lane saw at Tiffany’s were still the left and right earring that Cathy Lane saw at Van Cleef & Arpels. (ii) The left earring and right earring that Patty Lane saw at Tiffany’s were not still the left and right earring that Cathy Lane saw at Van Cleef & Arpels. If the nominal descriptive content in (i) and (ii) is taken to be the sometimes left earring and sometimes right earring at Tiffany’s, the derived adverbial describes the same events as the earrings at Tiffany’s. 25. Universal Coordinated Time. The International Telecommunication Union could not agree in 1970 on either the English Coordinated Universal Time (CUT) or the French Temps Universel Coordonné (TUC). 26. The identical tuplet now-en-scène cannot translate pronouns referring to an identical tuplet seen previously but no longer: (i) Here now is an identical tuplet princess. [Fade to castle tower on dark and stormy night]. She lives alone in her own dark tower. She is to be translated as the identical tuplet then en-scène, or the identical tuplet just now en-scène. Up to a complexity that does not overburden disambiguation, the translation of pronouns may help itself to tacit locutions such as the identical tuplet first en-scène (in this scene) or the identical tuplet first mentioned (i.e., the identical tuplet en-scène at the first token of “now” in this narration), etc. 27. The discourse “Now you see it, now you don’t” qua artifact contains two tokens of now, which are the same two tokens repeated whenever the discourse is. The first token is socalled to distinguish it from the second. A token is thus not to be confused with an utterance of it. 28. As Δ may in fact get its narratives from several sources, ‘Δµ©’ so indexed is essential to refer to that original narrative fragment containing µ. 29. “You Are There, CBS (1953–1957), used real network correspondents to tell historical stories as if they were being televised live. Its writers, all blacklisted in the McCarthy-era, used the tales of Joan of Arc, Galileo, and others to make thinly disguised points about contemporary witch-hunts” (Walter Cronkite, “Remembering ‘You Are There,’ ” npr.org, October 27, 2003). 30. The adverb now in (113) is not the now of (99)/(103) unless the narrator for whom the past tense had signaled that October 19, 1987 is remote shifts—if you were there—just when
892
Notes
the adverb is tokened. See the literature on free indirect discourse (e.g., Banfield 1982, Sharvit 2008, and the references cited there). 31. Roger Schwarzschild (p.c., May 2013) protests the use of the historical present with stative predicates, a restriction for which I have no explanation. But he concurs in the judgments for the eventive counterparts of these sentences: (i) OMG, it’s the Big Bang. In 2½ hours, Man will walk the Earth. (ii) OMG, it’s the Big Bang. In 14 billion years, Man will walk the Earth. (iii) OMG, it’s the Big Bang. In 2½ hours and 14 billion years later, Man will walk the Earth. (iv) OMG, it’s the Big Bang. In 2½ hours, Man walks the Earth. (v) OMG, it’s the Big Bang. In 14 billion years, Man walks the Earth. (vi) #OMG, it’s the Big Bang. In 2½ hours and 14 billion years later, Man walks the Earth. 32. These remarks about the spatiotemporal now, then, here, and there parallel demonstrative reference, this and that, to objects and events. It must be grasped what object or event that lies at the point demonstrated is intended—nose, face, head, upper torso, body, or person; kiss, embrace, tussle, or love life. 33. Both πi above and σi below are variables over scenes. 34. Right-Node Raising (see chapter 2) of the Tense Phrase it will be the Dawn of Man is assumed so that Tense is distributed to the scope of both of the conjoined temporal-frame adverbials, in accordance with the sentence’s plain meaning. 35. Right-Node Raising (see note 34). 36. Recall (p. 417) Gallistel’s (1990) emergence from a subway station first misoriented 180° and then realigned. 37. See, for example, Aloimonos 1997; Gallistel 1990; Gallistel and King 2009; Golledge 1999a, 1999b; Redish 1999. In what follows, I use polar coordinates for all frames of reference, a gross simplification from expert opinion that favors path integration into allocentric frames of reference with Cartesian coordinates (see Gallistel 1990; Gallistel and King 2009). I pay for this when the meaning of left (see (137)) has to translate from polar coordinates to Cartesian coordinates. 38. Obviously, none of these could be the coordinate system by which actual New Yorkers navigate the city streets. 39. The plane of the celestial meridian halves the frame of reference so that azimuths on the celestial horizon between 0° and 180° are in the right hemisphere and those between 180° and 360° are in the left hemisphere. In the Cartesian coordinate system assumed, the y-axis in the plane of the celestial horizon coincides with the line from to for any positive r, and the y-axis has positive, increasing values from to and beyond. The x-axis coincides in the plane of the celestial horizon with the line from to , and it has positive, increasing values in that direction. 40. λt.Tt: π( sɶ(t )) and λt.Tt: l( sɶ(t )) are functions from moments to coordinates, f in f and 〈alt(l sɶ ( t ) ), alt(l sɶ ( t ) )〉 fo in egocentric fo, respectively, and so continuity and differentiability are defined as they are for motion through any polar coordinate space.
Notes
893
41. (86) still[E,Σ,F] ↔df ∃e1∃e2(e1≠e2 & Ee1 & Ee2) & ∃A en-scène[E, Σ,A,F] & ∀σ∀f((Σσ & Ff) → orienting(σ,f)) (87) again[E,Σ,F] ↔df ∃e1∃e2(e1≠e2 & Ee1 & Ee2) & ∃A en-scène[E, Σ,A,F] & ∃f(Ff & ∀σ(Σσ → orienting(σ,f)) 42. The English is a scene of is itself intentional language in that it may be recognized that a scene is of a human face—that is, that there is a human face in it—long before the recognition that it is of Groucho Marx. I intend for my purpose a relation between scenes and the objects or events they are scenes of such as might play a role in spatial orientation and navigation. At the moment one recognizes a human face, such a relation obtains—the scene, as it were, “pops” into 3-D space. If that is not a de re moment, then nothing is. It would thus be mistaken to classify the recognition of an unknown human face as a de dicto thought in opposition to a recognition de re of Groucho Marx. Rather, recognizing in an object its humanity and recognizing in it its Groucho Marx-hood are both de dicto but different de dicto judgments that accompany the same perception and recognition de re of an object, which is to deny once again that names are Millian. 43. Given the ever-so-flimsy excuse of an epistemic condition taken to warrant knowledge de re that Aristotle was the teacher of Alexander, it can hardly be objected that I take the proximal visual experience of s to confer knowledge de re of its distal original projection s©, as intimate knowledge of the one is knowledge of the qualitatively identical other. 44. The Rashomon cinematography was a special effect the point of which was to deny the veridicality otherwise presupposed. 45. Note, important later, that a single cinematic scene zooming in or out (see section 9.4.0) or in panoramic motion and thus a continuum of scenes overlapping in space subtended and landmarks observed is exempt from the anticonvergence condition, an organizational principle regulating the use of multiple cameras. 46. A concatenation of scenes standardly, but I do not preclude multiple, simultaneous scenes distributed in a panorama. 47. It will shortly be helpful to separately reify resolution as a reticule (see Schein 2002), “a network of fine threads or lines of reference in the focal plane of a telescope or other optical instrument, serving to determine the position of an observed object,” especially if it is as much an index of the precision of the navigator’s intentions and attention as of any optical quality in the scene itself. 48. Also, a hurricane that has stalled, which remains in the same place, as its stalling implies, while many of its features are in uninterrupted motion. The connection among reticule or precision, optical resolution, and what is under observation can be illustrated further. The Great Spot can be said to occupy a region surrounding the position of its centroid at −22o N −40o E. No honest report of the Great Spot’s position will however name this same point, –22.00000000000000000 … 00000000000000000 oN – 40.00000000000000000 … 00000000000000000 oE, both because the optical resolution does not support measurement to that precision and because, even if it did, the centroid’s position at that precision shifts with any molecule added to or subtracted from the Great Spot and thus runs up against the vagueness of what constitutes it at that scale. 49. Consider graph paper, a rather minimal reticule that resolves squares equinumerous with its cells. The squares’ proxies are the cells optically separated from each other by an inked
894
Notes
line that must have nonzero width to satisfy the definition of resolution (198). Things are resolved only if there is everywhere “daylight” between them, here an inky daylight. Thus, the cells that are proxies for the squares do not include the points of space that the inked lines occupy. 50. Correlative to (212) and (213), however, is that there is a scene counting two triangles in (i) and another counting one, but no scene that counts three, a problem I will leave in place without resolution. (i)
◭
Sentence (221) and other recounts such as (ii), which we might think of as narration to a scene that zooms out from the narrow frame of reference in (205) to the wider one, resemble violations of the anticonvergence condition (section 9.4.1), violations of sequencing of tense so-called in which those counted many are said to participate in events in which they converge on the fewer: (ii) The two greens that were flanking white are one green surrounding it. (iii) count(e2 ) & At(α sɶ ( t ( e 2 )), e2, t(e2 ), sɶ(t(e2 ))) & count(e1 ) & At(α sɶ ( t ( e 1)), e1, t(e1 ), sɶ(t(e1 ))) Zooming in and out and thus varying scale, resolution, and the resultant counts, are scenes in which what is first counted one way is later counted fewer or more. Note that for recounts, nothing in the scene that zooms in or out to a different result implies a spatial disorientation or misrepresentation of the space subtended. This proves crucial in distinguishing cases such as (221) and (ii) from violations of anticonvergence (section 9.4.1). 51. Schein (2005, 19f.) argues that count morphology, pl, in two fires, imposing the conditions of countability, applies to all the fire rather than to just the two: (i) [∃ξ : two[ξ] & fire[ξ] & [℩ξ : fire[ξ]]pl[ξ]] (ii) *[∃ξ : two[ξ] & fire[ξ] & pl[ξ]] 52. If you feel yourself about to stumble on the question of how many waves a waving hand waves, go back to the USSR for a more totalitarian example in which every well-wishing comrade en route conveys her wishes to the comrade VIP in a single right-handed salute. 53. It ought to be a law that streets intersect only at right angles, and to hell with Broadway. 10
Adverbialization in Logical Form
1. On conditionals as event quantification, see Lycan 1984, 1991, 2001; Schein 2003; Herburger and Mauck 2015. 2. First presented under the title “Life among Subevents (La vie parmi les sous-évènements),” invited address, Journées Scientifiques Sémantique et Modélisation, ENST-INFRES, CNRS, Paris, France, March 21, 2003. Many thanks to the organizers, Francis Corblin and Alda Mari, and to the conference participants. Also presented at the Semantics Workshop, Rutgers Center for Cognitive Science, May 7, 2005, with the benefit of extensive commentary from Kathrin Koslicki. I am also indebted here for helpful suggestions and discussion to Elena Guerzoni, Martin Hackl, Sabine Iatridou, Kyle Johnson, Richard Larson, Roumyana Pancheva, Paul Pietroski, and Roger Schwarzschild.
Notes
895
3. All that needs to be said about the transitivity alternation with symmetric predicates does not end with the observation that the same lexical item may be embedded in constructions imposing different perspectival relations. For the sake of concreteness, I take up a geometric example in (i)–(vi), in which the same lexical item align(e) occurs throughout: (i) (ii) (iii) (iv) (v) (vi)
The points are aligned. The points are aligned with each other. The points A and the points B are aligned. The points that are either A or B are aligned. Point A is aligned with point B. *Point A is aligned.
First, note that the intransitive (i) (and (iv)) puts all the points on the same line, in contrast to the reciprocal transitive in (ii), where, even assuming that strong reciprocity enforces pairwise consideration of all the points, it is a vacuous truth that any two points indeed determine a line. Second, within the intransitive construction, (i) and (iv) contrast with (iii), where the latter allows that the points A are aligned with the points B, without either the points A or the points B themselves being aligned. Imagine that A and B are each a cloud of points. It suffices for (iii) (but for neither (i) nor (iv)) that the “center of gravity” of the points A is aligned with the “center of gravity” of the points B. Third, contrary to what might be expected from the simplest Davidsonian analysis as a conjunction of terms, the phrase with point B cannot be dropped salva veritate from (v) to derive (vi). In response, consider that an alignment is constituted by an essentially plural relation to events or states (cf. (vi)). Following prior discussion of statives and (semi)collective predication (see The grains of rice weighed a gram (Lasersohn 1989; Gillon 1990; Schwarzschild 1991; Schein 1993, 302f.)), absent an external basis for individuating states or events, the events or states that the points in (i) participate in correspond either to the one in which they all partake or to those individuated by the points themselves. The meaning of align(e) then requires that e is an alignment just in case its constitutive events, E in (vii), are in a line: (vii) Theme[E,X] & O[E,e] & align(e) A single event is not constitutive of a line, but then the events in which the points participate are so constitutive just in case every point is in fact on the line. The coordination in (iii) says in separate conjuncts that there were one or more events in which the points A participated and there were one or more in which the points B participated: (viii) ∃E Themei[E,X] & ∃E Themej[E,Y] & [ιE: proi j] O[E,e] & align(e) Again, consistent with what is known about stative predication, take it that the points A participate in one event and the points B participate in another one event, of which (viii) then asserts that these two events are constitutive of an alignment. All the points are not thereby aligned, and it correctly leaves open whether it is “center of gravity” or some other standard that decides how to align the points A with the points B.
896
Notes
The failure of a modifier to drop salva veritate has already been encountered in the comitative construction (section 2.6.0.0; see also Schein 2002, 327ff.): (ix) T Russell wrote the whole of Principia Mathematica with Whitehead. (Lasersohn 1995) (x) F Russell wrote the whole of Principia Mathematica. The comitative construction was said to be an instance of causal “chaining.” Only the event(s) E′ in (xi), in which both agents and accomplices participate, cause an effect of, say, Principia Mathematica being written: (xi) … Agent[E,X] & With[E′,E,Y] & Cause[E′,E″] … No logical form based on (xi) would validate the inference that what the agents did alone E causes like effect. A similar reanalysis of the prepositional phrase in (v) could vacate the unwanted inference to (vi), but, unlike the comitative construction, the reanalysis here would need to swallow several prepositions ad hoc, or so it seems: (xii) Triangle A is similar to triangle B. (xiii) *Triangle A is similar. (xiv) Triangle A is different from triangle B. (xv) *Triangle B is different. The following remarks suggest a way out of this dilemma while falling short of a formal proposal. Recall from the text that a difference between Figure and Ground informs the transitive variant of all symmetric predicates. Since the humblest citizen in (32) is the Figure in motion relative to the fixed Ground that is the President, the citizen is understood to have elevated into that equality. Of course the state of equality itself implies no prior or subsequent movement or orientation and is thus distinct from whatever events or states provide the basis for discerning Figure and Ground. But however Figure and Ground are discerned, it must be said that the equality is the resultant and persistent state that they frame. If so, if equality is the point at which a motion or change of state comes to rest, where the President is the Ground and the humblest citizen is the Figure in motion, then the only circumstance in which equality is the result of such a motion is if it is motion to the President. Correlatively, if an inequality or difference is reported as in (xiv) to be the state resultant of motion with respect to triangle B, it must be motion from triangle B. The observation is that the notions of Figure and Ground and the choice of resultant state as described by the symmetric relation decide the appropriate choice of preposition. Now the analogy to the comitative construction can be taken up more generally. There is the state or event of being a Figure and the state or event of being a Ground (no matter the preposition), and these two events constitute or end in an alignment, according to (v). It then does not follow that (vi) the event in which point A is alone constitutes an alignment, as desired. Something along these lines is what is required to make good on the promise that it is the very same lexical item and concept, align(e), that is tokened in all these examples. 4. In some special contexts discussed in section 10.0.9, it is possibly the étude that has undergone transposition.
Notes
897
5. See Musan 1997 for discussion of this qualification. Compare (from Musan 1997): (31) a. On that day I was introduced to Gregory and Eva-Lotta. Gregory was from America, and Eva-Lotta was from Switzerland. b. I had a chance to have a closer look at him. Gregory had blue eyes. c. Suddenly I realized a remarkable thing: Gregory resembled Jörg Bieberstein. 6. In 2003, when this section was first written, Ted Kennedy (d. August 25, 2009) was alive. 7. Now why—just when thought turns toward a Platonic resemblance that never withers— one should have to weigh how quick the subject is, is a question I will return to. 8. “You know you don’t have to act with me, Steve. You don’t have to say anything, and you don’t have to do anything … not a thing. Oh, maybe just whistle. You know how to whistle, don’t you, Steve? You just put your lips together and … blow.” 9. The contradiction I intend with these examples can of course be set aside if the reference of now is extended enough to allow for Linda Fiorentino to have changed within it. 10. Spared Broadway cabaret, the community for whom Lauren Bacall is only she of To Have and Have Not (1944) and The Big Sleep (1946) accepts (76) as true and unambiguous, and fixed by a resemblance to Lauren Bacall then. However, the reference of a name is fixed (see Trotsky [1924] 1971, Evans 1973, and Zimmermann 2005) and whatever temporal index is tacitly evoked, that index stands in no relation to Tense if the name is not in subject position. 11. How this is cashed out in the semantics is immaterial here. If ‘’ refers to the ordered pair of Lenin and 2003, which certainly exists, then resemble should be taken to denote only such ordered pairs that correspond to actual temporal slices. Alternatively, the semantics of the angled brackets, as it were, could be that of a definite description referring to such temporal slices, in which case both ‘’ and ‘’ among many others simply fail to refer. 12. Cf. Landman 1989a, 1989b, for whom as-phrases are term restrictors. Thus, (i) (ii) (iii) (iv) (v) (vi) (vii)
John as a judge is John. John as a judge is a judge. If John as a judge is P and John as a judge is Q, then John as a judge is P and Q. If John as a judge is P and P implies Q, then John as a judge is Q. It is not the case that John as a judge is both P and not-P. John as a judge is either P or not-P. If John as a judge is P, then John is a judge.
For important critical discussion and an alternative closer in spirit to the one offered here, see Szabó 2003. 13. If adnominal by hypothesis in (i), it remains sometimes adverbial at least when there is no nominal in sight as in (ii): (i) Lenin while he rests dead in the Kremlin resembles Queen Victoria. (ii) It is admitted while Lenin rests dead in the Kremlin that the Bolshevik Revolution did not create a workers’ paradise.
898
Notes
A sometimes adverbial use of along the edge of the wood in (iii) is also suggested by the sentence’s ambiguity: (iii) Bellerophon tracked Pegasus along the edge of the wood. It can be understood that either (1) Bellerophon himself moves along the edge of the wood as he tracks Pegasus, who may be in a clearing throughout, (2) Bellerophon remains in a clearing with an unobstructed view of Pegasus, who moves along the edge of the wood, or (3) neither Bellerophon nor Pegasus is along the wood’s edge, both in a clearing, but the line of sight, the tracking, is along the edge. Adnominal modification in (iii), if it exists, could only derive two of three interpretations, the first two, I guess, leaving the third to be derived by adverbial modification of track. Even if some of these reflect vagueness rather than ambiguity, it is hard to imagine that an adnominal modification could allow that only the tracking and neither Bellerophon nor Pegasus is along the edge of the wood. If so, then (iii) too presents a case of adverbial modification. Thanks to Andrew Ira Nevins (p.c., May 2003) for a question prompting this note. 14. I continue to suppress internal event quantifiers and cross-referring event pronouns, with the exception of ‘[℩E0:pro0]’ which translates number morphology (chapter 2). 15. Tense is shown applied to E0, but the argument allows that it apply to any Ei, in which the referent of the subject participates and the referent of the object of resemble does not: (i) W[E0,α] [℩E0:pro0] O[E0,E1] Past[E1] ∃XTheme[E1,X] … resemble[Ei] … TO[Ej,Y]. … Any association of Tense with what subjects do—directly or indirectly mediated through overlapping events—and its dissociation from what objects do suffice for the lifetime effect (section 10.0.7). Sentences about time travel (section 10.3), prompting the introduction of the dislocative morpheme − in (272) (section 10.3) and the location of adverbs below it, and sentences about intentional states (see (24)–(26)), comply as required. 16. It could hardly be otherwise under some plausible assumptions. Suppose sentences are parsed ‘… Tense-VP’, and VP contains in superior position some thematic relation describing the subject’s participation, ‘∃XTheme[E1,X]’ in (108). Now suppose further that Tense (and perhaps even adverbial modification) is local—especially if a bound morpheme. Tense is then constrained in (108) to apply either to E0 or to E1. In short, given the exploded structure that supermonadicity provides, it may be impossible—all the better—for Tense to apply to the resembling E2 or to the even more remote events of the object’s participation. 17. On phase sortals, see Burge 1975; Larson 1983, 1998; Enç 1986; Higginbotham 1987; Parsons 2000. In adopting the relations Lenin(e,x), Clark Kent(e,x), Superman(e,x), Hesperus(e,x), Phosphorus(e,x), Venus(e,x), etc., I do not deny their direct reference to what they denote, extending Burge’s (1973) treatment of names as austere predicates, which accommodate Kripkean objections to richer descriptive content and yet as predicates manage to render as analytic “Bruce, who teaches logical positivism, is a Bruce that teaches logical positivism,” “… (℩x)(Bruce(x)) … [∃x : Bruce(x) …]… .” (For further discussion see Platts [1979] 1997; Higginbotham 1987; Larson and Segal 1995; Longobardi 1994.) The relational view of names has repelled direct reference theorists I have met in bus stations, who believe that Kripkean arguments tell against even austere, rigidified descriptions and lead straightaway to the Millian logical form b for the name Bruce. Not being a philosopher, I take those argu-
Notes
899
ments to belabor the point that not only is a picture worth a thousand words, it’s worth more than any number of words, a valid point that however holds across the vocabulary. Thus to be competent to use to flutter, shimmy, shake, or shudder and to reliably discriminate among the motions they denote is not to command coextensive paraphrases of the verbs whether they are taken at large or in the jargon of car talk. It rarely gets beyond “I know it when I see it.” Even when the subject possesses vocabulary that could in principle provide complete description, as in musical or dance transcription, an exquisite sense for aesthetic judgments such as Slim swings East Coast style circa 1944 does not imply a torturous analysis into the vocabulary of constituent dance movements and description of the class of movements that constitute to swing. Contrary to what the Millian seems to take for granted, a word’s resistance to paraphrase and its meaning’s essential dependence on experience is no clue at all to its logical type—b, (℩x)(Bruce(x)), or [℩x : ∃E Bruce[E,x]]. Thanks to Paul Pietroski for discussion on this and much else. 18. Or instead of ‘N[E,E-1] bald[E-1]’, any of a number of more robust clauses such as N[E,E-1]∃XExperiencer[E-1,X] bald[E-1] N[E,E-1] ∃XW[E-1,X] ∃E(O[E-1,E] ∃XExperiencer[E,X] bald[E]) See note 19 for some considerations bearing on the choice between these variations and those where the objectual quantifier itself quantifies into the adverbial, as in [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]][℩E-1 : N[E,E-1] bald[E-1,x]] … [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]][℩E-1 : N[E,E-1] Experiencer[E-1,x] bald[E-1]] … [℩x : ∃E Lenin[E,x]][℩E : Lenin[E,x]] [℩E-1 : N[E,E-1] W[E-1,x] ∃E(O[E-1,E] ∃XExperiencer[E,X] bald[E])] … 19. What if, in response to (130), one concludes that Lenin dead is not the same thing as Lenin? Then the analysis of (111) according to (i) would lead straightaway to contradiction (see note 18): (i) [℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]] [℩E′ : N[E,E′] dead[E′,x]] [∃E0 : N[E′,E0]]W[E0,x] Past[E0] O[E0,E1]∃YTheme[E1,Y] … Instead, revise (i) to read as in (ii), so that Lenin is referred to only once and the rest of the sentence is about the something else into which he is transformed: (ii) [℩x : ∃ELenin[E,X]][℩E : Lenin[E,x]] [℩E′ : N[E,E′] R[E,E′] ∃Ydead[E′,Y]] [∃E0 : N[E′,E0]] R[E′,E0] ∃YW[E0,Y] Past[E0] O[E0,E1]∃YTheme[E1,Y] … No doubt there is a being dead with a corpse in it that is uniquely related to the Lenin-izing. By hypothesis, the corpse is not identical to Lenin, but it nevertheless suffices for the resemblance. In one way, (ii) is closer to the spirit of what is advocated elsewhere here, with frequent existential closure of objectual positions and relations between phrases carried out via reference to events. (ii) also allows the semantics to skate by a metaphysical dilemma unruffled. According to (ii), Lenin refers sure enough to Lenin, but (ii) escapes the commitment that Lenin dead is the same thing, although it is the latter that sustains the resemblance.
900
Notes
20. For Eventish conditionals, I say in Schein 2003 that the ceteris paribus condition is an overt relation to sentences or propositions Φ. That is, ‘If Φ0; if Φ1, Ψ’ is ‘If Φ0; ceteris paribus, if Φ1, ceteris paribus Ψ’: [℩E0 : ∀e(E0e ↔ Φ0)] [℩E′: N[E0,E′] & ∀e′(E′e′ ↔ ∀Φ(¬∀e(E0e → (True-at(Φ,t(e′)) ↔ True-at(Φ,t(e)))) → (True-at(Φ,t(e′)) ↔ True-at(Φ,t(u)))))] [℩E1 : ∀e(E1e → E′e) & ∀e(E1e ↔ Φ1)] [℩E′: N[E1,E′] & ∀e′(E′e ′ ↔ ∀Φ(¬∀e(E1e → (True-at(Φ,t(e′)) ↔ True-at(Φ,t(e)))) → (True-at(Φ,t(e′)) ↔ True-at(Φ,t(E0)))))] [∀e: E′e] Ψ ‘The events E0 that Φ0 are such that the events neighboring E0 where all things other than ┌ Φ0┐remain as they are in u are such that the events E1 that Φ1 are such that the events neighboring E1 where all things other than ┌Φ1┐ remain as they are in E0 are such that Ψ.’ 21. One may, consistent with my argument, insist that Lenin is a referring expression and that (109)–(111) express singular propositions, conceding only that the name is complex, ‘(℩x)(∃E Lenin[E,x])’, properly containing the predicate ‘Lenin[E,x]’, much as one concedes that the complex demonstrative that Bolshevik contains the predicate Bolshevik. As above, adverbialized logical form copies nominal content into an adverb: (109) Lenin resembled Queen Victoria. λx([℩E : Lenin[E,x]][∃E0 : N[E,E0]] (108))((℩x)(∃E Lenin[E,x])) (110) Lenin bald resembled Queen Victoria. λx([℩E : Lenin[E,x]][℩E′ : N[E,E′] bald[E′]] [∃E0 : N[E′,E0]] (108)) ((℩x)(∃E Lenin[E,x])) (111) Lenin dead resembled Queen Victoria. λx([℩E : Lenin[E,x]][℩E′ : N[E,E′] dead[E′]] [∃E0 : N[E′,E0]] (108)) ((℩x)(∃E Lenin[E,x])) Despite now the formal difference that segregates referring expressions from quantificational nominals, it would have to be said of both that their content restricts an adverbial. 22. The perspective from the past tense is no different. One can say of Franz Kafka (1883– 1924) either (ii) or (iii), as his hometown Prague was Austro-Hungarian at his birth and Czechoslovakian from 1918. On the other hand, Prague was Holy Roman Empire, and Kafka was from Prague, but Kafka was not from the Holy Roman Empire, (i). Again, as with the present tense, even had Kafka lived to see it from a safe distance, accompanying his friend Max Brod to Palestine, one could not say (iv) without him ever having lived under the occupation: (i) (ii) (iii) (iv)
#Kafka was from the Holy Roman Empire. Kafka was from the Austro-Hungarian Empire. Kafka was from Czechoslovakia. #Kafka was from occupied Czechoslovakia.
23. Here Trotsky anticipates Evans 1973. 24. I thank Roger Schwarzschild (p.c., May 1, 2010) for bringing this critical document in the history of linguistics and philosophy of language to my attention.
Notes
901
25. Take a hint from Latin, where a prepositional phrase involves preposition, Case, and DP, and let Case express the thematic relation proper while the preposition itself expresses a relation between events, as in (160). 26. Alternatively, in deference to the unaccusativity of the matrix verb, Theme may be located postverbally, or in both pre- and postverbal position. 27. Without further qualification, ‘from[Ei,Ej]’ will behave badly in sentences with plural arguments. Consider the Bialystok children, Leon, Max, Hannah, and Rosa. The boys were born in Lithuania, and the girls in Latvia. The oldest boy and girl were born in Soviet republics, but not the youngest. Hence, (i) is false. (i) The Bialystoks are from two Soviet republics. Yet, from within their collective extended now, one can find two births in two Soviet republics, which should be prevented from rendering (i) true. The further qualification demands that ‘from[Ei,Ej]’ preserve the participation relation, so that the only initial segments Ej of Ej considered are those in which all the children participate: (ii) From[Ei,Ej] → ∀Y(W[Ei,Y] ↔ W[Ej,Y]) [The X : ∃EBialystok[E,X]][℩E : Bialystok[E,X]][∃E0 : N[E,E0]] W[E0,X][℩E0 : pro0]Present[E0]O[E0,E1] ∃XTheme[E1,X]O[E1,E2] be[E2] [∃ X : 2(X) ∃EUSSR[E,X]][℩E : USSR[E,X]] [∃E0 : N[E,E0]] W[E0,X]O[E0,E3]∃XSource[E3,X]from[E2,E3] 28. If, for some reason, in, in contrast to from, is an intransitive particle, as in ‘In[E] O[E,E′] Location[E′,Y]’ or stands only for a thematic relation, ‘In[E,Y]’, it follows a fortiori that the lifetime of the USSR must include Gorbachev in the present: present events of being fall within the lifetime of Gorbachev, and these same events are then said to be or coincide with some being-in that falls within the lifetime of the USSR, contrary to fact. 29. For illustration, the logical form fixes details that the argument leaves open. The subject is here taken to be nonagentive and therefore a Theme. The direct object is then the Source and the distance between its participation and the Theme’s is provided by an abstract preposition SUB (serving the same purpose as from in (160)), with the suggestion that it is incorporated into the verb as a (pronounced) prefix. 30. An alternative would be to pursue more aggressively an analogy to causativization. If teach is ‘cause-learn’ (see (123)), as it were, then resemble is re-semble, where re- expresses some distal, stative relation like ‘to’ or ‘from’: (i) [℩x : ∃E EMK[E,x]][℩E : EMK[E,x]][∃E0 : N[E,E0]] W[E0,x] [℩E0:pro0] Present[E0] O[E0,E1] ∃XTheme[E1,X] re-[E1, E2] [℩x : ∃E JFK[E,x]][℩E : JFK [E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E2] ∃XGoal[E2,X]-semble[E2] Prior discussion has been explicit about two causal/topological relations between events, ‘C(e,e′’)’, a causative that appears in the decomposition of action verbs, and ‘O(e,e′)’, appearing in the decomposition of stative verbs and everywhere else where exact overlap between events is warranted. Obviously, resemble is not an action verb, excluding its translation as ‘C(e,e′) & resemble(e′)’. Given the gloss on ‘O(e,e′)’, translation as ‘O(e,e′) & resemble(e′)’ does not yet put any distance between the state of the resembler and that of the resembled,
902
Notes
which in (167) becomes the burden of ‘TO[E2,E3]’ treating the direct object as the goal of the resemblance. But I have not considered a systematic typology of verbs, and so perhaps the causal/topological relations that emerge from their decomposition should include along with ‘C’ and ‘O’ a noncausal topological relation that is other than exact overlap. Were this so, ‘resemble[E2]’ in (i) could appear as its complement. All the analysis demands is that some relation deviating from exact overlap intervenes somewhere in the structure between subject and object, with the proviso noted in the text concerning the placement of Tense. In (167), the necessary distance between what Ted Kennedy does and what JFK does emerges entirely from within the analysis of an abstract, embedded prepositional phrase, on analogy to the treatment of from and in, (162) and (163). The choice among grammatical analyses here is further confounded by the prospect of incorporating preposition into verb (implicit in the analysis (164) of succeed; see note 29), which could account for the failure of the abstract preposition postulated in (167) to appear overtly. With incorporation in the mix, it becomes that much more difficult to tell (i) and (167) apart. One consideration of immediate semantic relevance is that the nominalizations the resembling and the resemblance, when anaphoric to a present-tense assertion such as (166), denote a present state, and should therefore abstract on the subject’s participation, E1 in (i), and in (167), E1 or E2. I should think that this observation would favor (167), which in any case maintains the closer association between verb and theme. As a further variation, the conclusion from supermonadicity that there is only one way to participate in an event (and therefore sentences contain descriptions of many events) appears to render thematic relations superfluous, allowing (ii) as an economical alternative to (167): (ii) [℩x : ∃E EMK[E,x]][℩E : EMK[E,x]][∃E0 : N[E,E0]] W[E0,x] [℩E0:pro0] Present[E0] O[E0,E1] O[E1,E2] resemble[E2] [℩x : ∃E JFK[E,x]][℩E : JFK [E,x]][∃E0 : N[E,E0]] W[E0,x] O[E0,E3] TO[E2,E3] Again, I assert no preference here, but see Schein 2002 for a possible argument in support of thematic relations. 31. The interpretation of time-frame adverbials is dependent on their syntactic position in a way that confirms the outline of the analysis, raising at the same time an interesting problem. With the Spanish succession as it is in (152), Prince Fonsy does not become Alfonso XIII until 1902, 17 years after the death of Alfonso XII in 1885. Sentence (i) is an accurate report, and of course (ii) and (iv) are false, as nothing that could be mistaken for a succession occurred in 1885. Sentence (iii) is acceptable only to the extent that Alfonso XII remains a ghostly presence in 1902. The contrast between (i) and (iii) again illustrates the privileged position of the subject in locating the events denoted by Tense, which Kratzer (1995) and Musan (1997) first observed. (i) In 1902, Alfonso XIII succeeded Alfonso XII to the Spanish throne. (ii) *In 1885, Alfonso XIII succeeded Alfonso XII to the Spanish throne. (iii) ???In 1902, Alfonso XII was succeeded by Alfonso XIII. (iv) *In 1885, Alfonso XII was succeeded by Alfonso XIII. In the current setting, what these sentences show is that a frame adverbial on the left periphery is unsurprisingly subject-oriented, modifying what events the subject participated in. The
Notes
903
same adverbial relocated elsewhere either becomes object-oriented, as in (v) and (vi), where in 1902 applies accurately to Alfonso XIII’s participation, or remains subject-oriented as in (vii) and (viii) with its implication of a ghostly presence. (v) (vi)
(In 1885,) Alfonso XII died, and was succeeded by Alfonso XIII in 1902. (In 1885,) Alfonso XII died, and was succeeded in 1902 by Alfonso XIII.
(vii) ???(In 1885,) Alfonso XII died, and in 1902 was succeeded by Alfonso XIII. (viii) ???(In 1885,) Alfonso XII died, and was, in 1902, succeeded by Alfonso XIII. According to (ix), whatever Alfonso XII-ing there was in the past, and there was such, relates by the meaning of the participle succeeded or by SUB-Source (see (164)) to some future event in 1902. In contrast, the Alfonso XII-ing in (iii) is itself located in 1902. (ix) Alfonso XII was succeeded in 1902 by Alfonso XIII. But given license to modify different events of participation in the same sentence, what now accounts for the unacceptability of (iv) and (x)? (x) *In 1885, Alfonso XII was succeeded in 1902 by Alfonso XIII. Alfonso XII Alfonso XII-ing in 1885 is after all his last occasion that can be related to the future succession in 1902. Yet neither a time adverbial, as in (x), narrowing in on a crucial milestone in Alfonso XII’s lifetime, nor any other that frames part of his lifetime to the exclusion of his successor’s is acceptable ((xi) vs. (xii)): (xi) *Sometime in the 19th century, Alfonso XII was succeeded (in 1902) by Alfonso XIII. (xii) Within some period spanning the 19th and 20th centuries, Alfonso XII was succeeded (—in 1902, as a matter of fact—) by Alfonso XIII. A related worry concerns the contrast between (i) and (xiii)–(xvi) among the active sentences where the Theme is the subject. Here no adverbial may be introduced to locate the object’s participation even under appropriate conditions ((xii) vs. (xiv)/(xvi)): (xiii) *In 1902, Alfonso XIII succeeded in 1885 Alfonso XII to the Spanish throne. *In 1902, Alfonso XIII succeeded Alfonso XII in 1885 to the Spanish throne. *In 1902, Alfonso XIII succeeded Alfonso XII to the Spanish throne in 1885. (xiv) *Within some period spanning the 19th and 20th centuries, Alfonso XIII succeeded in 1885 Alfonso XII to the Spanish throne. *Within some period spanning the 19th and 20th centuries, Alfonso XIII succeeded Alfonso XII in 1885 to the Spanish throne. *Within some period spanning the 19th and 20th centuries, Alfonso XIII succeeded Alfonso XII to the Spanish throne in 1885. (xv) *By 1902, Alfonso XIII had succeeded in 1885 Alfonso XII to the Spanish throne. *By 1902, Alfonso XIII had succeeded Alfonso XII in 1885 to the Spanish throne. *By 1902, Alfonso XIII had succeeded Alfonso XII to the Spanish throne in 1885. (xvi) *Within some period spanning the 19th and 20th centuries, Alfonso XIII had succeeded in 1885 Alfonso XII to the Spanish throne. *Within some period spanning the 19th and 20th centuries, Alfonso XIII had succeeded Alfonso XII in 1885 to the Spanish throne. *Within some period spanning the 19th and 20th centuries, Alfonso XIII had succeeded Alfonso XII to the Spanish throne in 1885.
904
Notes
The observed restrictions on time adverbials fall under the following generalization. The time adverbial in all the sentences considered locates the events of the Theme’s participation. What varies with its syntactic position is whether it also locates the subject’s participation. Perhaps, as a point of grammar, the only underlying position for a time adverbial is one from which it modifies Theme and as a result of subsequent movement to higher position, it also comes to restrict the thematic relations that fall within its scope. These restrictions on time adverbials recall an old chestnut, the time of the killing (for recent discussion, see Pietroski 1998, 2000, and the references cited there): (xvii) #Booth killed Lincoln on April 14, 1865, but Lincoln did not die until April 15, 1865. (xviii) Booth shot Lincoln on April 14, 1865, but Lincoln did not die until April 15, 1865. (xix) a. #Booth killed Lincoln on April 15, 1865 (having shot him on April 14, 1865). b. #Before the Confederacy breathed its last gasp, Booth had killed Lincoln on April 15, 1865. As above, the time adverbial circumscribes the Theme’s participation in whatever goings-on are being reported. In (xvii), Lincoln’s participation in getting killed includes a change of state falling outside the circumscribed period. In contrast, (xviii) describes an event where all of what befalls Lincoln within that event falls on April 14. The false time adverbial in (xix) can also fall under this generalization under a precise meaning of ‘Agent(e,b) & C(e,e′) & kill(e′) & Theme(e′,l)’. It should not be enough for the truth of this expression that Booth’s action, e, the gunshot on April 14, caused at an arbitrary distance an effect, e′ (one among many), Lincoln’s death on April 15. Rather, the effect on Lincoln that is to be related to Booth’s action should take in the full measure of what happened to him, which begins when the bullet, the instrument of Booth’s action, enters Lincoln’s body. Lincoln’s participation from bodily injury to death does not in fact occur on April 15, falsifying (xix). Consider now murder by different means. On the morning of April 14, Booth poisons Lincoln’s stash of brandy. The fatal draught lies in wait in the liquor cabinet until the evening of April 15. That evening, Lincoln pours out a nightcap and succumbs shortly thereafter. To my ear, this scenario rather improves (xx), as it entirely confines Lincoln’s part in his killing to April 15, and it brings out a definite contrast with the time adverbial at the left periphery in (xxi): (xx)
a. Booth killed Lincoln on April 15, 1865. b. Before the Confederacy breathed its last gasp, Booth had killed Lincoln on April 15, 1865. (xxi) #On April 15, 1865, Booth killed Lincoln. (xxii) Within two days, Booth killed Lincoln. Pietroski’s (1998, 2000) analysis and argument complement mine, putting as he does distance between actions and effects. The former ground events that terminate in the latter: (xxiii) Nora melted the chocolate. (xxiv) ∃e{∃a[grounds(a,e) & action(a, Nora)] & ∃f[terminates-in(e,f) & melt(f) & Patient(f, the chocolate)]} The distances become considerable when the actions are tryings located in the head, as Pietroski persuasively argues. His argument, concerned with the individuation and identification of actions, ignores lifetime effects, and yet it happily converges on supermonadic logical forms, providing them independent support (for which I am grateful).
Notes
905
Perhaps to shoehorn the analysis as much as possible into citation Davidsonian logical form, the quantification over actions in (xxiv) is given limited scope, and the possibility is entertained that it need not appear there at all if, as in (xxv), lexical meaning spells out Nora’s relation to the superordinate e in terms of her action a. (xxv) a. ∃e{Agent(e,Nora) & ∃f[terminates-in(e,f) & melt(f) & Patient(f, the chocolate)]} b. Agent(e,x) ↔df ∃a(grounds(a,e) & action(a,x)) Throughout I have departed from (xxiv), replacing it, as it were, with (xxvi) making all the subevents available to cross-reference: (xxvi) ∃e∃a∃f [action(a, Nora) & grounds(a,e) & terminates-in(e,f) & melt(f) & Patient(f, the chocolate)] For further discussion see section 2.6.0.0 and Schein 2002, appendix. 32. See note 18. 33. Note in particular the existential event quantification in line (iv). I must assume here that no gloss on ‘to[E2,E3]’ will say, for example, that the route that a resemblance takes from Lenin now E1 to Queen Victoria then E2 implies that Queen Victoria finds herself in no other pose between then and now that does not uphold the resemblance. 34. Strictly speaking, no Lenin-izing does not entail that Lenin is not alive if it can be imagined how Lenin alive could nevertheless fail to Lenin-ize. I assume that speakers, for whatever reason, reliably fail to grasp this possibility and so conclude that Lenin’s death is the intended respect in which there is Lenin-izing no longer. 35. For expository convenience, I have contrasted how things are with the subject and how things are with objects. As should be clear from the discussion of adverbial modification, what is implicated to be no longer is rather how things were as described by the phrase preceding Tense. Musan (1997) observes that whereas the individual-level attribution in (i) implicates that Aunt Theresa is dead, sentence (ii) topicalizing the dative object implicates that Aunt Theresa or Grandmother is dead: (i) Tante Theresa ähnelte meiner Groβmutter. Aunt Theresa.nom resembled my grandmother.dat ‘Aunt Theresa resembled my grandmother.’ (ii) Meiner Groβmutter ähnelte Tante Theresa. my grandmother.dat resembled Aunt Theresa.nom ‘My grandmother, Aunt Theresa resembled.’ This follows on the present account provided that the verb raising is undone and reconstructed into its lower position in logical form, thus locating Tense within the scope of both the subject and dative object. 36. Alternatives to (179) are welcome provided that none implies that E0 occur in the present. So, for example, if one were to relinquish the spatiotemporal coincidence of E0 and E1 (‘O[E0,E1]’ in (179)), one could be explicit about present causation (and a chain reaction): (i) a. [℩x : ∃E Supernova[E,x]][℩E : Supernova[E,x]][∃E0 : N[E,E0]] W[E0,x] C[E0,E1] ∃Y Agent[E1,Y] Present[E1] Cause[E1,E2] bombard[E2] … b. [℩x : ∃E Supernova[E,x]][℩E : Supernova[E,x]][∃E0 : N[E,E0]] W[E0,x] C[E0,E1] ∃Y Agent[E1,Y] Cause[E1,E2] Present[E2] bombard[E2] …
906
Notes
The logical form (ia) asserts present cause and leaves present effect as inference from proximate causation, and (ib) asserts present effect, leaving present cause to be inferred from proximate causation. In (i), ‘C’ would have to be a relation of distal causation. The logical form (179) might have stood as is if ‘Cause’ is there understood to be distal causation and there is an inference from distal causation of present effect to present proximate causation, without explicit reference in (179) to the events that are the proximate cause. But insertion in (179) of an adverb “focused just now on a detection array” qualifies the present tense and thus demands explicit events in the present in which the supernova participates focused: (ii) Supernova GRB 090423 at the most distant edge of the universe 13 billion years ago is focused just now on a detection array, bombarding the Swift spaced-based observatory with gamma radiation. 37. The argument against nominal reference to temporal slices or aspects was that it falls short of an account of lifetime effects and the interaction with Tense and adverbial modification and that once such an account is provided, nominal reference to temporal slices or aspects becomes superfluous (and undesirable if semantic innocence is a concern). The appeal of temporal slices, aspects, or guises is otherwise obvious—appearances matter, as any reliance on perspectival notions merely confirms. The discussion in the text, however, suggests a possible further objection, which I will mention without working it out in detail. A temporal slice has exactly the same spatial contours as the object it is a slice of. It extends no further in space and deviates from the object only by coming up short in the temporal dimension. Similarly, one thinks of a guise or aspect as determined and circumscribed by the object from the surface inward. Yet, if one has already engaged scenes to distinguish (43) implicatures in different contexts, it may supplant direct reference to slices, guises, aspects, and the like. Why not always refer instead to the object itself in a scene? Figure and Ground are relational properties defined over the entire perceptual field and unless one has fallen into a cubist nightmare, principles of perspective must in any case govern when different aspects or guises can belong to the same coherent scene, if one has troubled to introduce them. Consider now the force of this last context for (43). If the proffered implicature is that the subject’s aspect changes, then in this context the subject’s aspect reaches beyond the prelude to include the étude. There is no useful notion of slice, guise, or aspect that does not dissolve into the larger scene. 38. As a suggestion for further research, I believe the tools are in place—perspectival thematic relations and supermonadicity, affording reference to participants isolated in their own (perspectival) subevents—for a fully semantic treatment of Goodman’s (1947) contrast: (i) If I were Julius Caesar, I wouldn’t be alive in the 20th century. (ii) If Julius Caesar were I, he would be alive in the 20th century. Example (i) comes close to translation as “If I were to go to Julius Caesar, …” and (ii) as “If Julius were to come to me, …” 39. See Stump 1985 on the semantic variability of the construction.
