Binominal Quantifiers in Spanish: Conceptually-driven Analogy in Diachrony and Synchrony 9783110406733, 9783110403718

Citation preview

Katrien Dora Verveckken Binominal Quantifiers in Spanish

Beihefte zur Zeitschrift für romanische Philologie

Herausgegeben von Claudia Polzin-Haumann und Wolfgang Schweickard

Band 391

Katrien Dora Verveckken

Binominal Quantifiers in Spanish Conceptually-driven Analogy in Diachrony and Synchrony

ISBN 978-3-11-040371-8 e-ISBN (PDF) 978-3-11-040673-3 e-ISBN (EPUB) 978-3-11-040683-2 ISSN 0084-5396 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.dnb.de abrufbar. © 2015 Walter de Gruyter GmbH, Berlin/Boston Druck und Bindung: CPI books GmbH, Leck ♾ Gedruckt auf säurefreiem Papier Printed in Germany www.degruyter.com

| To my real-life fairies Lieselotte, Marie, Klara and Peter

Acknowledgements I do believe in fairies…

Some time ago, I was delighted to hear my husband tell our little girls to believe in fairies, just like we do. With fairies we don’t mean the prototypical enchanting ones with shiny wings in fairy-tales the girls are so fond of, but the real-life versions on which the metaphor is based: those common people whose daily support, free help and kind advice give you wings and whose modesty wants them to be overlooked. It is no false modesty to state that many people have contributed in their own way in the present research project. I wish to thank them now. First and foremost, I want to express my deepest gratitude to my supervisor Nicole Delbecque. Her enthusiasm for Spanish and cognitive linguistics has caught me since the very first lecture on ser y estar. In addition to being an inspiring linguist, she excels at being a stimulating and dedicated supervisor. My writings have greatly benefited from her insights, eye for precision and finishing touch. Dear Nicole, it has been a great privilege to be one of your doctoral students. Thank you for your unbounded confidence and friendship. I would also like to thank Ignacio Bosque, Bert Cornillie, Béatrice Lamiroy and María José Rodríguez Espiñeira for their detailed and careful comments on an earlier version of this book. I owe a special word of gratitude to María José, for having been the most wonderful host my family and I could wish for in Santiago de Compostela. I am also grateful to Claudia Polzin-Haumann and Wolfgang Schweickard who, as editors of Beihefte zur Zeitschrift für romanische Philologie, have accepted my proposal in their book series. I thank Mouton de Gruyter’s editors Ulrike Krauß and Christine Henschel, Christine in particular for her patience, good cheer and professional guidance through the publication process. Thanks to Florian Ruppenstein for his swift and efficient help with the formatting issues. In the department of Linguistics I have found an excellent work environment. My research and writings have significantly benefited from the suggestions and supportive comments of many people. In addition to Nicole Delbecque, Bert Cornillie and Béatrice Lamiroy, these are Barbara De Cock, Katleen Van den Steen, Lise Van Gorp, Daniel Michaud Maturana, Hilde Hanegreefs, Hendrik De Smet, Hubert Cuyckens, Kristin Davidse, JeanChristophe Verstraete and Lot Brems. A special word of thanks to Ingeborg Verelst, Lydia Fernández Pereda, María Sol Sansiñena Pascual and Nuria Herrera Coronas, for their native-speaker judgments to enigmatic examples and help

VIII | Acknowledgements

in translating them. A kind word of thanks goes to all my great colleagues for their encouragements, sympathy and friendly words. I also want to thank Barbara, Katleen, Lise, Daniel and Hilde for their extremely valuable comments on the final version of the text, but, above all, for being true buddies, especially beyond the world of linguistics. For being delightfully non-linguistic, for many types of moral and practical support, I wish to thank my family and friends. I warmly thank my parents, for their unconditional moral as well as logistical support, and most prominently for the wonderfully warm and inspiring home I have enjoyed, in which no question was to remain unanswered. Finally, a very special word of thanks goes to the four most special people in my life. Without the enchanting smiles of my princesses and the unconditional love and inexhaustible care of my husband, this book would never have seen daylight. Thank you, Peter, Lieselotte, Marie and Klara, for coming so close to the winged fairies and for reminding me what really matters. This study is dedicated to you.