Notes
907
40. These Aristotelian examples are discussed in Szabó 2003. There are contrasts that reflect, I believe, the choice of underlying event operator: (i) ?An isosceles triangle, this polygon has angles that add up to two right angles. (ii) ?Being an isosceles triangle, that polygon has angles that add up to two right angles. (iii) ?If an isosceles triangle, this polygon has angles that add up to two right angles. (iv)
F By being an isosceles triangle, this polygon has angles that add up to two right angles. (v) F As an isosceles triangle, this polygon has angles that add up to two right angles. (vi) F Qua isosceles triangle, this polygon has angles that add up to two right angles. (vii) F In virtue of being an isosceles triangle, this polygon has angles that add up to two right angles. (viii) F Because of its being an isosceles triangle, this polygon has angles that add up to two right angles. The Aristotelian judgment that (iv)–(viii) are false seems muted in (i)–(iii), which would follow if the event operators in (iv)–(viii) but not those in (i)–(iii) were nonincreasing with respect to their second term. (Cf. ?Every isosceles triangle has angles that add up to two right angles vs. F Only an isosceles triangle has angles that add up to two right angles.) Note that nonincreasing and only nonincreasing adverbial quantification induces subjectauxiliary inversion (Liberman 1974): (ix) (x)
*(Only) an isosceles triangle does this polygon have angles that add up to two right angles. *(Only) being an isosceles triangle does this polygon have angles that add up to two right angles.
(xi)
(Only) by being an isosceles triangle does this polygon have angles that add up to two right angles. (xii) (Only) as an isosceles triangle does this polygon have angles that add up to two right angles. (xiii) (Only) qua isosceles triangle does this polygon have angles that add up to two right angles. (xiv) (Only) in virtue of being an isosceles triangle does this polygon have angles that add up to two right angles. (xv) (Only) because of its being an isosceles triangle does this polygon have angles that add up to two right angles. (xvi) *If an isosceles triangle does this polygon has angles that add up to two right angles. (xvii) Only if an isosceles triangle does this polygon have angles that add up to two right angles. Apparently absolutives, (i), (ii), (ix), (x), must be increasing, and the prepositional/ complementizer phrases, (iv)–(viii), (xi)–(xv), must not be. The null operator associated with bare tokens of if-clauses forms an increasing quantifier, (iii), (xvi), but if-clauses are themselves available to restrict other operators, (xvii), as the literature on unselective binding since Lewis 1975 has assumed.
908
Notes
41. As the puzzle is known from a recent exchange. Carlson (1982, 174ff.) recognized it earlier in (i)–(iv), and anticipating Moore’s (1999) position, concluded that “Reggie Jackson” refers to j, “the third batter” to a, and “the twelfth batter” to b and that j ≠ a ≠ b: (i) (ii) (iii) (iv)
Reggie Jackson was the third batter. Reggie Jackson was the twelfth batter. Bob Welch struck out the third batter. Bob Welch did not strike out the twelfth batter.
The puzzle survives explicit reference to past events in the following contrast: (iii′) Bob Welch struck out at that time, the third batter up, the third batter. (iv′) Bob Welch did not strike out at that (later) time, the twelfth batter up, the twelfth batter. (v) (vi)
Bob Welch struck out at that time, the third batter up, Reggie Jackson. Bob Welch did not strike out at that (later) time, the twelfth batter up, Reggie Jackson.
(vii) #Bob Welch struck out at that time, the third batter up, the twelfth batter. (viii) #Bob Welch did not strike out at that (later) time, the twelfth batter up, the third batter. 42. Moore’s (1999) laconic (17) raises the puzzle without the distraction of supernumeraries with attitudes of their own: (17) Yesterday, Phosphorus shone brightly, but clouds obscured Hesperus. And Barber (2000) notes the contextual dependence of such usage. Should it be discovered that Venus is 10 degrees hotter when it is dusk in North America than when it is dawn there, accurate reports of the discovery would be (i) addressed to a North American conference of astronomers and (ii) addressed to a conference in New Zealand. (i) Hesperus is hotter than Phosphorus. (ii) Phosphorus is hotter than Hesperus. 43. Barber 2000; Forbes 1997, 1999; Moore 1999; Saul 1997a, 1997b, 1999, 2007; Szabó 2003; Zimmermann 2005. An exception might be (5) from Forbes 1999: (5) Clark went to the fancy dress party as Superman. If monoclausal and descriptive of a single event, (5) should not be glossed as ‘Clark, as Clark, went to the fancy dress party as Superman’. The exponent of this view is then forced to treat the adverbialization ‘as Clark’ as merely optional and suppressed if sense requires it as in (5), or alternatively, to deny that as-phrases are synonymous with the covert adverbial co-occurring with nominals. The second sentence in Moore’s (1999) (11) is an especially robust instance of the problem arising from substitution into the same clause: (11) Superman is more successful with women than Clark Kent. But they both strike out when it comes to men. If a plural pronoun implies reference to more than one thing, then the pronoun in (11) must refer to two aspects rather than two males, as Moore contends.
Notes
909
44. As indicated by tensed verbs or (comparative) complementizers: … than Clark Kent (does). 45. I do not treat Superman as an agent in a contest, since the more natural reading here is that the speaker alone draws the comparison. Of course, the sentence could also be glossed as agentive and causative. 46. Barber 2000; Forbes 1997, 1999; Moore 1999; Predelli 1999; Saul 1997a, 1997b, 1999, 2007; Zimmermann 2005. 47. As remarked above, it does no good to talk about the same one event (in (190) or (196)) as being both Clark Kent-ish and Superman-ish. So, if it were nevertheless stipulated, contrary to supermonadicity, that there is just the one event, the resembling or the outscoring, it would fall to the nominals—no place else to go—to distinguish Clark Kent-izing and Supermanizing. Truth for these sentences requires an analog of reference to temporal slices. Moore’s (1999) solution is that the names, Clark Kent and Superman, themselves refer directly to aspects, or modes of presentation, in effect to the Clark Kent-ing and the Superman-izing, respectively. The fine-grained reference to aspects may purchase the truth of (190) and (196), but it gains nothing on the implicature that how things were with the subject is no longer, which persists into these sentences. In (i)–(viii), it is only the subject’s aspect that changes when the glasses are replaced: (i) (ii)
Clark Kent didn’t resemble Superman until the glasses came off. #Clark Kent didn’t resemble Superman until the glasses went on.
(iii) #Superman didn’t resemble Clark Kent until the glasses came off. (iv) Superman didn’t resemble Clark Kent until the glasses went on. (v) (vi)
Superman outscored Clark Kent when(ever) the glasses came off. #Superman outscored Clark Kent when(ever) the glasses went on.
(vii) #Clark Kent was outscored by Superman when(ever) the glasses came off. (viii) Clark Kent was outscored by Superman when(ever) the glasses went on. (ix) (x)
Superman does not resemble Clark Kent. Superman bespectacled resembled Clark Kent.
Deriving the implicature that privileges the subject revisits Kratzer’s (1995) dilemma. If there is only the one event, the resembling or the outscoring, then Tense becomes a temporal slicer of aspects, and Tense and adverbial phrases with it are shunted into the adnominal. Yet it remains open to Moore to adopt supermonadicity while the names continue to refer to aspects, in which case the logical form for (ix) becomes something along the lines of (xi), which should be compared to the adverbialized logical form for (ix) in (xii): (xi) [℩E : ∃xSuperman[E,x]] ¬ ∃E0 W[E0,E] Present[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2] [℩E : ∃xClark Kent[E,x]] ∃E0 W[E0,E] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3] ‘The Superman-izing does not participate as Theme in some resembling (i–ii) such that the Clark Kent-ing participates as Goal in some events (iii–iv), and the resembling is to the Clark Kent-ing’s events (iv).’
(i) (ii) (iii) (iv)
910
Notes
(xii) [℩x : ∃ESuperman[E,x]][℩E : Superman[E,x]]¬[∃E0 : N[E,E0]] W[E0,x] Present[E0]O[E0,E1]∃YTheme[E1,Y]O[E1,E2]resemble[E2] [℩x : ∃EClark Kent[E,x]][℩E : Clark Kent[E,x]] [∃E0 : N[E,E0]] W[E0,x] O[E0,E3]∃YGoal[E3,Y] TO[E2,E3]
(i) (ii) (iii) (iv)
‘Superman (in the neighborhood of) Superman-izing does not participate as Theme in some resembling (i–ii) such that Clark Kent (in the neighborhood of) Clark Kent-ing participates as Goal in some events (iii–iv), and the resembling is to Clark Kent’s events (iv).’ In both (xi) and (xii), it is reference to Superman-izing and Clark Kent-ing that rescues (ix) from contradiction. There is however a difference of translation. In (xi), the Superman-izing and Clark Kent-ing come from the names and in (xii) from tacit adverbial phrases. But if, on the one hand, such reference can come only from names (and definite descriptions), the translation will fail the generalized quantifiers in (208) and (209). On the other hand, if tacit adverbial phrases are admitted for the sake of (208) and (209), it becomes otiose for the names in (ix) (or definite descriptions in general) to deviate from their ordinary reference to objects. Moore (1999) defends ambiguous reference to objects and to aspects, contrasting two contexts in which (196), for example, might be uttered. Where all parties to the discourse are “enlightened” in the knowledge that Clark Kent is Superman, as the above discussion has assumed throughout, the utterance is judged true, and reference to aspects, one way or another, is necessary to secure it. (196) Superman outscored Clark Kent. In contrast, in “unenlightened” contexts, where the speaker of (196) mistakenly believes that Clark Kent is not Superman, her utterance is judged false, which would follow if she is taken to use the names to refer to the objects themselves and not to their aspects. Taking the altered truth conditions at face value, there would indeed have to be some ambiguity in (196). Yet, in adverbialized logical form, nominal expressions are themselves semantically innocent—that is, invariant in the object referred to—and the adverbial phrases constructed from their nominal content are always present. The semantic variability that alters truth conditions must reside elsewhere, for which I can think of two candidates. The first, which section 10.3 (see note 67) will reject on other grounds, locates an ambiguity in the manner the adverbialized phrases modify, in the choice of neighborhood relations, ⌜N[Ei,Ej]⌝. Let a phase-invariant neighborhood, Nnoφ, be one indifferent to the events that an object is described as participating in: (xiii) (Φ(e0,x) & Ψ(e1,x)) → (Nnoφ(e0,e) ↔ Nnoφ(e1,e)) That is, any event e is in the neighborhood of x’s Φ-ing just in case it is also in the neighborhood of x’s Ψ-ing. As neighborhoods are also relied on to distinguish the various conditional modalities, the phase-invariant neighborhoods presumably correspond to a class of relations. Now, if one is enlightened in knowing that Superman is Clark Kent, one will not combine the two names with phase-invariant modification, since doing so would render the choice of multiple names to refer to the same object entirely vacuous and gratuitous (see the constraint against “elegant variation,” section 10.3.1). If one is enlightened, one can felicitously utter (196) only when intending phase-dependent modification so that there is a point to the varia-
Notes
911
tion in referring to the same object. So much seems to follow from cooperative principles of conversation. In another context, however, one might intend phase-invariant modification, in which case (196) has said something (necessarily) false, exposing its latent ambiguity. Moore (1999) appears to suggest that in an unenlightened context, the speaker of (196) can only be taken to have uttered something false. That is, if unenlightened, a speaker defaults into intending phase-invariant remarks. If true, this looks to be more substantive than what can be extracted from conversational principles alone. Even if she believes on the basis of the evidence available to her that Clark Kent is not Superman, it is unclear to me why a cautious and compulsive witness would not use (196) to report the narrow facts of the scene before her—in Moore’s terms—that the Superman aspect outscored the Clark Kent aspect, which remains true whether or not she is later enlightened. No matter—whatever story is told to guide a speaker between reference to object and reference to aspect can be retold as a story choosing between phase-invariant and phase-dependent neighborhoods. But section 10.3 will find reason to reject phase-invariant neighborhoods altogether. More promising, the difference in truth in what the enlightened and unenlightened say may be located in the name itself, in the recognition that the name in the speaker’s private language may differ from our own, given our knowledge of the speaker’s unenlightened condition. Elsewhere in the vocabulary, if “That, my son, is a colt” is a first introduction for a city kid who does not know from a horse, we can with some confidence anticipate that he will mistake the extension of colt to include horses other than males under 4 years old, which explicit definition will have to correct. On the other hand, it is unlikely that “that, my child, is a caterpillar” will result in the overextension of caterpillar to adult butterflies. Crucially, a metamorphosis separates the juveniles from the adults. Similarly, “This is Clark Kent” tends not to prompt acquisition of a name for the day-job phases alone or any phases short of all those constituting the life: (Φ[e,x] & ∃e Clark Kent(e,x)) → Clark Kent(e,x). An explicit narrative is required to “enlighten” the learner otherwise. Absent “enlightenment,” we presume that the speaker has acquired the names Clark Kent and Superman, each of which denotes all phases of a someone. But then what this speaker intends by (196) is indeed literally false in her language, or so we presume given her ignorance of the facts and our knowledge of verbal behavior. If one retreats further, stepping away from Moore’s claim that the unenlightened speech is literally false, one can do without any ambiguity in either the neighborhood relation or the name if the correction the unenlightened speaker inevitably invites indicates a defect in her communicative intent other than the literal falsehood of her utterance. 48. More examples: (1) The planet we observed at the star party in January was closer to us than the planet we observed from the backyard in July. (2) Any planet we observed at the star party in January was closer to us than any planet we observed from the backyard in July. (3) Venus in January was closer than Venus in July. (4) #Venus was closer than Venus (was). (5) A youth who is in love is more alive than a youth who has love lost. (6) The teenager who was in love was a better dinner guest than the teenager who was heartbroken. (7) Romeo in love is more alive than Romeo heartbroken. (8) #Romeo was more alive than Romeo.
912
Notes
49. Assuming again an invariance of logical form throughout the paradigm (213)–(216), I owe an explanation for why (216) has no reading equivalent to (i): (i) Walden Pond some time in the past was colder than Walden Pond some (other) time in the past. Absent overt adverbs indicating the contrary, the comparative clause and the matrix clause are understood to be about the same time, which rests on details that I have not supplied about the logical syntax of tense, tense anaphora, and temporal modification. For related discussion, see section 10.3.1. 50. On reduced comparative clauses, see Hackl 2001a. 51. A point emphasized in Zimmermann 2005. 52. Zimmermann (2005) glosses (196) so that it expresses a truth only as a counterfactual speech act: If I spoke the language of someone who did not know that Clark Kent was Superman, I would have said (196). As conceded in the paper’s last paragraphs, the gloss misfires on (191), since the identity is exactly what someone who does not know it would not say. I have not fully understood the rationalization that follows for why a false counterfactual might be useful anyway, but it does not look as if it will offer any relief from the related objection to the negation of (191), which is not discussed. Clark Kent is not Superman is unequivocally and unambiguously false, in contrast to the counterfactual it allegedly may convey—If I spoke the language of someone who did not know otherwise, I would say that Clark Kent is not Superman. Note also that counterfactual speech acts are too blunt an instrument to discern the contrast below between (226) and (227). A speaker who does not know that Clark Kent is Superman (and presumes otherwise according to Zimmermann) has no grounds to recoil from the conclusion in (227). Yet the second and third premises of (226) escape self-contradiction and the reasoning comes out as valid only by appeal to what such a speaker would say. The account leaves no room for the actual speaker’s knowledge that Clark Kent is Superman and her theory of happiness and mental constitution to play any role in telling apart (226) and (227). 53. Parsons (2000, 92f.) remarks that if boy and man in (i) and all nouns are phase sortals, the inference in (ii) is correctly invalidated. If so, (iii) must be an enthymeme resting on unstated, default assumptions about the coincidence of states of being female and states of being a driver. (i)
Here you can see him a boy, and there you can see him a man.
(ii) She is a driver under the age of 25. She is a drunk. Therefore, she is a drunk driver under the age of 25. (iii) She is a driver under the age of 25. She is a female. Therefore, she is a female driver under the age of 25. 54. On this alternative, happiness itself of course remains dispositional rather than episodic. Even though a given sentence is always an attribution to an object at a certain phase, lexical meaning implies its true attribution at other phases: (Φ[e0,x] & N(e0,e) & happy[e,t]) → ((Ψ[e1,x] & Memory[e0,x,t] & Memory[e1,x,t]) → ∃e(N(e1,e) & happy[e,t]))
Notes
913
One could also go on to make the implicitly modal character of a dispositional attribution an explicit feature of its decomposition in logical form, so that Disp happy is true at a phase just in case happy is true at all nearby phases. 55. Performed live at the Semantics Workshop, Rutgers University Center for Cognitive Science, May 7, 2005, with Jeff King as the mild-mannered reporter and the Man of Steel. 56. In related discussion Moore (1999) measures the truth of (1) against a context where Clark Kent goes into the phone booth and comes out in purple swim trunks. (1) Clark Kent went into the phone booth, and Superman came out. 57. Compare the effect of overt adverbs in nondemonstrative DPs: (1) The liar fails a polygraph. Any liar fails a polygraph. Every liar fails a polygraph.
(2) The sometime liar fails a polygraph. Any sometime liar fails a polygraph. Every sometime liar fails a polygraph.
(3) The sinner regrets it. Any sinner regrets it. Every sinner regrets it.
(4) The sometime sinner regrets it. Any sometime sinner regrets it. Every sometime sinner regrets it.
As expected from adverbialized logical form, the polygraph failure (1) is an effect of lying, and any lie that escapes detection is a counterexample to what is asserted. For (2), it would suffice that sometimes lying makes one anxious enough to fail even when telling the truth. Similarly, (3) implies that every sin is regretted, while in (4) what may be regretted is not sinning enough. Faulting the examples in (2) as is, Katy McKinney-Bock, Saurov Syed, and Barbara Tomaszewicz (p.c., April 7, 2011) remark that the contrast with (1) is compromised if the sentences in (2) are understood to conceal a descriptive anaphor in the matrix referring back to the time of the lie—“Every liar at some time then fails a polygraph.” A more robust variant of the contrast opposes (1′) and (2′): (1′) Every liar always fails a polygraph.
(2′) Every sometime liar always fails a polygraph.
58. In Star Trek, the transporter beam that disassembles the traveler’s mass in one location and reassembles it elsewhere can also be used to modify reassembly. So suppose it disassembles Jules Bagwell and crashes for as long as you like before reassembly as Julia Bagwell. 59. A similar example concerns a constitution relation, the processing of grain into something nearly indigestible: (i) The matzoh meal that is in the silo was Meyer’s winter harvest. (ii) #Meyer’s winter harvest was the matzoh meal that is in the silo. 60. Independent reference to some past events—and not necessarily to specific or definite events—suffices to escape the lifetime effect of (251) (cf. note 5): (i) Once upon a time no one knows when, at a masked ball unknown to you and me and anyone else but him, Jules Bagwell was (already) Julia Bagwell.
914
Notes
61. The contrast holds equally of the historical present. Imagine Walter Cronkite narrating an episode of You Are There (CBS (1953–1957); see chapter 9, note 29) on the life of Bagwell: (i) T … Halloween 2005, Jules Bagwell is (now) Julia Bagwell … (ii) F … Halloween 2005, Julia Bagwell is Jules Bagwell … (iii) F … Carnaval 1995, Jules Bagwell is Julia Bagwell … (iv) T … Carnaval 1995, Julia Bagwell is Jules Bagwell … 62. In contrast, I find (i), without the explicit shift that now affords, infelicitous in implying present phases of Jules Bagwell. See discussion below. (i) #Jules Bagwell resembles Juliana Bagwell. 63. Recall that the structure for adverbial modification must for the general case interpolate some such relation between adverbial clause and matrix clause, ⌜[Q: Φ[E]] [∃E0 : N[E,E0]]Ψ[E0]⌝, if it is not always to imply that the Φ-ings are the very same events as the Ψ-ings, *⌜[Q: Φ[E]]Ψ[E]⌝. (For example, no analysis of If the parasite lives, the host dies should entail that the parasite’s living is an event identical to the host’s dying. If so, and if the adverbial clause is an event quantifier, as assumed, an accessibility relation must intervene. See Schein 2003 for further discussion.) By hypothesis, every DP adverbializes, populating (254) with the tokens of ⌜N[Ei,Ej]⌝ shown. 64. The translation of to Geneva (lines (iii)–(iv) in (2)) requires some device referring back to the going, a descriptive pronoun ‘[℩E2 : pro2]’ (shown arbitrarily in the first position of (iii)). (1) Gorbachev went to Geneva. (Cf. (161)) (2) [℩x : ∃EGorbachev[E,x]][℩E : Gorbachev[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0 : pro0]Past[E0]O[E0,E1] ∃XTheme[E1,X]O[E1,E2] go[E2] [℩E2 : pro2][℩x : ∃E Geneva[E,x]][℩E : Geneva[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E4]∃XGoal[E4,X]to[E2,E4]
(i) (ii) (iii) (iv)
Nothing has been said about the logical form of the preposed PP in (3): (3) To Geneva, Gorbachev went. Perhaps it is just the prepositioning of (iii)–(iv) and construal of ‘[℩E2 : pro2]’ as a backward pronoun, perhaps its construal requires a reconstruction to undo the movement in (3), or perhaps the preposed phrase forms an event quantifier, some events (that are) to Geneva, quantifying-in: (4) [∃Ei: [℩x : ∃E Geneva[E,x]][℩E : Geneva[E,x]][∃E0 : N[E,E0]] W[E0,x]O[E0,E4]∃XGoal[E4,X]to[Ei,E4]] [℩x : ∃EGorbachev[E,x]][℩E : Gorbachev[E,x]][∃E0 : N[E,E0]] W[E0,x][℩E0 : pro0]Past[E0]O[E0,E1] ∃XTheme[E1,X]O[E1,E2] go[E2] O[E2,Ei] Further alternatives involving a partial reconstruction can be imagined as well. In translating the temporal-frame adverbials, the text makes an arbitrary choice to quantify in, analogous to (4).
Notes
915
65. Recall the minimal scope convention for operators in parsing formulas without parentheses. Omitting all parentheses in ⌜[∃ν: Δ][∃υ: Γ]ΦΨ⌝ stands for ⌜([∃ν: Δ][∃υ: Γ]Φ)Ψ⌝. Otherwise, ⌜[∃ν: Δ]([∃υ: Γ]ΦΨ)⌝ stands for ⌜[∃ν: Δ](([∃υ: Γ]Φ)Ψ)⌝, and ⌜[∃ν: Δ][∃υ: Γ] (ΦΨ)⌝ is as is. 66. The formulation is deliberately coy since the spatiotemporal relation imposed when tense is not contrastive does not require overlap, let alone coincidence: (i) Finding a shilling outside the Stock Exchange, Scrooge gives it to charity at St. Paul’s. If Scrooge finds a shilling outside the Stock Exchange, he gives it to charity at St. Paul’s. St. Paul’s is a half mile and 10 minutes beyond the Stock Exchange, and of course the giving could only be a future consequence of the finding, and yet a certain proximity imposed by the lack of contrast in tense distinguishes (iii) from (ii): (ii) If Scrooge finds a shilling, he will(, being repentant,) give it to charity on his deathbed. If Scrooge finds a shilling, he will(, being a scrooge no more,) give it to charity on his deathbed. If Scrooge finds a shilling, he will(, being different,) give it to charity on his deathbed. (iii) #If Scrooge finds a shilling, he(, being repentant,) gives it to charity on his deathbed. #If Scrooge finds a shilling, he(, being a scrooge no more,) gives it to charity on his deathbed. #If Scrooge finds a shilling, he(, being different,) gives it to charity on his deathbed. Scrooge’s disposition on his deathbed of a lifetime’s treasure in found shillings, verifying (ii), is an event too remote for (iii), which seem to imply that Scrooge dies anew with every shilling found. Thanks to Elena Herburger for discussion. 67. Thus the grounds to reject the phase-invariant neighborhood relation discussed in note 47. 68. Recall Brentano’s distinction between thetic and categorical judgments, and its applications in the linguistics literature (von Fintel 1989; Kuroda 1972, 1992; Ladusaw 1994; McNally 1998; Sasse 1987). Across languages, it seems that sentences expressing categorical judgments dislocate a constituent to a higher, more peripheral position and the categorical judgment is judgment about whatever the dislocated constituent refers to. In verifying a categorical judgment about x, does the knowledge that it purports to be about x under mode of presentation E1 structure the search for confirming events E2 in ways spelled out by substantive conditions ? Note that (272) implies for syntactic structure an expanded left periphery outside TP, in that Tense applies to events described by the complement to . I suppose there are alternatives to (272) that conceal the effects of , the temporal shift, under more discreet operators if one is at war with abstract syntax. I also wonder whether seats the perspectival relations of Figure and Ground, earlier observed to induce asymmetries despite the embedded symmetric relations (see sections 10.0.0 and 2.5.1). Note that perspective constrains apparent identity statements, uses of the copula, even those relating events. Despite the spatiotemporal coincidence of the events, if Kate in fact sang words of praise, it should be reported as in (i), and if instead she dripped
916
Notes
poison, it should be reported as in (ii), according to the description demonstrative for the speaker in fixing the events’ position. (i) Kate’s praising Harry was her scorning Louis. (ii) Kate’s scorning Louis was her praising Harry. Similarly, the ordinary observer of the night sky says (iii) rather than (iiv): (iii) Last night, the northern sky glowing was the atmosphere ionizing. (iv) #Last night, the atmosphere ionizing was the northern sky glowing. Note also that the asymmetry in (v) and (vi) need not undermine the metaphysical identity of the reported actions if it is assumed that both Figure and Ground contain all the events (and objects) mentioned by the descriptions of the actions placed in them, and that the Figure cannot be larger than its Ground (v) Booth’s pulling the trigger was his shooting the gun. (vi) #Booth’s shooting the gun was his pulling the trigger. The action of the trigger pull and the shooting are exactly the same, but one should not try to locate that action and the gun shooting within a Ground that includes that action and only the trigger, as (vi) attempts. (See Pietroski 1998, 2000, for discussion.) 69. The target thought must instead be expressed with an overt adverb: Jules Bagwell was later Julia Bagwell. 70. Here too lexical meaning may overcome the temporal orientation on display in the contrast between (277) and (278). Afflicting the generations in (i) of the twin brothers Jacob and Meyer is a terrible genetic disease that causes them to perish at the moment a male child is conceived: (i) Hester Street Central Park West Bleeker Street East End Avenue
Jacob Julius Jon Joshua
Meyer Milton Mark Micah
As before, their lives do not overlap, but now one may say indifferently (ii) or (iii): (ii) Jacob was the father of Julius and the grandfather of Jon. (iii) Jon was the son of Julius and the grandson of Jacob. Sentence (ii), thanks to the relation “x is the father of y,” projects into the future. It is interesting to note that an asymmetric relation is not always sufficient to license a foray into the future. Suppose that none of the generations of (i) overlap, so that uncles did not coexist with nephews nor first cousins with second cousins, and so on. (iv) (What is remembered about him is that) Jacob was a teacher and the father of Julius (and the grandfather of Jon). (v) #(What is remembered about him is that) Jacob was a teacher and the uncle of Milton (and the great-uncle of Mark). Sentence (v) seems to contrast with (iv), which reprises (ii).*Being the uncle of Milton neither holds during Jacob’s lifetime nor is it a (future) effect of anything he did during his lifetime, unlike his being the father of Julius. One could, I suppose, construe Jacob’s living as the brother of Meyer as something he did that has among its (indirect) effects his becoming the
Notes
917
uncle of Milton. If the causal and thematic relations do not themselves exclude such indirection, it would suffice for the contrast between (iv) and (v) that speakers are simply slower to recognize the respect in which being the uncle of Milton could also characterize how Jacob was, where he was, or what he did, during his lifetime. *Note that the examples are designed to control for the intrusion of ahistorical tenses, the use of which allows us to say (vi) and (x) despite Russell being dead: (vi) Russell is an author of Principia Mathematica. (vii) Russell is the teacher of Wittgenstein. For reasons I don’t understand (see Roy 2001 and Ionin and Matushansky 2002 for relevant discussion), (viii) in contrast to (vi) and (vii) cannot be used without implying that Russell is still alive: (viii) Russell is a teacher. The first conjuncts of (iv) and (v), immune from ahistorical interpretation, are thus assured a lifetime effect and it extends to the second conjuncts falling under the same tense. Coordination of fully tensed clauses vacates the contrast: (ix) (What is remembered about him is that) Jacob was a teacher, and he was the father of Julius (and the grandfather of Jon). (x) (What is remembered about him is that) Jacob was a teacher, and he was the uncle of Milton (and the great-uncle of Mark). The second token of a past Tense need not be dependent on the first, and it may be understood as an ahistorical tense under tacit sequence of tense, as if to say one of the following: (xi)
(What is remembered about him is that) Jacob was a teacher, and (I saw on the genealogical chart that) he was the father of Julius (and the grandfather of Jon). (xii) (What is remembered about him is that) Jacob was a teacher, and (I saw on the genealogical chart that) he was the uncle of Milton (and the great-uncle of Mark). (xiii) (What is remembered about him is that) Jacob was a teacher, and (when the genealogical tree was finally established) he was the father of Julius (and the grandfather of Jon). (xiv) (What is remembered about him is that) Jacob was a teacher, and (when the genealogical tree was finally established) he was the uncle of Milton (and the great-uncle of Mark). Kyle Johnson (p.c., 2003) and I differ where the conjoined phrases are Tensed but subjectless: (xv) (What is remembered about him is that) Jacob was a teacher, and was the father of Julius (and the grandfather of Jon). (xvi) (#)(What is remembered about him is that) Jacob was a teacher, and was the uncle of Milton (and the great-uncle of Mark). I observe a contrast, so that (xvi) patterns with (viii), but Johnson views (xv) and (xvi) on a par with (xi) and (xii). 71. For definition of ⌜now(vi)⌝, see (103) in section 9.3. now[ER] ↔def ∀e(ERe → now(e)).
918
Notes
72. This default to a contemporary perspective and the consequent anomaly of (251) and (278) is out of the blue. I do not imagine that it would perturb a speaker uttering (251) or (278) in front of a genealogical tree or medical chart. 73. When (251) or (278) is uttered felicitously in front of a genealogical tree or medical chart, a tacit demonstrative, there or here, takes first position preserving the speaker’s frame of reference. 74. The adverbs now (now is not ⌜now(vi)⌝; see note 71) and then may order past events from within the speaker’s perspective on a moment in history from two decades ago, as in (i): (i) On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average was now / (at) first / at this time up a modest gain and then/later down a record-breaking 508 points by the closing bell. On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average was now / (at) first / at this time up a modest gain and soon down a record-breaking 508 points that ended trading. Yet in the historical present, the speaker becomes a contemporary of the events observed. If in that moment of observation the stock market is up, it is not down and it cannot be seen that even in the near future it will fall 508 points: (ii) #On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and then/later down (a record-breaking) 508 points (by the closing bell). #On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and soon down (a record-breaking) 508 points (that ends/ended trading). A subsequent token of the historical present advances to a subsequent moment of observation, also contemporary with the events observed, at which the stock market is no longer up and the crash is in plain sight: (iii) On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and is then/later down (a record-breaking) 508 points (by the closing bell). On the floor of the New York Stock Exchange, after the opening bell on Black Monday, October 19, 1987, the Dow Jones Industrial Average is now / (at) first / at this time up a modest gain and is soon down (a record-breaking) 508 points (that ends/ ended trading). See section 9.3 for further discussion. Unlike ‘Past[E0,ER]’, Present is never anaphoric to a preposed DP, since Jules Bagwell is unhappy as a man is false under all construals despite his having been unhappy while Jules Bagwell. This would not be if the present could shift to the time of Jules Bagwell-ing.
Notes
919
75. Negation, far above Tense in (292), does not appear to match its lower position in (291) or in any other English sentence. Its location faces a familiar problem. The forceful denials of (i) in (iii) are not the trivial truth that (iv) expresses as the result of dumb insertion of negation in what looks to be its pronounced position. It is rather that as in (v) negation is apparently removed to a position above Tense. (i)
A moonless night no one remembers, a flash flashed the southern sky above Roswell, NM. (ii) [∃t: moonless night(t)] [∃x: flash(x)] [∃e : during(e,t)] (Past(e) & flash(e) & …) (iii) A moonless night no one remembers, a flash did not flash the southern sky above Roswell, NM. A flash did not flash the southern sky above Roswell, NM, on a moonless night no one remembers. (iv) # [∃t: moonless night(t)][∃x: flash(x)][∃e : during(e,t)](Past(e) & not(flash(e) & …)) (v) [∃t: moonless night(t)][∃x: flash(x)] not [∃e : during(e,t)](Past(e) & flash(e) & …) The logical forms in (vi) and (vii) restore a linear correspondence to the sentences as pronounced: (vi) [∃t: … night(t)][∃x: flash(x)][℩E: Past[E] during[E,t]][∃e: Ee](flash(e) & …) (vii) [∃t: … night(t)][∃x: flash(x)][℩E: Past[E] during[E,t]] not [∃e: Ee](flash(e) & …) But (vi) and (vii) add another layer of event quantification: Tense is now taken up into the restriction of a (second-order) event quantifier. This is my official position and also that of Schein 1993, which, more extensively, replaces ‘∃e (Agent(e,x) & …)’ with ‘[℩E: Agent[E,x]] [∃e: Ee](…)’, etc. That said, it has to be immediately acknowledged that the sensible interpretation of (iii), which does not imply the existence of a flash that did not flash, places the negation even higher, perhaps an effect of lowering the quantifier below it: (viii) [∃t: … night(t)] not [∃x: flash(x)][℩E: Past[E] during[E,t]] [∃e: Ee](flash(e) & …) In the end, English may do little better in transparently representing the scope of negation than languages that pronounce it adjacent or attached to a verb. So as not to add here to the clutter created by supermonadicity and adverbialized logical form, I will favor the misrepresentations in (292) and (v). 76. That is, I assume the reader will not be distracted by those readings where Clark Kent does resemble Superman merely in virtue of neither being female and both being Metropolis taxpayers. 77. Note also that no interpretation of (i) is falsified by Clark Kent’s or Superman’s self-resemblance: (i) No civilian in street clothes resembles any superhero enough to be instantly recognized. 78. Note that a resemblance to Superman-ing does not imply Superman-ing at the time of the resemblance (in which respect, resemble Superman differs from be Superman), or else we would never resemble our noble but dead ancestors. 79. Fixing the reference time ‘[℩ER : then[ER,E]]’ is vacuous for the interpretation of Present Tense, which is never anaphoric to it. See note 74.
920
Notes
80. “Coat and tie” for “Clark Kent-ing” and “cape and tights” for “Superman-ing” in case the nonce words have finally become too irritating and also so I can reserve them for the meaning of the denominal adverbials. 81. As might be delivered by a bored logic instructor illustrating the reflexivity of identity. With what is known about the Bagwells, (i) and (ii) can again be heard as false: (i) At Halloween 2005, Jules Bagwell was JULES Bagwell. (ii) At Carnaval 2005, Julia Bagwell was JULIA Bagwell. A semanticist I know with little patience for logic instructors insists that (320) and (321) are also false despite any intonation that pretends to deliver something self-evident. If so, there is less for me to worry about than I assume. 82. The paraphrase sometime is not meant to preclude other means of temporal reference, which could just as well include that full complement that interprets Tense in main clauses. 83. “We include under this head [Elegant Variation] all substitutions of one word for another for the sake of variety. … Many writers of the present day abound in types of variation that are not justified by expediency, and have consequently the air of cheap ornament. It is impossible to lay down hard and fast rules, but two general principles may be suggested: (1) Variation should take place only when there is some awkwardness, such as ambiguity or noticeable monotony, in the word avoided. (2) The substitute should be of a purely pronominal character, a substitute and nothing more; there should be no killing of two birds with one stone. Even when these two requirements are satisfied, the variation is often worse, because more noticeable, than the monotony it is designed to avoid” (Fowler 1908, 175f.). 84. See Georgi 2011 for recent discussion. 85. Keep in mind that the first demonstrative is fixed in the current scene. In other interpretations, it too can be dependent on last New Year’s Eve. Flat intonation again favors interpretation with respect to the same frame of reference. It is a switch in frames of reference that needs to be signaled. But I feel an asymmetry, it requiring some extra effort or gesture to return to the current scene after the first demonstrative has been grounded elsewhere. (Cf. note 68.) 86. The paraphrase may be clumsier than the omitted logical form. Adverbialization, so it is maintained, applies without exception to [D NP] to derive an adverbial that restricts events reported to those while NP-ing, as it were, for the events NP describes. But which events or states does the NP in that guest who bit the hostess describe? Is it the guest’s state during the brief interval in which he bites, or his being a guest for the duration of the party at a certain moment of which he bites, or is it a lifetime that includes being a guest at this party at which he bites—H guest …? If the neighborhood is fixed to be local to the events described, then for the same description to include in the local neighborhood both this New Year’s Eve and last New Year’s Eve, the speaker must have in mind H guest in uttering (332) and intending both tokens of the demonstrative to be understood the same way. In the text, sometimes NP and having or being about to have NP stand for H NP . All this insinuates a fairly rich, unspoken aspectual morphology for nominals.
Notes
921
87. I skate by some problems of description and analysis: (i) Last week, Abraham Lincoln was Abraham Lincoln. Presumably, (i) is pathological in that the speaker represents herself as not knowing anything stronger than a fact about last week, and the sentence remains sick whether (i) is literally false because Lincoln did not exist then or literally true in asserting an eternal identity, merely the observation of which is reported to have been in the past. Sentence (ii) is unequivocally true but once again made anomalous by the adverb and the speaker’s epistemic stance: (ii) Yesterday, George Bush was George Bush. The case that is unclear to me is (iii) under the circumstances that George Bush was alive and well all day yesterday in Albania: (iii) In the White House yesterday, George Bush was George Bush. If the speaker was herself in the White House yesterday or represents her source as such, then (iii) joins (ii) in what goes wrong. If, on the other hand, the lights in the White House were out all day, I tend to think that (iii) is literally false, without escape from the implication that George Bush or something else relevant to this assertion was there. The text only takes up cases where the subject is alive and well on location, varying only whether he is there as described by nominal description. 88. Note the interaction of perspective and the temporal dimension of these names: (i) Phosphorus always first appears out over the ocean. (ii) Venus always first appears out over the ocean. (i) is true in Boston and Nantucket and false in San Francisco, and (ii) is true in Nantucket but in neither Boston nor San Francisco. 89. Cf. In North America, Venus is brighter in the evening than in the morning, which compares Venusian images in North America. 90. The logical form (340) comes naturally given the formal requirement on names, namely, that they adverbialize like other DPs and must thus provide an event description. Although I could wish for better, the facts of (338)–(339) do not themselves provide independent argument for names as complex descriptions if it is allowed that the PP’s formal requirements are met, quantifying into an explicitly represented higher verb or operator that shifts the context with respect to which a primitive, Millian name is interpreted: (i) [∃Ef: In North America[Ef]] it is said[Ef] that [[h]]Ef … [∃Ef: In North America[Ef]] it is observable[Ef] that [[h]]Ef … Of course, translating (338) as (i) confounds it with In North America, (it is said that) Hesperus is hotter than Hesperus, but this observation just returns to the earlier arguments in favor of adverbialized logical form. 91. Or perhaps the drive toward some meaning leads to understanding (335) frame-invariantly to mean: (i) The somewhere Hesperus is hotter than the somewhere Phosphorus. (ii) From anywhere, Hesperus is hotter than Phosphorus. But these are indeed contradictions rather than alternative expressions of (337).
922
Notes
92. “Bruces,” Monty Python’s Flying Circus, Episode 22, November 24, 1970. 93. Similarly for indexicals as when the von Trapp family children sing “Adieu, adieu, adieu to you and you and you” (Oscar Hammerstein II, “So Long, Farewell,” The Sound of Music, 1959). 94. For any frame of reference the conditions of epistemic salience for use of a name are not the same as those of demonstratives this or that (and neither are these the same as each other) (Higginbotham 1988). Suppose we have a mutual interest in Bruce, the phenomenologist, whose person I mistakenly confuse with that of Bruce, the epistemologist. Taking the epistemologist by the shoulder I say to you either (i) or (ii): (i) Bruce is a fine phenomenologist. (ii) This Bruce is a fine phenomenologist. Sentence (ii) is simply false in that something has been successively demonstrated, it is a Bruce, and he is not a phenomenologist. Sentence (i) rather fails a presupposition that the person demonstrated is who I think it is. 95. Suppose for example that deaccentuation indicated that the token names were exact copies, but the two tokens of a variable ranging over frames of reference are bound by different operators now that a second adverb intervenes. 96. Gawron and Kehler (2004) recognize the resemblance of their method to Schwarzschild’s (1996), who treats the problem that (i) and (ii) may be intended with distinct truth conditions despite analysis that the barn animals and the cows and the pigs refer to the same creatures: (i) The cows and the pigs were separated. (ii) The barn animals were separated. In effect, an adverb with a contextually determined parameter intervenes: (iii) The cows and the pigs were separated that way. (iv) The barn animals were separated that way. Mention of cows and pigs in (iii), as might anything else in the context—a chart, for example— raises to salience that way, the cow from pig way to separate them. In (iv), the linguistic object itself evokes only the barn animal from barn animal way. The relation of nominal description to adverb is no more structural than implied by the use of a demonstrative to pick out salient features of the environment, which exposes Schwarzschild’s (1996) treatment of nominal reference to the objection from chapter 1, note 2, that it effaces the contrast between (v) and (vi): (v) The elms and the beeches are equinumerous. (vi) *The trees (which are elms and beeches) are equinumerous. Whatever the cows and the pigs or the elms and the beeches refers to, now at issue is whether the influence a chosen nominal description has on the interpretation of the matrix sentence can be sloughed off onto a family of independent operators, respectively or Schwarzschild’s covering operators, with scope over the larger predicate and only an indirect connection to the nominal description itself. 97. See Schein 2003, 349–352, for an argument that reciprocal constructions also form adverbial event quantifiers restricted by and scoping over subatomic phrases.
Notes
923
98. Compare ordinary understanding of (i)–(ii), also with no direct mention of participants, and how one would go about verifying which alliance was which betrayal in verifying (iii): (i) Every alliance ended in betrayal. (ii) Every alliance was a betrayal. (iii) These alliances were these betrayals. But I slight here an important interaction between syntax and semantics. As remarked, adverbialized logical form posits an adverbial modifier ⌜[℩E : NP]⌝ derived from the NP restricting every DP. The neighborhood relation then comes with the Eventish analysis of any adverbial modifier, which, for example, finds in If the parasite lives, the host dies and in The parasite living, the host dies expression of a relation between events of the parasite living and events of the host dying. That relation in these sentences is plainly not identity, and it remains for a complete survey of conditionals and other adverbial modifiers to discover the range of spatiotemporal and causal relations that may be intended instead. (The neighborhood relations are thus the Eventish counterpart of modal accessibility relations. See Lycan 2001 for extensive discussion.) But the neighborhood relation—whatever it is in these sentences—is not one that implies that the participant in the living is the participant in the dying, contra (380). In contrast to these sentences composed from large clauses with overt subjects, the adverbial concealed in (378) is closer to the reduced adverbial clause in (iv): (iv) The allies allied are thwarted. Note that (380) happens to hold of all the events that (iv) implies are neighbors. The point of grammar slighted is whether this reflects that reduced clauses modify via a different, smaller class of neighborhood relations respecting (380) or that the neighborhood relations are the same for all adverbial clauses and the syntax (e.g., control) governing unpronounced subjects (and tense) in reduced clauses derives the truth of (380) for (iv) and (378). See Schwarz 1998a, 1998b, for some discussion. 99. See Koslicki 1999, 2005, and the references cited there. 100. Examples (i) and (ii) are anomalous in contrast to (387) and (388) for reasons that are obscure: (i) #The content of these three containers is insulated by its containers from (all) the containers that do not contain it. (ii) #The contents of the three containers is insulated by their containers from (all) the containers that do not contain them. 101. There are scopal differences, too: (i) (ii) (iii) (iv)
The drivers stood beside their two rented automobiles. The drivers stood beside their rented two automobiles. The drivers stood beside their two respective automobiles. The drivers stood beside their respective two automobiles.
Sentence (iv) implies that each driver has two automobiles, but (ii) need not. However, if (ii) is taken to be about two rentings and to imply that each driver stood beside what he himself rented, then he rented two automobiles. If two measures a collective event, then the standing beside is held to be in the neighborhood of this larger event and thus vague as to which driver stood beside which automobile.
924
11
Notes
Naive Reference for the Cinéaste
1. Gupta (1980, 24), exceptionally, does not believe in a bloated universe where there are passengers besides persons and avoids it in an account of the contrast between the sentences in (1) to which my proposal is similar in spirit. Already, Carlson (1982, 174ff.) concludes from different highlights in a baseball career that “Reggie Jackson” refers to j, “the third batter” to a, and “the twelfth batter” to b and that j ≠ a ≠ b (see chapter 10, note 41). Krifka (1990), on the other hand, manages without expanding the denotation of common nouns at the cost, it seems, of dissolving altogether the nominal reference of 4000 ships when something other than ships is to be counted. 2. With an implication of 4000 toots in mind for (i), suppose, as Moore (1994) and Barker (1999) implicitly do, that the pronoun is a device of direct reference to whatever its antecedent refers to and furthermore that the second sentence conceals no other expression to refer to the events antecedently described. If so, it looks like there is further argument that the antecedent 4000 ships—and with it the pronoun—refers to 4000 somethings other than the fewer ships: (i) Four thousand ships passed through the lock last year. They each tooted their horn when they cleared the last gate. (Barker 1999) By parity of reasoning, as Roger Schwarzschild remarks (p.c.), the 4000 bouquets implied by (ii) imply 4000 Bloombergs in bloom: (ii) Mayor Bloomberg welcomed 4000 ships to New York last year. He brought flowers. If the supposition about the pronoun’s and sentence’s impoverished structures is denied, and the pronoun taken to be descriptive so that (i) is like (iii), there is nothing more to this example beyond what has already been seen in (2)–(3) when nominal descriptions behave as if they were also describing events: (iii) The 4000 ships that passed through the lock last year each tooted their horn when they cleared the last gate. 3. Examples (4)–(6) are meant to threaten tamer versions of the view that nominals denote whatever is counted and thus to paint this view into an unattractive corner. There are things, there are times, and things persist through time. A phase or time slice of a thing, modeled innocently enough by an ordered pair , is no more daring than sliced bread, the slices related by constitution to the whole loaf. Nominals, on this view, denote by the slice. In (4)–(6), the (spatio)temporal slices described are, however, not as numerous as the count requires. It is not four distinct (spatio)temporal slices of the crook under investigation, 4000 distinct (spatio)temporal slices of city officials doing nothing, or three distinct (spatio)temporal slices of the pushcart vendor in violation of municipal codes. To extend the picture to (4)–(6), one should suppose that the one crook in this world enters into a constitutive relation with four counterparts in the narrative worlds of the investigations, a modal individual concept (Gupta 1980). Or one could just drop the pretense that the pairs (n-tuples) counted and denoted by nominal phrases need to constitute the things they are thought naively to refer to. This last
Notes
925
move is, however, hard to swallow if it has already been decided that person, passenger, manager, city official, and violator are all strictly monadic predicates. 4. The sentences in (16) are themselves allegedly ambiguous, referring either to passenger events, , or to passenger-event events, . That is, event-relativized reference for the descriptions in (12) should coincide with absolute reference for the descriptions in (16), and there should be some way to understand these sentences as both true. 5. Thanks for discussion to Roumyana Pancheva. 6. Out of the blue, sentences (i) and (ii) implicate that the HMS Pinafore no longer has a red mast or registration in Panama, or that it no longer exists (the “lifetime effect”) if mast color or registration is assumed to be permanent. (i) HMS Pinafore had a red mast. (ii) HMS Pinafore was registered in Panama. There is no such implicature if the past tense is understood to make definite reference to a past occasion on which the HMS Pinafore was observed (Musan 1997; see note 5 in chapter 10). Similarly, for Captain Corcoran, who has never left the deck and remains always in possession of the ship’s papers, to utter (iii) or (iv) implies that the HMS Pinafore has been without a red mast or registration in Panama often enough that four discrete episodes of having a red mast or being registered in Panama can be counted: (iii) Four times, HMS Pinafore had a red mast. (iv) Four times, HMS Pinafore was registered in Panama. The admiral, however, who boards the HMS Pinafore only for inspections, utters (iii) or (iv) without implying that the HMS Pinafore was ever other than as he found her on inspection. His inspections themselves rather than the HMS Pinafore’s state meet the conditions for counting four discrete events. 7. In fact, it suffices for a more minimal pair to have the harbormaster and the observer at port both uttering (69). The adverb by now in (93) is an expedient flag for the inferred assertion in that it seems to preclude report of direct perception. Once both interpretations are recognized, it is easy to see that the bare now in (69) supports both. 8. Thanks to Roger Schwarzschild for the example. 9. Count preservation does imply that (i) is not narration for a scene that begins from the perspective of the object-counting spy plane and zooms to that of the field officer’s in Red Square (ii) whose frame of reference is event counting (see (41)–(44)): (i) One hundred thousand Soviet troops flowed through Red Square in the May Day parade. (ii) Three hundred thousand Soviet troops flowed through Red Square in the May Day parade. The spy plane’s frame of reference serves well enough on its own, however, to frame events in Red Square, albeit in a long shot from a great distance. It thus proves difficult to construct an example that forces an eyewitness scene to zoom in and the sentence to therefore be rejected.
926
Notes
10. But the implication that every counted event must present a relevant effect is unwelcome in (i), where not every visit to a hotel room visits its bar: (i) (ii) (iii) (iv)
Four thousand hotel guests emptied 20 room bars in 6 days. Four thousand hotel guests clogged the plumbing in 6 days. Four thousand patients overwhelmed the ER in 6 days. #Four thousand patients overwhelmed the morgue in 6 days.
Rather, one might point more abstractly to a rate that characterizes average individual behavior and the average effect of the individual visit, which is to consume, say, 0.2 ounces from the hotel-room bar. In contrast, (121) and (iv) resist even such a reduction to aggregate, average individual behavior, according to which the effect of a bug’s visit to the screen door should amount to 1/40 of a death. If so, the events that (121) and (iv) purport to count do not have any local effects that, according to any natural theory of individual behavior and the effects of individual action, aggregate on the screen door or in the morgue. Note that if, as (i) suggests, Causes E may have effects E’ in virtue of aggregate, average individual behavior across events E, ‘Cause[E, E′]’ is not reducible to a first-order relation ‘Cause(e,e′)’. This is contrary to the speculation in Schein 2006 that the primitive vocabulary is first-order, unless a ditransitive, ‘Cause[E, E′, e+]’, includes reference to Pietroski’s (1998, 2000) fusion e+ of causes and effects. 11. In contrast to 4000 travelogues each about a single passage through the lock, imagine that four studios each commissioned 1000 travelogues each a documentary tracking a single ship’s adventures at sea, which include all its passages through the lock. The ships are still the subjects of 4000 travelogues, but now absent a correspondence between counting 4000 ships and framing 4000 passages, the event-counting reading fails. 12. See note 2. 13. More examples: i.
Three million passengers crowded National Airlines routes last year. (After Moore 1994) ii. Four hundred thousand elite Soviet troops streamed past the reviewing stand in 6 hours. iii. Much Soviet military hardware surged through the square during the parade. iv. Four thousand locusts mobbed autographs from Marilyn Monroe at the premiere of Gentlemen Prefer Blondes. v. Four thousand callers flooded the NBC switchboard in the first 5 minutes of The Tonight Show when Julia Roberts was the guest. vi. Forty thousand mosquitoes raked/strafed/blasted/pummeled/pounded an exposed arm in 3 minutes. vii. In the whirlwind, four billion hailstones had washed over/cascaded above the Arctic base camp in the first hour of the perfect storm. viii. In the whirlwind, much ice had washed over/cascaded above the Arctic base camp in the first hour of the perfect storm. 14. “Mary was wearing a green dress” is accepted without batting an eye, without knowing who said it or how it came to be known. But if the doorbell rings, and the blind neighbor says “it was Mary, and she was wearing a green dress,” our response is “Huh?” There is both a reluctance to accept without question in this case that he must have his reasons for saying
Notes
927
so and a blindness to the obvious conclusion that it is St. Patrick’s Day and Mary is Irish, without such considerations having been made explicit. 15. A last example. In a discussion of the quirky, superstitious rituals of ballplayers, one might remark on thirteen appearances at the plate, despite knowledge of the multiple appearances of some, that (i) a. Thirteen batters crossed themselves in the batter’s box. b. Thirteen batters scratched themselves at the plate. Although equally episodic and circumscribed by events at the plate, it is yet again odd to say, knowing that some were multiply injured, that (ii) Thirteen batters were injured at the plate struck by a wild pitch. Injuries belong to biography, it seems, and one cannot follow these biographies outside the ballpark without violating anticonvergence, although the scene of the literal report in (ii) is fully compliant. As these are editorial decisions, it may be that not all speakers are sensitive to a contrast between (i) and (ii). 16. Fixing the reference time ‘[℩ER : then[ER,E]]’ is vacuous for the interpretation of Present Tense, which is never anaphoric to it. See note 74 in chapter 10. 17. Thanks for the judgments to civilians Joe Andrews and Tom Buscher. 18. Adapted from Some Like It Hot (1959). 19. Paul Pietroski (p.c., January 28, 2007) offers his own version of Some Like It Hot: “Alice and Betty are sisters. They are also singers. The Smith Sisters perform, twice a year, at the local Rotary Club. Xavier has heard Alice and Betty sing there, many times. He especially enjoys their rendition of “Sweet Caroline.” Unbeknownst to Xavier, Alice and Betty have another act. They are jugglers who perform, in disguise, as Carla and Debra Johnson. The Juggling Johnsons appear only in venues far from where Alice, Betty, and Xavier live. The Smith Sisters have kept this part of their lives a secret. Not even their best friends suspect that Alice and Betty Smith might be Carla and Debra Johnson. Xavier travels a lot. One night, he sees The Juggling Johnsons and admires their skills. It never occurs to Xavier, or anyone in the audience, that Carla and Debra might be Alice and Betty. Indeed, the next day, Xavier forms a plan to get all four performers on the same stage: the Smiths could sing, while the Johnsons juggle. He makes a note to contact their respective agents, and goes back the next night to see the Johnsons again. There are, as Xavier notices on the second night, two distinct roles in the juggling act. Carla’s signature trick involves torches, Debra’s involves swords. Carla wears red, Debra does not. And so on. (There are also two roles in the singing act: Alice is an alto, Betty is a soprano.) But to keep all their skills sharp, and also to amuse themselves, Alice and Betty alternate juggling roles: one night, Alice appears as Carla, and Betty appears as Debra; the next night, Betty appears as Carla, and Alice appears as Debra. Xavier does not notice the switch. He assumes that the juggler in red, Carla, was the very person who juggled in red on the previous night. He also assumes that The Juggling Johnsons are sisters. (1) (2) (3) (4)
Xavier thinks that Carla and Debra are sisters. Carla and Debra are sisters. Xavier wants to hear Alice and Betty sing while Carla and Debra juggle. If Alice and Betty sing and juggle at the same time, then at that time, Alice and Betty are singing while Carla and Debra are juggling.
928
Notes
(5) (6) (7) (8)
Xavier would deny that Alice and Betty are Carla and Debra. Alice and Betty are Carla and Debra. Alice and Betty are The Juggling Johnsons. Alice is sometimes Carla and sometimes Debra, and Betty is sometimes Carla and sometimes Debra. (9) Alice is Carla, or Alice is Debra, or Betty is Carla, or Betty is Debra.” We concur that (1)–(3) and (5)–(8) are true, but (4) and (9) are not. For related discussion, see section 9.1.0. 20. Recall from section 9.4.2 that the optical counter on the ground at Red Square measures Soviet troopery at 300,000 strong, but the aerial scene from the spy plane measures it at 100,000. 21. I will leave aside the proper treatment of predicative bare plurals except to note that the contrast between them and their counting counterparts coincides with the syntactic and semantic differences distinguishing “characterizing” and “defining” nominal predicates in Roy’s (2006) crosslinguistic survey. For example, Russian obligatorily translates the bare plural with the instrumental and the counting plural with the nominative: (i) Oleg i Ivan byli dva xorošix rabotnika. Oleg.nom and Ivan.nom were two.nom good.gen workers.gen Oleg i Ivan byli (*dvumja)xorošimi rabotnikami. Oleg.nom and Ivan.nom were two.inst good.inst workers.inst (Pereltsvaig 2001, 79) The sense in which the bare plural is more predicative or characterizing rather than defining is suggested by the contrast between (ii) and (iiib): (ii) Rudyard Kipling and Sinclair Lewis were (both) Nobel laureates, which James Joyce and Franz Kafka were not. (iii) a. Rudyard Kipling and Sinclair Lewis were two Nobel laureates. b. *Rudyard Kipling and Sinclair Lewis were two Nobel laureates, which James Joyce and Franz Kafka were not. See Roy 2006 and the references cited for further discussion. 12
Measuring Events
1. Also, (i) Most first-class passengers who shared a magnum of champagne drank no more than five gulps of it. (ii) Most first-class passengers who sipped at a magnum of champagne swished before swallowing. 2. Thanks to Roger Schwarzschild (p.c., July 2009) for suggesting this accounting method. 3. See Blanchette 1999 for discussion of number and identity.
Notes
929
4. See section 9.4.2. ®[s, ζ, ξ, s ] ↔df ∀α s (Asα s → Rsα s ) & ∀α s ∀α s ′((Asα s & Asα s ′ & overlap(α s, α s ′)) → α s = α s ′) & ∀α s ∀α s ′((sα s & sα s ′ & At[α s,ζ , t( s), s] & At[α s ′,ζ , t( s), s]) → α s = α s ′) & ∃α s (sα s & At[α s,ζ , t( s), s] & ∀x(ξx → ¬∃y(overlap(x,y) & At(αs,y,t(s),s))) & ¬∃p(At(p,p,t(s), s) & Limit point(p, ζ, s) & Limit point(p, ξ, s)) 5. Thanks to Roger Schwarzschild (p.c., January–February 2009) for discussion of the following. 6. As Greg Scontras (p.c., November 13, 2015) points out, the numeric measurement itself, its value, does not need to be constant for the duration: (i) The suitcases are at least 30 pounds (empty or loaded). Alice is at least 2 feet tall (under the influence of a tea cake or not). What is meant by a stable dimension is illustrated in the following. Imagine a balloon that begins as a cylinder 1 foot in diameter, the length of which oscillates (like a spring) between 2 and 5 feet: It is at least 2 feet long. Imagine instead that it inflates from a 2-foot-long cylinder to a 5-foot-diameter sphere: *It is at least 2 feet long. It is awkward to attribute length to it for a period that includes it being a sphere, for which length is not defined. On the other hand, if at the same time its weight oscillates between 2 and 5 pounds, it is at least 2 pounds, and it doesn’t matter that it morphs into a geometric solid without length. The present tense qua generic attribution implies measurement of a dimension that is stable over some temporal interval under alternative conditions that otherwise preserve the “essential” or intrinsic properties of what is measured. In particular, the measure of these properties should be the same despite, say, the linear or rotational motion of the things so measured. A bouquet of flowers is large means that the bouquet displaces a continuous large volume, and it would do so under any displacement of the bouquet that preserves the bouquet—that is, under alternative conditions where the bouquet continues to exist as such. It cannot however be said that The flowers are large (when each flower is tiny), because the flowers are not the bouquet, and many movements of them disperse the flowers and then there is not a volume that is large let alone the same one. On the other hand, for a bouquet of flowers is heavy and the flowers are heavy, their weight is stable across the alternative conditions being independent of the flowers’ relative spatial position. When a stub predicate is not under a generic tense operator or any other means to attribute essential or dispositional properties, measurement under alternative conditions is not relevant: a long 613 popsicle sticks vs. 613 long popsicle sticks. For the first NP, there is a length that exists at this moment, constituted by the line of popsicle sticks, and it doesn’t matter that this length does not survive as a measureable dimension under alternative conditions. To talk about any of this, we need ‘long(e,x)’ rather than just ‘long(x)’.
930
Notes
7. At the margins of my dialect, a(n) in (i)–(v) gets pushed farther down, unless many in (122)–(126) is already derived from m(uch)+any (cf. K. A. Jayaseelan 2011, deriving every child from ever each a child): (i)
a. Most any a fielder is leaving tonight with a minor injury. b. *Most any a fielder are leaving tonight with a minor injury.
(ii) a. *Most (any) fielders is leaving tonight with a minor injury. b. Most (any) fielders are leaving tonight with a minor injury. (iii) a. *Most any an ill-matched seven fielders is leaving tonight with minor injuries. b. Most any an ill-matched seven fielders are leaving tonight with minor injuries. (iv) a. *Most any an (ill-matched) infielder and outfielder is leaving tonight with minor injuries. b. Most any an (ill-matched) infielder and outfielder are leaving tonight with minor injuries. (v) a. *Most any an ill-matched seven infielders and outfielders is leaving tonight with minor injuries. b. Most any an ill-matched seven infielders and outfielders are leaving tonight with minor injuries. 8. I have taken liberties with Heycock and Zamparelli’s (2003, 2005) discussion. The anticumulativity of a(n) is indeed claimed to explain *a veteran fielders, to which I have added that their condition is satisfied to good effect in a veteran seven fielders. It then however remains unexpected that a veteran seven or more fielders is acceptable given its cumulativity—if these are a veteran seven or more fielders and those are a veteran seven or more fielders, they all are a veteran seven or more fielders. 9. In Heycock and Zamparelli 2005, and means set product, the operation in (91), which applies to a coordination of singular NPs as exemplified in (92) and (93): (91) Set Product (sp) sp(S1, … , SN) =def {X : X = A1 ∪… ∪ AN, A1∈ S1, … , AN ∈ SN} (92) a. b. (93) a. b.
[[NPi]] = {{a}, {b}}, [[NPj]] = {{c}, {d}} [[NPi and NPj]] = sp([[NPi]], [[NPj]]) = {{a,c} {a,d} {b,c} {b,d}} [[NPi]] = {{a}, {b}, {c}}, [[NPj]] = {{c}, {d}} [[NPi and NPj]] = sp([[NPi]], [[NPj]]) = {{a,c} {a,d} {b,c} {b,d}{c,d}{c}}
Generalizing set product to conjunctions of higher type, Heycock and Zamparelli (p. 263) concede a problem with nonincreasing quantifiers. Their semantics would validate (159) and (160): (159) a. Exactly two children in the group were males. b. Exactly two adults in the group were females. c. Exactly two children and exactly two adults in the group were people. (160) a. No children in the group were males. b. No adults in the group were females. c. No children and no adults in the group were people.
Notes
931
For (159), if {x: male(x)} ∈ [[exactly two children in the group]] and {x: female(x)} ∈ [[exactly two adults in the group]], then {x: male(x)} ∪{x: female(x)} ∈ sp([[exactly two children in the group]], [[exactly two adults in the group]]); similarly for (160). Their response is to flee from the surface syntax of (159c) and (160c) in different directions. For (160), with the equivalences in (i) in mind, they suggest parsing (160c) as in (161) (simplified), relying on a popular decomposition of the decreasing quantifier No. The inference in (160′) based on the negations of sentences containing only increasing quantifiers becomes invalid as desired: (i)
a. [[ [[No NPi] and [No NPj]] ]] = [[ [No [NPi and NPj]] ]] b. [No [NPi and NPj]] Φ ↔ Not([∃ [NPi and NPj]] Φ)
(161) Not ([any children and any adults] were people). (160′) a. Not (any children were males). b. Not (any adults were females). c. Not (any children and any adults were people). Yet the set product of [[few children in the group]] and [[few adults in the group]] also validates (160″), which is no more valid than (160): (160″) a. Few children in the group were males. b. Few adults in the group were females. c. Few children in the group and few adults in the group were people. But here a forced decomposition conjoining only increasing quantifiers, thus translating (160″c) as (ii), fails to derive an equivalence (absent an analog of (i) for few): (ii) Not ([many children and many adults] were people). For the sake of concreteness, take few to be fewer than two. Sentences (iii) and (iv) fail to be equivalent, as any ten rabbis and a priest falsify (iii) but verify (iv): (iii) Fewer than two rabbis and fewer than two priests were Jewish. (iv) It is not the case that more than one rabbi and more than one priest was Jewish. Thus, the application of a set product and in (160″c) fails the meaning of few children in the group and few adults in the group, validating the false inference, and the sentence finds no equivalent in (ii) masquerading as its actual logical form. Generalizing set product and to higher types would either misinterpret (160″c) and like sentences or leave them without an interpretation. To avoid the like unwanted inference in (159), Heycock and Zamparelli speculate that (159c) does not contain a conjunction of DPs after all—that exactly like only is not a determiner but a sentential adverb. Spelling out what and should mean when either exactly or only appears as if in a coordination with DPs, as in (162) and (163), goes beyond the purview of this paper: (162) John and exactly one woman have met. (163) I interviewed ten men and only three women. The purview of chapter 2 here has included apparent DP coordinations interrupted by sentential adverbs of various sorts without harm to a collective interpretation. The thesis has been that such apparent DP coordinations, like all others, are clausal coordinations with and as the ordinary sentential connective.
932
Notes
As for when sentences are coordinated, as in (170a), if and always means set product, as Heycock and Zamparelli (2005) hope it does, then, as they concede (p. 266), sentential coordination itself becomes a problem: (170a) John left, and Mary returned. Their suggestion is to translate the sentences coordinated in (170a) as predicates denoting events coordinated with set product and: some events are John leaving and Mary returning. But what about (v) or (vi) when there are no events to speak of? (v) Nothing interesting happened, and nothing important occurred. (vi) Little of interest happened, and little of importance occurred. (Example (vi) is cited as counterpart to (160″c) in this discussion, as (v) is subject to an attempt similar to the reanalysis of (160c) as (161).) 10. Also, (i) a. In two separate trials in the same week, two sequestered juries have deadlocked after a fortnight deliberation. b. Two judges in two trials sequestered two deadlocked juries. (ii) a. F In two separate trials in the same week, a sequestered two juries have deadlocked after a fortnight deliberation. b. F Two judges in two trials sequestered a deadlocked two juries. The reported judgments come with a caveat not to mistake the false or infelicitous interpretation for a restrictive or contrastive reading that is more acceptable. In some sense modification of an indefinite is logically restrictive always, in that, for example, the further comment in (iii) that the juries were sequestered restricts the juries that can witness the truth of these sentences: (iii) a. There was a jury; (and) it was sequestered. There was a jury, which was sequestered. b. There were two juries; (and) they were sequestered. There were two juries, which were sequestered. But note that these sentences do not insinuate that there might also be juries that are not sequestered. A story that begins with (iiia) may very well be about a single jury with none other in sight on the narrative horizon or known to the speaker. A distinction between what I will call the appositive or noncontrastive use in (iii) and a restrictive or contrastive use of modifiers survives even in the absence of definite reference in whether or not the speaker is concerned to distinguish those referred to from others just like them also at hand—a sequestered two juries rather than a lunching-out two juries. A more complete statement of the facts reported for (127) and (128) now follows. In meeting the conditions for there being just one sequestration and one deadlock, the sentences in (iv) can be true without any other jurors in mind. This appositive or noncontrastive use of the modifier is reprised (v): (iv) a. In a trial last week, a sequestered twelve jurors have deadlocked after a fortnight deliberation. b. One judge sequestered a deadlocked twelve jurors. (v) a. In a trial last week, twelve jurors have deadlocked after a fortnight deliberation; (and) they were under a sequestration. b. One judge sequestered twelve jurors; (and) they were in a deadlock.
Notes
933
If, on the other hand, there is not just the one sequestration or deadlock, as in (128), the sentence is false as reported when the modifier is appositive or noncontrastive. It is not true, unless the speaker specifically has in mind a use that is restrictive or contrastive with respect to twenty-four jurors, where she means to distinguish a sequestered twenty-four jurors from, say, a lunching-out twenty-four jurors. If she has before her various groups of twenty-four jurors, she may use (128) intending that it is about a one of those groups with sequestered jurors. Given, on whatever basis, the prior organization into discrete groups of twenty-four, it need not be in this case that the twenty-four jurors come from the same sequestration or deadlock. I trust that the reader will have no trouble setting aside this reading, choosing a twenty-four from among many twenty-fours, from the one marked as false in the text. Speculating further on this restrictive or contrastive use, it may not signal any special mode of semantic composition if it can be viewed as the satisfaction of a(n)’s conditions on counting by a higher perspectival or presentational relation located above the pronounced modifier. The speaker’s take on the scene organizes it into discrete presentations a one of which is now being referred to and counted. 11. See caveat, note 10. 12. Note that the count morphology associated with a(n) is taken to apply to the AP alone. There could be just a single state of twelve skirts being long in a context with only twelve skirts. 13. Ionin and Matushansky (2004a, 2004b, 2006) claim that the modified cardinal construction excludes modifiers as in (i) that apply only to singular individuals (66), appealing to an intuition that a committee or a set is not a singular individual: (i)
a. b. (66) a. b.
*I met a very tall twelve committee members the other day. *She bought a blue twelve pencils. ??/*I met a very tall committee the other day. *She bought a blue set of pencils.
Example (132) joined with (ii) is intended as a counterexample to the claim, as is (iii) alongside (138) below: (ii) *A lengthened group of twelve skirts is/are hanging in the showroom. *A lengthened collection of twelve skirts is/are hanging in the showroom. *A lengthened rack of twelve skirts is/are in the showroom. (iii) *A bloodred group/sequence of seven sunsets preceded seven days’ bloody battle. 14. More examples: (i) The top drawer contains twelve paired socks. (ii) #The top drawer contains a paired twelve socks. (Unless one has just paired them off out of the laundry.) (iii) A parallelogram contains four parallel lines. (iv) #A parallelogram contains a parallel four lines. (v) Washington Square is bounded by four perpendicular streets. (vi) #Washington Square is bounded by a perpendicular four streets. 15. If one balks for a moment at the thought of a single being green whose parts never appear together, such has already been encountered and accepted in (138)/(139). There, however, it
934
Notes
makes sense that there was a single state of being bloodred and after that one event, seven days’ bloody battle, obtaining the only acceptable reading. 16. The conditional assertion of this last sentence concedes that the facts of (136)–(143) do not themselves close off a metaphysical out. I suppose the accumulated effects of twelve stoplights as twelve separate causes are different from the effects of twelve stoplights as a single cause. It could then be held that the description twelve green stoplights refers in (140)/ (141) to twelve causes and, now ignoring the plural number agreement, that the description a green twelve stoplights in (142)/(143) refers to something else altogether, a single cause. The twelve causes got me home, the one did not—similarly, for (136)–(139) about sunrises. If the facts of number agreement are not reason enough, the discussion surrounding (127)–(135) remains a more direct argument against gerrymandering the reference of the subject nominal while appealing to a desperate metaphysics. 17. Compare also (i), which can be a report of two lightning strikes damaging two buildings if the buildings are nearby the speaker’s current location (equidistant between the sites struck): (i) Last night in different parts of the city, two lightning strikes damaged a nearby two buildings. 18. Compare (i)’s report of damage to one water tower and one radio antenna near the speaker’s location: (i) Last night in different parts of the city, two lightning strikes damaged a nearby water tower and radio antenna. 19. Compare (i), where events of photographing and of touring happily coincide: (i) Last night in different parts of the city, before emergency services arrived, two accident victims flagged down a nearby photographer and tourist. [I do not understand why (ii) resists the conjunctive interpretation denoting a single person: (ii) ??Last night in different parts of the city, before emergency services arrived, two accident victims flagged down a nearby photographer and nearby tourist.] 20. See K. A. Jayaseelan 2011, deriving every child from ever each a child. 21. I will have little more to say about this divergence among singular plurals. See Ionin and Matushansky 2004a, 2004b, 2006, for discussion of why a modifier is obligatory with ‘∃-a(n)’. 22. Singular demonstratives show yet another pattern: (i) A star-crossed two lovers are embracing their doomed love. Every star-crossed two lovers are embracing their doomed love. Many a star-crossed two lovers embrace their doomed love. Most any (a) star-crossed two lovers embrace their doomed love. (ii)
*That star-crossed two lovers are embracing their doomed love.
(iii) That (star-crossed) boy and girl are embracing their doomed love. (iv) That (star-crossed) one boy and one girl are embracing their doomed love.
Notes
935
(v) That star-crossed one boy and two girls are out on a doomed adventure. (vi) *That star-crossed two boys and one girl are out on a doomed adventure. (vii) That boy and that girl who are locked in a deadly embrace are doomed. That one boy and that one girl who are locked in a deadly embrace are doomed. (viii) *That star-crossed boy and that girl are out on a doomed adventure. It is unclear why the demonstrative in (ii) should reject the singular plural construction accepted in (i). If I were not abandoning the problem to morphosyntax, I might linger awhile longer in a semantic idyll. Logical forms for the singular plural count first to one, ‘sg(E1),’ that is, ‘∃eμ1 count[eμ1,E11],’ in measurement eμ1, and then to k in measurement eμk, ‘∃eμk count[eμk, E,k]’.One can in extreme instances well imagine that a frame of reference, scene, and protocol for counting water droplets is one where the clouds do not exist let alone in an arrangement congenial for counting them, too; correlatively, a protocol for counting clouds does not also count water droplets. Suppose that an event and its participants share the same space and thus for any resolution for any scene of that space, if it is coarse enough to resolve a star-crossing or an ill-matching as one—say the entire event all falls within a single cell of the imaginary graph paper (see reticules in section 9.4.2; notes 47, 48 in section 9.4.2; and Schein 2002)—it is not fine enough to resolve and count its participants, proper parts in the space of that one event, as two or more. If so, then in accepting sentences such as those in (i), the speaker changes frame of reference and protocol, eμ1 ≠ eμk, zooming in, as it were, from one count to the next. This modal behavior with respect to protocols may just be part of the meaning of the indefinite article and distributive quantifiers. In contrast, a demonstrative anchors all protocols to a single one vivid in the context of utterance, in which case what measures one in that protocol cannot be measured as something greater under the same protocol. All this may provide some semantic excuse for the contrast between (ii) and (i), but it does not allow for a singular demonstrative with the coordinations in (iii)–(iv), predicting instead that they should be no better than (ii). The contrast between (v) and (vi) suggests first-conjunct agreement (see chapter 2), however, and therefore a syntactic solution in terms of coordination of DPs rather than NPs. The collective modifier is intended to discourage such a way out, but of course collective modification alone should not be such a deterrent in the face of (vii). Rather, if the singular agreement in (iii) is similar in derivation to that in (viii), it wants explanation why (viii) specifically should exclude the collective star-crossed. Leaning heavily on the syntax, suppose pronounced determiners, in contrast to those unpronounced or deleted, must move to higher position to license their pronunciation. A movement of the first determiner to higher position in (ix) and deletion in situ of the second determiner allows star-crossed to include within its scope a coordination of sub-DPs to derive (iii). In contrast, no prenominal modifier can occur before a pronounced determiner, *star-crossed those lovers, and thus it cannot include within its scope a coordination large enough to included full-blooded DPs (x). Any uttered prenominal modifier must be understood to be properly contained within the first conjunct, where it is not interpreted collectively. (ix) [Thati [star-crossed [[ti NP] and [thatj NP]]]] (x) *[[Thati [star-crossed [ti NP]]] and [thatj [ti NP]]] In chapter 3, pronounced subjects occupy higher position than their unpronounced counterparts and move asymmetrically out of first conjuncts to reach higher position, following Johnson’s ([1996] 2003, 2002) pioneering work on asymmetric movement. Contrasting
936
Notes
examples similar to (xi) and (xii), Heycock and Zamparelli (2003, 2005) argue that (xii) coordinates DPs equivalent to the necessarily false (xiii): (xi)
T In that cabinet, most of the cups and saucers are cups (the saucers having been decimated by loss or breakage). (xii) F In that cabinet, most cups and saucers are cups (the saucers having been decimated by loss or breakage). (xiii) F In that cabinet, most cups and most saucers are cups (the saucers having been decimated by loss or breakage). 23. See caveat, note 10. 24. Or a variant could say that W-ings are always counted and sometimes further described as ill-matchings, etc. 25. Here there is no dissent from the presumption that the (semi)distributivity of the DPs in (199) derives from a determiner many that is a quantifier. The conclusion of chapter 13 is rather that many fielders and some many fielders are both plural indefinite descriptions containing an arithmetic predicate many, differing only in that the former addresses plural scenes or frames of reference Σ and the latter a singular scene or frame of reference σ: (i) a. [∃X: ∃E∃σ(some[E] there[σ, E] many[E] fielders[E,X])] b. [∃X: ∃E∃Σ( there[Σ,E] many[E]) fielders[E,X])] The same holds mutatis mutandis for many a one or more fielders and some many a one or more fielders. 26. And depart from definite descriptions in a few contexts: (i) The elms are dense in the middle of the forest. (ii) *Some elms are dense in the middle of the forest. *One or more elms are dense in the middle of the forest. *Many a one or more elms are dense in the middle of the forest. 27. See (233)–(237) in section 12.3.2 for comment on how the variable of quantification may disagree in number with the observed number agreement. Compare also the treatment of the semidistributive operator in Schein 1993, chap. 8, and the treatment in section 2.6 of the explicit comitative constructions that affect number agreement in languages such as Spanish. 28. Similarly, similarity cannot hold of an event with only one participant: (i) Many mobile phones are alike. (ii) *Many a mobile phone is alike. 29. I am concerned here to contrast the distributive plural quantifiers many, most, any with their singular plural counterparts many a, most a, any n. Yet I will remark without explanation that the plural quantifiers in (i)–(iv) differ among themselves in the semidistributive interpretations under discussion: (i) (ii) (iii) (iv)
In Mobile, many mobile phones communicate with each and only each other. In Mobile, most mobile phones communicate with each other and only each other. In Mobile, any mobile phones communicate with each other and only each other. In Mobile, all mobile phones communicate with each other and only each other.
Notes
937
Suppose the many mobile phones of Mobile are divided among seven small networks. Sentence (i) remains true in that in Mobile, many mobile phones are in its networks. Yet it fails to rescue the false (ii)–(iv) that all its mobile phones are in networks. 30. Or [Many X : Ψ[X]] Φ[X] ↔ [Many Y : sg[Y] & [∀y: Yy]∃X(Xy &Ψ[X])][∀y: Yy]∃X(Xy &Φ[X]) 31. Or [Many Y : sg[Y] & [∀y: Yy][∃X: y = {X}] natural numbers[X])][∀y: Yy][∃X: y = {X}] (pl[X] & Φ[X]) 32. Here, for convenience, I assume that sg means “one” and pl, “one or more,” so that ‘sg[X]’ and ‘pl[X]’ are not inconsistent. 33. To be understood as exactly three times as many, so that the sentence is false if the pronoun is taken to refer to the kind line-segments or to all the segments on the canvas rather than to only the triangle-forming ones. 34. Cf. (i) #Many line segments form a triangle. When they do, they are three times as many as the triangle. (ii) Many line segments form a triangle. When they do, the meet at three vertices. Apart from the aspectual difference between (i) and (ii), the predicates also differ in that be three times as many … is antisemidistributive but meet … is not. I have yet to sort out the conditions on telescoping, nor why the singular plural is exempt from them. 35. In truth, a/an (event of) measurement eμ, a protocol for it π(eμ), its frame of reference f(eμ), and the scene of it σ(eμ), should perhaps all be separate parameters, which I abbreviate as the solitary index eμ, leaving it to context to disambiguate the intended parameter. 13
Antisemidistributivity vs. Conjunction Reduction Redux
1. For illustration, the examples to follow assume (7) true. But imagine a 4-minute track by a solo performer, the first half of which is solo piano and the second half solo voice. This 4-minute event is as single an event as any of music e by single participant x, and yet e is neither a 4-minute instrumental event nor a 4-minute vocal. For the sake of argument, assume either that the musical tradition proscribes performers from alternating between instrumentals and vocals in a single performance or that ‘instrumentalist(e,x)’ means that x has an instrumental part in e (and similarly for ‘vocalist(e,x)’). 2. Also known as genuine collectives. See Dowty 1987; Higginbotham and Schein 1989, 168ff.; Taub 1989; Brisson 1997, 1998, 2003; Winter 1998, 2002; also Hackl 2001a, 2001b, 151–186, as well as 2002. 3. Like many musicians, all is acceptable with collective predicates that are not antisemidistributive: (i) The musicians all pair up in dynamic duos. (ii) The musicians are all paired up in a dynamic duo. (iii) The musicians are all in dynamic duos.
938
Notes
4. Substitution of a coordinate NP does not repair the nondistributive collective interpretation, and so the contrast between bare and overt article survives: (i) #Innumerable stars and black holes light-years away are that galaxy at the edge of the universe. (ii) Some innumerable stars and black holes light-years away are that galaxy at the edge of the universe. Rather, substitution of a coordinate NP rescues only the semidistributive interpretation distributing stars and black holes among the galaxies: (iii) Innumerable stars and black holes light-years away are a galaxy at the edge of the universe. Innumerable stars and black holes light-years away are galaxies at the edge of the universe. (iv) Many stars and black holes light-years away are a galaxy at the edge of the universe. Many stars and black holes light-years away are galaxies at the edge of the universe. The rejected bare plurals presume a neutral intonation, without focus on the bare plural. Otherwise, (v) is construed in logical form (see Herburger 1997, 2000) so that the bare plural is in predicative position, as it were, where its null article is anaphoric to scenes or frame of reference antecedently referred to (see section 11.1.1): (v) Innumerable STARS light-years away are that galaxy at the edge of the universe. ‘That galaxy at the edge of the universe is innumerable stars light-years away.’ 5. Recall from the discussion of (154) in section 11.0.1 that the lexical ambiguity of airlift operation and cold war delivery service illustrates how concepts may differ in the conditions they impose on the constituents taken to fall within the single frame of reference for which they are defined: (154) (The) 3000 aircraft that landed at Tempelhof in 1948 were an airlift operation. (The) 3000 aircraft that landed at Tempelhof in 1948 were a cold war delivery service. If airlift operation (or cold war delivery service) is understood to denote an (ongoing) process, an airlift operation of 3000 aircraft may count 3000 discrete events of being a (landing) aircraft that are merely temporally separated within the same spatiotemporal frame of reference within which the airlift operation and its constituent events are all located. If, on the other hand, airlift operation or cold war delivery service refers to a standing company or squadron, its constituent aircraft, to amount to a full-fledged operation or delivery service at any time, must consist of concurrent and coordinated events of being so many aircraft. There are 3000 of these only if they are spatially separated, corresponding, that is, to object-counting 3000 aircraft lifetimes. 6. Adverbialization and further details are suppressed. For purposes of illustration, the obvious intervention suffices. But existential quantification over pluralities is mistaken for reasons mentioned below, demanding an alternative semidistributive operator such as the one proposed in Schein 1993, chap. 8.
Notes
939
7. Also known as genuine collectives. See Dowty 1987; Higginbotham and Schein 1989, 168ff.; Taub 1989; Brisson 1997, 1998, 2003; Winter 1998, 2002; also Hackl 2001a, 2001b, 151–186, and 2002. 8. The collective events quantified over in the semidistributive interpretation meet spatiotemporal conditions at least resembling those needed to count items rather than measure a mass (see section 13.2). For further discussion see Brisson 1997, 1998, 2003; Dowty 1987; Gillon 1990; Schein 1993, 302f., 320–322n16, and 2006, 752ff.; Schwarzschild 1991, 1996; Taub 1989. More examples of the crucial contrast: *Many musicians are an ensemble [i.e.,*unless implying that there are many ensembles of one]. Many a one or more musicians are an ensemble. Some/the musicians are an ensemble. *Many machine parts are a machine [i.e., *unless implying that many machines have only one part]. Many a one or more machine parts are a machine. Some/the machine parts are a machine. *Many atoms are a molecule [i.e., *unless implying that many molecules are of only one atom]. Many a one or more atoms are a molecule. Some/the atoms are a molecule. *Many little white pickets are a little white fence around a little white cottage. Many a one or more little white pickets are a little white fence around a little white cottage. Some/the little white pickets are a little white fence around a little white cottage. 9. Winter (1998, 2002) remarks in particular on the contrast between complex cardinals such as more than three and simple three, unexpected when so similar: (i) Three elms are a cluster in the middle of the forest. (ii) Three elms that are a cluster cluster in the middle of the forest. (iii) #More than three elms are a cluster in the middle of the forest. (iv) #More than three elms that are a cluster cluster in the middle of the forest. The contrast finds explanation in Hackl (2001a, 2001b), who abandons the pretense that more-than-3 is a primitive expression in the logical syntax, either a quantifier like all or cardinal predicate like three. Hackl resolves the internal structure of more than three in the course of reducing (1)–(4) to the same construction. Providing a uniform syntax and semantics for more … than, its occurrence in all the comparative quantifiers of (1)–(4) is assimilated to comparative constructions in general, including (5): (1) John bought more than three books. (2) John bought more books than papers. (3) John bought more books than Bill (did/bought). (4) John bought more books than there are planets in the solar system. (5) John is more bookish than Bill (is).