Contents List of abbreviations | XV

Introduction  1 1.1 1.2 1.3

Introduction: from quantifying nouns to binominal quantifiers | 3  Aim 1: Towards a constructional network model of the development of BQs | 5  AIM 2: Towards a cognitive-functional model of the organization of BQs | 6  Aim 3: Towards a reassessment of persistence and analogy | 7

Part 1 Preliminaries   2 Framework, state of the art and methodology | 11  2.1 Description of the framework | 11 2.1.1 Definition of five interrelated key-concepts | 12  2.1.1.1 Iconicity | 13  2.1.1.2 Construal | 14  2.1.1.3 Schematicity | 15  2.1.1.4 Subjectivity | 17  2.1.1.5 Construction | 21  2.1.2 Towards a cognitive-functional model of binominal quantifiers | 24  2.1.2.1 Nouns and noun phrases | 24  2.1.2.2 Quantification and quantifiers in noun phrases | 26  2.1.2.3 A cognitive-functional model of binominal quantifiers | 27  2.2 State of the art | 33 2.2.1 Description of QNs in GDLE (Bosque 1999) and in NGLE (RAE 2009) | 34  2.2.2 Synchronic case-studies | 37  2.2.3 Diachronic case-studies | 40  2.2.4 Identification of the head of the construction | 42  2.2.5 Gradient boundaries between partitive, pseudopartitive and appositional binominals | 47  2.2.5.1 Partitive and pseudopartitive constructions | 48  2.2.5.2 Pseudopartitive and appositional constructions | 51  2.3 Methodology | 53

X | Contents

3 3.1 3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 3.1.6 3.1.7 3.2 3.2.1 3.2.2 3.2.3 3.2.4 3.2.5 3.3 3.3.1 3.3.2 3.3.3

Binominal quantifiers as a locus of grammaticalization | 58  Grammaticalization and grammaticalization theory | 58 Definition | 58  Characteristics | 60  Conditions | 61  Desemanticization and decategorialization | 62  Mechanisms | 64  Parameters | 65  Discussion | 68  Lexical and functional uses | 69 Head uses of binominal quantifiers | 71  Quantifier uses of binominal quantifiers | 74  Two-way specifier uses of binomianl quantifiers | 79  Ambivalent uses | 83  Concluding remarks | 86  The grammaticalization of binominal quantifier constructions | 89 The underlying conceptual motivation of GR in BQs | 89  The Spanish BQ as a locus of GR: towards the case-studies | 92  Concluding remarks | 96 

Part 2 Diachronic case-studies – towards a constructional network model   4 4.1 4.2 4.3 4.3.1 4.3.2 4.3.2.1 4.3.2.2 4.3.2.3 4.3.2.4 4.3.3

The development of montón de, a typical locus of grammaticalization | 101  Methodological background | 103 Encyclopaedic information on montón | 104 A diachronic corpus analysis of montón | 106 Medieval Spanish: the emergence of quantifying pragmatic inferences | 107  Classical Spanish: the emergence of the specifier use | 115  Head uses of montón de in Classical Spanish | 116  Quantifying uses of montón de in Classical Spanish | 119  Specifying uses of montón de in Classical Spanish | 124  The combinatorial pattern of montón de in Classical Spanish | 127  Modern Spanish: partial continuation and gradual extension | 131 

Contents | XI

4.3.3.1 4.3.3.2 4.3.3.3 4.3.3.4 4.3.4 4.3.4.1 4.3.4.2 4.3.4.3 4.3.4.4 4.3.5 4.4 5 5.1 5.2 5.2.1 5.2.2 5.2.3 5.2.4 5.2.5 5.2.6 5.2.7 5.3 5.3.1 5.3.2 5.3.3 5.4 6 6.1 6.1.1 6.1.2 6.1.3 6.2 6.3 6.3.1 6.3.2 6.3.3