940
Notes
Canonical translations for (1), (2), and (4) turn out to be something like (1′), (2′), and (4′): (1′) To a greater degreei than that if John buy 3-many books John (in fact) bought thati-many books. (2′) To a greater degreei than thatj if John buy soj-many papers John bought thati many books. (4′) To a greater degreei than that to which there are planets in the solar system John bought thati many books. The talk about degrees in (1′)–(4′) purchases a common measure for books (1)–(4) and bookishness (5). As in these paraphrases, a comparative quantifier, more than three but not three, translates into two quantifiers, one over degrees and one over so-many or that-many books. Unpacking the degree quantifier, the comparative phrase than … proves always to be a clause even if much of it is unspoken as in (1)–(3). (A difference remains between spoken clauses as in (4) and unspoken in that the latter for reasons both syntactic and semantic are always tenseless (prompting my use of the subjunctive in (1′) and (2′)). Pertaining to Winter’s (1998, 2002) observed contrast between three and more than three, the comparative conceals a relation, many to degrees, appearing as indicated in the paraphrases (1′)–(4′), 3-many, that-many, so-many. This many is taken to be the same as the one pronounced in (87)–(90) and thus the comparative quantifier and many show the same antisemidistributivity. Absent many, simple three is not antisemidistributive. The analysis extends to most (Hackl 2003, 2009), according to which (6) decomposes into (7), a comparative that again conceals the relation to degrees, many, from which it too inherits antisemidistributivity: (6) Most books are paperback. (7) More books are paperback than not. In the text above, example sentences are given for most, more … than, and many, but logical forms only for many. Given Hackl’s reduction, what is said about many carries over to the comparative quantifiers, and I would rather not clutter logical form any more with the comparative quantification of degrees. The many spoken in many elms and concealed elsewhere, Hackl (2001a, 2001b) assumes to be a determiner (relativized to degrees) that is intrinsically distributive with respect to both its restriction and nucleus, from which it follows that many, spoken and concealed, is antisemidistributive in both arguments. For now, I will instead cling to familiar habit in assuming that many is a vague cardinality predicate with the same logical syntax as, say, three, representing (semi)distributivity by a separate operator quantifying over that which is many. I return in section 13.1 to the syntax of many. 10. Semidistributivity (Schein 1993, chap. 8) is yet further argument for a logical syntax with separate, independent thematic relations occurring as such as the only content of these descriptive anaphora. The pronoun they in (101) is an early example of a pronoun that must be descriptive despite its occurrence within the scope of its antecedent no vaudevillians. 11. The event quantification intrudes ad hoc following Schein 1993 and most approaches that do not view semidistributivity as basic. Note however that the term referring to events in (103) resembles the event pronoun that has been said to translate number agreement (see
Notes
941
also section 12.3.0). Let us then identify the term in (103) as number agreement and propose that ⌜many fielders Φ⌝ always translates in effect as “many a fielder is such that what he and others who together are fielders did Φ.” That is, the plural NP fielders always induces as an immediate result of number agreement the logical form directly reflecting semidistributivity. In contrast, the translation of a singular NP ⌜every fielder Φ⌝ as “every fielder is such that what he and others who together are a fielder did Φ” rules out the semidistributive a priori. Having declared the semidistributive as basic logical form for many fielders, the fully distributive interpretation is obtained under three different circumstances: (i) it arises that the only events satisfying the conditions on count event quantification (section 13.0.1) are those with a solitary participant, (ii) it is analytic that the given predicate is itself true of several if and only if true of each (e.g., many fielders are fast), and (iii) the speaker intends with a tacit restriction that the events quantified over are only those with solitary participants. This last is deployed in representing the ambiguity of Many boys lifted the piano. Note that even though many boys is singular, distributive quantification in its force, it is now the fully distributive interpretation of lift the piano that needs supplement to be derived from the semidistributive. 12. There is an unarticulated assumption in play, namely, that whatever is to be said about the subject’s participation in events, what vaudevillians did, is to be achieved by a term that refers to these events and occurs adjacent to the subject no vaudevillians rather than occurring at a distance from the subject, say, within the scope of the direct object no more than three ballads. This particular assumption is in fact necessary given a logical syntax that respects separation as discussed in Schein 1993, 139ff. A more general (1993, 150ff.) result pertinent here is that a simple clause tokens no more than one existential event quantifier, with all others being definite descriptions. 13. The exposition in the text develops the argument as if many students and few students were first-order distributive quantifiers, ‘[Many x : ∃X∃E(Xx & students[E,X])]’ and ‘[Few x : ∃X∃E(Xx & students[E,X])]’. But, as anticipated earlier, many students and few students are themselves plural indefinite descriptions. The contrast below between few students and no students or not any students must instead contrast the former’s reference to plural frames of reference and the latter’s to the singular: (i) a. ¬[∃X: ∃E∃σ(there[σ, E] any[E] students[E,X])] not any students, no students b. ¬[∃X: ∃E∃Σ(there[Σ,E] many[E] students[E,X])] not many students, few students 14. A paraphrase more faithful to (122)’s actual logical form goes: (i) Few(er than two) thieves who never planned (their heists) while being thieves who never planned, having never planned, (have) gathered (up) / collected / rounded up / reaped no more than five gemstones. (i) is equivalent to its abbreviation in (122). The first adverbial phrase in (i) corresponds to the adverbialization of the entire NP restricting the quantifier. The second adverbialization reflects the interpretation of the matrix’s tense anaphoric to tense in the antecedent relative clause.
942
Notes
15. Similarly, the truth of (113) in (112) is made false in (i) with a relative clause that is almost vacuous in its contribution to the sentence: (113) Few students gathered only once. (i) Few students who gathered gathered only once. Few students who gathered, having gathered, gathered only once. (ii) Few students gathered ONLY ONCE. With focus (ii), (113) itself can be coerced into expressing the equivalent of (i), but according to Herburger (1997, 2000), focus affects logical form, in effect transforming (ii) into (i). 16. If a tense relative clause is adverbialized, then the expectation is that it relate to the matrix as tensed adverbial clauses do. If an untensed NP, thieves, is adverbialized, the range of interpretations it shows modifying the matrix should coincide with those of untensed modifiers, while thieves, being thieves, etc. 17. Cf. (i)
Every thief emptied a jewelry store, and the jewelry store was one of many on the 700 block of Fifth Avenue. (ii) Every thief emptied a jewelry store, and the jewelry stores filled the 700 block of Fifth Avenue. (iii) Every thief emptied a jewelry store, and the jewelry stores were each one of the many on the 700 block of Fifth Avenue. 18. There is no principle favoring a defined operator rather than the expansion of logical form itself in a way that parallels the definition, except to spare the reader the longer logical forms and to facilitate comparison of this semidistributive operator to the one in Schein 1993, 157f., 311. 19. See note 13. There is yet less to tell apart the logical forms for (136) and (137) than appears for now in (138) and (139), as both many passengers and some many passengers are plural indefinite descriptions, ‘[∃X: ∃E∃Σ(there[Σ,E] many[E] students[E,X])]’ and ‘[Some X: ∃E∃σ(there[σ, E] many[E] students[E,X])]’, respectively, differing in whether plural or singular frames of reference are addressed. Continuing, however, to humor the readership for whom many passengers is a distributive, first-order quantifier, the logical form in (138) stumbles over the remark that the events of passengering are counted many. Perhaps then (138) should read ‘[Many e : ∃E(Ee & ∃X passengers[E,X])] …’ The problem here is that what is counted, always events, need not be what a DP refers to or quantifies over, as sections 12.1–12.2 argue. The representation ⌜[Many v : Φ[v]]⌝ denies them their divorce. 20. But suppose that 3000 passengers are 10 frequent fliers who ordered champagne only on their maiden voyages, so that (i) is false: (i) Every passenger ordered champagne. Full distributivity may be only over the 10, but for any x, ordering champagne during the events of x being a passenger, [℩E0: passenger.sg[E0, x]], is champagne on thirty flights, falsifying (i), as desired. 21. Thanks to Pranav Anand (p.c., October 2009) and Roger Schwarzschild (p.c., July–October 2009) for discussion.
Notes
943
22. Arguing that many is a determiner, Hackl (2001b, 71f.) rejects that it is predicative, citing its distributional contrast with tall: (i) John looks tall. (ii) *The guests look many. *The Yankees fans looked fewer than 200. (iii) Mary considers the guests tall. (iv) *Mary considers the guests many. (v) Mary considered the Red Sox fans more numerous than the Yankees fans. (vi) *Mary considered the Red Sox more than the Yankees fans. If so, then it is also to be rejected that twelve is predicative: (vii) *The Apostles look twelve. (viii) *Peter considers the Apostles twelve. Some difference of syntactic and semantic classification between tall and twelve can be conceded without denying that twelve is predicative, especially as there is not much of an alternative for the underlying concept of number; similarly for many. NPs and APs share logical type too without identical distribution. 23. Internal to Hackl’s (2001a, 2001b, 2009) analysis of comparative quantification, which is assumed here (see note 9), there is further argument that the morpheme many is not itself distributive. If it is tokened in the than-clauses (… than [d-many NP] …) as the analysis proposes and it is distributive, then than-clauses should be antisemidistributive: (i)
?More elms are randomly scattered throughout the park than are a cluster that Olmsted & Vaux planted. (ii) *More elms are a cluster that Olmsted & Vaux planted than are randomly scattered throughout the park. (iii) ?More lovers’ souls burn bright on Earth than are as many as the stars in heaven. (iv) *More lovers’ souls are as many as the stars in heaven than burn bright on Earth. As clumsy as tensed, unreduced than-clauses are in the first place, it is much easier for them to contain a true collective predicate (i, iii) than for the matrix. Sentences (ii) and (iv) are unremittingly incoherent in asserting, respectively, that the individual elm is a cluster and that the individual soul is as many as the stars. If antisemidistributivity is rather the effect of an unspoken operator analogous to each, (i)–(iv) in effect assimilate to (i′)–(iv′), where the than-clauses are unaffected by the presence of each: (i′)
Each of more elms than are a cluster that Olmsted & Vaux planted are randomly scattered throughout the park. (ii′) *Each of more elms than are randomly scattered throughout the park are a cluster that Olmsted & Vaux planted. (iii′) Each of more lovers’ souls than be/are as many as the stars in heaven burn bright on Earth. (iv′) *Each of more lovers’ souls than burn bright on Earth are as many as the stars in heaven. Arguing the contrary that many’s antisemidistributivity is not the effect of each or any other quantifier, Hackl (2001b, 246) points to generic contexts (29b, 30a) that suspend
944
Notes
antidistributivity for many and its allied comparative quantifiers but not for most or any other quantifier (29c), thus dissociating the antisemidistributivity of the one from the others’: (29) a. #More than three students constituted a majority / weigh 800 pounds / be a team that participates in the race. b. More than three students can constitute a majority / weigh 800 pounds / be a team that participates in the race. c. #All the / no / none of the / most of the students can constitute a majority / weigh 800 pounds / be a team that participates in the race. (30) a. More students than professors can constitute a majority / weigh 800 pounds / be a team that participates in the race. b. More students than John had expected can constitute a majority / weigh 800 pounds / be a team that participates in the race. The absence of number agreement in (29b) and (30) masks the possibility that the subject is a tacit counterpart to Having more than three students … , There being more than three students, denoting a singular situation or perhaps even a number: (v)
More than three students constitutes a majority. More than three students weighs at least 800 pounds. More than three students is a team eligible to participate in the race.
(vi) *More than three students constitute a majority. *More than three students weigh at least 800 pounds. *More than three students are a team eligible to participate in the race. (vii) Three students constitute(s) a majority. Three students weigh(s) at least 800 pounds. Three students are/is a team eligible to participate in the race. Strong quantifiers cannot be embedded in phrases denoting a singular situation, as if in a context for the definiteness effect: (viii) *All/the/most (of the) students constitutes a majority. *All/the/most (of the) students weighs 800 pounds. *All/the/most (of the) students is a team eligible to participate in the race. What is at issue is directly reflected in the scope of a weak, decreasing quantifier: (ix) Fewer than twelve guilty verdicts constitutes a hung jury. Fewer than twelve guilty verdicts is going to constitute a hung jury. (x) *Fewer than twelve guilty verdicts constitute a hung jury. *Fewer than twelve guilty verdicts are going to constitute a hung jury. It’s the having fewer than twelve guilty verdicts that ends in mistrial, with scope of fewer than twelve guilty verdicts confined to the description of a singular situation, (ix). Embedding within such a description is an island for extraposition (xii), unlike elsewhere (xiii): (xi)
Fewer guilty verdicts than not-guilty verdicts constitutes a hung jury. Fewer guilty verdicts than not-guilty verdicts is going to constitute a hung jury.
(xii) *Fewer guilty verdicts constitutes a hung jury than not-guilty verdicts. *Fewer guilty verdicts is going to constitute an hung jury than not-guilty verdicts.
Notes
945
(xiii) Fewer guilty verdicts than not-guilty verdicts are in an acquittal. Fewer guilty verdicts are in an acquittal than not-guilty verdicts. Crnič (2009) proposes a formally similar analysis, according to which many and the comparative quantifiers are embedded in the restriction to an adverb of quantification, a silent universal in the case of generic statements, as in (8): (1) c. More than seven students are a good team. c′. More than seven students are usually a good team. (8) Always/Usually when there is something that is more than seven students, there is a good team. The proposals disagree on number agreement. Adverbs of quantification, while rearranging the logical syntax, do not disturb the surface where the subject remains just the DP, more than seven students, resulting in plural number agreement. To my ear, the sentences in (1) are unacceptable, requiring singular number agreement, as expected if the phrase properly containing more than seven students, a description of a situation, is the subject proper. But either way, the proposals agree in removing the offending predicate from within the scope of the distributive quantifier. Hackl’s contrast between many, more … than and all, the, most, etc. in certain contexts reflects just that many, more … than can in these contexts occur in alternative structures that all, the, most, etc. cannot, without it indicating that the distributivity of many is intrinsic to its lexical meaning rather than the effect of an associated tacit each. Crnič’s (2009) proposal does not readily extend to the weak, decreasing quantifiers, (ix) and (xi). Suppose it has always been that every grandparent has many grandchildren to brag about. Sentence (xiv) is then unequivocally false, including that interpretation in which a few grandparents occurs within the restriction of usually, (xv). Sentence (xvi), on the other hand, is unequivocally true. (xiv) Usually a few grandparents have only a few grandchildren to brag about. (xv) Usually when there are a few grandparents, they have only a few grandchildren to brag about. (xvi) Usually few grandparents have only a few grandchildren to brag about. (xvii) *Usually when there are few grandparents, they have only a few grandchildren to brag about. Yet, if the decreasing few grandparents were felicitous in the restriction to usually, the sentence should also have a false interpretation, as (xvii) is hardly different from (xv). Since decreasing quantifiers are infelicitous in the restriction to an adverb of quantification, some other construction must enable antisemidistributivity to be suspended in (ix) and (xi), such as the one first proposed. Crnič’s (2009) proposal, however, fares just as well in blocking extraposition in that there appears to be no extraposition from within the restriction to an adverb of quantification, taking care to construct examples that favor construing the host comparative within the restriction to the adverb: (xviii) Usually more grandparents than grandchildren in a recreation center play more shuffleboard than table tennis. (xix) *Usually more grandparents in a recreation center play more shuffleboard than table tennis than grandchildren.
946
Notes
24. Counting in the present, the sentences say that baseball legends by the hundreds, or hundreds at a time, are turning out to be Jews, with measurement in progress and also without implying that being Jews is in progress. As is well known (Dowty 1977), progressive aspect on VP shows a similar effect. In many a neutral context, it is infelicitous to say “Interstate 80 is leading to the George Washington Bridge,” the naked facts about road and bridge failing to satisfy the conditions for progressive aspect. Yet the sentence becomes felicitous as I am driving east toward home, not because of any change in the objective condition of the participants, but because the surrounding conditions of observation meet the requirements for progressive aspect. It is also enough for my finger to be tracing out Interstate 80 on a map, or to do so with the mind’s eye, wherever I may be. 25. Note that in many untallied baseball the missing or failed tally is again a third-party incident: (i) Many untallied baseball legends by the hundreds were/are Jews. (ii) Many untallied baseball legends hundreds at a time were/are Jews. Sentence (ii) means that many, hundreds at a time, among the previously untallied baseball legends were/are Jews. I assume that modifiers at the left edge, many here, preempt indexical reference to events (see section 12.1.0) so that the speaker’s counting is of many, which is not a leaving untallied. 26. See section 9.4.1 (190)–(194) and the discussion. Recall that the cinerama now in progress comprising Sɶ may be scenes from memory or in reprise with locations for the events E they are scenes of fixed at the scenes Sɶ of their original projection. Whether to introduce new vocabulary such as passim or instead to introduce its definition as itself a phrase in logical form is an arbitrary guess at compromise between a logical form that is brief and more or less transparent to the surface syntax and one that is explicit in its logical and analytic commitments. Analogous arbitrary choices are made throughout. 27. I can see no harm in explicit indexical reference to the current scene, but I have not thought through an advantage for it: (i) [∃X: ∃E (∃F passim[E,F] now-en-scène[E] elms[E,X])] (see section 9.4.1, (178)–(180)) (ii) [∃X: ∃E (∃F some[E,F] now-en-scène[E] elms[E,X])] One consideration (section 11.1.0) is that exemption of event counting from the anticonvergence condition when epistemic necessity warrants the assertion, as in (iii) in contrast to (iv), derives from the warranted omission of ‘now-en-scène’ from such assertions. (iii) (The) 30,000 passports that passed through Passport Control at JFK in January 1995 are expired by now (in accordance with international law). (iv) #(The) 30,000 passports that passed through Passport Control at JFK in January 1995 have been renewed this year by their bearers. As this concerns omissions of ‘now-en-scène’ outside DP and later in the sentence, I do not see that it should have any bearing on reference to the current scene within the AdrP. 28. The further conditions against convergent histories remain an effect of the surrounding clausal structure (see section 11.1.0). 29. Counting by sight the parallel bamboo reeds of a bamboo curtain is a function (among other things) of visual distance from the curtain, the thickness of the reeds, and the spacing
Notes
947
between them (see Pylyshyn 2001). Or imagine counting a dense scatter of points using transparent graph paper that is constantly slipping. 30. As Koslicki (1997, 1999, 2005) remarks, so-called count nouns are felicitous even under conditions hostile to counting, where it is indeterminate what to count. Nevertheless, this does not impede many: (i)
The branches on that tree steal each other’s sunlight and water and should be pruned back. (ii) Some waves have overtaken each other before reaching shore. (iii) ???613 branches on that tree steal each other’s sunlight and water and should be pruned back. (iv) ???613 waves have overtaken each other before reaching shore. (v) Many branches on that tree steal each other’s sunlight and water and should be pruned back. (vi) Many waves have overtaken each other before reaching shore. If all there is to many is some vagueness about number—how many is many—it is unclear why the conditions for 613 should be more stringent than for many. It is unclear why the indeterminate individuation of X, branches or waves, should defeat one and not the other: 613(X) ↔ card(X, 613) many(X) ↔ ∃k (many(n) & n ≤ k & card(X, k)) 31. Below however are cited only examples of nuclear antisemidistributivity. 32. So, the answer to the question why antisemidistributivity attaches to many and not to twelve is that the former presupposes many frames of reference and the latter one (per perfective event of measurement). Recall from section 9.4.2.0 that the cardinal number may be the summation of several determinate, perfective measurements if so understood from context). This itself is of course a stipulation, but one that correlates with a felt difference of meaning opposing vagueness and exactness, here formalized in terms of frames of reference. Given the observation in note 30, something along these lines is needed anyway. 33. As expected, too, analogous contrasts occur with bare plurals. In referring to multiple frames of reference with a temporal distribution, one can be confident that they frame equally numerous short events of NP-ing and thus confident of a judgment about the temporal distribution of these short events. The multiple frames of reference obstruct getting a fix on whether the participants in these events are the same or on how many they are and thus confound any judgment about their spatial distribution. This contrast is reflected in the use of antisemidistributive dense to describe temporal and spatial distributions: (i) (ii)
*Stars are dense in the Milky Way. The stars are dense in the Milky Way.
(iii) (iv) (v)
*Shooting stars are dense in the Milky Way. Shooting stars are dense in a meteor shower. Shooting stars are dense in August and November.
(vi) *Spikes/surges are dense in a regional power grid. (vii) The spikes/surges are dense in a regional power grid. (viii) Spikes/surges are dense in a brownout.
948
(ix) (x) (xi) (xii)
Notes
*Rockets are dense in Gaza. *Rockets are dense in a rocket assault. *Rocket bursts/launches are dense in Gaza. Rocket bursts/launches are dense in a rocket assault.
(xiii) *Hailstones are dense in an ice storm. (xiv) *Hail strikes are dense in an open field. (xv) Hail strikes are dense in an ice storm. (xvi) *Raindrops are dense in a cloudburst. (xvii) *Sparkles are dense in a burst of fireworks. (xviii) *Earthquakes are dense in a fault zone. (xix) Aftershocks/seismic waves are dense in an earthquake. (xx) (xxi)
Potholes were dense on the Cross Bronx Expressway. Potholes were frequent on the Cross Bronx Expressway.
Schein (1993, 366–367n8) notes the following contrasts: (xxii) *Houses are dense in the middle of the Bronx. (xxiii) The houses are dense in the middle of the Bronx. (xxiv) Accidents were frequent in the Bronx. Crimes were frequent in the Bronx. (xxv) *The accidents were frequent in the Bronx. *The crimes were frequent in the Bronx. The contrast between (xxii) and (xxiii) is as above. That between (xxiv) and (xxv) perhaps reflects a conflict between the perfective nominal aspect imposed by the in (xxv) with the temporal dilation of being frequent, which requires the imperfective nominal aspect of the bare plural in (xxiv). See the discussion of (199)–(222) and (303)–(308). 34. Not to say we don’t have—Moe Berg, Hank Greenberg, Sandy Koufax, and more. 35. When the Determinerator applies to a relation such as many that sets its own minimum, the result is a nonincreasing quantifier, “the Fs that G are many” for Many F G. 36. See section 9.4.2. ® [ s, ζ , ξ , s ] ↔ df ∀α s (Asα s → Rsα s ) & ∀α s ∀α s ′((Asα s & Asα s ′ & overlap(α s, α s ′)) → α s = α s ′) & ∀α s ∀α s ′((sα s & sα s ′ & At[α s,ζ , t( s), s] & At[α s ′,ζ , t( s), s]) → α s = α s ′) & ∃α s (sα s & At[α s,ζ , t( s), s] & ∀x(ξx → ¬∃y(overlap(x,y) & At(αs,y,t(s),s))) & ¬∃p(At(p,p,t(s), s) & Limit point(p, ζ, s) & Limit point(p, ξ, s)) 37. Explicit restriction, in contrast to (303) uttered out of the blue, allows the natural numbers to be revisited as in any event counting, subject to the conditions on sequencing of events: (i) Most natural numbers sampled were prime. (ii) Most natural numbers sampled are prime. Sentence (i) reports the result of sampling, which may very well have seen the same natural numbers sampled more than once. Note that the determination that a sampled number is prime is taken to be contemporaneous with its sampling. If it is allowed that there may have been resampling, sentence (ii) is acceptable only as an inference from (i) and that, if deter-
Notes
949
mined to be prime, a natural number is always prime. Alternatively, (ii) may be parsed as in (iii), which are true only under very biased sampling: (iii) Sampled, most natural numbers are prime. Most natural numbers, if sampled, are prime. 38. Not the same as the landmarks of sections 9.3–9.4. Perhaps those here should be renamed beacons or waypoints. 39. τ()l ↔df ∃l′(l ′ & l = τ(l′)) 40. ℜ(f, ζ, f ) ↔df ∃ξ(∀x∀y((ξx & ξy & x ≠ y) → ®(f, x, y, f ]) & ∀x(ζx → ∃y(ξy & overlap(x,y) & ∀z((ζz & overlap(z,y)) → z = x))) & ∀x(ξx → ∃y(ζy & overlap(x,y) & ∀z((ξz & overlap(z,y)) → z = x))) ℜ[ f , ζ, f ] ↔ df ℜ( f , ζ, f ) & ∀ f ′((ℜ( f , ζ, f ′) & ∀α f (f α f → f ′α f )) → f ′ = f ) 41. I have not said much about the meaning of among, which I assume provides explicit reference to the space of landmarks when it occurs. Note (Schein 2005, 57–58n13) that (i) with among is ambiguous between a truth of 1853 reporting Burton’s pilgrimage in disguise to Mecca and a falsehood that he has converted to become one of the faithful: (i) Sir Richard Francis Burton is among the faithful. If it is a persistent feature of the world’s languages to confound these two senses in a single lexical item, one might hope for a univocal meaning. Perhaps the variation is just in whether the faithful are the landmarks for a discrete space or for a continuous one interpolated from the landmarks. If it is discrete, then to be located at a landmark implies identity with it— hence, the usage of among that resembles is one of. If other points fill in the space around the landmarks, then of course there is no implication that the subject located is one of them. Contexts that do not make salient an interpolated space result in minimal contrasts such as between (ii) and (iii), which are alleviated as soon as one is provided as in (iv): (ii) The tsunami of December 26, 2004, is among the great natural disasters. The tsunami of December 26, 2004, is among several great natural disasters. (iii) *The tsunami of December 26, 2004, is among the other great natural disasters. *The tsunami of December 26, 2004, is among several other great natural disasters. (iv) The tsunami of December 26, 2004, is among the other great natural disasters in the (recent) history of the Indian Ocean. The tsunami of December 26, 2004, is among several other great natural disasters in the (recent) history of the Indian Ocean. Notice that the metaphor of navigation by beacons in a landscape shrouded in night and fog is an example of navigation through continuous space only by reference to discrete landmarks. Positions and distances within that space can only be approximated—“two beacons north and three beacons east and then some before the next beacon.” 42. Also known as genuine collectives. See Dowty 1987; Higginbotham and Schein 1989, 168ff.; Taub 1989; Brisson 1997, 1998, 2003; Winter 1998, 2002; Hackl 2001a, 2001b, 151–186, 2002.
950
Notes
43. More examples: (i)
61,300 trees were a perfect circle along the rim of a volcano. 61,300 trees were perfect circles along the rims of volcanoes.
(ii)
The many trees were a perfect circle along the rim of a volcano. The many trees were perfect circles along the rims of volcanoes.
(iii)
Some trees were a perfect circle along the rim of a volcano. Some trees were perfect circles along the rims of volcanoes.
(iv)
#Many trees were a perfect circle along the rim of a volcano. #Many trees were perfect circles along the rims of volcanoes.
(v)
#Trees were a perfect circle along the rim of a volcano. #Trees were perfect circles along the rims of volcanoes.
(vi)
Cf. Many trees formed a perfect circle along the rim of a volcano. Many trees formed perfect circles along the rims of volcanoes. Trees formed a perfect circle along the rim of a volcano. Trees formed perfect circles along the rim of a volcano.
(vii) Thirty passengers were a secret quality control team. Thirty passengers were secret quality control teams. (viii) The many passengers were a secret quality control team. The many passengers were secret quality control teams. (ix)
Some passengers were a secret quality control team. Some passengers were secret quality control teams.
(x)
#Many passengers were a secret quality control team. #Many passengers were secret quality control teams.
(xi)
#Passengers were a secret quality control team. #Passengers were secret quality control teams.
(xii) Cf. Passengers passed secret notes / were secretive / furtive. Passengers were secret quality control team members. 44. It is unclear to me whether (i) can also be a felicitous report that many vegetables were among trials where the vegetables weighed turned out to be too heavy for the laboratory scale or too heavy for the bathroom scale. Similarly, (ii) may not be a felicitous report that many vegetables were among trials where the vegetables weighed 1 kilogram. (i) Many vegetables were too heavy for the laboratory scale and too light for the bathroom scale. (ii) Many vegetables weighed 1 kilogram. If not, countable events are a necessary but insufficient condition for semidistributivity.
Notes
951
45. More examples where antisemidistributive predication is rescued by an NP or relative clause that can be construed to denote discrete countable events (sometimes requiring an assumption that only maximal events as described are denoted): (1) (2)
#Most wives are no more than one. Most wives of the same aspiring polygamist are no more than one.
(3) (4)
#Most children are two. Most children of the same parents are two.
(5) (6)
#Most loved ones are only two. Most lovers are only two.
(7) (8)
Children trust indiscriminately. Many adults trust only themselves. Children trust indiscriminately. Many adult relatives trust only themselves.
(9) Most barflies order drinks only for themselves. (10) Most barflies who sit together order drinks only for themselves. (11) #Most barflies are only two. (12) Most barflies who sit together are only two. (13) #Most pedestrians are one or (even) more. (14) Most pedestrians at a busy red light are one or (even) more. (15) Most lottery winners are only one. (16) Most lottery winners are one or more or maybe none. (17) #Most campers at that campfire were a scout troop. #Most campers at that campfire have been a scout troop. (18) Most campers at a campfire were a scout troop. Most campers at a campfire have been a scout troop. (19) #Many gals are a bridal party. (20) Many gals in the same dress are a bridal party. (21) #Most guys are a groomsmen brigade. (22) Most guys in the same tuxedo are a groomsmen brigade. 46. Taub (1989) and Brisson (1997, 1998, 2003) have the good sense not to be gulled by the red herring of logical type, emphasizing the role of Aktionsart in characterizing the contrast between (i) and the antisemidistributive (ii): (i)
The senators all cosponsored the amendment. Many/Most senators cosponsored the amendment.
(ii) #The senators all passed the amendment. (iii) #Many/Most senators passed the amendment. In demonstrating the role of Aktionsart, Taub (1989) expands Dowty’s (1987) original list of antisemidistributives, predicates of exclusively geometric or arithmetic content like be a cluster and be twelve. Brisson’s (1997, 1998, 2003) suggestion is that collective predication can marry distributive quantification only if there is a separate thematic relation to occur within the scope of the distributive quantifier while the collective verb remains safely outside. Within a standard neo-Davidsonian analysis without supermonadicity, this is a plausible distinction
952
Notes
classifying predicates, but supermonadicity never leaves any structure lacking for separate thematic relations. Also, it isn’t as if the meaning of (i) and (ii) lacks something the senators each do alone, in virtue of which and only in virtue of which the amendment passes. Why shouldn’t (i) and (ii) mean that (i′) The senators all voted for it and the amendment passed. (ii′) Many/Most senators voted for it and the amendment passed. Passing the amendment implies a measurement even if it is not only measurement: (iv) *The senators all totaled a majority. (v) *Many/Most senators totaled a majority. (vi) *The senators all racked up a majority. (vii) *Many/Most senators racked up a majority. So, the amendment passes only if there is a majority under a single frame of reference, as a determination of such requires. The single frame of reference framing a single scene, as it were, of the amendment passing is to be reconciled to the multiple frames of reference introduced by distributive quantification framing individual senators voting. The circumstance recalls (98) from section 11.0.0: (98) Four thousand passengers have (4000) opinions about their flights last year. The utterance is infelicitous as an event count when addressed to a focus group of the 1000 frequent fliers whose events it reports, but it becomes acceptable spoken while alone in the office thumbing through the 4000 customer surveys collected. The problem was that the scene of the focus group does not itself resolve into 4000 presentations. The 4000 separate opinions that past experience generated now converge in presentations of their havers, and nothing presented to the speaker now can be thought to be of one opinion and not of another held by the same person, unlike the scene with 4000 customer surveys in hand. If the many scenes of senators voting are also to converge on as many presentations within the scene of the amendment’s passage, it would seem that the ballots or votes themselves that constitute passage are exactly like the customer surveys in meeting all conditions on sequencing of events, perspective, and aspect. Here is where Taub (1989) and Brisson (1997, 1998, 2003) might step in. Sentences (ii) and (iii) are achievements, punctual in their extent whatever their constituent subevents might be. There are not, in fact, discrete, countable subevents within the achievement in which the senators participate that are as many as the senators voting. The events or states of being a vote or a ballot are as many but they are not events or states in which the senators participate—rephrasing Brisson’s point—although they are the effects of such, as are the customer surveys. But here, being an achievement, the Aktionsart requires that the many discrete countable events in which the senators participate both have as many effects in the scene of the amendment’s passage and occur within an event, an achievement, of no greater temporal extent than the moment of the amendment’s passage. The senators’ actions are not so confined, and their effects, although confined, are not events or states in which the senators participate, as any analysis of (ii) and (iii) requires there to be. 47. In the text, I speculate that the intervening structure present in (352)–(357) smuggles in a displacement from frames of reference F to frame of reference f. Alternatively, the grammar and syntax for reference to frames of reference may be common to all these sentences and formally allow to the copula construction too a displacement from the F to the f. Instead,
Notes
953
analogous to the suggestion in note 46, it could be that the lexical meanings of the copula, tense, and aspect combine to confine the F and the f to the same spatiotemporal region. Then, as a matter of human perception and cognition, it could be than no one can concurrently grasp scenes of the same spatiotemporal region disoriented (i.e., failing to path-integrate) for the frames of reference F for some of those scenes while oriented to f in virtue of grasping other scenes of the same. 14
[DP D AdrP and AdrP]
1. Event counting can combine with disjunctive interpretations, nine infielders and outfielders, 4000 cutters and ketches. Since the mereological relation ‘O’ (“overlap”) fails to preserve cardinality, this logical form may be mistaken. I am rather hopeful that the rigors of counting (section 9.4.2 and chapter 12) would pragmatically strengthen the overlap to identity, as how else are these events of infielder-ing and outfielder-ing to be counted gerrymandered into something else than as they are? If not, the logical form should be strengthened, replacing ‘O[E1,E2]’ with ‘W[E1,E2]’ or with ‘scenes-of[E1,E2]’. 2. Is Lenin the Agent, Patient, or Theme of his own Lenin-izing? Or would it be more straightforward to think that ‘Lenin[E,X]’ is itself a thematic relation? 3. As expected, (i) contrasts with (30) and (31): (i) #The morning star and evening star are tracing the ecliptic from one day to the next. #The morning star and evening star trace the ecliptic from one day to the next. #The morning star and evening star have traced the ecliptic since Creation. (ii) The morning star and evening star is tracing the ecliptic from one day to the next. The morning star and the evening star traces the ecliptic from one day to the next. The morning star and the evening star has traced the ecliptic since Creation. 4. That is, it relies in context on a “wide” neighborhood (see section 10.1, note 48), the one called on to validate If Clark Kent is Superman, then Clark Kent has leaped over as many tall buildings as Superman. 5. With no protium and no deuterium radioactive, Determiner-sharing, deriving in effect “most (of the) protium, most (of the) deuterium, and most (of the) tritium …,” would derive a false interpretation. One stumbles into an illusion of equivalence choosing the wrong predicate. The property attributed, like radioactivity, needs to be true of that just in case some of it is, and not only if all of it is. “Most students and professors came to the lecture” implies that most students came to the lecture and most professors did too, but only because it is the case that they came only if they all did (cf. Heycock and Zamparelli 1999, 132). Even if, contrary to the spirit here, one were inclined to recognize a special-mixture and, in whiskey-and-water or franks-and-beans, there is no mixture and substance derived from just the hydrogen in water, which is dispersed among larger atoms of oxygen. Later examples will illustrate the divergence of coordinated NPs and simple NPs without this distraction. The point of (35) and (36) is that they have the true interpretations reported, and not to deny them an ambiguity where they can also mean the equivalent of Determiner-sharing. To my ear, the latter seems possible, adjusting the intonation, where the respondent believes that there are two (or three) answers to the question Most WHAT in seawater is radioactive? “Well, most protium and (also) tritium.” Answering so, she is of course mistaken.
954
Notes
6. As be radioactive is not antisemidistributive, it isn’t only the antisemidistributive predicates that occasion the substitution puzzle for coextensive simple and coordinate NPs—#Many musicians are an ensemble vs. Many instrumentalists and vocalists are an ensemble. 7. Note that seawater differs from hydrogen, at least to my ear. Minimal amounts of seawater must preserve its profile of containing traces of tritium and are therefore always radioactive: (i) T Most seawater is radioactive. Such are the vagaries of primitive concepts to oppose hydrogen and seawater in their relation to hydrogen isotopes. 8. See (6) and the surrounding discussion. 9. (See Hackl 2003, 2009, deriving most from many/much, as discussed in section 13.0, note 9, and section 12.0.) I assume without definition ‘measure[eμ,E0,n],’ the real-number analog to counting. 10. In light of what is to be observed about coordination with nondenoting AdrPs (section 14.1.2), the sampling is rather never-one-without-the-other-unless-there-isn’t-oneor-the-other. 11. Recall from the discussion above of (47) and (48) that temporal distributivity might be suppressed. 12. See sections 9.4.1 and 13.1: (i)
f © orient(ṡ,f, n) ↔def ṡ© = projection(t(ṡ©),σf(ṡ©),πf(ṡ©), f, l(ṡ©), No( sɺ ) ,o(ṡ )) & © © © f n sees at t(ṡ) that ∃πf ∃l ∃N o ∃o ṡ = projection(t(ṡ ),σf(ṡ ),πf, f, l, N of ,o)
(ii) orient( sɶ, f , n) ↔ def ∀t(LTsɶ ( sɶ)(t ) → orient( sɶ(t ), f , n)) ɶ ɶ][∃f : Ff ][∃n : Nn] orient( sɶ, f , n) & (iii) orient[Sɶ , F , N ] ↔ def [∀sɶ : Ss ɶ ɶ][∃n : Nn] orient( sɶ, f , n) & [∀f : Ff ][∃sɶ : Ss ɶ ɶ][∃f : Ff ] orient( sɶ, f , n) [∀n : Nn][∃sɶ : Ss See section 13.1 for remarks that sanction an anonymous narrator, leaving it to an inference that the narrator for whom the scenes of measurement are orienting is the current narrator. Natural language and thought is rather glib in its reference to then & there as to whether the events so demonstrated, the scene thereof, or the address in the frame of reference thereto is intended, which my logical form clumsily unpacks. Much is arbitrary for the sake of concreteness. 13. Omitted is [DP some [streams AdrP]].. and [DP some [dry creeks AdrP]] …, some streams on the left and some dry creeks on the right. From what follows, I hope it will be obvious what to expect in this case and note here that it is as expected. 14. If the cyclist is at the center of her cycling, it can hardly be that her visual experience should hold her own cycling en-scène. The logical forms (66)–(67) revert to the official idiom for adverbial modification in which the modifier cycling at twilight denotes events related by the neighborhood relation ‘N[Ec,E]’ to the events that the modified phrase describes. These events near the cyclist and her cycling are those said to be en-scène. Logical forms for the earlier sentences about sampling hydrogen or its isotopes could be abbreviated in that
Notes
955
sampling can easily be thought of as an action in view involving little more than the sampled and a sampling instrument or hand. 15. Recall that (i) as it occurs in (44) is meant to finesse imperfect aspect for both interrupted, continuous measurement and a prolonged series of discrete, punctuated measurements: (44) [∃Eμ: (i now-en-scène[Eμ] measuring[Eμ])] … [∀tμ: t(Eμ)tμ][℩Eμ: proi & t(Eμ)tμ] … (i) … [∀t: t(E)t][℩Et: proi & t(Et)t] … (ii) … [∃E: (i now-en-scène[E] N[Ec,E])], … [∀t: t(E)t][℩Et: proi & t(Et)t] … Despite emphasis, (iii) doesn’t really say how often this hypochondriacal wimp suffers: (iii) Cycling at twilight, throughout / constantly / at every conceivable moment, a pinched nerve screamed in three different bodily locations. It is left to a contextual neighborhood relation ‘N[Ec,E]’ in (ii) and in (66)–(67) to parse the cycling into as many episodes as deserved to be called constant or throughout if at every one there is pain (iii) or the events are as (66)–(67) describe. It may be that dedicated aspectual morphology ‘Imperfective[Ec,E] & N[Ec,E]’ has something to say about when the neighborhood relation is an appropriate parse. 16. Recall from section 13.1 that all DPs with overt articles contain the epistemic condition ‘some[E0,F]’, which is here suppressed. 17. See section 9.4.0: (156) left(e, α, f) iff ∃αf′(At(α,e,t(e),f) & αf′ = f & r(α) * cosine(alt(α)) * cosine(az(α)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) (i)
left[E, , F ] ↔df ∀e(Ee → ∃α ∃f(Aα & Ff & left(e, α,f))) & ∀α(Aα → ∃e∃f(Ee & Ff & left(e, α,f))) & ∀f(Ff → ∃e∃α(Ee & Aα & left(e, α,f)))
18. As defined in section 9.4.0, (172)–(173), and in section 9.4.1, (190)–(194), if the events E now-en-scène[E] are just one it must reappear in all the scenes now en-scène if there are more than one: (i) en-scène(e, sɶ, α f , f ) ↔df ∃π(cinerama(π) & Sɶπ sɶ ) & At(αf, e, t(e), f) & ∀ṫ((moment(ṫ) & overlap(ṫ, t(e))) → sɶ(tɺ) = projection(tɺ, α f , π f ( sɶ)(tɺ), f , l( sɶ)(tɺ), Nof ( sɶ ), o( sɶ)(tɺ) )))
ɶ ɶ → Sɶπ sɶ)) & (ii) en-scène[E, Sɶ , AF , F ] ↔df ∃π(cinerama(π) & ∀ sɶ(Ss ɶ ɶ & AF α f & Ff & en-scène(e, sɶ, α f , f ) )) & ∀e(Ee → ∃sɶ∃α f ∃f (Ss ɶ ɶ → ∃e∃α f ∃f (Ss ɶ ɶ & AF α f & Ff & en-scène(e, sɶ, α f , f ) )) & ∀sɶ(Ss ɶ ɶ & AF α f & Ff & en-scène(e, sɶ, α f , f ) )) & ∀α f ( AF α f → ∃e∃sɶ∃f (Ss ɶ ∀f(Ff → ∃e∃sɶ∃α f (Ssɶ & AF α f & Ff & en-scène(e, sɶ, α f , f ) )) (iii) now-en-scène(e) ↔ ∃sɶ∃α f ∃f (now( sɶ) & en-scène(e, sɶ , α f , f ) ) (iv) now-en-scène[E] ↔ ∃Sɶ∃AF ∃F (now(Sɶ ) & en-scène[E, Sɶ , AF , F ]) (v) For discourse Δ and token µ of ⌜now(vi)⌝ in Δ such that ltΔµ©(µ) = δt, Σ satisfy u ↔ Within(nowΔµ©(δt), Σ(vi)).
956
Notes
Every such scene would still be of stream and dry creek, but I suppose the logical form in the text might allow in principle for the stream to be on the left only in some of the scenes and elsewhere in others and a similar confusion for the location of the dry creek. If this is a worry, then (ii) can be strengthened without harm elsewhere to demand that no event E appear in more than one of the scenes Sɶ . I should think the coherence and anticonvergence conditions on cinerama would suffice to exclude the same event from multiple scenes. 19. The constraint on (66) asserts itself more subtly than the text discloses. Suppose that every cyclist in mind cycles alternating exclusively between the two lines of sight perpendicular to her course, in confirmation of (67) and (68). Given what is known about cycling, cyclists, roads, streams, and dry creeks, there is a ready inference from (67)–(68) and the conviction that it was the same stream and dry creek under observation to the conclusion that an imaginary cyclist with eyes locked straight ahead could have given a direct report of her experience in the words of (66). As remarked, an utterance is often a reprise of original narration. Especially if it is to be judged true or false in the absence of concurrent cinerama or navigation, it is judged so in deference to its original narration as imagined. I myself as one of the actual cyclists might retell the story using (66) either to vividly evoke the scene ahead or to forget what my cycling style unsafe at any speed entitles me to report direct witness of. I would like to find an example where worldly knowledge would not support inference from the likes of (67)–(68) to (66)’s analog. 20. As in This book has taken two decades and This damn book has been in preparation for two decades. 21. Given the overt article, the full DP includes the epistemic condition, ‘some[E0, F]’ in (i), that the frames of reference F are path-integrated to some single frame of reference f (section 13.1). All of this is material when plural frames of reference may be in play, as allowed to plural or mass terms, some accompaniments, accompaniments, some accompaniment, accompaniment, that may scatter their denotata. But any so-and-so such that there is a so-and-so must, I assume, be located whole at a single address in a single scene in a single frame of reference, and so the logical forms in the text cut to the chase: (i) … [∃F: ∏] … [∃X: ∃E0 (Δan(E0) some[E0, F] now-en-scène[E0] [∃ : VF[E0, ]](ADR[E0, , F ] ∃X accompaniment.sg[E0,X]))] … 22. If even in simple AdrPs a token of ‘W’ were to occur innocuously as in (i), the logical form would be as in (ii): (i) [[DPAn.sg W accompaniment.sg] … W … θ … Tense …] (ii) … [∃F: ∏] [∃X: ∃E0 (Δan(E0) now-en-scène[E0] W[E0,X] [℩E1: Φ]O[E0,E1] ( Φ ∃E1 ∃F (ADR[F , E1 ] ∃X accompaniment.sg[E1,X]))] [℩E0 : Δ] [∃E3 : N[E,E3]] … W[E3,X] O[E3,E4] ∃X θ[E4,X] … [∃F: ∏] (… Tense[Ei,F] …) 23. See the structure for coordinate singular plural in section 12.2, (154). 24. Presumably, both the initial and final 40-minute events are both while there be appetizer and while there be dessert.