Head uses of montón de in Modern Spanish | 134  Quantifying uses of montón de in Modern Spanish | 138  Specifying uses of montón de in Modern Spanish | 140  The combinatorial pattern of montón de in Modern Spanish | 143  Present-Day Spanish: advanced grammaticalization stage | 145  Head uses of montón de in Present-Day Spanish | 146  Quantifying uses of montón de in Present-Day Spanish | 151  Specifying uses of montón de in Present-Day Spanish | 156  Conventionalization of the grammaticalized uses | 159  Intermediary conclusion | 162  Conclusions | 167 The role of conceptual persistence and analogy in the development of BQs | 173  Methodological background | 175 The development of binominal quantifiers | 178 Outline of the history of pila de | 180  Outline of the history of aluvión de | 184  Outline of the history of letanía de | 188  Outline of the history of (h)atajo de | 192  Outline of the history of barbaridad de | 198  Outline of the history of mar de | 202  Concluding remarks | 219  The role of persistence and analogy | 225 Persistence and analogy in the extension towards new N1s | 226  Persistence and analogy in the extension towards new N2s | 228  Concluding remarks | 240  Conclusion | 244 Theoretical reflection: a constructional network model of the development of BQs | 247  The challenges for the diachronic account of BQs | 249 The complex frequency pattern | 249  The formal and semantic persistence | 253  The exceptional role of (un) montón de N2 | 259  The explanatory scope of existing accounts | 264 The constructional network model revisited | 269 Revisiting the analogical model(s) | 271  The explanatory scope of the redefined analogical model | 274  Revisiting the macro-constructional level | 277 

XII | Contents

6.3.3.1 6.3.3.2 6.3.4 6.4 6.4.1 6.4.2 6.5

Binominal quantity and quality assessment as comparison construals | 277  Co-extensiveness as the motivation of the comparison construal | 281  The explanatory scope of the revisited macro-constructional level | 284  The grammaticalization of the BQ-construction | 291 BQs as the locus of grammaticalization | 291  GR of the binominal quantifier macro-construction | 296  Conclusion | 298

Part 3 Synchronic case-studies – towards a constructional network model   QN-related schematization and N2-profiling | 303  Methodological background | 305 Quantification construal in BQs | 307 Construal difference in the minimal pair un alud / aluvión de N2 | 312  7.2.1.1 Lexicographical accounts | 312  7.2.1.2 The conceptual image of alud and aluvión de | 313  7.2.1.3 The conceptual images broken down into conceptual facets | 316  7.2.1.4 Differences in the image schemas of alud de and aluvión de | 331  7.2.1.5 Context and combinatorial pattern of alud de and aluvión de | 333  7.2.1.6 Concluding remarks | 338  7.2.2 Construal difference in the minimal pairs una pila/un montón de N2 and un montón/mogollón de N2 | 339  7.2.2.1 Lexicographical accounts | 340  7.2.2.2 Conceptual image of (una) pila de N2 | 342  7.2.2.3 Conceptual image of (un) montón de N2 | 347  7.2.2.4 Conceptual image of (un) mogollón de N2 | 359  7.2.3 Outline of the conceptual images of the remaining QNs | 368  7.2.3.1 Conceptual image of (un) racimo de N2 | 369  7.2.3.1.1 Lexicographical accounts | 369  7.2.3.1.2 The conceptual image of (un) racimo de N2 | 369  7.2.3.2 Conceptual image of (un) hatajo de N2 | 374  7.2.3.2.1 Lexicographical accounts | 374  7.2.3.2.2 The conceptual image of (un) hatajo de N2 | 374  7 7.1 7.2 7.2.1

Contents | XIII

7.2.3.3 7.2.3.3.1 7.2.3.3.2 7.2.3.4 7.2.3.4.1 7.2.3.4.2 7.2.4 7.3

Conceptual image of (una) letanía de N2 | 381  Lexicographical accounts | 381  The conceptual image of (una) letanía de N2 | 381  Conceptual image of (una) barbaridad de N2 | 387  Lexicographical accounts | 387  The conceptual image of (una) barbaridad de N2 | 388  The contribution of the plural morpheme | 392  Conclusion | 397

8 The motivated nature of the QN-related co-selection patterns | 400  8.1 Morphosyntactic description of the BQ in the literature | 401 8.1.1 Collective or distributive reading | 404  8.1.2 Pluralization of N1 | 406  8.1.3 Suffixation of N1 | 408  8.1.4 Determiner pattern of N1 and N2 | 408  8.1.5 Adjectival modification to N1 and N2 | 411  8.1.6 (Lexical) restriction on N2 | 412  8.1.7 Verb agreement | 413  8.1.7.1 Traditional account of the verb agreement of collective nouns | 413  8.1.7.2 Studies on verb agreement of (pseudo-)partitive constructions | 416  8.1.8 Topicalization and extraposition of the prepositional phrase | 421  8.1.9 The scope of quantification by BQs | 422  8.1.10 Conclusion | 425  8.2 Morphosyntactic variation of BQs in the corpus data | 426 8.2.1 Fluctuation in the N1-determiner pattern | 426  8.2.2 Fluctuation in the premodification pattern | 432  8.2.3 Fluctuation in the N2-pattern | 435  8.2.4 Fluctuation as to verb agreement | 443  8.2.4.1 Methodological note | 445  8.2.4.2 Explanatroy scope of the existing accounts | 446  8.2.4.3 Towards the case-study | 449  8.2.4.3.1 Definiteness of N1 | 450  8.2.4.3.2 Distance between the subject and the verb | 451  8.2.4.3.3 Position of the subject with respect to the verb | 452  8.2.4.3.4 Partitive versus pseudopartitive constructions | 453  8.2.4.3.5 Verb agreement in relative clauses | 455  8.2.4.3.6 Type of verb and contextual cues | 456 

XIV | Contents

8.2.4.3.7 8.2.4.3.8 8.2.4.4 8.2.4.5 8.3

Function of N2 | 459  Concluding remarks | 460  CIP as a complementary factor in verb agreement | 461  Agreement hierarchy of BQ-constructions | 463  Conclusion | 465

9 Theoretical reflection: the notion of CIP revisited | 468  9.1 The notion of CIP revisited | 468 9.1.1 CIP as a gradual phenomenon | 468  9.1.2 CIP as an unpredictable phenomenon | 475  9.1.3 CIP as two-way discourse phenomenon | 477  9.2 Theoretical implications for the notion of analogy | 482 9.2.1 The concept of analogy in a nutshell | 483  9.2.2 The notion of analogy revisited | 485  9.2.2.1 The usage-based model of analogy (De Smet 2013) | 485  9.2.2.2 Discourse level as an additional motivation for analogy | 488  9.3 Conclusion | 490

Part 4 Conclusions   10 10.1 10.2

Conclusion | 495  Concluding remarks | 495 Further research | 498

References | 503 

List of abbreviations Abbreviations used in the analysis and frequency tables A Adj Adv BQ CI CIP Def. Dem. Det. Excl. H I Indef. Interr. G GR GT Lm N N1 N2 NP Pl PNP Q Qadj Qadv QN S Sg Tr

ambiguous uses adjective adverb binominal quantifier conceptual image conceptual image persistence definite demonstrative determiner exclamative head (use) indeterminate (use) indefinite interrogative ground grammaticalization grammaticalization theory landmark noun first noun in binominal construction [N1 de N2] second noun in binominal construction [N1 de N2] noun phrase plural prepositional noun phrase quantifier (use) adjective intensifying use adverb intensifying use quantifying noun specifier (use) singular trajector

XVI | List of abbreviations

Abbreviations for data sources CLAVE CORDE COREC CREA DUE DRAE GDLE NGLE

Clave-Diccionario de USO Del Español (Maldonado González (ed.) 1999) Corpus diacrónico del español Corpus oral de referencia del español contemporáneo Corpus de referencia del español actual Diccionario de uso del español (Moliner 2008) Diccionario de la lengua española (RAE 2001) Gramática descriptiva de la lengua española (Bosque & Demonte (dirs.) 1999) Nueva gramática de la lengua española (RAE 2009)

Symbols used in the examples * ?