Notes
957
25. Alternatively (see note 22), (i) W[E0,X][℩E1: Φ]O[E0,E1] (Φ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X accompaniment.sg[E1,X]))] 26. As mentioned in note 1, I will need to restrain ‘O[E,X]’ from exercising mereological fusion to defeat cardinality preservation. As before e2, the second event of active appetizer preparation is excluding not being active preparation of dessert. What about that event of 30 minutes spanning e2 and the second event of active dessert preparation? It is while being appetizer and dessert throughout given the persistence of appetizer and dessert through periods of inactive preparation, but it is not an event of active preparation. It is or contains two events of active preparation, and two events are not in the neighborhood of one event of there being appetizer and dessert, given the cardinality preservation of the neighborhood relation N. Note that if this were not the case, it would mistakenly be rendered true that an appetizer and dessert took 240 minutes of active preparation. 27. See note 26. 28. I had thought to exploit that lexical roots, thematic relations in particular, occur in logical form only pluralized (Schein 2006), which implies their (asymmetric) exhaustivity with respect to participation in events. That is, (i) implies (ii): (i) θ[E,X] ↔df ∃xXx & ∃eEe & ∀x(Xx ↔ ∃e(Ee & θ(e,x))) (Pietroski 2003, 282) (ii) θ[E,X] → ∀Y(θ[E,Y] → ∀x(Yx ↔ Xx)) To put the point in relief, first rewrite ‘left[A,E,F]’ as ‘leftA[F,E]’ and understand it to say that the events E are exhaustive of the frames of reference F. Of course, if appetizer events fill their left hemispheres, then dessert must be elsewhere, and coherent lexical choice is appropriately forced elsewhere. The obvious flaw is that it empties these frames of reference of anything except for appetizer and dessert. Even to impose as much on scenes thereof makes for very empty images, unless something more intentional and attentional is taken on. Recall that scenes are further individuated by their reticulation, which is defined both by optical resolution and a designated addressing representing the attentional salience of some locations and not others. One could then define left to imply that what is said to be left exhausts the left-hemispheric addresses in this scene’s reticulation. If all that is appetizer, then dessert must again be elsewhere in this reticulation, as desired, and without emptying frame of reference or visible scene of all else. Rather, appetizer and dessert are all there is of interest under scrutiny on the left or right. So might a syntactic stipulation derive from arcane meaning, which I will defer for now. 29. Alternatively (see note 22), (i) W[E0,X][℩E1: Φ]O[E0,E1] (Φ∃E1 [∃ : VF[E0, ]](ADR[E1, , F ] ∃X fielders[E1,X]))] 30. Another example: (i) Many machine parts are in a machine. (ii) #Many machine parts are a machine. (iii) Many widgets and gadgets are a machine. Sentence (ii) means only that many are such that the part is the machine; (iii) allows for machines with several parts.
958
Notes
31. If my earlier argument has been insufficient that instrumentalists and vocalists contains the sentential connective along with the rest of the W-based apparatus for deriving the disjunctive interpretation, it does not greatly add to it that there is an Eventish account of the contrasts between (345) and (99) and between (35) and (37). An adversary might return to an obvious alternative. Suppose antisemidistributive predicates denote only truly irreducible, echt (genuine) collectives (Dowty 1987; Higginbotham and Schein 1989; Winter 1998, 2002; Hackl 2001a, 2001b, 151–186, 2002). Then the contrast between (345) and (99) can be taken to confirm that and in instrumentalists and vocalists derives an NP that denotes genuine collectives, unlike the simple musicians (cf. the set-product operator in Heycock and Zamparelli 2005, discussed above in section 12.2, note 9). The contrast between (35) and (37) is just a further consequence of and’s formation of a NP denoting genuine collectives, which are the only elements that (35) quantifies over and applies radioactive to. The response (sections 13.0 and 13.2) here is twofold. First, to deny the alleged logical-type distinction for antisemidistributive predicates, pointing to contrasts such as between (345) and (346) and between (i) and (ii) to urge a more supple account of the variation in terms of reference to discrete, countable events: (i) *Many musicians are a chamber group. *Most musicians are a chamber group. (ii) Many musicians who are on stage at a chamber recital are a chamber group. Most musicians who are on stage at a chamber recital are a chamber group. Whatever is offered to explain the contrast between (iii) and (iv) should not be so brittle it rules out (v), and correlatively, whatever lets pass (v) (and (iii)) should not lose the antisemidistributivity of (iv): (iii) Many people in the offices of the Empire State Building first met in an elevator—but not the same elevator. (iv) #Many people in the offices of the Empire State Building are a team. (v) Many people in the same office in the Empire State Building are a team. If reference to events is the gateway to all variation in (anti)semidistributivity, as argued in sections 13.0 and 13.2, then any such variation that tells apart instrumentalists and vocalists from musicians must be cashed in as variation in their description of events for whatever apparatus makes that description of events matter for the interpretation of are an ensemble or are a chamber group. Second, it is denied (section 14.1.2) that and itself could be an operator demanding the presence of both instrumentalists and vocalists in the genuine collectives allegedly referred to—not all NP-coordinations imply the existence of witnesses for both conjuncts. The required presence of both instrumentalists and vocalists is rather an illusion of the locative relations imposed on the conjuncts and the optimization of the scans that parse the scenes realizing them. 32. More examples. The distributive quantification in (i) and (iii), in contrast to (ii) and (iv), is only about things each of which is cup and saucer in one: (i)
a. Most things on the table are cups and saucers. b. Many things on the table are cups and saucers. c. Few things on the table are cups and saucers.
Notes
959
(ii) a. The things on the table are cups and saucers. b. Some things on the table are cups and saucers. (iii) a. Most things on the table that are cups and saucers are glass. ?Most things on the table that are cups and saucers are glued together. b. Few things on the table that are cups and saucers are glass. ?Few things on the table that are cups and saucers are glued together. (iv) a. The things on the table that are cups and saucers are glass. The things on the table that are cups and saucers are glued together. b. Some things on the table that are cups and saucers are glass. Some things on the table that are cups and saucers are glued together. 33. The contrast between distributive quantification and plural (in)definite description persists attenuated when Hungarians and Romanians with nouns replaces the adjectives in Hungarian and Romanian: (i)
a. Most Transylvanians are Hungarians and Romanians. b. Many Transylvanians are Hungarians and Romanians. c. Few Transylvanians are Hungarians and Romanians.
(ii) a. The Transylvanians are Hungarians and Romanians. b. Some Transylvanians are Hungarians and Romanians. (iii) a. Most Transylvanians who are Hungarians and Romanians are rootless cosmopolitans. b. Many Transylvanians who are Hungarians and Romanians are rootless cosmopolitans. c. Few Transylvanians who are Hungarians and Romanians are rootless cosmopolitans. (iv) a. The Transylvanians who are Hungarians and Romanians are rootless cosmopolitans. b. Some Transylvanians who are Hungarians and Romanians are rootless cosmopolitans. At issue and consonant with what follows is that one can evoke an experience of countable, discrete presentations of there being Hungarians and Romanians, as one might come upon them roaming Transylvania. Evoking such a context, (iiia) can be understood as if to say that most Transylvanians gathered here and there as Hungarians and Romanians are rootless cosmopolitans, interpreting the quantifier semidistributively with respect to contextually supplied collective events. In contrast, it defies imagination to evoke discrete encounters with states of being Romanian and Hungarian. These observations also take in a contrast between (iii) in this note and (iii) in note 32, where despite the coordination of nouns, cups and saucers, there is less escape from implying that anything talked about is cup and saucer in one. Without a temporal dilation to supply several discrete encounters with Hungarians and Romanians, there is no obvious or intended organization of the scatter on the table into discrete events or states of there being cups and saucers. (See Schwarzschild 1996, 82f., 92f., and Schein 2006, 754f., on parsing a scene into spatially discrete events.) 34. Note that be a cluster, be twelve, and be Hungarian(s) and Romanian(s) also all happen to lack “subentailments” (Dowty 1987), to the extent one can make sense of the notion. But, as Taub (1989) trenchantly observes, this fails to be a general characteristic of antisemidistributive predicates; see note 46 in section 13.2 for discussion.
960
Notes
35. These sentences are better than those above in escaping the temptation to lapse into a “respectively” reading. 36. As Gillon (1990) and Schwarzschild (1991, 1996) point out (see section 13.2), the semidistributive interpretation becomes acceptable with a specific context in mind that, for example, corresponded to encountering clusters of Christians and Jews on one’s travels (see note 33). 37. The above remarks are directed against treating and as a set-product operator (e.g., Heycock and Zamparelli 2005; also see the discussion in chapter 12, note 9). Some may also be tempted by a pairing function for the meaning of (i): (i) (ii)
Many smartphones and cellphones communicate with only each other. Many a smartphone and cellphone communicate with only each other.
(209) In Mobile, many mobile phones communicate with (each other and) only each other. But the context mentioned in the discussion surrounding (209) (see section 12.3.1) verifies (i) and (209) and falsifies (ii), where pairing is not in doubt. 38. Thus, in two stages: (137) is the nonzero pluralization of the primitive, first-order relation, and then the relation derived via (138) admits empty E and empty X just in case both are. Two stages are not necessary, but (138) is an emendation to (137), which has appeared up until this point throughout. The notation ‘‖’ is however a nuisance, and it would be better from the start to define pluralization to itself allow the relation between empty plurals. 39. Except for this section, ‘‖’ is suppressed elsewhere although it should be presumed present. 40. Examples (145)–(148) are constructed to preempt Determiner-sharing, More/Fewer than n … (,more/fewer than n …,) and more/fewer than n … , which in every example would reverse the reported truth value. 41. Here Determiner-sharing derives an equivalent result. The question is rather why the sentences do not have a vacuously true interpretation if and made existential commitments. Since no account is yet on offer for the distribution of existential commitments among contexts for NP-coordination, it could very well be that some sentence is ambiguous in this respect, indicating a structural ambiguity in the coordination. For the lexical item and to be unambiguous, it must not impose existential commitments, for the benefit of sentences where none is imposed. 42. See section 9.4.0: (156) left(e, α, f) iff ∃αf′(At(α,e,t(e),f) & αf′ = f & r(α) * cosine(alt(α)) * cosine(az(α)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) (i) left[E, , F ] ↔df ∀e(Ee → ∃α ∃f(Aα & Ff & left(e, α,f))) & ∀α(Aα → ∃e∃f(Ee & Ff & left(e, α,f))) & ∀f(Ff → ∃e∃α(Ee & Aα & left(e, α,f))) 43. Assuming that NP-coordination is without existential commitment, Jim Higginbotham (p.c., 1992) has suggested that it is only a pragmatic implicature that there are both custards and blancmanges when using twelve custards and blancmanges. Notice that if one knows (i), Gricean considerations will block both (ii) and (iii), even though the meaning of
Notes
961
the disjunction plainly does not commit the sentence to the existence of both books and newspapers: (i) The boxes contain only books. (ii) #The boxes contain sm books and newspapers. (iii) #The boxes contain sm books or newspapers. Indeed such a context provides no argument for a semantic difference between (ii) and (iii). But if instead one knows (iv), one may say (v), but (vi) still violates a Gricean maxim. (iv) The boxes contain sm books and sm newspapers. The boxes contain twenty-five books and twenty-five newspapers. (v) The boxes contain sm books and newspapers. The boxes contain fifty books and newspapers. (vi) %The boxes contain sm books or newspapers. %The boxes contain fifty books or newspapers. Why should this be if the NP-coordination with and, without an existential commitment from both NPs, is equivalent in this context to the NP-disjunction with or? 44. For the opposition in the text between definite and indefinite description to be maintained, the decreasing indefinites in (i) and (ii) are included among the indefinite descriptions, perhaps as not-many French peers and not-any French peers, and these may of course occur with expressions of the irrealis: (i) Few French peers that there are if any hold legal titles. (ii) No French peers that there are if any hold legal titles. 45. As in Hackl 2001a, 2001b, 2003, 2009 (also see chapter 13, note 9, and chapter 12, section 12.0), Most F G derives from a comparative, More F G than F not G, comparing the n-measured F that G (cf. (153)) to the m-measured F that do not G. Given that existential commitment in both conjuncts derives from the presence of a morpheme to assert it, its absence from AdrP-coordination with most implies its absence from the comparative constructions as well if Hackl is correct in deriving the one from the other. As implied, in (145)–(148), more than n F and G and fewer than n F and G are not committed to the existence of both Fs and Gs. For Hackl, than always introduces a clause, so that the truth of (i) derives from (ii): (i) Fewer than 700 French peers are celebrating today defeat at Agincourt in 1415. (ii) To a lesser degreei than if 700 French peers be celebrating today defeat at Agincourt in 1415 thati-many French peers are (in fact) celebrating today defeat at Agincourt 1415. As Hackl shows on both syntactic and semantic grounds, the clausal complement to than unspoken in (i) is untensed in support of a counterfactual reading as my use of if and the subjunctive in (ii) paraphrase. 46. If a formal ambiguity is to be recognized one way or another, an alternative formulation could relocate it to the definite description itself. The operator would remain without existential commitment, and (131) and (163) would instead alternate between ‘[The X: ∃E Φ[E,X]]’ in (131) and ‘[The X: ∃E Actual[E,A,F ] Φ[E,X]]’ in (163). Apart from forgoing the direct connection between event quantification and the speaker’s intentions in referring to events, this remains an alternative to the one considered in the text.
962
Notes
47. With the relative clause as an appositive containing a plural predicate All-Star teams, (i) has an acceptable, distracting interpretation that both of the ball clubs, the Apostles and the Saints, were All-Star teams: (i) Some Apostles and Saints, who once were both All-Star teams, warmed the benches at Madison Square Garden. All-Star teams are a selection of the best players, and (i), like (181c–e), cannot mean that some of the Apostles and some of the Saints were All-Star teams. Similarly, if it were not known that the active roster of a major league baseball team, such as the Angels or Padres, was twenty-five, an appositive who were both nine in number would be subject to the same distraction. 48. The text does not cite (i) since its parse is ambiguous between NP-coordination and DPcoordination. If one fixes intonation to favor the former, (i) is unacceptable: (i) Some twelve Apostles and twelve Saints who once were both an All-Star team warmed the benches at Madison Square Garden. 49. Recall the monitors at the harbor at the end of the voyage, following the observations of passages at the point of departure. 50. Imagine that all that is visible above the subway crowd are hands gripping a rail. One may still count persons recognizing left and right hand from the same despite the intervention of others’ hands on the rail. Similarly, a fire, as defined by its development, may express itself as several discrete regions ablaze. 51. Despite his superpowers, Superman is no better in this respect than ordinary mortals conforming to physical laws. Equipped with a transporter beam borrowed from Star Trek, he could of course dematerialize in one location and rematerialize at a distance. The point is that a frame of reference is a complete representation of motion within its space. 52. Among the kinematic scenes or frames of reference are those that are object-tracking, a stronger condition that becomes necessary below. 53. I say ‘only if’ in the preceding sentence so as not to dispossess a mereologist who would like to count under a different protocol a single discontinuous green region even under the narrow frame of reference. 54. Recall also that the two motorists of (144) may be dispersed to the lightning strikes, each nearby only one of them, but in (145)/(147), the travelers move as one, a single nearby, which cannot be near both lightning strikes: (144) Last night in different parts of the city, two lightning strikes in heavy traffic injured two nearby motorists. (145) Last night in different parts of the city, two lightning strikes in heavy traffic injured a nearby two motorists. (147) Last night in different parts of the city, two lightning strikes in heavy traffic injured a nearby motorist and pedestrian. 55. One cannot substitute photographed for pictured without interference from its transitive usage (Stieglitz photographed them all vs. *Stieglitz pictured them all), where a single action of photographing may result in many photographs. See the discussion of long vs. lengthened and duplicate vs. duplicated in section 12.2.
Notes
15
963
The Ordered-Pair Illusion
1. For those grasping at “salience,” I do not find (i) and (ii) any friendlier toward finer-grained identity despite the “salience” of pitcher and catcher: (i) a. An ill-matched two teammates who were pitcher and catcher for three innings started for the Red Sox yesterday, and another equally ill-matched two teammates who were pitcher and catcher for two innings started for them tonight. b. An ill-matched two teammates who were pitcher and catcher for three innings started for the Red Sox yesterday, and another equally ill-matched two teammates who were catcher and pitcher for two innings started for them tonight. (ii) a. Two teammates who were pitcher and catcher for three innings started for the Red Sox yesterday, and another two teammates who were pitcher and catcher for two innings started for them tonight. b. Two teammates who were pitcher and catcher for three innings started for the Red Sox yesterday, and another two teammates who were catcher and pitcher for two innings started for them tonight. 2. (i) A homeowner and brotherly neighbor cleared one home’s yard, and another homeowner and brotherly neighbor cleared another home’s yard. (ii) #A homeowner and brotherly neighbor tore down one fence between them, and another homeowner and brotherly neighbor tore down another. 3. I owe thanks for extensive discussion of this section to Pranav Anand, Paul Pietroski and Roger Schwarzschild. 4. The plural number agreement is crucial. The sentences become true (although less than fully grammatical) in the singular: (33) These arranged this way [as in photo 1] is not them arranged that way [as in photo 2]. These arranged as they are is not them arranged in this other way. In this case, I assume that these arranged this way, them arranged that way, etc., are indeed a small clause denoting an event, scene, situation, or fact. 5. Crucial to the discussion that follows is the conclusion that negation in the copula construction includes within in its scope modifiers that apply to local events as modifiers do in other constructions. Less important is whether a single meaning for the lexical copula can be distilled from its various uses. The copula construction in the text should not be confounded with a resultative copula construction. As above, (35) is true, referring to the photos as indicated: (35) These1 arranged here1 just so1 are not them2 arranged there2 just so1 There is also a false reading of (i) that purports that what is seen in photo 1 is the result of rearranging them in photo 2 to appear as they do in photo 1: (i) These1 arranged here1 just so1 are not them2 arranged just so1
964
Notes
It is clearer in (ii), which is true of (iv) and false of (v). In contrast, (iii), which reverses it, is false of (iv) and true of (v) (see section 10.3.0): (ii) These1 ((arranged) here1) are them2 (arranged) further apart. These1 are them2 separated (further). (iii) These1 ((arranged) here1) are them2 arranged there2 further apart. These1 are them2 separated there2 (further). (iv)
Photo 1
Photo 2
(v)
Photo 1
Photo 2
6. One may also wish to restrict similar orientation to compare just the locations of the events’ participants, as in (i), or of some other events or objects that the scenes are said to highlight for attentional salience, a question left for further research. (i) similarly orienting[E0,E1] ↔df ∃Sɶ0 ∃Sɶ1 ( ∃0 ∃F0 en-scène[E0, Sɶ0 , 0 , F0] & ∃1∃F1 en-scène[E1, Sɶ1 , 1 , F1] & ∀X(W[E0,X] ↔ W[E1,X]) → ( ∀t ([∀ζ: [∀x: ζx][∃X: W[E0,X]]Xx] [∀ξ: [∀x: ξx][∃X: W[E0,X]]Xx] ( ∃ ®[Sɶ0 (t ), ζ, ξ, ] ↔ ∃ ®[Sɶ1 (t ), ζ, ξ, ]) → ∀ζ∀ξ∀ ((®[Sɶ0 (t ), ζ, ξ, ] & ∃F0 en-scène[ζ, Sɶ0 (t ), , F0]) ↔ (®[Sɶ0 (t ), ζ, ξ, ] & ∃F0 en-scène[ζ, Sɶ0 (t ), , F0])))))
Notes
965
7. See section 9.4.2: (200) ®[ṡ, ζ, ξ, Aṡ] ↔df ∀αṡ(Aṡαṡ → sɺα sɺ ) & ∀αṡ∀αṡ′((Aṡαṡ & Aṡαṡ′ & overlap(αṡ,αṡ′)) → αṡ=αṡ′) & ∀αṡ∀αṡ′((Aṡαṡ & Aṡαṡ′ & At[αṡ,ζ,t(ṡ), ṡ] & At[αṡ′,ζ,t(ṡ), ṡ]) → αṡ=αṡ′) & ∃αṡ(Aṡαṡ & At[αṡ,ζ,t(ṡ), ṡ] & ∀x(ξx → ¬∃y(overlap(x,y) & At(αṡ,y,t(ṡ),ṡ))) & ¬∃p(At(p,p,t(ṡ), ṡ) & Limit point(p, ζ, ṡ) & Limit point(p, ξ, ṡ)) 8. The visual experience may reflect either the motion of at least one sock to new coordinates within the frame of reference or a different scale (magnification) for the visual experience of stationary objects such as might accompany a change in the observer’s position or instrument. Similar orientation across frames of reference requires an identity in the coordinates addressed, independent of scale. 9. See section 9.4.0: left(e, α, f) iff ∃αf′(At(α,e,t(e),f) & αf′ = f & r(α) * cosine(alt(α)) * cosine(az(α)) < r(αf′) * cosine(alt(αf′)) * cosine(az(αf′)) left[E, ,F] ↔df ∀e(Ee → ∃α ∃f(Aα & Ff & left(e, α,f))) & ∀α(Aα → ∃e∃f(Ee & Ff & left(e, α,f))) & ∀f(Ff → ∃e∃α(Ee & Aα & left(e, α,f))) top(e, α, f) iff At(α,e,t(e),f) & alt(α) > 0 10. See note 9. 11. This is not to slight the problem of recovering unspoken content, which is assisted by narrative conventions, semantic or pragmatic, that prescribe a stepwise march across space or time (first, second, third,…) when the vocabulary is left unspoken. The use of spoken vocabulary is obviously less constrained. It may turn out that narrative progress distinguishes conjunction (⌜S1, and S2, and S3.⌝) from full stops (⌜S1. S2. S3.⌝), which may also then distinguish conjunctions of smaller phrases from structures in which conjunctions are absent. 12. See note 9. 13. “Is not the whole land before thee? Separate thyself, I pray thee, from me: if thou wilt take the left hand, then I will go to the right; or if thou depart to the right hand, then I will go to the left” (Genesis 13:9). 14. Moreover, a single spoken adr is not license for such quantification into an unspoken one. If it were, (i) and (ii) could mean (87) and (88), respectively: (i) They first are not them. (ii) They on the left are not them. 15. Rather, sentences (89) and (90) make the point. A linguistic antecedent for the comparison class can be denied (21) if left and right are taken to mean the left and right sides (halves) of the photos. In contexts where it is, however, clear that the speaker does not or could not mean the left and right sides (halves) of the night sky, (90) is forced to rely on linguistic antecedents as in (93) to express relative position. With the ordinal relations in (89), there is no distraction from the conclusion that the comparison class is provided by the socks themselves.
966
Notes
16. The understood anaphor in other than them is itself descriptive and thus copying the content of its antecedents preserves the contrast between (5) and (8) (see appendix 1, note 2). 17. See “Archaeology Today,” episode 21, Monty Python’s Flying Circus, 1970. 18. Without further strengthening, this objection to (105) is pragmatic. Compare (i), which is perverse but literally true: (i) During the ownership of the one home, the front-yard fence was torn down, and during the ownership of the other, the backyard fence was torn down. Appendix 1 Descriptive Anaphora and Nonmaximal Reference under Selective Perspectives 1. See Elbourne 2005 and the references cited there. 2. There is further argument against relying on wide-scope indefinites to derive a dependent pronoun’s nonmaximal reference: (i) On the first day of kindergarten, Sally made a new friend. He lived next door, and he had a red wagon and a red dog. Sally may have made other friends the first day of kindergarten, other than the boy who turned out be a neighbor. This example, however, defeats a claim that the indefinite therefore includes within its scope the entire discourse, since it occurs in a position that discourages quantifying in, a characteristic of arguments of what Quine (1960) calls notional verbs and Kratzer (1995) verbs with ill-behaved objects (see also Heim 1987): (ii) *For some new friend x (of Sally’s)—[Sally made x. He … ] In the plural also, Sally and Jane may have made other friends the first day of kindergarten, other than those who turned out to be neighbors: (iii) a. On the first day of kindergarten, Sally made a new friend, and Jane made a new friend. The boys turned out to live next door, and they had a red wagon and a red dog. b. On the first day of kindergarten, Sally made a new friend, and Jane made a new friend. They turned out to live next door, and they had a red wagon and a red dog. The boys and they refer to just two. They refer to only two because in some sense only two have been previously mentioned. Yet, as in the singular, these examples discourage deriving the reference to just two from indefinites that include within the scope the entire discourse and quantify into notional verbs. Rather, the coindexation of plural anaphora should be taken to indicate the embedding of singular descriptive anaphora rather than singular variables of direct reference: (iii) a. On the first day of kindergarten, Sally made a new friendi, and Jane made a new friendj. Thei,j boys turned out to live next door, and they had a red wagon and a red dog. (iv) On the first day of kindergarten, Sally made a new friendi, and Jane made a new friendj. [The X: boys[X] & [℩x: new friend of Sally’s[x]]Xx & [℩x: new friend of Jane’s[x]]Xx] …
Notes
967
The embedded singular definite descriptions manage singular reference to the extent that the speaker and hearer can accommodate a perspective that narrows in on the events reported, on the making of one new friend among the many of the day, which then subsequently restricts the singular definite descriptions (see Schein 1993, 219ff.). The plural reference to exactly two is an effect of the logical syntax, namely, that the plural anaphor has just two such singular descriptions for its antecedents and plural definite description as defined in Sharvy 1980 (see appendix 2). (Thanks here to Roger Schwarzschild for discussion.) An example with another notional verb illustrating the same point: (v) Manny developed a rashi last month, and Sammy woke up with a scaly patchj. Thei,j sores were from poison ivy at day camp. 3. This is essentially Heim’s (1990) answer, too: the singular pronoun is a singular incomplete definite description to be completed by a locative relation to somewhere else to vary its reference, the chair of Leif’s there. 4. In this regard, consider also (i) and (ii): (i) #Yesterday, 613 was greater than another natural number. Tomorrow it will still be greater than it. (ii) #In Yorktown Heights, 613 is greater than another natural number. In Almaden, it is greater than it too. It may be a literal truth that yesterday, today, and tomorrow 613 > 248. But even if one is inclined to accept the spatiotemporal location of the abstract, the last it in (i) and (ii) fails singular reference—the natural number that 613 was greater than yesterday, which one was that? Sentences (i) and (ii) are however rehabilitated, once it is realized that they report the readings of experiments in nanocomputation at IBM Research Centers. 5. The perspective Π0 and the imagined scene of a presentational event are of a solitary tuplet. The having an identical tuplet relates to the presentational event without supposition and reference to another perspective in which an identical tuplet is presented either alone or with the tuplet of Π0. She has an identical tuplet, who is not here in Π0 and perhaps somewhere not to be encountered. The identical tuplet, her in the third sentence, is the identical tuplet that the tuplet of Π0 has at the time of the presentational event of Π0, which implies her uniqueness as desired. (The translation of her in (12), the identical tuplet that she is identical to, takes a shortcut in translating she as the variable x0 rather than expanding its descriptive content, too.) Having fixed singular reference to the tuplet of Π0, successful singular reference to the identical tuplet depends on the irreflexivity of having an identical tuplet. Were it otherwise—Once upon a time, there was a woman. She closely resembled a member of the royal family. She was really, really just like her—successful singular reference would need to be supplemented contextually with an irreflexive relation, the family member closely resembled who is not the woman of Π0 (see Heim 1990). Note the infelicity of (i), where the spoken pronoun diverges from the unspoken tense anaphor in (10) in a way reminiscent of Iatridou’s (1994) discussion contrasting if p, q and if p, then q: (i) Once upon a time no one knows when, there was a tuplet. Then, she had an identical tuplet. Making it available to focal stress, choosing to pronounce what could be left implicit is contrastive—then and perhaps not at other times—which a silent continuation with reference to the occasion already under discussion is not.
968
Notes
6. Heim’s (1990) citation examples, exemplify the problem of indistinguishable participants within single clauses, and thus for her too, “the twin who has an identical twin here is really, really identical to the identical twin some twin has there” for here ≠ there. 7. That is, monozygotic, as opposed to a fraternal or dizygotic twin. 8. Thanks to Elena Herburger (p.c., May 2011) for discussion of this point. 9. The text contrasts settings for (22): one in which the camera begins with a close-up of one identical twin and then sweeps to a close-up of the other or zooms out to a view of them together, and the other of a single shot of the two together in equal prominence, without visual cue as to which is first described. Of course an initial close-up of just one and hiding the other from view is not necessary. A single static shot of both of them would do if it were supplemented with, say, a spotlight on just one, or if it were a photograph with a gesture from the narrator pointing to one and then the other. The scene is to be taken here as delimited by the visual boundaries of the focus of attention. 10. In uttering the abbreviated example (i) rather than (ii), the speaker violates Gricean principles, and uttering self-evident truths such as (i) and (ii) is never a big hit anyway: (i) Some twin has an identical twin. (ii) The twins have identical twins. Yet (i) and (ii) are begrudgingly true, as is (iii): (iii) In this photo, some twin has an identical twin that she is really, really identical to. The use of singular pronouns in (22) in this context seems to be a further insult to good sense. 11. Chapter 1, see note 50. 12. A purely spatial example illustrates the important point that supermonadicity is necessary if a selective perspective on a scene is sometimes the only route to singular reference, and scenes are just like the scenes of ordinary visual experience. C. S. Lewis, author of the allegory The Chronicles of Narnia, let it be known that Aslan the Great Lion is Christ in Narnia. But Christ does not Himself appear in the allegory: (i) In The Chronicles of Narnia, a deity has an incarnation. There is a God, and he has an incarnation in Narnia. The scenes that are a presentation of Christ to the narrator are not Narnian scenes. Thus, the spatiotemporal adverb, “in The Chronicles of Narnia,” locates the state of having an incarnation separate from and without locating there being a deity. 13. As remarked in the text, given supermonadicity and the proliferation of subatomic clauses and subatomic descriptive anaphora, there would seem to be little to tell apart (10) and (i): (10) #Once upon a time no one knows when, there was a tuplet. She had an identical tuplet. She was really, really identical to her. … (i) Once upon a time no one knows when, a tuplet was really, really identical to an identical tuplet. Yet, (i) escapes the implication that the identical tuplets are just two—a tuplet initially presented and the one and only identical tuplet she has. It cannot be that (i) finds felicitous a
Notes
969
spatiotemporal-frame adverbial that (10) does not, as there is nothing about the metaphysics of events, states, or spatiotemporal location to distinguish these sentences. But of course a difference of scope remains. In (i), a tuplet may include within its sentential scope all the rest, but in (10), it includes none of it, where nonmaximal reference is achieved only via a failed spatiotemporal adverbial. A discourse that begins with (i) continues in (ii) with the expected effects on extrasentential descriptive anaphora: (ii) Once upon a time no one knows when, a tuplet was really, really identical to an identical tuplet. She loved her as she loved herself. It is inherited from (i) that the tuplet presented may have many identical tuplets, but the pronoun her refers to the one and only one that she is really, really identical to, without recourse to a spatiotemporal adverbial to derive nonmaximal reference. Again, within a sentence, the implication of uniqueness disappears: (iii) Once upon a time no one knows when, a tuplet was really, really identical to an identical tuplet and loved her as she loved herself. As in the discussion of (10), the pronoun she that refers to the subject, the twin or tuplet initially presented in (ii) and (22), refers to her in a scene in which she stands alone, as much a requirement of the subatomic presentation as the presentational sentences in (9) and (10). Consider a minimalist sculpture on display, Love’s Triangle, a congruent triangle with 2-meter sides, of which a tedious docent may say any of (iv)–(vi), all of which are true: (iv) In the minimalist sculpture, Love’s Triangle, a PVC tube connects an identical PVC tube to another identical PVC tube. (v) In the minimalist sculpture, Love’s Triangle, a PVC tube connects to an identical PVC tube and connects to another identical PVC tube. (vi) In the minimalist sculpture, Love’s Triangle, a PVC tube connects to an identical PVC tube and is connected to another identical PVC tube. In contrast to she referring to the twin or tuplet initially presented, the singular reference of it stumbles in (vii) and (viii): (vii) In the minimalist sculpture, Love’s Triangle, a PVC tube connects to an identical PVC tube, and it connects to another identical PVC tube. (viii) In the minimalist sculpture, Love’s Triangle, a PVC tube connects to an identical PVC tube, and it is connected to another identical PVC tube. The other twin or tuplets are easily elsewhere, but what is this scene at the museum of Love’s Triangle presenting one of its sides without the others? One imagines that the utterance is supplemented with a gesture, or the speaker expects the hearer to eye-track along and share her fixations, or maybe one side of the sculpture is under a spotlight. On the other hand, if (vii) or (viii) is the utterance of a blind visitor to a tactile museum, the isolation of Love’s Triangle’s sides in tactile scenes of their own is obvious and the felicity of singular reference assured. 14. Events without definite spatiotemporal location, the other side of the coin, are expected to be irrelevant, not being defined by spatiotemporal location.
970
Notes
15. See Burge 1986. The same landmass, molecule for molecule, is an island here, but in a counterfactual geological history, it is indistinguishable as a geological feature, buried in a continent’s interior, and so is not an island. 16. Existential event quantification is plural count quantification (see Schein 2006, 752ff., and the references cited there; also see section 13.0 of the present book). 17. See Schein 2006. Countable events are resolved (see section 9.4.2). 18. Decreasing quantification qualified out of the blue by perspective is self-dealing in that there is always a perspective narrow enough for nothing much to be going on within it, thus vacuously verifying the assertion. With this in mind, the constraint on perspective (Schein 1993, 234) says that whatever its use in securing reference, perspective should not intrude on the description of how things are in the world; a perspective-laden assertion should always yield a perspective-free truth. 19. A free parameter for demonstrative perspectives, and existentially quantified, as shown, when no particular perspective is at hand. 20. The logical form (43) withholds from the clausal architecture of tensed clauses any declaration that the subject centers the adopted perspective. Besides the temporal relations that Tense describes, in such a proposal a higher relation would be explicit in its declaration that events in which the subject participates center or foreground the perspective at hand. In such a proposal, the higher relation describes presentational events that are robust as presentations, as if to assert on a subatomic level that there were—behold—18 barrels, of which it is true that so-and-so. At issue is the extent to which differences between (in)definite description and other quantification intrude into the sentence structure to correlate with the prima facie differences between the DPs. Note that (in)definite descriptions as the foci of presentational sentences and as the subjects of simple sentences pattern together and contrast with other quantification, even to belie the alleged equivalence of few and not many: (i) (ii)
Not many square miles were flooded. There were not many square miles flooded.
(iii) *Few square miles were flooded. (iv) *There were few square miles flooded. (v) (vi)
Not many feet are left before we strike bedrock. There are not many feet left before we strike bedrock.
(vii) *Few feet are left before we strike bedrock. (viii) *There are few feet left before we strike bedrock. (ix) (x)
Not many rows and columns are in the table on page 18. There are not many rows and columns in the table on page 18.
(xi) *Few rows and columns are in the table on page 18. (xii) *There are few rows and columns in the table on page 18. Wine stains, but since it was not served at tables set with this heirloom tablecloth, (xiii) a. b. (xiv) a. b.
Not many spills stained it. Not many stains stain it. There were not many spills staining it. There are not many stains staining it.
Notes
(xv) a. b. (xvi) a. b.
971
*Few spills stained it. *Few stains stain it. *There were few spills staining it. (Cf. There were few spills to stain it.) *There are few stains staining it.
The contrasts are reminiscent of discussion relating the strong/weak contrast among DPs to their partitivity—that is, to the understanding that the NP restriction in [D NP] itself bears definite reference and the relation of these DP properties to stage- and individual-level predication so-called, where the latter implicates some correlation between DP and clausal structure (see Herburger 1997 and the references cited there). Appendix 2
Eventish
1. See also Cartwright 1996.
Bibliography
Abusch, Dorit. 2012. Applying discourse semantics and pragmatics to visual narrative in silent comics and Indian art. Talk at “Visual Narrative: An Interdisciplinary Workshop at UCLA,” June 20–22, 2012. Aloimonos, Yiannis. 1997. Visual Navigation: From Biological Systems to Unmanned Ground Vehicles. Mahwah, NJ: Erlbaum. Aoun, Joseph, and Elabbas Benmamoun. 1996. Agreement, coordination and gapping. Fragments Workshop. SOAS, September 2, 1996. Handout. Aoun, Joseph, Elabbas Benmamoun, and Dominique Sportiche. 1994. Agreement, word order and conjunction in some varieties of Arabic. Linguistic Inquiry 25 (2): 195–220. Aoun, Joseph, Elabbas Benmamoun, and Dominique Sportiche. 1999. Further remarks on first conjunct agreement. Linguistic Inquiry 30 (4): 669–681. Bach, Emmon, and Robert Harms, eds. 1968. Universals in Linguistic Theory. New York: Holt, Rinehart & Winston. Bach, Kent. 1994. Conversational impliciture. Mind & Language 9:124–162. Bach, Kent. 2011, August 31. Review of François Recanati, Truth-Conditional Pragmatics Notre Dame Philosophical Reviews. http://ndpr.nd.edu/news/25657-truth-conditional -pragmatics-2. Banfield, Ann. 1982. Unspeakable Sentences: Narration and Representation in the Language of Fiction. London: Routledge & Kegan Paul. Barber, Alex. 2000. A pragmatic treatment of simple sentences. Analysis 60 (4): 300–308. Barker, Chris. 1999. Individuation and quantification. Linguistic Inquiry 30 (4): 683–691. Barwise, Jon. 1979. On branching quantifiers in English. Journal of Philosophy 78:369–397. Barwise, Jon, and Robin Cooper. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4:159–219. Bäuerle, R., and Urs Egli. 1985. Anapher, Nominalphrase und Eselsätze. Papier 105 des Sonderforschungsbereichs 99, Universität Konstanz. Beardsley, Monroe. 1975. Actions and events: The problem of individuation. American Philosophical Quarterly 12 (4): 263–276.
974
Bibliography
Bezuidenhout, Anne L. 2002. Truth-conditional pragmatics. In James E. Tomberlin, ed., Philosophical Perspectives 16: Language and Mind, 105–134. Malden, MA: Blackwell. Blanchette, Patricia. 1999. Relative identity and cardinality. Canadian Journal of Philosophy 29 (2): 205–224. Bolinger, Dwight L. 1967. Adjectives in English: Attribution and predication. Lingua 18:1–34. Brentano, Franz. [1874] 1973. Psychology from an Empirical Point of View. Trans. A. C. Rancurello et al. London: Routledge & Kegan Paul. Bresnan, Joan, and Hoskuldur Thràinnson. 1990. A note on Icelandic coordination. In Joan Maling and Annie Zaenen, eds., Syntax and Semantics 24: Modern Icelandic Syntax, 355–365. New York: Academic Press. Brisson, Christine. 1997. Distributivity without (very much) scope. In Rutgers Working Papers in Linguistics, vol. 1, 29–58. New Brunswick, NJ: Rutgers Linguistics Graduate Student Association (LGSA), Department of Linguistics, Rutgers University. Brisson, Christine. 1998. Distributivity, Maximality, and Floating Quantifiers. Doctoral dissertation, Rutgers University. Brisson, Christine. 2003. Plurals, all and the nonuniformity of collective predication. Linguistics and Philosophy 26 (2): 129–184. Browman, Catherine P., and Louis Goldstein. 1986. Towards an articulatory phonology. In C. Ewen and J. Anderson, eds., Phonology Yearbook 3, 219–252. Cambridge: Cambridge University Press. Browman, Catherine P., and Louis Goldstein. 1992. Articulatory phonology: An overview. Phonetica 49 (3–4): 155–180. Burge, Tyler. 1972. Truth and mass terms. Journal of Philosophy 69 (10): 263–282. Burge, Tyler. 1973. Reference and proper names. Journal of Philosophy 70:425–439. Reprinted in Peter Ludlow, ed., Readings in the Philosophy of Language, 593–608 (Cambridge, MA: MIT Press, 1997). Burge, Tyler. 1975. Mass terms, count nouns, and change. Synthese 31:459–478. Burge, Tyler. 1986. Individualism and psychology. Philosophical Review 45 (1): 3–45. Büring, Daniel, and Katharina Hartmann. 1998. Asymmetrische Koordination. Linguistische Berichte 174:172–201. Burton, Strang, and Jane Grimshaw. 1992. Coordination and VP-internal subjects. Linguistic Inquiry 23 (2): 305–313. Burzio, Luigi. 1986. Italian Syntax: A Government-Binding Approach. Dordrecht: Reidel. Camacho, José. 1996. Comitative coordination in Spanish. In Claudia Parodi, Carlos Quicoli, Mario Saltarelli, and Maria Luisa Zubizarreta, eds., Aspects of Romance Linguistics: Selected Papers from the Linguistic Symposium on Romance Languages XXIV. Washington, DC: Georgetown University Press. Camacho, José. 1997. The Syntax of NP Coordination. Doctoral dissertation, USC. Camacho, José. 2000. Structural restrictions on comitative coordination. Linguistic Inquiry 31 (2): 366–375.
Bibliography
975
Camacho, José. 2003. The Structure of Coordination: Conjunction and Agreement Phenomena in Spanish and Other Languages. Studies in Natural Language and Linguistic Theory. Berlin: Springer. Cappelen, Herman, and Ernie Lepore. 2004. Insensitive Semantics: A Defense of Semantic Minimalism and Speech Act Pluralism. Oxford: Blackwell. Carlson, Greg N. 1977. Reference to Kinds in English. Doctoral dissertation, University of Massachusetts, Amherst. Carlson, Greg N. 1982. Generic terms and generic sentences. Journal of Philosophical Logic 11 (2): 145–181. Carston, Robyn. 1988. Implicature, explicature and truth-theoretic semantics. In Ruth Kempson, ed., Mental Representations, 155–181. Cambridge: Cambridge University Press. Reprinted in S. Davis, ed., Pragmatics, 33–51 (Oxford: Oxford University Press, 1991). Cartwright, Helen Morris. 1996. Some of a plurality. In James E. Tomberlin, ed., Philosophical Perspectives 10: Metaphysics, 137–157. Malden, MA: Blackwell. Cartwright, Richard. 1994. Speaking of everything. Noûs 28:1–20. Castañeda, Hector-Neri. 1967. Comments. In Nicholas Rescher, ed., The Logic of Decision and Action. Pittsburgh: University of Pittsburgh Press. Chomsky, Noam. 1957. Syntactic Structures. The Hague: Mouton. Collins, Chris. 1988a. Part I: Conjunction adverbs. Ms., MIT. Collins, Chris. 1988b. Part II: Alternative analyses of conjunction. Ms., MIT. Collins, Chris. 1988c. Part III: The structure of CONJP. Ms., MIT. Crnič, Luka. 2009. Counting and collective predication. Abstract, Penn Linguistics Colloquium. Dalrymple, Mary, Irene Hayrapetian, and Tracy Holloway King. 1998a. The semantics of the Russian comitative construction. Natural Language and Linguistic Theory 16 (3): 597–631. Dalrymple, Mary, Makoto Kanazawa, Yookyung Kim, Sam Mchombo, and Stanley Peters. 1998b. Reciprocal expressions and the concept of reciprocity. L&P 21 (2): 159–210. Davidson, Donald. 1967. The logical form of action sentences. In Nicholas Rescher, ed., The Logic of Decision and Action. Pittsburgh: University of Pittsburgh Press. Reprinted in Donald Davidson, Essays on Actions and Events, 105–122 (Oxford: Oxford University Press, 1980). Davidson, Donald. 1980. Essays on Actions and Events. Oxford: Oxford University Press. Davidson, Donald. 1985. Adverbs of action. In B. Vermazen and M. B. Hintikka, eds., Essays on Davidson: Actions and Events, 230–241. Oxford: Clarendon Press. Davies, Martin. 1981. Meaning, Quantification, Necessity: Themes in Philosophical Logic. London: Routledge and Kegan Paul. Davies, Martin. 1991. Acts and scenes. In N. Cooper and P. Engel, eds., New Inquiries into Meaning and Truth, 41–82. New York: St. Martin’s Press. Dayal, Veneeta. 1998. Any as inherently modal. Linguistics and Philosophy 21 (5): 433–476. De Swart, Henriette. 1991. Adverbs of Quantification: A Generalised Quantifier Approach. Doctoral dissertation, Rijksuniversiteit Groningen.
976
Bibliography
Doetjes, Jenny, and Martin Honcoop. 1997. The semantics of event-related readings: A case for pair-quantification. In Anna Szabolcsi, ed., Ways of Scope Taking, 263–310. Dordrecht: Kluwer Academic. Donnellan, Keith. 1978. Speaker reference, descriptions, and anaphora. In Peter Cole, ed., Syntax and Semantics 9: Pragmatics, 47–68. New York: Academic Press. Doron, Edit. 1999. VSO and left-conjunct agreement: Biblical Hebrew vs. Modern Hebrew. In Andrew Carnie and Eithne Guilfoyle, eds., The Syntax of Verb-Initial Languages, 75–95. Oxford: Oxford University Press. Dowty, David. 1977. Toward a semantic analysis of verb aspect and the English “imperfective” progressive. Linguistics and Philosophy 1 (1): 45–77. Dowty, David. 1979. Word Meaning and Montague Grammar: The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ. Studies in Linguistics and Philosophy, vol. 7. Dordrecht: Reidel. Dowty, David. 1987. Collective predicates, distributive predicates and all. In F. Marshall, ed., Proceedings of the Third ESCOL, Ohio State University. Dresner, Eli. 2001. Tarski’s restricted form and Neale’s quantificational treatment of proper names. Linguistics and Philosophy 24 (4): 405–415. Elbourne, Paul. 2005. Situations and Individuals. Current Studies in Linguistics 41. Cambridge, MA: MIT Press. Elbourne, Paul. 2008. Implicit content and the argument from binding. In Tova Friedman and Satoshi Ito, eds., Proceedings of Semantics and Linguistic Theory (SALT) 18, 284–301. Ithaca, NY: Cornell University. Enç, Mürvet. 1986. Towards a referential analysis of temporal expressions. Linguistics and Philosophy 9 (4): 405–426. Evans, Gareth. 1973. The causal theory of names. In Proceedings of the Aristotelian Society 47, 187–225. Reprinted in Gareth Evans, Collected Papers, 1–24 (Oxford: Oxford University Press, 1985). Evans, Gareth. 1977. Pronouns, quantifiers and relative clauses (I). Canadian Journal of Philosophy 7 (3): 467–536. Reprinted in Gareth Evans, Collected Papers, 76–152 (Oxford: Oxford University Press, 1985). Evans, Gareth. 1980. Pronouns. Linguistic Inquiry 11 (2): 337–362. Reprinted in Gareth Evans, Collected Papers, 322–342 (Oxford: Oxford University Press, 1985). Evans, Gareth. 1985. Collected Papers. Oxford: Oxford University Press. Fiengo, Robert, and Robert May. 1994. Indices and Identity. Linguistic Inquiry Monograph 24. Cambridge, MA: MIT Press. Finer, Daniel. 1985. The syntax of switch-reference. Linguistic Inquiry 16:35–55. von Fintel, Kai. 1989. Theticity in generative grammar. In Emmon Bach, Angelika Kratzer, and Barbara Partee, eds., Papers on Quantification. NSF Grant BNS 8719999. Amherst: Department of Linguistics, University of Massachusetts. Fodor, Janet Dean. 1970. The Linguistic Description of Opaque Contexts. Doctoral dissertation, MIT. Forbes, Graeme. 1997. How much substitutivity? Analysis 57 (2): 109–113.
Bibliography
977
Forbes, Graeme. 1999. Enlightened semantics for simple sentences. Analysis 59 (2): 86–91. Fowler, H. W. 1908. The King’s English. 2nd ed. Oxford: Oxford University Press. Fox, Danny. 1995. Economy and scope. Natural Language Semantics 3 (3): 283–341. Fox, Danny. 2000. Economy and Semantic Interpretation. Linguistic Inquiry Monograph 35. Cambridge, MA: MIT Press. Gajewski, Jon. 2005. On the excluded middle. In Christopher Davis, Amy Rose Deal, and Youri Zabbal, eds., Proceedings of NELS 36, 279–290. Amherst: GLSA, University of Massachusetts. Gallistel, C. R. 1990. The Organization of Learning. Cambridge, MA: MIT Press. Gallistel, C. R., and Adam Phillip King. 2009. Memory and the Computational Brain: Why Cognitive Science Will Transform Neuroscience. Malden, MA: Wiley-Blackwell. Gawron, Jean Mark, and Andrew Kehler. 2002. The semantics of the adjective respective. In L. Mikkelsen and C. Potts, eds., Proceedings of WCCFL 21, 85–98. Somerville, MA: Cascadilla Press. Gawron, Jean Mark, and Andrew Kehler. 2004. The semantics of respective readings, conjunction, and filler-gap dependencies. Linguistics and Philosophy 27:169–207. Geach, Peter T. 1962. Reference and Generality. Ithaca, NY: Cornell University Press. Georgi, Geoff. 2011. Demonstratives in Logic and Natural Language. Doctoral dissertation, USC. Gillon, Brendan. 1990. Plural noun phrases and their readings: A reply to Lasersohn. Linguistics and Philosophy 13:477–485. Giorgi, Alessandra, and Fabio Pianesi. 1997. Tense and Aspect: From Semantics to Morphosyntax. Oxford Studies in Comparative Syntax. Oxford: Oxford University Press. Gleitman, Lila R. 1965. Coördinating conjunctions in English. Language 41:260–293. Reprinted in D. A. Reibel and Sanford A. Schane, eds., Modern Studies in English: Readings in Transformational Grammar, 80–112 (New York: Prentice-Hall, 1969). Gleitman, Lila R., Henry Gleitman, Carol Miller, and Ruth Ostrin. 1996. Similar, and similar concepts. Cognition 58 (3): 321–376. Godard, Danièle. 1989. Empty categories as subjects of tensed Ss in English or French? Linguistic Inquiry 20:497–506. Goldsmith, John. 1985. A principled exception to the Coordinate Structure Constraint. In William Eilfort, Paul D. Kroeber, and Karen L. Peterson, eds., CLS 21, Vol. 1: Papers from the General Session, 133–143. Chicago: Chicago Linguistic Society, University of Chicago. Golledge, Reginald G. 1999a. Human wayfinding and cognitive maps. In Reginald G. Golledge, ed., Wayfinding Behavior: Cognitive Mapping and Other Spatial Processes, 5–45. Baltimore: Johns Hopkins University Press. Golledge, Reginald G., ed. 1999b. Wayfinding Behavior: Cognitive Mapping and Other Spatial Processes. Baltimore: Johns Hopkins University Press. Goodman, Nelson. 1947. The problem of counterfactual conditionals. Journal of Philosophy 44:113–128. Reprinted in Frank Jackson, ed., Conditionals, Oxford Readings in Philosophy, 9–27 (Oxford: Oxford University Press, 1991).
978
Bibliography
Gupta, Anil. 1980. The Logic of Common Nouns: An Investigation in Quantified Modal Logic. New Haven, CT: Yale University Press. Hackl, Martin. 2001a. Comparative Quantifiers. Doctoral dissertation, MIT. Hackl, Martin. 2001b. Comparative quantifiers and plural predication. In K. Megerdoomian and L. A. Bar-el, eds., Proceedings of WCCFL 20, 234–247. Somerville, MA: Cascadilla Press. Hackl, Martin. 2002. The ingredients of essentially plural predicates. In Mako Hirotani, ed., Proceedings of NELS 32, 171–182. Amherst: GLSA, University of Massachusetts. Hackl, Martin. 2003. On the internal composition of comparative quantifiers: Some semantics and some processing. Ms., Pomona College. Hackl, Martin. 2009. On the grammar and processing of proportional quantifiers: Most versus more than half. Natural Language Semantics 17 (1): 63–98. Haiman, John. 1983. On some origins of switch-reference marking. In John Haiman and Pamela Munro, eds., Switch Reference and Universal Grammar: Proceedings of a Symposium on Switch Reference and Universal Grammar, Winnipeg, May 1981, Typological Studies in Language, 105–128. Amsterdam: John Benjamins. Haiman, John, and Pamela Munro, eds. 1983. Switch Reference and Universal Grammar: Proceedings of a Symposium on Switch Reference and Universal Grammar, Winnipeg, May 1981. Typological Studies in Language. Amsterdam: John Benjamins. Hale, Kenneth, and Jeanne LaVerne. 1976. Notes for Hopi Workshop. Ms., University of Arizona, Tucson. Harman, Gilbert. 1972. Deep structure as logical form. In Donald Davidson and Gilbert Harman, eds., Semantics of Natural Language. Dordrecht: Reidel. Heim, Irene. 1982. The Semantics of Definite and Indefinite Noun Phrases. Doctoral dissertation, University of Massachusetts. Heim, Irene. 1987. Where does the definiteness restriction apply? Evidence from the definiteness of variables. In Alice ter Meulen and Eric Reuland, eds., The Linguistic Representation of (In)definiteness, 21–42. Cambridge, MA: MIT Press. Heim, Irene. 1990. E-type pronouns and donkey anaphora. Linguistics and Philosophy 13 (2): 137–177. Heim, Irene, Howard Lasnik, and Robert May. 1991. Reciprocity and plurality. Linguistic Inquiry 22 (1): 63–101. Herburger, Elena. 1997. Focus and weak Noun Phrases. Natural Language Semantics 5 (1): 53–78. Herburger, Elena. 2000. What Counts: Focus and Quantification. Linguistic Inquiry Monograph 36. Cambridge, MA: MIT Press. Herburger, Elena. 2015. Only if: If only we understood it. In Sarah D’Antonio, Mary Moroney, and Carol Rose Little, eds., Proceedings of Semantics and Linguistic Theory (SALT) 25, 284–301. Ithaca, NY: Cornell University. Herburger, Elena, and Simon Mauck. 2015. Conditionally strict conditionals. Ms., Georgetown University. Heycock, Caroline. 1995. Asymmetries in reconstruction. Linguistic Inquiry 26 (4): 547–570.
Bibliography
979
Heycock, Caroline, and Anthony Kroch. 1993. Verb movement and coordination in Germanic languages: Evidence for a relational perspective on licensing. Ms., Yale University and University of Pennsylvania. Heycock, Caroline, and Anthony Kroch. 1994. Verb movement and coordination in a dynamic theory of licensing. Linguistic Review 11:257–283. Heycock, Caroline, and Roberto Zamparelli. 1999. Towards a unified analysis of DP conjunction. In Paul Dekker, ed., Proceedings of the Twelfth Amsterdam Colloquium, 127–132. Amsterdam: Institute for Logic, Language and Computation (ILLC), University of Amsterdam. Heycock, Caroline, and Roberto Zamparelli. 2003. Coordinated bare definites. Linguistic Inquiry 34 (3): 443–470. Heycock, Caroline, and Roberto Zamparelli. 2005. Friends and colleagues: Plurality, coordination, and the structure of DP. Natural Language Semantics 13:201–270. Higginbotham, James. 1987. Indefiniteness and predication. In Alice ter Meulen and Eric Reuland, eds., The Representation of (In)definiteness, 43–70. Cambridge, MA: MIT Press. Higginbotham, James. 1988. Contexts, models, and meaning: A note on the data of semantics. In Ruth Kempson, ed., Mental Representations: The Interface between Language and Reality, 29–48. Cambridge: Cambridge University Press. Higginbotham, James. 2003. Conditionals and compositionality. In John Hawthorne and Dean Zimmerman, eds., Philosophical Perspectives 17: Language and Philosophical Linguistics, 181–194. Malden, MA: Blackwell. Higginbotham, James, and Barry Schein. 1989. Plurals. In Juli Carter and Rose-Marie Déchaine, eds., Proceedings of NELS 19, 161–175. Amherst: GLSA, University of Massachusetts. Hinterhölzl, Roland. 1999. Restructuring Infinitives and the Theory of Complementation. Doctoral dissertation, USC. Hinterhölzl, Roland. 2006. Scrambling, Remnant Movement, and Restructuring in West Germanic. Oxford: Oxford University Press. Hoeksema, Jack. 1983. Plurality and conjunction. In Alice ter Meulen, ed., Studies in Modeltheoretic Semantics, GRASS 1, 63–83. Dordrecht: Foris. Hoeksema, Jack. 1988. The semantics of non-Boolean “and.” Journal of Semantics 6:19–40. Höhle, Tilman. 1990. Assumptions about asymmetric coordination. In Joan Mascaró and Marina Nespor, eds., Grammar in Progress: GLOW Essays for Henk van Riemsdijk, 221–235. Dordrecht: Foris. Höhle, Tilman. 1991. On reconstruction and coordination. In Hubert Haider and K. Netter, eds., Representation and Derivation in the Theory of Grammar, 139–197. Dordrecht: Kluwer. Hornstein, Norbert. 1995. Logical Form: From GB to Minimalism. Cambridge, MA: Blackwell. Hornstein, Norbert, and Amy Weinberg. 1990. The necessity of LF. Linguistic Review 7 (2): 129–167. Iatridou, Sabine. 1994. On the contribution of conditional then. Natural Language Semantics 2:171–199.
980
Bibliography
Iatridou, Sabine, Elena Anagnastopoulou, and Roumyana Izvorski 2001. Observations about the from and meaning of the perfect. In Michael Kenstowicz, ed., Ken Hale: A Life in Language, 189–238. Cambridge, MA: MIT Press. Ionin, Tania, and Ora Matushansky. 2002. Encasing the time: Temporal effects on Russian predicative Case. Paper presented at NELS 33, MIT, Cambridge, MA. Ionin, Tania, and Ora Matushansky. 2004a. The non-existence, syntax and semantics of numerals. Invited talk at the Studientag: “Nouns and Pronouns—Typology and Theoretical Problems,” Freie Universität Berlin, April 2, 2004. Ionin, Tania, and Ora Matushansky. 2004b. A singular plural. In Benjamin Schmeiser, Vineeta Chand, Ann Kelleher, and Angelo Rodriguez, eds., Proceedings of WCCFL 23, 101–114. Somerville, MA: Cascadilla Press. Ionin, Tania, and Ora Matushansky. 2006. The composition of complex cardinals. Journal of Semantics 23 (4): 315–360. Jackendoff, Ray. 1971. Gapping and related rules. Linguistic Inquiry 2:21–35. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press. Jackendoff, Ray. 1977. X′ Syntax: A Study of Phrase Structure. Linguistic Inquiry Monograph 2. Cambridge, MA: MIT Press. Jayaseelan, K. A. 2011. Comparative morphology of quantifiers. Lingua 121:269–286. Johannessen, Janne Bondi. 1998. Coordination. Oxford Studies in Comparative Syntax. New York: Oxford University Press. Johnson, Kyle. [1996] 2003. In search of the English middle field. Ms., University of Massachusetts, Amherst. Johnson, Kyle. 2000a. Few dogs eat Whiskas or cats Alpo. In Kiyomi Kusumoto and Elisabeth Villalta, eds., UMOP 23: Issues in Semantics and Its Interface, 59–82. Amherst: GLSA, University of Massachusetts. Johnson, Kyle. 2000b. How far will quantifiers go? In Roger Martin, David Michaels, and Juan Uriagereka, eds., Step by Step, 187–210. Cambridge, MA: MIT Press. Johnson, Kyle. 2002. Restoring exotic coordinations to normalcy. Linguistic Inquiry 33 (1): 97–156. Johnson, Kyle. 2009. Gapping is not (VP) Ellipsis. Linguistic Inquiry 40 (2): 289–328. Johnson, Kyle, and Satoshi Tomioka. 1997. Lowering and mid-size clauses. In Graham Katz, Shin-Sook Kim, and Heike Winhart, eds., Reconstruction: Proceedings of the Tübingen Workshop, 185–206. Universität Stuttgart and Universität Tübingen. Kadmon, Nirit. 1990. Uniqueness. Linguistics and Philosophy 13 (3): 273–324. Kamp, Hans. 1981. A theory of truth and semantic representation. In Jeroen Groenendijk, Theo M. V. Janssen, and Martin Stokhof, eds., Formal Methods in the Study of Language, Mathematical Centre Tracts 135, 277–322. Amsterdam: Mathematisch Centrum. Reprinted in Jeroen Groenendijk, Theo M. V. Janssen, and Martin Stokhof, eds., Truth, Interpretation and Information, 1–42 (Dordrecht: Foris, 1984). Kamp, Hans, and Uwe Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer.
Bibliography
981
Kaplan, David. 1989. Demonstratives. In J. Almog, J. Perry, and H. Wettstein, eds., Themes from Kaplan, 481–563. Oxford: Oxford University Press. Karttunen, Lauri. 1977. Discourse referents. In James D. McCawley, ed., Syntax and Semantics 7: Notes from the Linguistic Underground, 363–386. New York: Academic Press. Kathol, Andreas. 1992. On coordination and constituency in German. In C. Canakis, G. Chan, and J. Denton, eds., Proceedings of the Chicago Linguistic Society 28, 267–281. Chicago: Chicago Linguistic Society, University of Chicago. Kathol, Andreas. 1993. Linearization grammar and coordination in German. Ms., Ohio State University. Kayne, Richard S. 1993. Toward a modular theory of auxiliary selection. Studia Linguistica 47:3–31. Kayne, Richard S. 1994. The Antisymmetry of Syntax. Linguistic Inquiry Monograph 25. Cambridge, MA: MIT Press. Kitagawa, Yoshi. 1986. Subjects in Japanese and English. Doctoral dissertation, University of Massachusetts, Amherst. Klinedinst, Nathan. 2011. Quantified conditionals and Conditional Excluded Middle. Journal of Semantics 28:149–170. Koopman, Hilda, and Dominique Sportiche. 1991. The position of subjects. Lingua 85 (2): 211–258. Koslicki, Kathrin. 1997. Isolation and non-arbitrary division: Frege’s two criteria for counting. Synthese 112 (3): 403–430. Koslicki, Kathrin. 1999. The semantics of mass-predicates. Noûs 33 (1): 46–91. Koslicki, Kathrin. 2005. Nouns, mass and count. In Donald M. Borchert, ed., Encyclopedia of Philosophy, 2nd ed., 659–666. New York: Macmillan Reference. Koslicki, Kathrin. 2008. The Structure of Objects. Oxford: Oxford University Press. Kratzer, Angelika. 1995. Stage-level and individual-level predicates. In G. N. Carlson and F. J. Pelletier, eds., The Generic Book, 125–175. Chicago: University of Chicago Press. Kratzer, Angelika. 1996. Severing the external argument from the verb. In Johan Rooryck and Laurie Zaring, eds., Phrase Structure and the Lexicon, 109–137. Dordrecht: Kluwer. Kratzer, Angelika. 2003. The Event Argument. Unpublished ms. http://semanticsarchive.net/ Archive/GU1NWM4Z. Krifka, Manfred. 1990. Four thousand ships passed through the lock. Linguistics and Philosophy 13 (5): 467–520. Kripke, Saul. 1980. Naming and Necessity. Oxford: Oxford University Press. Kuroda, Shige-Yuki. 1972. The categorical and the thetic judgment: Evidence from Japanese. Foundations of Language 9:153–185. Kuroda, Shige-Yuki. 1988. Whether we agree or not. Linguisticae Investigationes 12:1–47. Kuroda, Shige-Yuki. 1992. Judgment forms and sentence forms. In Shige-Yuki Kuroda, ed., Japanese Syntax and Semantics: Collected Papers, 13–77. Dordrecht: Kluwer. Ladusaw, William A. 1994. Thetic and categorical, stage and individual, weak and strong. In Mandy Harvey and Lynn Santelmann, eds., Proceedings of Semantics and Linguistic Theory (SALT) 4, 220–229. Ithaca, NY: Cornell University.
982
Bibliography
Lakoff, George. 1965. On the nature of syntactic irregularity. Mathematical Linguistics and Automatic Translation, Report No. NSF-16. Cambridge, MA: Computation Laboratory of Harvard University. Lakoff, George. 1986. Frame semantic control of the Coordinate Structure Condition. In Anne M. Farley, Peter T. Farley, and Karl-Erik McCullough, eds., CLS 22, Vol. 2: Papers from the Parasession on Pragmatics and Grammatical Theory, 152–167. Chicago: Chicago Linguistic Society, University of Chicago. Lakoff, George, and Stanley Peters. 1969. Phrasal conjunction and symmetric predicates. In D. A. Reibel and Sanford A. Schane, eds., Modern Studies in English: Readings in Transformational Grammar, 113–142. Englewood Cliffs, NJ: Prentice-Hall. Landman, Fred. 1989a. Groups I. Linguistics and Philosophy 12 (5): 559–605. Landman, Fred. 1989b. Groups II. Linguistics and Philosophy 12 (6): 723–744. Langendoen, Terence D. 1978. The logic of reciprocity. Linguistic Inquiry 9:177–197. Larson, Richard. 1983. Restrictive Modification. Doctoral dissertation, University of Wisconsin, Madison. Larson, Richard. 1998. Events and modification in nominals. In D. Strolovitch and A. Lawson, eds., Proceedings of Semantics and Linguistic Theory (SALT) VIII. Ithaca, NY: Cornell University. Larson, Richard. 2003. Time and event measure. In John Hawthorne and Dean Zimmerman, eds., Philosophical Perspectives 17: Language and Philosophical Linguistics, 247–258. Malden, MA: Blackwell. Larson, Richard, and Franc Marušič. 2004. Indefinite pronoun structures with APs. Linguistic Inquiry 35:268–287. Larson, Richard, and Gabriel Segal. 1995. Knowledge of Meaning. Cambridge, MA: MIT Press. Larson, Richard, and Naoko Takahashi. 2007. Order and interpretation in prenominal relative clauses. In M. Kelepir and B. Öztürk, eds., MIT Working Papers in Linguistics: Proceedings of the Workshop on Altaic Formal Linguistics II, 101–120. Cambridge, MA: MITWPL, Department of Linguistics and Philosophy, MIT. Lasersohn, Peter. 1989. On the readings of plural Noun Phrases. Linguistic Inquiry 20 (1): 130–134. Lasersohn, Peter. 1995. Plurality, Conjunction and Events. Dordrecht: Kluwer Academic. Leslie, Sarah Jane. 2009. If, unless and quantification. In R. J. Stainton and C. Viger, eds., Compositionality, Context and Semantic Values: Essays in Honour of Ernie Lepore, Vol. 85: Studies in Linguistics and Philosophy, 3–30. Dordrecht: Springer Netherlands. Levine, Robert D. 1984. Against “reanalysis transformations.” Linguistic Analysis 14:3–30. Levine, Robert D. 1985. Right Node (Non-)Raising. Linguistic Inquiry 16 (3): 492–497. Liberman, Mark. 1974. On conditioning the rule of subject-auxiliary inversion. In Ellen Kaisse and Jorge Hankamer, eds., Proceedings of NELS 5, 77–91. Amherst: GLSA, University of Massachusetts. Lin, Vivian. 2000. Determiner sharing. In R. Billerey and B. D. Lillehaugen, eds., Proceedings of WCCFL 19, 274–287. Somerville, MA: Cascadilla Press.
Bibliography
983
Lin, Vivian. 2001. A way to undo A-movement. In Karine Megerdoomian and Leora A. Barel, eds., Proceedings of WCCFL 20, 358–371. Somerville, MA: Cascadilla Press. Lin, Vivian I-Wen. 2002. Coordination and Sharing at the Interfaces. Doctoral dissertation, MIT. Löbner, Sebastian. 1985. Definites. Journal of Semantics 4:279–326. Lohndal, Terje. 2014. Phrase Structure and Argument Structure: A Case Study of the SyntaxSemantics Interface. Oxford Studies in Theoretical Linguistics. Oxford: Oxford University Press. Longobardi, Giuseppe. 1994. Reference and proper names. Linguistic Inquiry 25:609–665. Lønning, Jan Tore. 1987. Mass terms and quantification. Linguistics and Philosophy 10 (1): 1–52. Loomis, Jack M., Roberta L. Klatzky, Reginald G. Golledge, and John W. Philbeck. 1999. Human navigation by path integration. In Reginald G. Golledge, ed., Wayfinding Behavior: Cognitive Mapping and Other Spatial Processes, 125–151. Baltimore: Johns Hopkins University Press. Ludlow, Peter. 1994. Conditionals, events and unbound pronouns. Lingua e Stile 29 (2): 165–183. Ludlow, Peter. 2008. Cheap contextualism. Philosophical Issues 18: Interdisciplinary Core Philosophy, 104–129. Ludlow, Peter. 2014. Living Words: Meaning Underdetermination and the Dynamic Lexicon. New York: Oxford University Press. Lycan, Willam G. 1984. A syntactically motivated theory of conditionals. In P. French, T. E. Uehling, and H. Wettstein, eds., Midwest Studies in Philosophy IX: Causation and Causal Theories, 437–455. Minneapolis: University of Minnesota Press. Lycan, William G. 1991. Even and even if. Linguistics and Philosophy 14 (2): 115–150. Lycan, William G. 2001. Real Conditionals. Oxford: Clarendon Press. Manzini, Maria Rita, and Leornardo M. Savoia. 2005. I dialetti italiani e romance: Morfosintassi generative. 3 vols. Alessandria, Italy: Edizioni dell ’Orso. Martí, Luisa. 2006. Unarticulated constituents revisited. Linguistics and Philosophy 29 (2): 135–166. May, Robert. 1977. The Grammar of Quantification. Doctoral dissertation, MIT. McCawley, James D. 1968. The role of semantics in grammar. In Emmon Bach and Robert Harms, eds., Universals in Linguistic Theory, 125–169. New York: Holt, Rinehart & Winston. McCawley, James D. 1971. Tense and time reference in English. In Charles J. Fillmore, and D. Terence Langendoen, eds., Studies in Linguistic Semantics, 97–113. New York: Holt, Rinehart and Winston. McCawley, James D. 1981a. Everything That Linguists Have Always Wanted to Know about Logic: But Were Ashamed to Ask. Chicago: University of Chicago Press. McCawley, James D. 1981b. Notes on the English perfect. Australian Journal of Linguistics 1: 81–90. McCawley, James D. 1982. Parentheticals and discontinuous constituent structure. Linguistic Inquiry 13 (1): 91–106.
984
Bibliography
McCawley, James. D. 1988. The Syntactic Phenomena of English. Vol. 2. Chicago: University of Chicago Press. McCawley, James D. 1993. Gapping with shared operators. In Joshua S. Guenter, Barbara A. Kaiser, and Cheryl C. Zoll, eds., Proceedings of the Nineteenth Annual Meeting of the Berkeley Linguistics Society, 245–253. Berkeley: Berkeley Linguistics Society, University of California. McCloskey, James. 1986. Right Node Raising and Preposition Stranding. Linguistic Inquiry 17 (1): 183–186. McCoard, R. W. 1978. The English Perfect: Tense Choice and Pragmatic Inferences. Amsterdam: North-Holland. McNally, Louise. 1992. VP coordination and the VP-internal subject hypothesis. Linguistic Inquiry 23 (2): 336–341. McNally, Louise. 1993. Comitative coordination: A case study in group formation. Natural Language and Linguistic Theory 11:347–379. McNally, Louise. 1998. Stativity and theticity. In Susan Rothstein, ed., Events and Grammar, 293–307. Dordrecht: Kluwer. Milsark, Gary. 1977. Toward an explanation of certain peculiarities of the Existential Construction in English. Linguistic Analysis 3:1–29. Mitchell, Jonathan. 1986. The Formal Semantics of Point of View. Doctoral dissertation, University of Massachusetts, Amherst. Moltmann, Friederike. 1994. Together and alone. Ms., UCLA. Moltmann, Friederike. 1997. Parts and Wholes in Semantics. Oxford: Oxford University Press. Moore, Joseph G. 1999. Saving substitutivity in simple sentences. Analysis 59 (2): 91–105. Moore, Steven. 1994. Events, objects and eventual objects. In Erin Duncan, Donka Farkas, and Philip Spaelti, eds., WCCFL 12: The Proceedings of the Twelfth West Coast Conference on Formal Linguistics, conference held at UCSC, 1993, 577–590. Stanford, CA: CSLI, Stanford University. Munn, Alan. 1993. Topics in the Syntax and Semantics of Coordinate Structures. Doctoral dissertation, University of Maryland. Munn, Alan. 1999. First conjunct agreement: Against a clausal analysis. Linguistic Inquiry 30 (4): 643–668. Musan, Renate. 1997. Tense, predicates, and lifetime effects. Natural Language Semantics 5 (3): 271–301. Oehrle, Richard T. 1987. Boolean properties in the analysis of gapping. In G. J. Huck and A. E. Ojeda, eds., Syntax and Semantics 20: Discontinuous Constituency, 203–240. San Diego: Academic Press. Pancheva, Roumyana. 2003. The aspectual makeup of perfect participles and the interpretations of the perfect. In A. Alexiadou, M. Rathert, and A. von Stechow, eds., Perfect Explorations, 277–306. Berlin: Mouton de Gruyter. Parsons, Terence. 1990. Events in the Semantics of English: A Study in Subatomic Semantics. Cambridge, MA: MIT Press.
Bibliography
985
Parsons, Terence. 2000. Underlying states and time travel. In James Higginbotham, Fabio Pianesi, and Achille C. Varzi, eds., Speaking of Events, 81–93. Oxford: Oxford University Press. Partee, Barbara. 1973. Some structural analogies between tenses and pronouns in English. Journal of Philosophy 70 (18): 601–609. Partee, Barbara. 1984. Nominal and temporal anaphora. Linguistics and Philosophy 7:243–286. Partee, Barbara. 1989. Binding implicit variables in quantified contexts. In Caroline Wiltshire, Randolph Graczyk, and Bradley Music, eds., Proceedings of CLS 25, 342–365. Partee, Barbara, and Mats Rooth. 1983. Generalized conjunction and type ambiguity. In Rainer Bäuerle, Christoph Schwarze, and Arnim von Stechow, eds., Meaning, Use, and the Interpretation of Language, 361–383. Berlin: Walter de Gruyter. Peacocke, Christopher. 1981. Demonstrative thought and psychological explanation. Synthese 49:187–217. Pereltsvaig, Asya. 2001. On the Nature of Intra-Clausal Relations: A Study of Copular Sentences in Russian and Italian. Doctoral dissertation, McGill University. Perlmutter, David M., and John Robert Ross. 1970. Relative clauses with split antecedents. Linguistic Inquiry 1 (3): 350. Pesetsky, David. 1995. Zero Syntax: Experiencers and Cascades. Current Studies in Linguistics 27. Cambridge, MA: MIT Press. Pietroski, Paul M. 1998. Actions, adjuncts and agency. Mind 107:73–111. Pietroski, Paul M. 2000. Causing Actions. Oxford: Oxford University Press. Pietroski, Paul M. 2003. Quantification and second-order monadicity. In John Hawthorne and Dean Zimmerman, eds., Philosophical Perspectives 17: Language and Philosophical Linguistics, 259–298. Malden, MA: Blackwell. Pietroski, Paul M. 2005. Events and Semantic Architecture. Oxford: Oxford University Press. Pietroski, Paul M. 2006. Interpreting concatenation and concatenates. Philosophical Issues 16 (1): 221–245. Platts, Mark. 1979. Ways of Meaning: An Introduction to a Philosophy of Language. London: Routledge & Kegan Paul. Platts, Mark. 1997. Ways of Meaning: An Introduction to a Philosophy of Language. 2nd ed. Cambridge, MA: MIT Press. Poesio, Massimo, and Alessandro Zucchi. 1992. On telescoping. In Chris Barker and David Dowty, eds., Proceedings of Semantics and Linguistic Theory (SALT) 2, 347–366. Published as Ohio State University Working Papers in Linguistics 40. Columbus: Ohio State University. Postal, Paul M. 1998. Three Investigations of Extraction. Current Studies in Linguistics 29. Cambridge, MA: MIT Press. Predelli, Stefano. 1999. Saul, Salmon, and Superman. Analysis 59 (2): 113–116. Predelli, Stefano. 2005. Contexts: Meaning, Truth and the Use of Language. Oxford: Oxford University Press. Progovac, Ljiljana. 1998a. Structure for coordination, Part I. Glot International 3 (7): 3–6.
986
Bibliography
Progovac, Ljiljana. 1998b. Structure for coordination, Part II. Glot International 3 (8): 3–9. Pylyshyn, Zenon W. 2001. Visual indexes, preconceptual objects, and situated vision. Cognition 80:127–158. Quine, W. V. O. 1960. Word and Object. Cambridge, MA: MIT Press. Recanati, François. 1993. Direct Reference. Oxford: Blackwell. Recanati, François. 2007. It is raining (somewhere). Linguistics and Philosophy 30 (1): 123–146. Redish, David A. 1999. Beyond the Cognitive Map: From Place Cells to Episodic Memory. Cambridge, MA: MIT Press. Reibel, D. A., and Sanford A. Schane, eds. 1969. Modern Studies in English: Readings in Transformational Grammar. New York: Prentice-Hall. Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York: Dover. Reinhart, Tanya. 1987. Specifier and operator binding. In E. Reuland and A. ter Meulen, eds., The Representation of (In)definiteness, 130–167. Cambridge, MA: MIT Press. Reinhart, Tanya, and Eric Reuland. 1993. Reflexivity. Linguistic Inquiry 12:657–720. Rescher, Nicolas, ed. 1967. The Logic of Decision and Action. Pittsburgh: University of Pittsburgh Press. Richards, Barry. 1984. On interpreting pronouns. Linguistics and Philosophy 7:287–324. Roberts, Craige. 1989. Modal subordination and pronominal anaphora in discourse. Linguistics and Philosophy 12:683–722. Roeper, Peter. 1983. Semantics for mass terms with quantifiers. Noûs 17 (3): 251–265. Rögnvaldsson, Eiríkur. 1982. We need (some kind of) a conjunction reduction rule. Linguistic Inquiry 13:557–561. Reprinted in Joan Maling and Annie Zaenen, eds., Syntax and Semantics 24: Modern Icelandic Syntax, 349–353. New York: Academic Press, 1990. Rögnvaldsson, Eiríkur. 1990. Null objects in Icelandic. In Joan Maling and Annie Zaenen, eds., Syntax and Semantics 24: Modern Icelandic Syntax, 367–379. New York: Academic Press. Root, R. 1986. The Semantics of Anaphora in Discourse. Doctoral dissertation, University of Texas, Austin. Rooth, Mats. 1987. NP Interpretation in Montague Grammar, File Change Semantics and Situation Semantics. In P. Gärdenfors, ed., Generalized Quantifiers, 237–269. Dordrecht: Reidel. Ross, John Robert. 1967. Constraints on Variables in Syntax. Doctoral dissertation, MIT. Roy, Isabelle. 2001. Predicate nominals in French. Ms., USC. Roy, Isabelle. 2006. Non-Verbal Predications: A Syntactic Analysis of Predicational Copular Sentences. Doctoral dissertation, USC. Russell, Bertrand. 1903. The Principles of Mathematics. Cambridge: Cambridge University Press. Russell, Bertrand. 1938. The Principles of Mathematics. 2nd ed. New York: Norton. Sasse, Hans Jürgen. 1987. The thetic categorical distinction revisited. Linguistics 25:511–580. Sauerland, Uli, and Paul Elbourne. 2002. Total reconstruction, PF movement, and derivational order. Linguistic Inquiry 33 (2): 283–319.
Bibliography
987
Saul, Jennifer M. 1997a. Substitution and simple sentences. Analysis 57 (2): 102–108. Saul, Jennifer M. 1997b. Reply to Forbes. Analysis 57 (2): 114–118. Saul, Jennifer M. 1999. Substitution, simple sentences, and sex scandals. Analysis 59 (2): 106–112. Saul, Jennifer M. 2007. Simple Sentences, Substitution, and Intuitions. Oxford: Oxford University Press. Scha, Remko. 1981. Distributive, collective and cumulative quantification. Reprinted in J. Groenendijk et al., eds., Truth, Interpretation and Information, GRASS 2, 131–158 (Dordrecht: Foris, 1984). Schein, Barry. 1993. Plurals and Events. Cambridge, MA: MIT Press. Schein, Barry. 2002. Events and the semantic content of thematic relations. In Gerhard Preyer and Georg Peter, eds., Logical Form and Language, 263–344. Oxford: Oxford University Press. Schein, Barry. 2003. Adverbial, descriptive reciprocals. In John Hawthorne and Dean Zimmerman, eds., Philosophical Perspectives 17: Language and Philosophical Linguistics, 333–367. Malden, MA: Blackwell. Schein, Barry. 2005. Plurals. www-rcf.usc.edu/~schein/Plurals2005.pdf. Abridged as Barry Schein, “Plurals,” in Ernest Lepore and Barry C. Smith, eds., The Oxford Handbook of Philosophy of Language, 716–767 (Oxford: Oxford University Press, 2006). Schein, Barry. 2006. Plurals. In Ernest Lepore and Barry C. Smith, eds., The Oxford Handbook of Philosophy of Language, 716–767. Oxford: Oxford University Press. Schein, Barry. 2012. Event semantics. In Gillian Russell and Delia Graff Fara, eds., The Routledge Companion to Philosophy of Language, 280–294. New York: Taylor and Francis. Schlenker, Philippe. 1999. Propositional Attitudes and Indexicality: A Cross-Categorial Approach. Doctoral dissertation, MIT. Schlenker, Philippe. 2004. Minimize Restrictors! (Notes on definite descriptions, Condition C and epithets). In Emar Maier, Corien Bary, and Janneke Huitink, eds., Proceedings of Sinn und Bedeutung 9, 385–416. Nijmegen: University of Nijmegen. Schwarz, Bernhard. 1998a. On odd coordinations in German. Journal of Comparative Germanic Linguistics 2:195–219. Schwarz, Bernhard. 1998b. Reduced conditionals in German: Event quantification and definiteness. Natural Language Semantics 6:271–301. Schwarz, Bernhard. 1999. On the syntax of either … or. Natural Language and Linguistic Theory 17:339–370. Schwarz, Bernhard. 2000. Topics in Ellipsis. Doctoral dissertation, University of Massachusetts, Amherst. Schwarzschild, Roger. 1991. On the Meaning of Definite Plural Noun Phrases. Doctoral dissertation, University of Massachusetts, Amherst. Schwarzschild, Roger. 1994. Plurals, presuppositions and the sources of distributivity. Natural Language Semantics 2 (3): 201–248. Schwarzschild, Roger. 1996. Pluralities. Dordrecht: Kluwer. Schwarzschild, Roger. 2008. Can grammar be wrong about the molecular theory of matter? IRCS Linguistics Speaker Series, University of Pennsylvania.
988
Bibliography
Schwarzschild, Roger. 2009. Stubborn distributivity, multiparticipant nouns, and the count/ mass distinction. In Suzi Lima, Kevin Mullin, and Brian Smith, eds., Proceedings of NELS 39, vol. 2, 661–678. Amherst: GLSA, University of Massachusetts. Searle, John. 1978. Literal meaning. Erkenntnis 13:207–224. Searle, John. 1980. The background of meaning. In J. Searle, F. Kiefer, and M. Bierwisch, eds., Speech Act Theory and Pragmatics, 221–232. Dordrecht: Reidel. Sennet, Adam. 2008. The binding argument and pragmatic enrichment, or, why philosophers care even more than weathermen about “raining.” Philosophy Compass 3 (1): 135–157. Sharvit, Yael. 2008. The puzzle of Free Indirect Discourse. Linguistics and Philosophy 31 (3): 353–395. Sharvy, Richard. 1980. A more general theory of definite descriptions. Philosophical Review 89 (4): 607–624. Sher, Gila. 1990. Ways of branching quantifiers. Linguistics & Philosophy 13 (4): 393–422. Reprinted (with minor revisions) in Gila Sher, The Bounds of Logic: A Generalized Viewpoint, 105–129. Cambridge, MA: MIT Press, 1991. Siegel, Muffy E. A. 1984. Gapping and interpretation. Linguistic Inquiry 15:523–530. Siegel, Muffy E. A. 1987. Compositionality, case and the scope of auxiliaries. Linguistics and Philosophy 10 (1): 53–76. Sjoblom, Todd. 1980. Coordination. Doctoral dissertation, MIT. Sommers, Fred. 1982. The Logic of Natural Language. Oxford: Oxford University Press. Sperber, Dan, and Deirdre Wilson. [1986] 1995. Relevance: Communication & Cognition. 2nd ed. Malden, MA: Blackwell. Sportiche, Dominique. 1997. Reconstruction and constituent structure. Talk at MIT. Sportiche, Dominique. 2005. Division of labor between Merge and Move: Strict locality of selection and apparent reconstruction paradoxes. In Philippe Schlenker and Dominique Sportiche, eds., Proceedings of the Workshop Divisions of Linguistic Labor, La Bretesche Workshop, Nantes, France. http://www.linguistics.ucla.edu/general/Conf/LaBretesche. Stalnaker, Robert. 1968. A theory of conditionals. In Nicholas Rescher, ed., Studies in Logical Theory, American Philosophical Quarterly Monograph Series 2, 98–112. Oxford: Blackwell. Reprinted in Frank Jackson, ed., Conditionals, Oxford Readings in Philosophy, 28–45 (Oxford: Oxford University Press, 1991). Stanley, Jason. 2000. Context and logical form. Linguistics and Philosophy 23:391–434. Stirling, Lesley. 1993. Switch-Reference and Discourse Representation. Cambridge: Cambridge University Press. Stone, Matthew. 1999. Reference to Possible Worlds. RuCCS Report 49. New Brunswick, NJ: Rutgers University. Stump, Gregory T. 1985. The Semantic Variability of Absolute Constructions. Dordrecht: Reidel. Svenonius, Peter. 2007. Spatial P in English. Ms., CASTL, University of Tromsø. Reprinted in Guglielmo Cinque and Luigi Rizzi, eds., Mapping Spatial PPs: The Cartography of Syntactic Structures, vol. 6, 127–16. Oxford: Oxford University Press, 2010.