an asterisk at the beginning of the example indicates ungrammaticality (unless stated otherwise in footnote) a question mark at the beginning of the examples indicates its artificial or highly marked nature; though the example is not strictly ungrammatical, it deviates from usage standards

| Introduction

1 Introduction: from quantifying nouns to binominal quantifiers This study presents a cognitive-functional analysis of quantifying nouns within binominal quantifier constructions in Spanish. Quantifying nouns (henceforth QNs) are nouns which display a quantifying potential in addition to their lexical meaning. For instance, literally, montón ‘heap’ designates a collection of things lying in a disorderly way one upon another so as to form an elevated conical configuration. Yet since the heap-like mass is necessarily composed of more than one constituent, the focus may shift from the unitizing constellation or whole to the plurality of constituting entities. Binominal quantifiers (henceforth BQs) are two-constituent patterns which consist of a determiner, a QN (N1), the preposition de ‘of’ and a plural or mass noun (e.g. libros ‘books’, gente ‘people’) which designates the constituting entities (N2), as in (1). The claim will be made that QNs can acquire quantifying uses within the binominal syntagm. (1)

un montónN1 de amigasN2 ‘a heap of friends’

The cognitive-functional starting assumption of this analysis is that BQs constitute a specific way to conceptualize, i.e. to construe, the quantity assessment of N2, next to the canonical quantification by absolute quantifiers (e.g. cardinal numbers, mucho/a(s) ‘many, much’). In other words, a speaker who utters un montón de amigas has opted for the quantifier un montón de for a specific reason and, by doing so, (s)he conceptualizes the (number of) friends in a manner which is distinct from the conceptualization evoked by muchas amigas ‘many friends’ for instance. An obvious theoretical challenge presented by the research topic is the identification of the head or nucleus of the binominal construction. Traditionally, the prepositional phrase is considered to introduce a complement to the first noun – or head – of the construction, yielding the underlying structure [Det. N1 [de N2]]. The analysis is workable in instances such as un montón de libros: the speaker profiles a heap of books, instead of a heap of paper or of dishes for instance. However, the underlying structure with the QN as the syntactic head clashes with instances such as un montón de amigas: here, the speaker certainly does not wish to picture an elevated mass of friends, yet a large number of them. In other words, amigas seems to function as the semantic core – or head – of the construction, yielding the structure [[Det. N1 de] N2]. The matter becomes even more complicated in metaphorically loaded occurrences such as provocó un alud de protestas ‘he provoked an avalanche of protests’, where the identification of the semantic head is not straightforward either. Although the structur-

4 | Introduction: from quantifying nouns to binominal quantifiers

al difference is invisible in the surface manifestation of the construction, the alternative dependency analysis entails major consequences in the domains of verb agreement and anaphoric reference. Equally eye-catching is the variation observed in the QN-paradigm: in addition to highly frequent and/or conventionalized QNs such as montón ‘heap’, mogollón ‘mess’ and mar ‘sea’, the QN-paradigm is open to any kind of noun with a quantifying potential. In addition, the study of BQs verges on the interface between syntax and semantics. The quantifying potential or inherent ambiguity of typical QNs raises the question whether QNs presenting quantifying uses (e.g. una pila de años ‘a lot of years’, un alud de críticas ‘an avalanche of criticisms’) are polysemic lexical items, or whether the quantifying interpretation is to be attributed to the occurrence in the binominal construction. In the former case, the quantifying meaning ‘a lot’ is an additional sense to the literal configuration-reading (respectively ‘pile’ and ‘avalanche’) and captured by a separate dictionary definition. However, while the dictionary entry of the DRAE1 for pila encloses the quantifying reading, the quantifying reading of alud is not stipulated. By contrast, in the latter case, i.e., if the quantifying reading is attributed to the constructional semantics, the continuation of the literal interpretation as in una pila de libros ‘a pile of books’ and un alud de nieve ‘a slide of snow’ calls for an answer. Further, the fact that some QNs have given rise to additional quantifying uses outside the binominal construction as well (as in me gusta mogollón ‘I like it a lot’ or está la mar de bien ‘(s)he is doing perfectly fine’), seems hard to reconcile with the second train of reasoning. The aim of the corpus-based analysis is threefold. First, the study seeks to provide an accurate and fine-grained analysis of the development of BQs (AIM 1). In addition, it tries to shed light on the functional organization of BQs and on the range of uses the construction is apt to display (AIM 2). Finally, the study intends to reassess the secondary status generally attributed to analogy and persistence in semantic changes (AIM 3). The guideline crossing the diachronic and synchronic case-studies is the iconicity-principle, i.e. the motivated, non-arbitrary nature of syntax. Diachronically, the claim will be made that formal changes lag behind semantic changes to the extent of being triggered by them. Synchronically, it will be argued that both the variation in the QN-paradigm and the morphosyntactic behavior of the QN-construction are conceptually motivated.

|| 1 DRAE stands for the Diccionario de la lengua española of the Real Academia Española.

Aim 1: Towards a constructional network model of the development of BQs | 5

1.1 AIM 1: Towards a constructional network model of the development of BQs In the literature, the BQ-construction is repeatedly claimed to be the locus of grammaticalization processes, both in Spanish as cross-linguistically: within the binominal construction, the QN shifts from noun-status to the functional quantifier-status. Yet in contrast to genuine cases of grammaticalization, the BQ-construction remains highly analyzable. In spite of obviously specifying the quantity of N2, the QN is pluralized in (2) and combines with a premodifying adjective in (3), which are both typical noun-features. Likewise, the predicate inundados in (3) recalls the source semantics of the literal avalanche, thereby suggesting that the QN has not desemanticized and the BQ remains compositional to a large extent. The previous observations point to an impure or hybrid status as grammaticalized item: they question not only the reanalysis of N1 as a true quantifier, but also the analysis of the entire string [un N1 de] as a single chunk. However, Brems’ (2007b; 2011) fine-grained analysis of the BQ-construction in English, which has been the starting point of this book, has shown that the grammaticalization of BQs can be operationalized in terms of changes in combinatorial pattern. The present investigation will illustrate that in Spanish as well, all constructional slots have become more constrained in the grammaticalized uses. (2)

Ha tenido ocho o diez esposas, montones de hijos y nietos... (CREA, 2003, press) ‘He has had eight or ten wives, heaps of children and grandchildren…’

(3)

¿No nos hemos visto “inundados” por un repentino alud de productos agrícolas e industriales, (…)? (CREA, 1995, novel) ‘Did we not see ourselves “inundated” by a sudden avalanche of agricultural and industrial products, (…)?’

My interest lies not only with the grammaticalization per se, but also, and most prominently, with the semantic-pragmatic motivation for such change as well as the mechanisms that operate in it. A complex interplay between analogical thinking and conceptual persistence will be put forward at different levels of schematicity. Although the early emergence and dramatic increase in frequency of the prototypical QN montón invites to picture the grammaticalization of the remaining QNs in terms of a snowball-effect, this study will focus on the divergent pathways of change followed by the individual QNs. Finally, a constructional network model of the BQ-construction will be posited in an attempt to

6 | Introduction: from quantifying nouns to binominal quantifiers

account for the complex frequency pattern, the different degrees of grammaticalization, the variety of pathways followed and the functional overlap with expressive binominal constructions.

1.2 AIM 2: Towards a cognitive-functional model of the organization of BQs The synchronic description deals with the extensive variation in the QNparadigm. From a usage-based perspective, the claim that all individual grammaticalizing QNs run the desemanticization-course completely and end up expressing ‘much, many’ – which would be a fairly uneconomic solution – is unacceptable. The present study will explain the immense variation in the QNparadigm in the light of their pragmatic utility. The claim will be made that QNs do not merely express the quantity of N2, but also characterize N2. By way of illustration, the selection of the QN in (4) entails a slightly different conceptualization of the N2 gente: while montón profiles a chaotic group of (different kinds of) people, pila profiles a kind of organized sequence or row of people and aluvión rather profiles a group of invaders which show up suddenly and simultaneously. The potential to categorize N2 is overtly realized in the second type of grammaticalized uses, viz. the (two-way) specifier use, where quantity assessment is backgrounded to the benefit of type specification, as in (5). (4)

Se me acercó(/aron) un montón de / una pila de / un aluvión de gente. ‘A heap / pile / flood of people came close to me.’