Bibliography
989
Szabó, Zoltán Gendler. 2003. On qualification. In John Hawthorne and Dean Zimmerman, eds., Philosophical Perspectives 17: Language and Philosophical Linguistics, 385–414. Malden, MA: Blackwell. Szabolcsi, Anna, ed. 1997. Ways of Scope Taking. Dordrecht: Kluwer. Szabolcsi, Anna. 2009. Quantification. Cambridge: Cambridge University Press. Talmy, Len. 1978. Figure and ground in complex sentences. In J. Greenberg, C. Ferguson, and M. Moravcsik, eds., Universals of Human Language (IV), 625–649. Stanford, CA: Stanford University Press. Talmy, Len. 1983. How language structures space. In H. Pick and L. Acredolo, eds., Spatial Orientation: Theory, Research, and Application, 225–281. New York: Plenum Press. Tancredi, Christopher D. 1992. Delection, Deaccenting, and Presupposition. Doctoral dissertation, MIT. Taub, Alison. 1989. Collective predicates, Aktionsarten and all. In Emmon Bach, Angelika Kratzer, and Barbara Partee, eds., Papers on Quantification, 122–152. Amherst: Department of Linguistics, University of Massachusetts. Taylor, Barry. 1985. Modes of Occurrence. Oxford: Blackwell. Thiersch, Craig. 1994. Asymmetrical coordination as adjunction. Revision of GLOW version. Ms. Tomioka, Satoshi. 1993. Verb movement and tense specification in Japanese. In Jonathan Mead, ed., Proceedings of WCCFL 11, 482–494. Stanford, CA: CSLI, Stanford University. Tortora, Christina. 2010. Domains of clitic placement in finite and non-finite clauses: Evidence from a Piedmontese dialect. In R. D’Alessandro, A. Ledgeway, and I. Roberts, eds., Syntactic Variation: The Dialects of Italy, 135–149. Cambridge: Cambridge University Press. Trotsky, Leon. [1924] 1971. Lenin: Notes for a Biographer. Trans. T. Deutscher. New York: Putnam. Tversky, Amos. 1977. Features of similarity. Psychological Review 84 (4): 327–350. Tversky, Amos, and I. Gati. 1978. Studies in similarity. In Eleanor Rosch and B. Lloyd, eds., Cognition and Categorization, 79–98. Hillsdale, NJ: Erlbaum. Van Oirsouw, Robert. 1987. The Syntax of Coordination. London: Croom Helm. Vlach, Frank. 1983. On situation semantics for perception. Synthese 54:129–152. Westerståhl, Dag. 1987. Branching generalized quantifiers and natural language. In Peter Gärdenfors, ed., Generalized Quantifiers: Linguistic and Logical Approaches, 269–298. Dordrecht: Reidel. Wexler, Kenneth, and Peter W. Culicover. 1980. Formal Principles of Language Acquisition. Cambridge, MA: MIT Press. Wilder, Chris. 1994. Coordination, ATB and ellipsis. Ms., Arbeitsgruppe Strukturelle Grammatik, Max-Planck-Gesellschaft, Berlin. Wilder, Chris. 1995a. Rightward movement as leftward deletion. Ms., Arbeitsgruppe Strukturelle Grammatik, Max-Planck-Gesellschaft, Berlin. Wilder, Chris. 1995b. Some properties of ellipsis in coordination. Ms., Arbeitsgruppe Strukturelle Grammatik, Max-Planck-Gesellschaft, Berlin.
990
Bibliography
Wilder, Chris. 1997. Some properties of ellipsis in coordination. In A. Alexiadou and T. A. Hall, eds., Studies on Universal Grammar and Typological Variation, 59–107. Amsterdam: John Benjamins. Wilder, Chris. 1999. Right node raising and the LCA. In Sonya Bird, Andrew Carnie, Jason D. Haugen, and Peter Norquest, eds., Proceedings of WCCFL 18, 586–598. Somerville, MA: Cascadilla Press. Williams, Alexander. 2015. Arguments in Syntax and Semantics: Key Topics in Syntax. Cambridge: Cambridge University Press. Winter, Yoad. 1998. Flexible Boolean Semantics: Coordination, Plurality & Scope in Natural Language. Doctoral dissertation, University of Utrecht. Winter, Yoad. 2002. Flexibility Principles in Boolean Semantics: The Interpretation of Coordination, Plurality, and Scope in Natural Language. Cambridge, MA: MIT Press. Zimmermann, Thomas Ede. 2005. What’s in two names? Journal of Semantics 22:53–96. Zoerner, Cyril Edward, III. 1995. Coordination: The Syntax of &P. Doctoral dissertation, University of California, Irvine. Zwart, J.-W. 1991. Subject deletion in Dutch: A difference between subjects and topics. In M. Kas, E. Reuland, and C. Vet, eds., Language and Cognition I, 333–350. Groningen: University of Groningen. Zwarts, Joost. 1997. Vectors as relative positions: A compositional semantics of modified PPs. Journal of Semantics 14:57–86.
Name Index
Note: Locators for endnotes are of the form 000n00, where 000 is the page number on which the note occurs. A locator of the form 000n indicates the incomplete note that occurs at the top of page 000 (whose beginning is on a preceding page). Abusch, Dorit, 889n9 Aloimonos, Yiannis, 393, 436, 448–449, 892n37 Andrews, Joe, 927n17 Aoun, Joseph, 120, 122, 123, 125, 127, 131, 173–174, 191, 197, 837n27, 838n32, 838n33, 842n43, 850–851n80, 855n Aristotle, 1, 2, 3, 5, 8, 10, 22, 40, 60, 94, 125, 209, 644, 820n3, 821n, 836n25, 907n40 Bach, Kent, 832n54 Banfield, Ann, 892n Barber, Alex, 508, 509, 536, 908n42, 908n43, 909n46 Barker, Chris, 402, 551, 586, 924n2 Barwise, Jon, 76, 698 Bäuerle, Rainer, 829n Beardsley, Monroe, 16 Benmamoun, Elabbas, 120, 122, 123, 125, 127, 131, 173, 191, 197, 837n27, 838n32, 838n33, 842n43, 854–855n80, 855n Bezuidenhout, Anne L., 74, 832n54, 832n56 Bittner, Maria, 197, 855n98 Blanchette, Patricia, 928n3 Bolinger, Dwight L., 616 Brentano, Franz, 837n30, 915n68 Bresnan, Joan, 213, 274, 275, 276, 366 Brisson, Christine, 819n2, 827n36, 847n, 937n2, 939n7, 939n8, 949n42, 951–952n46 Browman, Catherine P., 889n7
Burge, Tyler, 469, 536, 833n8, 898n17, 970n15 Büring, Daniel, 260, 860n17, 869n Burton, Strang, 824n15, 858n4, 858n45, 953n41 Burzio, Luigi, 352 Buscher, Tom, 927n17 Camacho, José, 87, 99, 107, 109, 120, 122, 180, 193, 195–196, 278, 656, 821n, 837n, 837n30, 845n60, 850n, 851n, 856n106, 857n108 Cappelen, Herman, 832n54 Carlson, Greg N., 402, 551, 586, 690, 908n41, 924n1 Carston, Robyn, 832n54 Cartwright, Helen Morris, 96, 827n32, 971n1 Cartwright, Richard, 621 Castañeda, Hector-Neri, 1, 11, 15 Chomsky, Noam, 2 Choueiri, Lina, Nada, and Leila, 139, 189, 837n27, 838n32, 851n Collins, Chris, 856n Cooper, Robin, 698 Crnič, Luka, 945n Culicover, Peter W., 107 Dalrymple, Mary, 161, 164, 180, 193, 849n79, 850n, 857n107 Davidson, Donald, 1, 11, 15 Davies, Martin, 24, 85, 96, 469 Dayal, Veneeta, 232, 859n15
992
de Swart, Henriette, 96, 469 Doetjes, Jenny, 56, 386, 402, 551, 554, 561–562, 563, 564, 586, 595, 597, 681, 737, 738, 789 Donnellan, Keith, 30 Doron, Edit, 120, 121, 122 Dowty, David, 315, 819n2, 827n36, 846– 847n64, 881n19, 937n2, 939n7, 939n8, 946n24, 949n42, 951n46, 958n31, 959n34 Dresner, Eli, 834n15, 835n16, 844n53 Egli, Urs, 829n Elbourne, Paul, 96, 284, 832n55, 866–867n37, 966n1 Enç, Mürvet, 53, 469, 898n17 Evans, Gareth, 23, 28, 30, 37, 96, 469, 793, 826n27, 897n10, 900n23 Fiengo, Robert, 319 Finer, Daniel, 821n Fodor, Janet Dean, 886n8 Forbes, Graeme, 505, 908n43, 909n46 Fowler, Henry Watson, 920n83 Fox, Danny, 213, 220, 223, 249, 261, 280, 281, 286, 291, 292–293, 295, 301, 364, 365, 858n7, 870n57, 874n61, 881n16 Gajewski, Jon, 830n, 886n8 Gallistel, C. R., 393, 417, 436, 448, 888n, 892n36, 892n37 Gati, Itamar, 182 Gawron, Jean Mark, 53, 539–540, 546, 922n26 Geach, Peter T., 94 Georgi, Geoff, 920n84 Gillon, Brendan, 699, 847n, 895n3, 939n8, 960n36 Giorgi, Alessandra, 879n9, 879n11, 881n19 Gleitman, Lila R., 15, 67, 119, 182, 183, 185, 186, 190, 192, 194, 474, 502, 853n Godard, Danièle, 227, 273, 276, 277, 366 Goldsmith, John, 869n Goldstein, Louis, 889n7 Golledge, Reginald G., 393, 436, 448, 892n37 Goodman, Nelson, 906n38 Grimshaw, Jane, 824n15, 858n4, 868n45 Groenendijk, Jeroen, 859n15
Name Index
Guerzoni, Elena, 20 Gupta, Anil, 56, 385, 551, 552, 554, 555, 601, 608, 924n1, 924n3 Hackl, Martin, 604, 671, 684, 699, 819n2, 827n36, 912n50, 937n2, 939n7, 939–940n9, 943n22, 943–945n23, 949n42, 954n9, 958n31, 961n45 Haiman, John, 870n51 Hale, Kenneth, 820–821n3 Hansson, Gunnar Ólafur, 275, 833n11, 869n47 Harbour, Daniel, 867n Harman, Gilbert, 67 Hartmann, Katharina, 260, 860n17, 869n Hayrapetian, Irene, 161, 164, 180, 193, 850n Heim, Irene, 96, 405, 428, 469, 793, 794, 799–800, 809, 829n, 841n38, 844n53, 853n39, 863–864n30, 966n2, 967n3, 967n5, 968n6 Herburger, Elena, 262, 295, 689, 692, 830n, 894n1, 915n66, 938n4, 942n15, 971n Heycock, Caroline, 19–20, 621–622, 626–627, 631, 824n16, 860n17, 869n, 930n8, 930– 932n9, 936n, 953n5, 958n31, 960n37 Higginbotham, James, 295, 469, 702, 819n2, 827n36, 830n, 870n55, 898n17, 922n94, 937n2, 939n7, 949n42, 958n31, 960n43 Hinterhölzl, Roland, 847n65 Hoeksema, Jack, 86, 111, 159, 714, 842–843n43 Höhle, Tilman, 260–261, 860n17, 863n28, 868n44, 869n Honcoop, Martin, 56, 386, 402, 551, 554, 561–563, 564, 586, 595, 597, 681, 737–738, 789 Hornstein, Norbert, 253, 285, 286, 833n11 Iatridou, Sabine, 577, 967n5 Ionin, Tania, 621–622, 626, 627, 631, 917n, 933n13, 934n21 Jackendoff, Ray, 4, 75, 107, 217, 224, 262 Jayaseelan, K. A., 930n7, 934n20 Jeanne, LaVerne, 820–821n3 Johannessen, Janne Bondi, 860n17, 869n
Name Index
Johnson, Kyle, 212, 214, 217, 224, 228, 230, 236, 253–254, 285, 286, 288, 290, 297–298, 322, 354, 822n10, 859n9, 867n39, 869n, 876n2, 877n1, 878n4, 917n, 935n Kadmon, Nirit, 405, 794 Kamp, Hans, 271, 793, 799–800, 809, 841n38 Kaplan, David, 887n3 Karttunen, Lauri, 841n38 Kathol, Andreas, 860n17, 869n46 Kayne, Richard S., 107, 166, 308, 352, 876n4, 884n11 Kehler, Andrew, 53, 539–540, 546, 922n96 King, Adam Phillip, 892n37 King, Tracy Holloway, 161, 164, 180, 193, 850n Kitagawa, Yoshi, 824n15, 856n4, 868n45 Klinedinst, Nathan, 830n Koopman, Hilda, 824n15, 858n4, 868n45 Kortobi, Ibtissam, 837n27 Koslicki, Kathrin, 693, 923n99, 947n30 Kratzer, Angelika, 17, 473, 476–477, 478–482, 484, 489, 499, 546, 902n31, 909n47, 966n2 Krifka, Manfred, 551, 610, 924n1 Kripke, Saul, 887n3, 898n17 Kroch, Anthony, 860n17, 869n Kuroda, Shige-Yuki, 824n15, 837n30, 858n4, 868n45, 915n68 Ladusaw, William A., 837n30, 915n68 Lakoff, George, 15, 869n Landman, Fred, 820n, 897n12 Langendoen, Terence D., 853n89 Larson, Richard, 469, 616, 689, 847n68, 885n2, 886n8, 898n17 Lasersohn, Peter, 1, 86, 111, 159, 180, 227, 273, 277, 366, 657, 699, 847n, 856n100, 895n3, 896n Lasnik, Howard, 853n89 Lepore, Ernie, 832n54 Leslie, Sarah Jane, 830n Levine, Robert D., 107 Liberman, Mark, 907n40 Lin, Vivian, 216, 228, 312, 322, 822n10, 877n1, 878n4 Löbner, Sebastian, 886n8 Lohndal, Terje, 21, 824n15, 857n2, 858n4
993
Longobardi, Giuseppe, 898n17 Lønning, Jan Tore, 886n8 Loomis, Jack M., 449, 887n4 Ludlow, Peter, 72, 73, 96, 98, 230, 281, 469, 605, 832n54, 832n56, 833n9, 834n13, 864n Lycan, William G., 894n1, 923n98 Manzini, Maria Rita, 352, 882n4 Martí, Luisa, 832n55 Marušic, Franc, 616 Matushansky, Ora, 621, 622, 626, 627, 631, 917n, 933n13, 934n21 Mauck, Simon, 894n1 May, Robert, 287, 319, 853n89 McCawley, James D., 4, 7, 15, 75, 107, 109, 228, 577, 822n10, 845n60, 877n1 McCloskey, James, 107 McCoard, Robert W., 332, 577 McKinney-Bock, Katy, 20, 913n57 McNally, Louise, 164, 180, 193, 195–196, 202, 656, 824n15, 837n30, 850n, 855n97, 857n107, 858n4, 868n45, 915n68 Milsark, Gary, 295 Mitchell, Jonathan, 72 Mohammad, Mohammad, 837n29 Moltmann, Friederike, 180 Moore, Joseph G., 256, 504, 505, 508, 509, 908n41, 908n42, 908n43, 908n46, 909– 911n47, 913n56 Moore, Steven, 56, 385, 402, 551, 552, 554, 555, 561, 562, 571, 581, 586, 601, 608, 610, 612, 924n2, 926n13 Munn, Alan, 119, 120, 121, 122, 123, 171, 237, 837n, 837n27 Musan, Renate, 473, 476–478, 482, 483, 485, 490, 499, 500, 501, 546, 547, 897n5, 902n31, 905n35, 925n6 Oehrle, Richard T., 322 Ouwayda, Sarah, 139, 837n27, 843n46, 851n81, 853n90, 854n92 Parsons, Terence, 1, 11, 15, 200, 469, 549, 855n95, 856n102, 896n17, 912n53 Partee, Barbara, 9, 13, 72, 281, 379, 661, 697, 858n7, 870n53
994
Peacocke, Christopher, 848n78 Pereltsvaig, Asya, 928n21 Perlmutter, David M., 4, 75 Pesetsky, David, 313, 857n2 Peters, Stanley, 15 Pianesi, Fabio, 879n9, 879n11, 881n19 Pietroski, Paul M., 307, 741, 812, 814, 815, 824n14, 824n18, 825n19, 825n24, 832n5, 845n61, 846n62, 847n68, 856n102, 876n1, 879n10, 904n, 916n, 926n10, 927n19, 957n28 Platts, Mark, 898n17 Poesio, Massimo, 94, 98, 230, 281, 664, 859n14 Postal, Paul M., 834n12, 843n45, 869n Predelli, Stefano, 887n3, 909n46 Progovac, Ljiljana, 833–834n11 Pylyshyn, Zenon W., 947n Quine, W. V. O., 966n2 Recanati, François, 832n54, 832n55 Redish, David A., 393, 436, 448, 892n37 Reichenbach, Hans, 522, 881n19 Reinhart, Tanya, 170, 829n Reuland, Eric, 170 Reyle, Uwe, 271 Richards, Barry, 605 Rizzi, Luigi, 882n4 Roberts, Craige, 98, 230, 232, 281 Roeper, Peter, 886n8 Rögnvaldsson, Eiríkur, 213, 273, 277, 366 Root, Rebecca, 829n Rooth, Mats, 9, 281, 829n, 858n7, 870n53 Ross, John Robert, 3, 4, 15, 75, 869n Roy, Isabelle, 917n, 928n21 Russell, Bertrand, 621, 870n55 Sasse, Hans Jürgen, 837n30, 915n68 Sauerland,Uli, 866n37 Saul, Jennifer M., 490, 505, 513–514, 547, 908n43, 909n46 Savoia, Leonardo M., 352 Scha, Remko, 700
Name Index
Schein, Barry, 21, 22, 24, 25, 33, 35, 85, 96, 130, 131, 315, 389, 405, 624, 626, 634, 653, 655, 699, 702, 741, 794, 812, 814, 815, 819n2, 820n, 822n9, 825n, 825n19, 825n20, 825n21, 825n24, 827n36, 830n49, 832n1, 834n14, 835n17, 845n61, 847n, 848n73, 848n74, 849–850n79, 852n85, 852n86, 853n89, 855n94, 856n99, 856n103, 885n15, 886n8, 890n, 893n47, 894n51, 894n1, 895n3, 896n, 900n20, 902n, 905n, 914n63, 919n75, 922n97, 926n10, 935n, 936n27, 937n2, 938n6, 939n7, 939n8, 940n10, 940n11, 941n12, 942n18, 948n, 949n41, 949n42, 957n28, 958n31, 959n33, 967n, 970n16, 970n17, 970n18 Schlenker, Philippe, 73, 534 Schwarz, Bernhard, 213, 271–272, 366, 860n17, 860–861n19, 868n44, 869n, 890n20, 923n98 Schwarzschild, Roger, 180, 538, 587, 618, 699, 700, 814, 820n, 847n, 892n31, 895n3, 922n96, 924n2, 939n8, 959n33, 960n36 Scontras, Greg, 929n6 Searle, John, 73, 832n56 Segal, Gabriel, 689, 847n68, 898n17 Sennet, Adam, 832n55 Sharvit, Yael, 892n Sharvy, Richard, 96, 812, 827n32, 967n Sher, Gila, 76 Siegel, Muffy E. A., 322 Sjoblom, Todd, 109, 149, 157, 192, 846n63 Sommers, Fred, 794 Sperber, Dan, 69, 70, 73 Sportiche, Dominique, 120, 122, 123, 125, 127, 131, 173, 191, 197, 824n15, 837n27, 838n32, 838n33, 842n43, 850n80, 855n, 858n4, 868n45, 876n1, 882n4 Stalnaker, Robert, 488 Stanley, Jason, 69, 70, 72, 73, 857n3 Stirling, Lesley, 278, 821n Stone, Matthew, 697 Stump, Gregory T., 906n39 Svenonius, Peter, 391 Syed, Saurov, 20, 913n57 Szabó, Zoltán Gendler, 897n12, 907n40, 908n43 Szabolcsi, Anna, 76, 133, 271, 841n41
Name Index
Takahashi, Naoko, 616 Talmy, Len, 67, 119, 182, 183, 184, 185, 186, 192, 194, 474, 502, 852–853n88 Tancredi, Christopher D., 535 Taub, Alison, 315, 819n2, 827n36, 847n, 876n7, 937n2, 939n7, 939n8, 949n42, 951n46, 952n46, 959n34 Taylor, Barry, 24, 85 Thiersch, Craig, 860n17, 869n Thràinnson, Hoskuldur, 213, 274, 275, 276, 366 Tomaszewicz, Barbara, 913n56 Tomioka, Satoshi, 213, 253–254, 277–278, 285, 286, 288, 290, 354, 366 Tortora, Christina, 352, 884n11 Trotsky, Leon, 490, 547, 897n10, 900n23 Tversky, Amos, 182 Uriagereka, Juan, 882n4 van Oirsouw, Robert, 841n29 Verkuyl, Henk, 478, 495 Vlach, Frank, 15–16 von Fintel, Kai, 841n30, 919n68 Weinberg, Amy, 833n11 Westerståhl, Dag, 76 Wexler, Kenneth, 107 Wilder, Chris, 107, 166, 311, 312, 868n46, 869n48 Williams, Alexander, 21, 828n, 830n47 Wilson, Deirdre, 69, 70, 73 Winter, Yoad, 256, 702, 819n2, 827n36, 937n2, 939n7, 939–940n9, 949n42, 958n31 Zamparelli, Roberto, 621–622, 626–627, 631, 930n8, 930–932n9, 936n, 953n5, 958n31, 960n37 Zimmermann, Thomas Ede, 402, 504, 586, 897n10, 908n43, 909n46, 912n51, 912n52 Zoerner, Cyril Edward, III, 166, 876n4 Zucchi, Alessandro, 94, 98, 230, 281, 664, 859n14 Zwart, Jan-Wouter, 868n44 Zwarts, Joost, 390
995
Subject Index
Note: Locators for endnotes are of the form 000n00, where 000 is the page number on which the note occurs. A locator of the form 000n indicates the incomplete note that occurs at the top of page 000 (whose beginning is on a preceding page). About (intentional relation), 501, 573–574 “Aboutness,” 837n30 Absolutives, 248–249, 504, 868n43, 886n8, 907n40 Abstracta, 426 Accessibility relations, 203, 887n12 Accommodation, 83 Accomplices, 68, 86, 119. See also Cum(e,x) Acquisition, 196, 197, 206 Across-the-board movement clause size limiting, 213, 280 of clitics, 352 as a construction-specific stipulation, 278 of subjects, 272–273, 274, 276 of Tense, 277, 355, 876n5 undone by reconstruction, 213, 280–281, 288–289, 290 Actions and effects, 904–905n Addresses adverbialization and, 63–64, 379 in canonical phrase structure, 13, 709 coordinated NPs/AdrPs and, 406–408 in identity statements, 66 narrative intentions and, 63 nonmaximal reference and, 800 number agreement and, 115 one per NP/AdrP, 710 responsible for apparent effects of coordination, 786–787 sameness of between scenes, 771–772
sequencing of, in conjoined DPs, 82–83 substitutivity failures and, 50–51, 52, 378–383, 642, 704–705, 710–711, 719 AdrPs. See also NPs adverbialization of, 399, 413–417 conjoined, disjunctive interpretation of, 707–708, 787–789 conjoined, existential commitments, 712, 740–750 in the resolution of substitution puzzles, 710–711 semidistributivity and, 651, 704–705 as sentences, 707 adrs, 709, 731–734, 743–744, 776–783. See also Locatives and locative relations Advancement of narrative, 9, 225, 324, 432 Adverbialization, 12 as absolutive modification, range of relations denoted, 886n8 addresses and, 63–64 of AdrPs, 399, 413–417 assimilated to overt adverbial modification, 504, 507 as conditional modification, semantic variability of, 504, 507 conjoined DPs and, 81 counting and, 56–60 cued by pronunciation of subject, 210–211, 226–227, 324, 326–327, 362–363 descriptive anaphora and, 52–53
998
Adverbialization (cont.) effect on clause size, 227 event counting and object counting and, 557, 610 lifetime effects and, 485–500 locality, 518–519, 571 modification of identity statements under, 65–68 naive reference and, 55 of nonsubjects, 493–497 NPs as event descriptions and, 542–543 one per DP, 713 of relative clauses, 659–661 respective and, 53–54, 384, 542–543 scope of, 53–54, 241–242, 542–543, 546 scope of and clause size, 212, 234, 350, 351, 363 semidistributivity and, 659–661, 666 “sloppy” interpretation of event pronouns and, 110, 112–113 of subjects, 485–489, 495–500 substitutivity failures and, 50–52, 54–60, 378–383, 505–509, 513, 549, 610 supermonadicity and, 55, 373, 384–385, 506, 546–548 temporal reference and, 324, 326–327 Adverbs and adverbials as adnominal, one account of the lifetime effect, 481–484, 909n47 collective, 25, 825n20 in conditioned identity statements, 65–66 contrary, 11, 17, 21, 433–436, 588 event quantifiers introduced by, 162–163, 222 interfering with scope inversion, 252–253 lexical, 54 lifetime effects and, 481–484, 495–496, 497–500, 902–903n31 parity between spoken and unspoken, in polysemy, 830n43 with present perfect participles, simple and coordinated, 332–333 quantificational, 151–152, 152–157 in Right-Node Raising, 4–5, 75–76 scope of, and clause size, 224, 236
Subject Index
tacit, licensing “telescoping,” 98–99, 137–149, 230 temporal, 433–436 Affixes. See Bound morphemes Again, 400, 417–418, 421–422 Ago, 72 Agreement with first conjunct (see First-conjunct agreement) in number (see Number agreement) with objects, in Basque, 845n60 Ahistorical tenses, 917n Aktionsarten, 876n7, 951–952n46. See also Aspect, lexical Alphabetic identity of variables, RightNode Raising and, 88–89, 92, 106, 203, 214 Already, 532, 537 Altitude, 437 Ambiguity, subatomic, of sparse and cover, 158 Amharic first-person pronoun, 73 Among, 949n41 Analytic truths, 469, 470, 546 Anaphoric intentions, 105 Anticonvergence condition, 401–403, 436–437, 451–456, 567–569, 575–580, 582–587, 593–597, 894n50 Anticumulativity, 621–622 Antisemidistributivity. See also Genuine collective predicates Aktionsarten and, 951–952n46 descriptive event anaphora and, 699, 703 as a difference in logical type, 643, 644, 651, 702–703, 958n31 excluded by null determiner, 671–675 frames of reference and, 647–648, 680–687, 703–704, 728, 737, 751–752 with present tense, 684–685 recounts as, 681 substitutivity failures and, 645, 649–651, 704–705, 735–740, 751–752, 954n6 A(n) with a plural NP. See Singular plurals Appositive modifiers, 932–933n10 Arabic. See Lebanese Arabic; Moroccan Arabic; Palestinian Arabic Arithmetic predicates. See Cardinality predicates; Counting
Subject Index
Artifactual nature of narrative, 427 Aspect imperfective, 718 (see also Imperfective measurement) lexical, reconstruction and, 261 (see also Aktionsarten) perfect, 239, 327–333 perfective (see Perfective measurement) progressive, 323–327, 862n21 viewpoint, 558–561, 577–578 Aspects (of individuals), 552, 908n43, 909–910n47 Asymmetric (conditioning) relation between predicative conjuncts, 231, 233, 272 Asymmetric coordination, 249, 860n17, 863n28. See also Mixed coordinations Asymmetric extraction across-the-board movement versus, 868–869n46 of direct objects in German, 860–861n19 out of first conjunct, 935n quantifier scope and, 236–237 of subjects, 236–237, 859n9 Asymmetric relations, 67–68, 89, 92, 831n51 Asymmetry of perspective first-conjunct agreement and, 175–180 on symmetric relations, 67, 119, 181–186, 194, 206, 474–476, 502–504, 524, 915–916n68 At (locative relation), 438 Austere predicates, 898n17 Auxiliary verbs collectivized Right-Node Raising and, 243–244 in conjuncts, quantifier scope and, 229–234 modal, with conjoined complements, interpretation, 322–323 participation in events and, 239, 344 raising analysis of, 862n22 shared (see Aux sharing) tensed and untensed, and disjunctive interpretation, 340–341, 342–348 unspoken, 224–225, 310, 322 Aux sharing consequent states and, 327–329 of modal auxiliary, 322–323
999
by participial phrases with or without pronounced subjects, 224–225 in the perfect, 327–333 possible despite opacity of and to complementation, 310–311 in the progressive, 323–327 scope of decreasing quantifiers and, 321–322 temporal reference and, 324–327 Azimuth, 437–438 “Backtracking,” 844n53 Backward anaphora of events, 29, 31–32, 38–39, 46, 94, 102, 336–339, 807 ruled out for quirky-Case pronouns, 275–276 typological differences from forward anaphora, 869n48 Ballistic causation, 825n21 Bare plurals. See also Null article derived distributivity in, 688–691 focus and distributivity and, 689–692, 938n4 multiple frames of reference presupposed by, 467, 599–600, 645, 647, 648, 677–679, 686–687, 936n25, 947n33 perfective and imperfective measurement and, 675–677 Basque event quantifiers and thematic relations in, 157, 846n63, 848n72 number agreement in, 109, 149, 837n28 Be equinumerous, 819–820n2 Be sparse, 158, 822n6, 846–847n64 Before, 72 Binary quantification. See Branching quantification Binding arguments, 832n55 Both, tacit, 163, 846n64 Bound morphemes, 166, 216, 311–319, 367, 857n2 Branching quantification, 76–77, 78–79, 832n1 Brazilian Portuguese, estar and ser, eventive–stative contrast, 837n30 British plural, 866–867n37
1000
C (causal relation), 139–141, 144, 149, 901–902n30 Cardinality predicates. See also Numerals; Counting complex, 939–940n9 many versus numerals, differences, 676–677, 680, 698, 947n30, 947n32 Cardinality preservation, 567–569, 573 Cartesian coordinates, 440, 892n37 Case nominative, association with Tense, 224 nominative, not dispositive of gapping, 332, 876n2 of overt subject of second conjunct, assigned by and, 237 of participial subject, position, 241, 253 and Quantifier Lowering, 253 in resumptive absolutive construction, 248, 249, 868n43 Categorical judgments, 837n30, 844n54, 915n68 Causality first-conjunct agreement and, 126 in the relation between conjuncts, and “telescoping,”, 137, 331–332, 350, 843n44 in the relation between conjuncts, with pronounced participial subject, 247 spatiotemporal proximity contrasted with, 126, 843n44 Causal/topological relations, 318–319, 342, 344, 347–348, 360, 901–902n30 Causative analysis of transitives, 15, 16, 339, 471, 901–902n30, 904–905n Cause-and-effect relation between events, 15–16, 572–573 Ceteris paribus laws, 328, 879n10 Chains, causal/topological, 200, 202, 205, 896n Characterizing nominals, 928n21 Cinematic experiences, 63 Cinerama and cinerama semantics, 396, 397–398, 418, 436–456, 582–585, 720 Clause size auxiliary verbs and, 217 conjunctive or disjunctive interpretation of predicative coordination and, 279–280
Subject Index
disjunctive interpretation and, 337–341, 344–347 distributive quantification and, 133 effect of adverbialization on, 227 limitation on across-the-board movement, 213, 280 limitation on reconstruction, 213, 280, 286 meaning differences and, 217–218 number agreement and, 90, 91–92 opacity and, 88 presence of auxiliary verb and, 229–234 pronunciation of subject and, 217–218, 223–227, 228, 233, 236, 240–241 quantifier scope and, 7–8, 211–212, 227–234, 236–237, 262–270, 278–279, 286 simple verbs and, 217, 233–234 subject–adverb inversion in English and, 151 tense and, 228, 232–233 Clause structure, canonical, perspectives and frames of reference in, 63, 64, 66–67 Clitics and clitic climbing, 352–353, 362 Clocks, 427–431, 455 Co-. See Semidistributivity, operator for Code switching, 73 Coextensive NPs/AdrPs, 47–48, 50–51, 373–376, 377–381, 553–554, 602–603, 641–644, 649–650, 704–705, 710, 711, 716–717, 719, 726, 732, 733, 734–736, 764–767, 922n96, 938n4 Cognition, human, 437 Cognitive science, 831n51 Coincidence, condition on conjoined participial phrases, 329–331 Collective predicates distributive quantification and, 633–634 event counting and, 555, 611–612 first-conjunct agreement and, 123–125 genuine, 61, 380–381, 819–820n2, 822n6, 827–828n36 (see also Antisemidistributivity) semidistributive interpretation with, 314–318 and switch reference, 821n
Subject Index
Collective reference coordination of NPs/AdrPs not an example of, 707–708 in nominal conjunction, 1–2, 40, 46–47 to objects, and coordination of NPs, 370–371 in predicative coordination, 218, 324, 337–339, 342, 343–347, 884n12 (see also Disjunctive interpretation, of predicative coordination) Collectivized Right-Node Raising. See Right-Node Raising, collectivized Comitative condition on first-conjunct agreement, 127, 128, 135–149, 845n56 Comitative construction acquisition, 196, 197 as an argument for supermonadicity, 120 first-conjunct agreement distinguished from, 193–198 number agreement and, 87, 120, 193, 197–198, 855–856n98, 936n27 in Spanish, 87, 120, 193, 195–196, 197, 656–657, 936n27 Comitative phrases causal and spatiotemporal proximity and, 180–181 modifying thematic relations, 201–202, 856n104 scope interactions with instrumentals, 199–200 subordinate status of, 164–170, 195, 198–199, 308 truth conditions and, 199–200, 896n Comitativity, 134–135, 656–657 Communication, navigational, 392–398, 400–404, 409–410, 417, 426–468, 584 Communicative intentions frames of reference and, 410 narrative and, 62–64, 396 navigation and, 62–63, 392–396 object tracking and, 757 Comparative meaning of directionals, 440, 780 Comparatives, 509, 912n49, 939–940n9, 943–945n23 Compilation, 401, 452
1001
Complementary, 538–539 Complementation of auxiliary verbs by (conjoined) participial phrases, 310–311 with bound morphemes, 311–319, 367 as constraint on logical form, 215–216, 367 implemented in Eventish as locality conditions on anaphora, 339–340 logical structure of, 29 opacity of and to, 88, 166, 170, 207, 308, 310–319, 847–848n69 Complex cardinals, 939n9 Compositionality, 40, 68, 122, 836n20 Concatenation, 307–308, 832n5, 847n68 Concept reciprocity, 820n Conditionals and conditional modification, 488, 504, 507, 742, 894n1, 900n20 Conditioned identity, 65–66, 768 Conditioning relation between predicative conjuncts, 231, 272 Conjunction Reduction, 2 causative analysis of transitives and, 15 Hopi switch reference and, 820–821n3 participation relation in, 14 supermonadicity and, 14, 15, 21 translation into logical form and, 5 Conjunctivism, 876n1 Consequent states, 327–329 Constraints on movement, 292–302 on perspective (Schein, 1993), 970n18 on runaway translation, 6, 8 on switching frames of reference, 62 Constructional meaning, 474 Context of utterance, 62, 63, 395–396, 699–700 Contextualism, 68, 857n3 Continuity between scenes, and number agreement, 116 Continuity conditions on frame of reference, 396–398, 402, 417–426, 568, 714, 890n16 Continuous causation, 825n21 Contrastive modifiers, 932–933n10 Control, 851n Coordinate NPs/AdrPs. See NPs, conjoined; AdrPs, conjoined
1002
Coordinate Structure Constraint, 212–213, 834n12, 843n45, 861n, 869n Coordinate systems, 437–443, 713–714, 719, 891n21 Coordination-specific morphemes and processes, 278. See also Coordinative pronouns Coordinative pronouns exempt from Quantifier Lowering, 285–286 referring to events, 277–279, 283 referring to objects, 213, 271–277, 283–284, 365–366 Copula, 9, 528, 704, 824n14, 963–964n5 Copying descriptive anaphora not equivalent to in meaning, 9–10, 282–283 of event pronouns, 103, 110, 133 exactness of, 88, 92, 106, 110, 133, 214, 243 of (in)definite descriptions, 85 short and long, 112, 833n9, 834n13 unmotivated, constraint against, 166 Coreference, 19–20 Count, 60–61, 553, 608, 614–616, 626 Countability, 460, 614, 623, 635, 650–651, 667–671, 699–701, 752–755, 804, 939n8, 959n33, 970n17. See also Discreteness Counterdemonstrative contexts, 513–515, 533–534 Counterfactuals, 431, 912n52 Counting adverbialization and, 56–60 cinerama and, 404 epistemic conditions for, 56–61, 385–388, 402–403, 468, 586–587 epistemic necessity and, 564–566, 568 of events and objects, 461–464, 468, 563, 568–569, 574, 580–582, 612–613, 753, 953n1 frames of reference and scenes and, 57–61, 386–388, 402–403, 459–461, 556–558, 567–568, 575–580, 582–587, 593–597, 599–600, 677–679, 679–687, 694–696 lexical semantics and, 578–579 narrator’s (dis)orientation and, 454–455 of phases or temporal slices, 551–552
Subject Index
present tense and, 56, 386, 554–555, 562, 566, 574 protocols for, 56–57, 60–61, 385–386 of proxies, 457–459, 462 by quantifiers, 604–607 resolution and, 457–461 sequencing of events and, 561–582 spatial orientation and, 465–468 substitutivity and, 553–554, 601–603 viewpoint aspect and, 558–561, 577–578 visual, 460–466, 615, 694–695 Count preservation. See Cardinality preservation Count quantification, and nonmaximal reference, 804 Covariance of pronouns with quantifiers, 39–40, 90–91, 98–99, 126–127, 128, 135–149, 838n33 Cover, subatomic ambiguity of, 158 Cross-reference to events in Davidsonian semantics, 22, 26 in Eventish, 24 using null pronouns or variables, 11–12, 36 Cum(e,x), 173–175, 184, 193–194, 197. See also Accomplices Cumulativity, 621–622 Davidsonian semantics ‘&’ in, versus Eventish, 832n5 basic clause structure, revised by supermonadicity, 471–472 causative analysis of transitivity in, 905n collective predicates in, 702 Conjunction Reduction and, 1, 2–3, 6 decomposition and bound morphemes in, 312 distributive quantification in, 130, 134, 222, 874–875n65 event pronouns in, 22–23, 28–29 events and thematic relations in, 15 Prepositional Phrases in, 536 sentential analysis of predicative conjunction in, 209–210 subject copying and event quantification in, 209–210, 305, 364 thematic relations in, 22 VP-internal subject hypothesis in, 219
Subject Index
Decomposition, 11, 15, 28–29 Decreasing quantifiers anaphora to, 31, 36, 37, 793, 804–806 indefinite descriptions contrasted with, 30, 826n27 scope of, and Aux sharing, 321–322 De dicto and de re beliefs and intentions, 451, 867n, 893n42, 893n43 Deduction, 69–70 Defining nominals, 928n21 Definite descriptions with conjoined quantifiers, 80, 83 contrasting with distributive quantifiers with respect to event quantification, 85–86, 111–113, 129–131, 132–133, 162, 809, 874–875n65 of events, 827n31 existential commitments of, 96–97, 827n32 of frames of reference, 410–413 naive reference of, 55 nonmaximal reference and, 31 nonrigid interpretation of, 35–36, 411–413 in paraphrases of quantifier conjunction, 79–83 perfective and imperfective measurement and, 675–677 plural, compared with distributive quantifiers, 643 pronouns as (see Descriptive anaphora) rigid interpretation of, 35–36 single and multiple frames of reference with, 645, 647, 648, 677–679 Degrees and degree quantifiers, 940n Deletion site in Right-Node Raising. See Gap in Right-Node Raising Demonstration and demonstratives, 391–392, 433, 513–515, 533–535, 768, 778–779, 892n32, 934–935n22 De re possibility, 827n33 Descriptive anaphora adverbialization and, 52–53, 469–470, 546 copying/reconstruction not equivalent to in meaning, 9–10, 282–283 to decreasing quantifiers, 30, 793 distributive quantification and, 633–634, 637–640
1003
of events, 11, 22–30, 31–47, 94–95, 361, 655–656, 699–701, 806–807 glibness, 105 to indefinite descriptions, 30, 793 lack of existential commitment, 94–97 nonmaximal and maximal reference in, 405–406, 793–809 of objects, 23, 104 in paraphrases of quantifier conjunction, definite or indefinite, 79–83 proportion problem for, 605–607 vagaries of reference in, 23, 30, 84 Descriptive semantics, 69 De se knowledge, 451 Determiner-Sharing, 822n10, 953n5, 960n40, 960n41 Dialects of Basque, 845n60 Different, 612–613 Direct-reference theory of proper names, 893n42, 898–899n17 Discourse-level quantification over individuals, 793 over perspectives, 794–797, 799, 809, 830n49 as prerequisite for nonmaximal reference, 799 over spatiotemporal regions, 798–799, 800, 804 Discreteness. See also Countability of events, 699–701, 752–755, 803–804 of frames of reference, and semidistributivity, 651, 705 Disjoint reference of DPs in sentences, 20 Disjunction, 960–961n43 Disjunctive interpretation of AdrP-coordination, 707–708, 787–789 of predicative coordination, 213, 235, 238–240, 279, 315–318, 337–358, 359–361, 736, 740, 861n20, 876n3 Dislocative morpheme. See ℋ Displacement of narration from original navigation, 401, 427, 429, 584 Distinctness, of scenes. See Anticonvergence condition
1004
Distributive operators. See also Each, tacit; Both, tacit preverbal subjects and, 159–164 sparse and, 846–847n64 tacit, and clause size, 313 tacit, opacity of and to, 314 Distributive quantifiers in Davidsonian semantics, 130, 134, 222, 874–875n65 event quantifiers introduced by, 24–29, 222 and frames of reference for measurement, 61 plural, compared with singular plurals, 632–640, 643 singularity of variables, 877n8 Distributivity in bare plurals, operator to derive, 688–689 comitative phrases’ failure to support, 164 as ingredient in semidistributivity, 656–657 preverbal subjects and, 159–164 Divided reference, 2, 8–9, 40, 46–47 Do (auxiliary verb), 234, 312, 349 Documentary, 401, 453–454, 455 Domain restriction, event pronouns and, 152, 168–169 Double quantification, 833n Downward Distributive Determinerator, 688–691, 698 DPs conjoined (see next entry) participation relation within, 369–372 two classes with respect to event quantification, 85–86, 111–113, 129–131, 132–133, 162, 809, 874–875n65 weak and strong, 971n DPs, conjoined agreement with first (see First-conjunct agreement) analysis as Right-Node Raising, 5, 76, 93, 263–264 and as set product and, 930–931n9 corresponding to a conjoined NP/AdrP, 47, 48–49, 51–52, 58–60, 371–372, 376–377, 381–382, 601–602, 710–711, 712–716, 722–733, 746–752
Subject Index
Davidsonian analysis of, 1–2 dependencies between, 40, 89–91, 98–99, 116, 126 number agreement and, 40, 91–92, 263–264, 268–270, 366–367 possibly in, 40–42, 42–44 quantificational, 76–86 scope of quantification in, 39–40, 89–90, 263–264, 266–270 singular/partial agreement with, 121 Each clause size and, 314 tacit, 118, 159–164, 841n, 848n70, 868n42 Each other, 853n89. See also Reciprocals Economy constraint on gratuitous ellipsis, 311 Economy of expression, 54 Economy of movement and scope reconstruction, 292–302 Elegant variation, 534 Ellipsis clause size and, 224 of coordinations, 310 identity under, interaction with Economy, 293, 294–302 English, coordinative pronouns in, 213 En-scène, 422, 447, 455 Enthymemes, 549, 912n53 Epistemic conditions for counting, 56–61, 385–388, 468, 677–678, 754 imposed by overt articles, 724, 956n21 nonmaximal reference and, 801–802, 809 on present-tense narration, 401 on reference imposed by null and overt articles, 645, 647, 648 Epistemic grounds for assertion, eventivity in comitatives and, 196, 206 Epistemic necessity, 564–566, 568, 946n27 Epistemic stance of the speaker, 12–13 Epithets, 833n11 Equinumerous, 819–820n2 Essential separation, 25, 825n19, 825n20 Event identity, 852n85, 890n
Subject Index
Event pronouns in backward anaphora, 29, 31–32, 102, 336–339 conjoined AdrPs and, 707–708 conjoined DPs and, 1–2 coordinative, 277–279, 283 in Davidsonian semantics, 26 descriptive, 11, 22–30, 31–47, 94–95, 361, 655–656, 699–701, 806–807 disjunctive interpretation of predicative coordination and, 239–240, 279 emulating complementation, 319, 339–340 in forward anaphora, 29, 33–34 introducing second conjunct in predicative coordination, 231–232, 233 lacking existential commitment, 94–96, 97 locality constraints on reference, when unspoken, 337–339, 342, 343–347 logical form of, 106 mass/count distinction and, 833n8 nonmaximal reference with, 35, 806–807 number agreement as, 85–86, 109, 831n52 parity between spoken and unspoken, 28, 38–39, 806 plural, as bearers of collective reference, 361 plural, collectivized Right-Node Raising and, 83–84 plural, conjoined DPs and, 11, 78 plural, predicative conjunction and, 8–9, 11 plural, reconstruction and, 218 plural, in Right-Node Raising, 78 plural, univocal and and, 2–3 possibly and, 40–47 predicative conjunction and, 2 resumptive, introducing second conjunct with unpronounced subject, 231–232, 233, 243–247 room for in the structure, 344, 359–360 in semidistributivity, 650–651, 699–701 spoken, 23, 24, 28, 30, 32, 34–35, 37, 38, 46, 97, 102, 806 structural location of, and collective/ distributive meaning in Right-Node Raising, 84–85 under supermonadicity, 836n24
1005
unspoken and spoken, differences, 339 varying according to pronunciation of coordinated auxiliaries, 878n3 Event quantifiers absolutive modification and, 907n40 adverbialization and, 52–53, 469–471 distributive quantification and, 129–131, 132, 222 generic, 469 introduced by participation relation, 158–159 number agreement and, 132–133 pronunciation and silence of copied subjects and, 209–210 scope of, and nonmaximal reference, 35, 37–38, 118 single, with conjoined (in)definite descriptions, 170–172 “sloppy” interpretation and, 110, 129–131, 132, 221–223, 284, 291–292 subordination and, 168–169, 308 tense and, 7, 128, 243 thematic relations and, 157, 158–159, 846n62, 846n63 uniform definiteness of null elements and, 284–285 unspoken subjects and, 6 varying according to pronunciation of coordinated auxiliaries, 878n3 Event variables, 23, 36 Eventish, 3, 791, 811–817 conditionals in, 900n20 conjoined quantifiers in, 83–84 cross-reference to events in, 24 descriptive event pronouns in, 24, 31, 806 disjunctive interpretation of predicative conjunction and, 359–361 emendations to, 11 identity relation in, 65–67 intereventive relations in, 21–22 neighborhood relations in, 923n98 nondenoting NPs/AdrPs in, 741–742 realis relation in, 745 Right-Node Raising and, 361–362 sentential analysis of predicative conjunction in, 210
1006
Eventish (cont.) subject copying and event quantification in, 210, 305, 363–364 supermonadicity as feature of, 11 thematic relations in, 21–22 translation into, 85, 87–88, 363–367 Eventive predicates in the historical present tense, 892n31 Eventivity requirement for first-conjunct agreement, 125, 131, 189–190, 196, 205, 845n56 Every, 627–628 Exhaustivity, thematic, 845n61 Existential closure of thematic positions, 18–20, 238, 341–342, 852n87 Existential commitment in conjoined NPs/AdrPs, 740–750 of conjoined AdrPs, 712 of definite descriptions, 96–97, 827n32 descriptive anaphora and, 94–97 event pronouns and, 94–96, 97 Extraposition, 944n, 945n Failure to refer. See Referential failure Familiarity condition (Heim, 1982), 844n53, 864n Feature resolution rules, 836–837n26 Few, as arithmetic predicate rather than quantifier, 941n13 Figure and Ground, 67, 119, 184, 474–476, 502–504, 524, 896n, 915–916n68 First-conjunct agreement acquisition, 196, 197 asymmetry of perspective as meaning of, 188–192 causality and, 126 comitative construction distinguished from, 193–198 comitativity condition, 127 with demonstratives, 935n Eventish treatment of, 839–840n35 eventivity requirement, 125, 131, 189–190, 196, 205, 845n56 feature resolution rules and, 836–837n26 in Lebanese Arabic, 86, 120–149, 172–180, 188–192 limited to postverbal subjects, 121
Subject Index
spatiotemporal-proximity requirement, 125–126, 131, 135, 138 with statives, 189–190 typological diversity, 836–837n26 Focus, 262, 689–692, 938n4 Forward anaphora, 29, 33–34, 38–39, 869n48 Frames of reference. See also Perspectives; Scenes adverbialization of, 413–417 allocentric, 393, 396–397, 417–418 annotations to logical form to indicate, 12–13 antisemidistributivity and, 647–648, 651, 680–687, 703–704, 704–705, 728, 737, 751–752 communicative intentions and, 410 continuity conditions on, 396–398 counting and, 57–61, 386–388, 402–403, 459–461, 556–558, 567–568, 575–580, 582–587, 593–597, 599–600, 677–679, 679–687, 695–696 definite descriptions of, 410–413 as discourse parameter, 63, 395, 406 distinguished from perspectives and scenes, 889n12 distributive quantifiers and, 638 egocentric, 393, 396–397, 417–418, 438, 441–442, 721–722 formal definition, 438 inertial conditions on, 396–397 kinematic, 755–759 measurement and, 57–58, 720 metaphysics and, 57 in motion, 390–392 narrative intentions and, 63 number agreement and, 115–118, 268–270 one per DP, 651, 710, 724, 751–754 orbital, 62, 389–390, 410 quantification over, 81–83, 115, 407–409, 638–640 requantification over, 115–116, 268–270 segmentation conditions on, 397–398 single or multiple, according to determiner, 467, 599–600, 645, 647, 648, 677–679, 936n25 spatiotemporal orientation and, 62–64
Subject Index
substitutivity failures and, 50–52, 642, 704–705, 710–711, 713–716 tangential, 410 unbounded, 721 Free indirect discourse, 892n French, disjunctive interpretation of predicative coordination in, 882n4 Fronting, clause size and, 223 Future tense, sequencing of events and, 571–572 Gap in Right-Node Raising, 3, 88, 93–94, 100–104, 164–169 Gapping of auxiliary verbs (see Auxiliary verbs, unspoken) of main verbs, 878–879n4 of subjects (see Coordinative pronouns) Generalized Conjunction, 365, 708, 870n53 Generative semantics, 15, 68, 217, 313 Generic quantification over events, 469 Genuine collective predicates, 61, 380–381, 819–820n2, 822n6, 827–828n36. See also Antisemidistributivity German asymmetric coordination in, 863n28 asymmetric extraction of direct objects in, 860–861n19 coordinative pronouns in, 213 reconstruction of V2 topicalization in, 260, 867n41 V2 word order, 863n28 Germanic languages, asymmetric coordination in, 860n17 Gerunds, 823n13 Gestalt object recognition, 439 Glibness, 23, 37, 97, 105, 954n12 Gricean principles, 960–961n43, 968n10 Group formation, 836n20 ℋ, 520, 525–527, 574–575, 589–590 Have (auxiliary verb), participation in events and, 239, 344 Heterogeneity, of events referred to by pronouns, 86–87 Historical present tense, 401, 431–436, 914n61, 918n74
1007
Homogeneity Condition, 886n8 Hopi, switch reference in, 820–821n3 Hyperquotational logical forms, 889n8 Icelandic coordinative pronouns in, 213, 273–277 possessive reflexives in, 833n11 ℐ(dentity), 65–67, 528–529, 589–590 Identity, alphabetic, of variables, 88–89, 92, 106, 203, 214 Identity statements adverbialization and, 55, 509–510 frames of reference and, 418–426 ℋ in, 525, 589–590 involving counting, 588–600, 608–610, 681–682, 686 modification by adverbialization, 65–68 modification by adverbs, 524–525, 529–530, 531–537, 548, 769–770 modification by Tense, 65 phases and Tense and, 516–517, 518–520 substitutivity failures and, 48, 375–376, 382–383, 711 If-clauses, conjoined, singular agreement in, 865–866n36 Ill-behaved objects, 966n2 Imperfective aspect, 718 Imperfective measurement, 61, 388, 645, 675–677, 692–694, 716, 717–719 Implicit arguments, 71–73 Incorporation, 902n Increasing quantifiers, in first-conjunct subject position, 229, 826n27 Indefinite descriptions anaphora to, 30, 793, 804–806, 970–971n20 contrasting with distributive quantifiers with respect to event quantification, 85–86, 111–113, 129–131, 132–133, 162, 809, 874–875n65 existential commitments with, 744–745 perfective and imperfective measurement and, 675–677 plural, compared with distributive quantifiers, 643 single and multiple frames of reference with, 467, 645, 647, 648, 677–679, 936n25 Indexicality, 72–73
1008
Subject Index
Indexical reference to perspectives, 796–797, 799–800 Indistinguishable participants, 405–407, 428–429, 794–795, 800, 802, 863–864n30 Individual concepts, 924n3 Individual-level attributions, 477–478, 485 Inertial conditions on frame of reference, 396–397, 410, 424, 426 Inference of identity between event participants, 18, 20, 239 Inferential competence, 69–73 Infinite regress, 834n13 Initialization, of perspective, 795–796, 799 Instrumental phrases, 855n94, 856n103, 856n99 Instrumental relation, 824–825n18 Intentional events, reports of, 500–502, 547, 573–574 Intereventive relations, 15, 29 Interlinguistic communication, 71–72 Intervals, quantification over, 415–416 Intonation. See also Stress bare plurals and, 938n4 coreference or disjoint reference and, 830n47 demonstrative reference disambiguated by, 530, 534–535, 537, 920n81 parenthetical, 150, 257–258 Inversion of subject and adverb, 150–152 Inversion of subject and predicate. See Subject inversion Iota-operator, 31, 96, 827n32, 870n55 Island conditions on distributive quantifiers, 836n22 Italian, disjunctive interpretation of predicative coordination in, 882n4 Iteration of conditionals and other adverbials, 488 Iteration of “telescoping,” 99
Landmarks (for measurement), 695–697, 949n38 Landmarks (for orientation and navigation), 117–118, 391, 438–439, 453–454, 781, 949n38 Latin, prepositions in, 901n25 “Lazy” interpretation. See “Sloppy” interpretation Lebanese Arabic adverbs licensing “sloppy”/distributive readings in, 155–157 comitative construction, 198 distributive operators in, 159–161, 163 first-conjunct agreement, 86, 119, 120–149, 172–180, 188–192 number agreement in, 86, 119, 188–192, 831n52 preverbal subjects, 121, 146–148, 149, 159–160, 197, 842n43 Right-Node Raising in, 6–7 symmetry of nonperspectival thematic relations in, 187–188 Left and right, 62, 389–390, 410–412, 414–417, 440–443, 780–781 Lifetime effects, 226, 355–357, 473, 476–504, 515–516, 519, 521, 546–548, 552, 862n23 Line of sight, 444 Locality of adverbial modification under supermonadicity, 571, 770 of comitative phrases, 202–205 of event quantifiers introduced by thematic relations, 158–159, 846n63 unspoken event pronouns and, 337–339, 342, 343–347 Locatives and locative relations, 709, 722, 731, 733–734, 967n3. See also adrs Logical form, 66–73, 85, 87–88 Logical type, 651, 702–703, 899n, 943n22, 958n31 Logics, 69
Japanese apparent Tense sharing in, 277, 354 coordinative pronouns in, 213, 277–278
Many with a(n), 621 antisemidistributivity and, 943–945n23 as arithmetic predicate, 645–647, 648–649, 671–672, 936n25, 941n13
Kinematics, 59, 755–759
Subject Index
logical type and syntactic distribution, 943n22 as quantificational determiner, 940n, 943n22 Mass/count distinction, event pronouns and, 833n8 Maximal events, 28 Maximal reference, 793–794, 826n27, 826n28, 826n29 Measure phrases, 608 Measure spaces, 694–695, 716–717 Measurement. See also Counting addresses and, 716–720 imperfective, 61, 388, 645, 716, 717–719 perfective, 61, 388, 645, 675–677 protocols for, 56–57, 385–386 scanning, 675–677, 692–694 substitution puzzles and, 724–734 Metaphysics frames of reference and, 57, 61 number and, 607–608, 934n16 scenes and, 57, 386 substitutivity failures and, 49, 50, 56, 377 Minimalism, 217, 313 Misorientation, 417, 449 Mixed coordinations. See also Asymmetric coordination mixing quantificational and nonquantificational subjects, in Lebanese Arabic, 133, 842n43 mixing quirky and unquirky subjects, in Icelandic, 273–276 mixing tense and auxiliary types, 353–358, 860n17 Modal auxiliary verbs, 322–323 Modal individual concepts, 924n3 Modal (in)subordination, 232, 245, 248, 281–282, 350–351 Modality, inherent, of negative-polarity items, 232 Modes of presentation, 507, 909n47 Modification, clausal, 65–67 Modification of thematic relations, 201–202, 202–205, 206, 856n104 Modified cardinal construction, 620–624, 626, 759–761. See also Singular plurals
1009
Molecularity of event quantifiers introduced by adverbs, 845n55, 848n72 Moment of evaluation, advancement of, 9, 225, 324, 327, 432 Montage, 401, 429, 436–437 Moroccan Arabic, number agreement in, 850n81 Morphological agreement, 836–837n26 Morphosyntax, relationship of logical form to, 312, 314, 318 Most any a(n), 930n7 Movement across-the-board (see Across-the-board movement) asymmetric, out of first conjunct, 212, 935n, 859n9 of coordinations, 310 coreference and, 19–20 of determiners, 935n distributive quantification and, 841n41 opacity to, and clause size, 88 stylistic, 53–54, 542 Multiple conjunction, in Right-Node Raising, 101, 289, 290, 873–874n59 Multiple narrators or perspectives, integration of, 436–437 N (neighborhood relation), 413, 470–471, 485–486, 510–515, 887n12. See also Neighborhood relations Naive reference, 55, 56–57, 66, 553, 555–557 Names context dependence on frames of reference, 535–537 demonstrative use of, 514–515, 533–535 direct-reference theory of, 893n42, 898–899n17 lifetime effects and, 487, 489, 490, 506, 508, 897n10, 911n like other quantifiers, 508, 514 logical type, 899n relational approach to, 898–899n17 Naming conventions in narrative, 857n1 Narration of somatosensory experience, 399–401
1010
Narrative cinematic nature of, 557, 795 communicative intentions and, 62–64, 396 context and, 395–396 conventions, 857n1, 965n11 displacement from original navigational communication, 427 Navigation communication and, 62–63, 392–398, 400–404, 409–410, 417, 426–468, 584 counting and, 579–580, 677 Nav-topographical survey, 450 Necker cube, 448 Negation, 322–323, 770, 919n75, 963n5 Negative-polarity items with conjoined DPs, 89–90 inducing referential failure of pronouns, 231–232 inherently modal, 232 triggering referential failure of event pronouns, 245, 248 Neighborhood relations. See also N (neighborhood relation) counting and, 570–571 distributive and nondistributive, 543–544 lifetime effects and, 488 locality, 518–519, 571 nonspatiotemporal, 785–786 phase-dependent and phase-invariant, 533, 910–911n, 915n67 “No longer” implicature of past tense, 476–484, 498–500, 502–504, 519, 552, 909n47 Nominal conjunction. See DPs, conjoined Nominal coordination, adrs and, 783 Nominalization, 823n13 Nominative Case, 224, 332, 876n2 Nonconstituent Coordination, 101 Noncontrastive modifiers, 932–933n10 Nondecreasing quantifiers, 27 Nondenoting NPs/AdrPs, 740–744, 954n10 Nonincreasing quantifiers, 26–27, 28, 229, 653–658, 826n27 Noniterativity of “telescoping,”, 99 Nonmaximal reference, 30–36, 81, 405–406, 527, 793–809, 833n7
Subject Index
Nonrigidity in definite descriptions, 35–36, 411–413, 415. See also “Sloppy” interpretation Notional verbs, 966–967n2 Now-en-scène, 394, 455–456, 718 NPs. See also AdrPs canonical structure, addresses and frames of reference in, 63, 64, 709 coextensive, 47–48, 50–51, 373–376, 377–381, 553–554, 602–603, 641–644, 649–650, 704–705, 710, 711, 716–717, 719, 726, 732, 733, 734–736, 764–767, 922n96, 938n4 conjoined (see next entry) NPs, conjoined a(n) with, 620–621, 624–625, 626–627 addresses in frames of reference and, 406–408 and as set product and, 930n9, 960n37 coextensive with a simple NP/AdrP, 47–48, 373–376, 378–381, 602–603, 641–644, 649–650, 704–705, 710, 711, 716–717, 719, 726, 732, 733, 764–767, 938n4 corresponding to a conjoined DP, 47, 48–49, 51–52, 58–60, 371–372, 376–377, 381–382, 601–602, 710–711, 712–716, 722–733, 746–752 participation relation and, 371–372 Null article. See also Bare plurals antisemidistributive predicates incompatible with, 671–675 multiple frames of reference presupposed by, 645, 647, 648, 677–679, 686–687, 697–698 perfective and imperfective measurement and, 675–677 Number agreement comitative construction and, 193, 197–198, 201, 205, 206, 656–657, 855–856n98, 936n27 comparative quantification and, 944n, 945n compositionality and, 40 with conjoined DPs and conjoined NPs/ AdrPs, 40, 49, 371, 711, 713–714, 714–715, 731–732 dependent on semantic reference, 108
Subject Index
event pronouns as, 85–87, 109, 266–267, 632, 634–635, 831n52, 834n13, 940–941n11 frames of reference and, 115–118, 268–270 in Lebanese Arabic, 86, 119, 170–180, 188–192, 831n52 lowers to Case position, 255 with objects, in Basque, 845n60 partial (see First-conjunct agreement) plural, with conjoined distributive quantifiers, 138 with preverbal subjects, 197, 842–843n43 pronominal, 108 quantification in conjoined DPs and, 40, 89–90, 91–92 quantifier scope and, 263–270, 366–367 scene-counting, 842–843n43 scenes framing measurement events and, 58 semantic, 120, 121–122, 255–256, 831n52, 837–838n31 semidistributivity and, 940–941n11 singular, with plural weak DPs, 828n singular, with split antecedency, 256–260 in SOV languages like Basque, 109, 149, 837n28 spatiotemporal proximity and, 135 subject–adverb inversion in English and, 150–152 Tense’s interaction with, 484 with two classes of DPs, 111–113 Number morphology, 626, 628, 631, 632–633, 835n17 Numerals, 60–62, 553, 607–627, 628–629, 676–679, 943n22 Nusquam operator, 741–742, 743–744 O (overlap relation) as bound morpheme or freestanding verb, 318–319, 882n3 in causative analysis of transitives, 16, 901–902n30 coordinated NPs/AdrPs and, 953n1, 957n26 implication of identity between event participants, 239 meaning of, 190
1011
Objects direct, lifetime effects with, 491–492, 494–497 null, as coordinative pronouns, 277 prepositional, lifetime effects with, 490–491, 492–494 Object tracking, 59–60, 757–759, 761–762 Observatories, 437 Occasion sensitivity, 832n56 Opacity of and to bound morphemes, 311–319 of and to complementation, 88, 166, 170, 207, 308, 310–319, 847–848n69 to movement, and clause size, 88 to quantification, of tensed clauses, 230–233, 262–263, 877n2 of tensed auxiliaries to clitic climbing, 352–353 Open texture, 73, 832n56 Operators, distributive, preverbal subjects and, 159–164 Or, 960–961n43 Ordered pairs, 48, 375, 382–383, 763–764, 768–769, 788 Orientation. See Spatiotemporal orientation Overlap, lack of, between histories, 490–491, 516–517, 522, 526 Pairing functions, 960n37 Palestinian Arabic, first-conjunct agreement in, 837n29 Panoramas, 443 Parallelism in coordinate structures, 106, 112, 116, 133, 142, 293–294, 299–302, 787, 862n23, 870–871n57 Parameters of discourse, 62–63, 406 Parenthetical intonation, 150, 257–258 Parsing minimization of, 216, 367 of predicative coordination, 309–310 of Right-Node Raising, 107, 166, 308–309 Partial agreement. See First-conjunct agreement Participation relation (W) adverbs and, 845n57 auxiliary verbs and, 239, 344 conjoined DPs and, 77
1012
Participation relation (W) (cont.) disjunctive interpretation and, 708–709 event quantifier introduced by, 158–159 in first-conjunct agreement, 173–175 lifetime effects and, 502–504 meaning of coordinated AdrPs and, 708 not a bound morpheme, 216, 876n3 with NPs/AdrPs, 370–372 number agreement and, 109, 119, 173–175, 859n object-tracking, 761–762 possibly and, 41 in Right-Node Raising, 77 scenes framing counting and, 58–60 singular plurals and, 708–709 vagueness of, 14–15, 19, 21, 235, 238 Participial phrases conjoined, adverb scope and, 236 conjoined, asymmetric extraction of subjects from, 236–237 conjoined, disjunctive interpretation and, 340 conjoined, quantifier scope and, 228 conjoined, sharing an auxiliary verb, 224–225, 310 dependent nature of, 246 exposing an event variable, 246, 248–249 passive, in conjunctions, semantic conditions on, 329–331 subjects pronounced and silent in, 236–243, 246–252 Partitives and partitivity, 295, 297–298, 748–750, 971n passim (epistemic condition), 678 Past tense’s implicature of “no longer,” 476–484, 498–500, 502–504, 519, 552, 909n47 Path integration, 393–394, 436, 448–449, 585–586, 677–678 Perfect aspect, 239, 247, 327–333, 558–561, 577–578 Perfective measurement, 61, 388, 645, 675–677 Perseverance under change, 439 Perspectival relations, 186, 355–356, 360, 474–475, 502–504, 715 Perspective Phrases, 841–842n41
Subject Index
Perspectives. See also Frames of reference; Scenes annotations to logical form to indicate, 12–13 asymmetry of, in first-conjunct agreement, 119, 175–180, 188–192 distinguished from frames of reference and scenes, 889–890n12 initial, 795–796 lifetime effects and, 355 quantification over, and nonmaximal reference, 809 selective, nonmaximal reference as an effect of, 33, 35–36, 405, 794–809, 826n29 on symmetric relations, asymmetry of, 119, 181–186, 194, 206, 474–476, 502–504, 524, 915–916n68 Phases of individuals, 510, 551, 910–911n Phase sortals, 469, 487, 898n17, 912n53 Piedmontese, opacity of tensed auxiliaries to clitic climbing in, 352 pl[E], 109 Plurality conjoined DPs and, 85–86 of event pronouns, 2, 11, 22, 39, 44, 221 meaning of, 835n17 Point of view, 437, 889n12 Polish, comitative construction, 193, 195, 197, 656–657 Portuguese. See Brazilian Portuguese Possessive reflexives, 164–165, 170, 833n11 Possibilia, 472, 500–502, 862n21 Possibly disjunctive interpretation of predicative conjunction and, 882–883n5 in DP conjuncts, 38, 40–44 event pronouns and, 40–47 in predicate conjuncts, 40, 42, 44–46, 882–883n5 Postverbal subjects, 121 Pragmatics, 831n51 Predication, 47, 54, 373, 377 Predicative conjunction as account of adverb-scope facts, 860n18 as account of quantifier-scope facts, 228, 229–230
Subject Index
as coordination of VPs or PredPs, 10, 876n4 disjunctive interpretation of, 213, 235, 238–240, 279, 315–318, 337–358, 359–361, 736–740, 861n20, 876n3 divided reference and, 2 as explanation for lack of overt subject, 209. 220 possibly in, 40, 42, 44–46, 882–883n5 semidistributive interpretation with, 315 Prenominal modification, 616–618, 946n25 Prepositions and PPs, 390–391, 490–491, 492–494, 536 Present perfect puzzle, 332–333, 879n11 Presentations, 357–358, 566, 713, 796, 802, 967n5, 970n20 Presuppositions accommodation of, 83 of continuity of frame of reference, with adverbs, 417–426 in definite descriptions, 826n28 reconstruction and, 262 Preverbal subjects agreement with, 842–843n43 comitative, 193, 197–198, 201, 205, 855–856n98 distributive operators and, 159–164 in Lebanese Arabic, 121, 146–148, 149, 159–160, 197 number agreement and, 121, 193, 197–198, 201, 205, 855–856n98 “telescoping” and, 146–148, 149 Principle of Full Interpretation, 167 Progressive aspect, 323–327, 862n21 Projection, 443–447, 771 Pronouns coordinative, 213, 271–277, 277–279, 283, 285–286, 365–366 coreferent, 19–20 covariant, 39–40, 90–91, 98–99, 126–127, 128, 135–149, 838n33 glib, 23, 37 null, 5, 639–640 ostensibly referring to phases or temporal slices, 924n2 overt, in “telescoping,” 229–230, 281–282
1013
referring to events (see Event pronouns) (spatio)temporal, 665–666 Pronunciation and silence of adverbials, and polysemy, 830n43 of event pronouns, 35, 38–39, 46, 806, 831n52 glibness of pronouns and, 23 grammatical correlates of, 6–7, 8–10 of participial subjects, 236–243, 246–252, 324, 326–327, 331–332 of quirky subjects, 276 of subjects, adverbialization and, 227 of subjects, Case licensing and, 237 of subjects, disjoint reference and, 237–241 of subjects, effect on clause size, 223–227, 236, 240–241 of subjects, event quantification and, 209 of subjects, meaning differences, 5–6, 8–10, 209, 366 of subjects, quantifier scope and, 227, 276 of tense pronouns/anaphors, 967n5 Proper names. See Names Proportion problem, 605–607 Protocols, 56–57, 60–61, 385–386, 462–464, 553, 603–607, 626, 638–640, 647, 667–668, 935n Proxies, counting by, 457–459, 462, 464–465 Quantification. See also Quantifiers branching, 76–77, 832n1 count, and nonmaximal reference, 804 distributive, 24–29, 61 double, 833n over events, 6, 7, 35, 37–38, 52–53, 469 existential, scope of, 35 over frames of reference, 61, 64, 81–83, 115, 407–409 generic, 469 over intervals, 415–416 over perspectives, nonmaximal reference and, 809 in Right-Node Raised phrases, 834n14 scope of, and adverbialization, 52–53 scope of, and clause size, 7–8 scope of, and nonmaximal reference, 809 scope of, in conjoined DPs, 39–40, 89–90 structure of, 47, 373, 377
1014
Quantifier Lowering, 252–254, 285–286, 286–290, 293–294, 296 Quantifier Raising, 866n26 Quantifiers. See also Quantification in collectivized Right-Node Raising, 78–86, 286–290 conjoined, agreement with first (firstconjunct agreement), 126–127, 129–131 conjoined, analyzed as Right-Node Raising, 76–86 conjoined, dependencies between, 40, 89–91, 98–99, 116, 126 conjoined, number agreement and, 40, 91–92 decreasing, 30, 36, 37, 231–232, 321–322, 826n27 distributive, contrasting with (in)definite descriptions with respect to event quantification, 85–86, 111–113, 129–131, 132–133, 162, 874–875n65 distributive, in Davidsonian semantics, 130, 134, 222, 874–875n65 distributive, plural, compared with singular plurals, 632–640, 643 increasing, 229, 826n27 nondecreasing, 27 nonincreasing, 26–27, 28, 229, 653–668, 826n27 number agreement and, 632 phrasing of, 307 types of, in first-conjunct subject position, 229–230, 231–232, 271–272 types of, scopal behavior, 229–230, 231–232, 271–272 Quantifying in, 914n64, 966n2 “Quirky” predicates, 273–276 Raising analysis, of auxiliary verbs, 862n22 Realis relation, 745 Reciprocals, 634–635, 849n79, 922n97 Recitation, 401, 427, 429–433 Reconstruction. See also Quantifier Lowering across-the-board movement undone by, 213, 280–281 clause size limiting, 213, 280, 286, 364–365
Subject Index
deriving conjunctive interpretation of predicative coordination, 240 distinguished from “telescoping,” 282–283, 299–300 Economy and, 292–302 false examples of, 281–283 focus and, 262 of indefinite descriptions, descriptive anaphora not equivalent to in meaning, 9–10, 282–283 lexical aspect and, 261 necessary but not sufficient for “sloppy” interpretation, 222–223, 284 presupposition and, 262 into Right-Node Raised constituents, 218, 286–290, 361–362 of subjects in predicative coordination, 220, 223 substitutional predication as alternative to, 858n7 uniform definiteness of null elements and, 284–285 of V2 topicalization in German, 260 Recounts, 588–600, 681–682, 686 Recovery of unspoken content, 965n11 Reduced clauses, neighborhood relations in, 923n98 Reduplication of thematic relations, 202 Reference, vagaries of, 23, 30, 84 Reference events, 522–524, 927n16 Referential failure, of event pronouns, 245, 248 Reflexivity, 170, 851n Relational approach to proper names, 898–899n17 Relative clauses, adverbialization of, 659–661, 665–671, 701–702, 851n Reprise, 401, 427, 452 Requantification over frames of reference, 115–116, 268–270 Resemble, 476–484, 489, 494–495, 517–518, 520–521 Resolution, 456–461, 771–775, 890n15 Respective and respectively, 53–54, 384, 540–542, 544–546 Restrictions of quantifiers, partitivity in, 295, 297–298, 971n
Subject Index
Restrictive modifiers, 932–933n10 Resultative constructions, 963–964n5 Resumption, cued by nonpronunciation of subject, 210 Reticules, 458–459, 614, 771–772, 775, 890n, 893n47, 893n48 Right and left, 62, 389–390, 410–412, 414–417, 440–443, 780–781 Right-Node Raising collectivized, 3–5, 37, 75–77, 242, 244, 286–290 comitativity condition and, 141–142, 145 event pronouns in, 94–95 gap in, 3, 88, 93–94, 100–104, 164–169 in Lebanese Arabic, 6–7 parsing of, 107, 166, 308–309 semantic independence of conjuncts in, 127, 128 size of remnant, 100–102 translation schema for, 106 Rigidity in definite descriptions, 35–36 Russian characterizing and defining nominals in, 928n21 comitative construction, 193, 197, 202, 308 “Salience,” 963n1 Scanning measurement. See Imperfective measurement Scene analysis, 393–394, 436, 447–448 Scenes. See also Frames of reference; Perspectives annotations to logical form to indicate, 12 cinematic, 446–447 correspondence between participants in, 771–772 counting and, 57–61, 386–388, 402–403, 459–461, 556–558, 567–568, 575–580, 582–587, 593–597, 599–600, 614, 679–687, 694 event identity and, 890n frames of reference and perspectives distinguished from, 889n12 kinematic, 59 metaphysics and, 57 number agreement and, 842–843n43 object-tracking, 59–60, 757–759, 761
1015
orienting for some frames of reference and not others, 417–426, 448, 449 orienting for the narrator, 449–451 projection of, 443–447 resolution of objects or events by, 457 spatiotemporal orientation and, 62–63 still-life, 444, 772 substitutivity failures and, 50–52 uniqueness in, and nonmaximal reference, 794, 796–797, 802 visual and auditory, 62–63 Scope of adverbialization, 53–54, 542–543 of adverbs, and clause size, 236 boundedness of, for (in)definite descriptions and distributive quantifiers, 836n22 of DPs, and adverbialization, 55 inversion, 252–254, 287–302 of quantifiers, adverbialization and, 52–53 of quantifiers, asymmetric extraction and, 236–237 of quantifiers, clause size and, 7–8, 211–212, 227–234, 236–237, 262–270, 278–279, 286 of quantifiers, in conjoined DPs, 39–40, 89–90 of quantifiers, in first-conjunct subject position, 227–234 of quantifiers, nonmaximal reference and, 35, 809 of quantifiers, number agreement and, 263–270, 366–367 subject–adverb inversion and, 151–152 subject–verb inversion and, 146–147 Scrambling, of quantifiers, 322 Segmentation conditions on frame of reference, 397–398 Selective lifetime effect. See Lifetime effects Semantic objects, 70, 73 Semidistributivity with conjoined NPs in the subject of an antisemidistributive predicate, 649–651, 704–705 defined and analyzed, 314–315, 652, 653–671
1016
Semidistributivity (cont.) descriptive event anaphora and, 655–656, 699–701 distributivity plus comitativity, 656–657 incompatible with disjunctive interpretation of predicative coordination, 315–318, 360–361 not quantification over pluralities, 653–656, 657–658 operator for, 315, 317–318, 650, 652, 653, 662–664, 672, 936n27 with predicative conjunction, 315 substitution failures with, 704–705 variation among distributive plural quantifiers, 936–937n29 Separation, 941n12 Sequencing of events, 56, 386, 462, 554–555, 561–582, 597, 681, 789 of spatiotemporal address (spatiotense), 82–83, 115, 406, 800, 886n10 of tense, 324–325, 894n50 Serbo-Croatian, possessive reflexives in, 833n11 Set product, 930–932n9, 960n37 sg[E], 109, 615–616, 626, 628, 631, 632–633 Similarly oriented, 382–383, 418–419, 423–426, 711, 763, 770–776 Singular morphology meaning of, 626, 631, 835n17 pronounced “a(n),” 626, 628, 631, 632–633 Singular plurals, 59, 371–372, 601–602, 617, 618–619, 620–640, 643, 701–702, 708–709, 759–761 Size of clauses. See Clause size Size of coordination. See Clause size Slavic comitative construction, 656–657 (see also Russian, comitative construction) lack of tacit distributive operators, 161–162, 163, 164 possessive reflexives in, 164–165, 833n11 Slices, temporal, 480–483, 496, 551–552, 906n37, 909n47 “Sloppy” interpretation. See also Nonrigidity in definite descriptions of copied subjects, 210
Subject Index
of definite descriptions of frames of reference, 890n18 distributive quantification and, 132 of event pronouns, 109–110 event quantification and, 221–223, 284 of identical copies of phrases, 214 licensed by quantificational adverbs, 152–157 of overt pronouns, 94 of Right-Node Raised constituent, 128, 129–131, 838–839n34 of unpronounced subjects, 220 some (epistemic condition), 678 Some, single and multiple frames of reference with, 467, 936n25 Soundness, 69–70 SOV languages, number agreement in, 109 Spanish comitative construction, 87, 120, 193, 195–196, 197, 656–657, 936n27 disjunctive interpretation of predicative coordination in, 882n4 estar and ser, eventive–stative contrast, 837n30 Sparse, 158, 822n6, 846–847n64 Spatiotemporal-frame adverbials, 795–796, 798, 802, 805–806, 808, 968n12, 968– 969n13. See also Temporal-frame adverbials Spatiotemporal orientation apparatus for, 12–13, 437–447 conjoined quantifiers and, 81–83 counting and, 403–404, 585–586, 677–678, 679–680 to egocentric and allocentric frames of reference, 393, 396–397, 417–418 frames of reference and, 62–64 scenes and, 62–63 Spatiotemporal proximity, 126, 137, 843n44 Spatiotemporal-proximity requirement for first-conjunct agreement, 125–126, 131, 135, 138, 179–180, 196 Specificity condition of the British plural, 866–867n37 Speech production and perception, analogy to communicative intention, 889n7 Split antecedents, 96, 106, 256–260, 851n
Subject Index
Stage-level attributions. See Slices, temporal States, selective perspectives and nonmaximal reference and, 794, 796, 797, 801–803, 804 Stative predicates, in the historical present tense, 189–190, 892n31 Still, 400, 417–418, 419–421, 532, 537 Stress, 830n47, 887n2. See also Intonation Strong DPs, 971n Stubbornly distributed predicates (stubs), 618–620 Stylistic movements, 53–54, 542 Subentailments, 959n34 Subevents, and descriptive event pronouns, 24–29 Subject Gapping. See Coordinative pronouns Subject inversion in English, focus and, 867n41 in Icelandic, with unquirky and quirky predicates, 274–276 increasing adverbial quantification inducing, 907n40 in Lebanese Arabic (see Lebanese Arabic, preverbal subjects) Subjects copied, and event quantification, 209 of participial phrases, Case position, 241, 253 of participial phrases, pronounced and silent, 236–243, 246–252 of participial phrases, scope of, 241–242 pronunciation and silence, and disjoint reference, 237–241 pronunciation and silence, effect on clause size, 236, 240–241 quirky, 273–276 reconstructed, 220, 223 special relationship to Tense (lifetime effect), 476–489, 498–502, 516–524 spoken, and adverbialization, 227 spoken or unspoken, differences, 5–6, 324, 326–327, 331–332 spoken or unspoken, effect on clause size, 223–227 spoken or unspoken, quantifier scope and, 227
1017
unspoken, analysis as coordinative pronouns, 271–277, 283–284 unspoken, in apparent predicative coordinations, 213 unspoken, threat of runaway translation with, 209 Subordination contrasted with coordination, 164–170, 194–195, 307–308 logical form for, 307–308, 856n101 as phrasal concatenation, 307–308, 847n68 Substitutional predication, 858n7 Substitution salva veritate, 47–51, 373–377 Substitutivity, failures of addresses and, 50–51, 52, 378–383, 642, 704–705, 710–711 adverbialization and, 50–52, 54–60, 378–383, 505–509, 513, 549, 610 with conjoined and simple NPs and DPs, 47–52, 373–377, 377–378, 553–554, 601–603, 642–644, 649–651, 704–705, 710–711, 712–716, 719, 722–733, 734–736, 750–752, 938n4 demonstratives and, 513–515 with distributively quantified bare and singular plurals, 634 frames of reference and, 50–52, 642, 704–705, 710–711 in identity statements, 48, 375–376, 382–383, 609, 711, 763, 766–767 involving antisemidistributive predicates, 649–651, 704–705, 735–740, 751–752, 954n6 involving cardinality predicates, 553–554 involving semidistributive interpretation, 649–651, 704–705 supermonadicity and, 506, 909–910n47 variation in neighborhood relations and, 510–515 Summations of measurements, 462–463, 947n32 Supermonadicity, 11, 12, 14–22 adverbialization and, 55, 373, 384–385, 471, 485–486, 506, 546–548 causative analysis of transitivity and, 471 comitatives as an argument for, 120 Davidsonian semantics and, 471–472
1018
Supermonadicity (cont.) descriptive event pronouns and, 25, 29, 825n19 event-counting cardinality predicates and, 555, 571, 610–612 grammar of modification under, 66–67 intereventive relations under, 825n19 more events available for quantification over, 472, 473, 484 nonmaximal reference and, 800, 802, 968n12 opacity to quantification and, 267 possible parses of predicative coordination under, 309–310 sentential analysis of predicative conjunction under, 210 subject copying and event quantification under, 210 substitutivity failures and, 506, 909–910n47 Tense and, 472–473, 484 translation of event pronouns under, 836n24 Survey, 63, 393–394, 396–397, 450–453, 677 Switch reference, 278, 820–821n3 Symmetric relations, 67–68, 181–186, 474–476, 502–504, 515–538, 548–549, 831n51 “Telescoping” between sentences, 94, 98, 99, 838n33 of coordinative pronouns, 272, 282–283 distinguished from reconstruction, 282–283, 299–300 with distributive quantifiers, 638 iteration of, 99 of overt pronouns, 90–92, 229–230, 281–282 quantifier types and, 229–230 subatomic, 98–100, 137, 639–640, 832n3 tacit adverbial phrases licensing, 98–99, 137–149 Telicity, 825n21 Temporal anaphora, 13–14, 378–379, 661–662, 665–666, 697, 967n5 Temporal distributivity, 718–719, 954n11 Temporal orientation, 521–524 Temporal reference, 324, 326–327 Temporal sequencing, 13–14, 378–379
Subject Index
Temporal-frame adverbials, 433–436, 902–905n31, 914n64. See also Spatiotemporal-frame adverbials Tense across-the-board movement of, 277, 355, 876n5 as adnominal, accounting for the lifetime effect, 481–484, 909n47 ahistorical, 917n as anaphoric, 13–14, 312, 378–379, 661–662, 665–666, 697, 918n74, 919n79, 927n16 apparent sharing of, in Japanese, 277, 354 asymmetric meanings with symmetric predicates and, 475–476, 502–504 as bound morpheme, 312, 878n3, 898n16 clause size and, 7–8 in comparative clauses, 509, 912n49 conjoined clauses and, 13–14, 378–379, 661–662, 665–666 event quantifiers introduced by, 7, 146, 151, 222, 313 exposing an event variable (not referring to or quantifying over events), 246 historical present, 431–436, 914n61, 918n74 independent, escape from the lifetime effect with, 516–517, 521 modification of identity statements by, 65 narration and, 401, 429–436 nominative Case associated with, 224 past, implicature of “no longer,” 476–484, 498–500, 502–504, 519, 552, 909n47 present, and antisemidistributivity, 684–685 present, and measurement, 56, 386, 554–555, 562, 566, 574 quantifier scope and, 228–234, 877n2 relative to discourse local time, 429–433 Right-Node Raising and, 7, 243–244 sharing of, and do, 234, 312, 349 special relationship to subjects (lifetime effect), 476–489, 498–502 supermonadicity and, 472–473, 484 and tenselessness, of adverbials, 942n16 Thematic relations in DPs, 708 event quantifiers and, in Basque, 157, 846n63 in Lebanese Arabic, 187–192
Subject Index
modification of, 201–202, 202–205, 206, 856n104 nonperspectival (Agent, Patient, Theme), 187–188 one token of each per simple clause, 168, 203 perspectival, 186, 355–356, 360 reduplication of, 202 Thematic separation, 21 Theta criterion, 861n20 Thetic judgments, 837n30, 844n54, 915n68 Time, displacement of narrative and, 427–436 Time travel, 516–524, 526–527, 783–785 Tokens, distinguished from utterances, 891n27 Topicalization, as evidence against acrossthe-board movement, 868n46 Topological/causative relations, 318–319, 342, 344, 347–348, 360, 901–902n30. See also O (overlap relation) Traces of movement, 18–19, 239 Tracking, 453–454, 455 Transitivity alternation with symmetric predicates, 895–896n3 Translation into logical form for Right-Node Raising, 85, 87–88, 105–106 runaway, 6, 8, 209, 214–215 sufficient expressive power for, 13 unspoken subjects and, 5 Triadic intereventive relations, 824n18, 825n24 Truth conditions, 68–73, 116 Turkish, coordination-specific morphemes in, 870n51 Unarticulated constituents, 70 Uniqueness. See Maximal reference V2 word order in German, 863n28 Vagaries of reference, 23, 30, 84 Vagueness, 14–15, 19, 21, 235, 238, 438 Variables bound or existentially closed, and syntactic movement, 18, 19 of events, 23, 36, 246, 248–249
1019
or pronouns, device for cross-reference, 11–12, 36 of quantification, and number agreement, 936n27 Variation, elegant, 534 Variation, interspeaker, with regard to parallel scope inversion, 874n61 Verb Raising, 838n, 840n Verbs, auxiliary clause size and, 217, 229–234 quantifier scope and, 229–234 tensed and untensed, disjunctive interpretation and, 340–341, 342–348 Verbs, simple clause size and, 217, 233–234 disjunctive interpretation and, 341, 349–353 gapping of, 878–879n4 quantifier scope and, 228 Viewfinding, 716, 721–724 Viewpoint aspect, 558–561, 577–578 Visual counting, 694–695 vP, 858n4 VP-Internal Subject Hypothesis, 19, 219, 858n4, 868n45 W. See Participation relation Weak crossover, 301 Weak DPs, 971n Well-formed proofs, 69 West Germanic Verb Phrase, 847n65 Wide-scope event quantification, 35 Wide-scope indefinites, 793, 796 With (thematic relation), 184–186, 193–194 With, instrumental and comitative, 855n94 Word choice, pragmatics of, 831n51 Word formation, 312 ⌜x = y⌝, 64–67, 509–510, 530–531, 537–538 Yes and no, meaning and pragmatics of, 888–889n5 Zero morphology, 313, 857n2