(5)

Para ella, no éramos más que un hatajo de arribistas y traidores. (CREA, 2002, novel) ‘To her, we were nothing more than a bunch (lit. herd) of careerists and traitors.’

This study will argue that the QN imposes a specific conceptual image on N2. The source semantics of the QN can shade through in the grammaticalized uses, yet displays various levels of abstraction or schematization. The differences in conceptualization may be very subtle: regarding the near-synonymous alud ‘avalanche’ and aluvión ‘flood’, I will show that receiving un aluvión de críticas ‘a flood of criticism’ will be experienced as less face-threatening than un alud de críticas ‘an avalanche of criticism.’ In addition to turning the BQ-construction in a useful tool for expressive and hyperbolic quantification, the persistence of the source semantics in the

Aim 3: Towards a reassessment of persistence and analogy | 7

functional uses or the QN-related construal of N2 also constrains the grammaticalization of the BQ-construction in many ways, most prominently in its coselection pattern. By way of illustration, while montón seems free to combine with any type of N2 (count or mass nouns as well as concrete and abstract entities), hatajo is restricted to N2s designating unpleasant human entities. Further, the in-depth corpus-study will illustrate that the morphosyntactic behavior of the BQ-construction is largely conceptually motivated.

1.3 Aim 3: Towards a reassessment of persistence and analogy A close examination of the concepts of lexical persistence and analogy is the main theoretical concern of the present work. In theorizing on grammaticalization, they are usually ascribed secondary importance only. Lexical persistence refers to the tendency of some grammaticalizing items to retain particular features of their original lexical use, which may continue to influence their further development in various ways. This tendency goes hand in hand with desemanticization or the loss of concrete meaning which is commonly considered to be both a prerequisite and the most important semantic change in grammaticalization. By contrast, analogy refers to the process whereby a specific construction is perceived as structurally or semantically similar to another construction. The mechanism is usually considered less ‘necessary’ than reanalysis, i.e. the mechanism whereby a specific construction is assigned a new underlying structure without overtly modifying it. In line with the recent trend in grammaticalization studies to stress the gradual nature of changes and the gradient nature of category-boundaries, this study will claim that analogical thinking is an essential trigger and analogy a crucial mechanism in the grammaticalization of BQs. Both closely interact with conceptual persistence. To conclude this general introduction, I will briefly outline the structure of this volume. Part I provides the necessary theoretical background to the casestudies and consists of two chapters. Chapter 2 first introduces the cognitivefunctional framework I subscribe to and outlines the cognitive-functional model of BQs, which forms the point of departure for the case-studies. Second, as the BQ-construction is generally overlooked in the literature, the section concerning the state of the art is rather limited yet traces the major research questions posited so far. The third and final section of Chapter 2 delineates the research topic and introduces the methodology. Chapter 3 discusses the starting assumption of

8 | Introduction: from quantifying nouns to binominal quantifiers

this investigation, viz. the grammaticalization of BQ-constructions which results in quantifying and premodifying uses. Part II presents the diachronic description of BQ-constructions. While Chapter 4 zooms in on the prototype among the QNs, viz. montón de, Chapter 5 analyzes the emergence and the pathways of changes followed by another six QNs, viz. pila ‘pile’, aluvión ‘avalanche’, letanía ‘litany’, (h)atajo ‘herd’, barbaridad ‘barbarity’ and mar ‘sea’. Both chapters repeatedly stress the importance of persistence and analogy in the development towards new uses. Chapter 6 is devoted to the theoretical implications of the diachronic case-studies and puts forward a constructional network model of the BQ-construction in Spanish. Part III turns to the synchronic description. Chapter 7 is mainly concerned with the semantic-pragmatic aspect of the BQ-construction and characterizes the QN-related construals as gradual abstractions of the QN’s source semantics. Chapter 8 focuses on the morphosyntactic characteristics of the BQ-construction and the underlying conceptual motivations for the various co-selection patterns. Chapter 9 returns to the necessary and conceptual refinements of persistence and analogy and attempts to harmonize the synchronic findings with the theoretical implications of the diachronic casestudies. To conclude, Part IV briefly overviews the findings and claims presented in this volume and highlights some prospects for further research.

| Part 1 Preliminaries

2 Framework, state of the art and methodology The present chapter aims to motivate the choices made as to the framework adopted, the delimitation of the research topic and the methodology. Section 2.1 justifies the cognitive-functional approach of the BQ-construction. The compact State of the Art in Section 2.2 sheds light on the raison d’être of this monograph. Section 2.3 briefly introduces the selection of the dataset and the methodology adopted.

2.1 Description of the framework This volume is not to be located within a specific framework, yet at some kind of interface between functional, cognitive and constructional linguistics. Although the three clusters of linguistic theorizing can be associated with one or more particular seminal formulation,2 there is a lot of convergence as to the basic assumptions of the three models (Nuyts 2007; Langacker 2007; 2009b). By way of illustration, one the founders of cognitive linguistics argues that his Cognitive Grammar is “strongly functional”, provided that the two basic functions of language are symbolic (…) and communicative / interactive. The symbolic function is directly manifested in the very architecture of Cognitive Grammar, which posits only symbolic structures for the description of lexicon, morphology, and syntax. A manifestation of the communicative/interactive function is the fundamental claim that all linguistic units are abstracted from usage events. (Langacker 2007, 422);

yet also more radical than (radical) construction grammar (Langacker 2005, 106), in that grammatical form is considered to be meaningful in itself (whereas in the constructional model, grammatical form symbolizes meaning). The fuzzy boundaries between cognitive and constructional approaches are however more commonly acknowledged than the crossing between functional and cognitive

|| 2 In terms of ground-breaking theory-building, my view on binominal quantifiers is mainly inspired by the functional approach of Halliday/Matthiesen (2004), the cognitive model as founded by Langacker (1987; 1991) and the constructional guidelines formulated by Goldberg (1995; 2006) and Croft (2001; 2007). For detailed literature reviews, see Croft/Cruse (2004), Cuenca/Hilferty (1999), Dirven/Verspoor (1998), Geeraerts/Cuyckens (2007), Langacker (2005; 2009b), Nuyts (2007), Schönefeld (2006) and Ungerer/Schmid (1996).

12 | Framework, state of the art and methodology

principles.3 This is probably due to the fact that constructional linguistics is repeatedly presented as under construction.4 The basic tenets underlying the present investigation can be summarized as follows: grammar is intrinsically meaningful, iconic to a large extent and also usage-based. The present study is functional in that the BQ-construction is analyzed in the light of the uses – both semantic and pragmatic – it is apt to display in actual contexts. Further, the triggering factors in the grammaticalization of the construction are related to the communicative needs of the speaker. The present study is cognitive-constructional by the importance it attributes to the semantics of the BQ-construction: BQs are characterized as a specific way to construe a set of entities, i.e. to conceptualize quantity assessment, and the morphosyntactic make-up of the construction is considered to be conceptually motivated. Further, all constructional slots of the partially filled BQ-construction are analyzed both syntagmatically and paradigmatically. Finally, the construction is considered symbolic at different levels of schematization. In the remainder of this section, I will briefly touch upon five key-concepts which form the backbone of the descriptive analysis (Section 2.1.1). However, the notion of conceptual persistence, which is the actual keynote of the present volume, will be specified in due time. Section 2.1.2 will outline the cognitivefunctional model of binominal quantifiers which is taken for granted in this study.

2.1.1 Definition of five interrelated key-concepts The present section outlines my understanding of five key-concepts in this book which fit the cognitive-functional framework I subscribe to. The closely related concepts of iconicity, construal, schematization, subjectification and construction will be treated in this order.

|| 3 I therefore opt to define my analysis as cognitive-functional, using cognitive as a cover term for both cognitive and constructional approaches. 4 Four prominent lines of research are usually distinguished within constructional approaches, associated respectively with Fillmore and colleagues' seminal construction grammar (1988), Goldberg’s foundational Construction Grammar (1995; 2006), Langacker’s Cognitive Grammar (1987; 1991; 2005) and Croft’s Radical Construction Grammar (2001; 2007; Croft/Cruse 2004). They differ as to how formalist, how usage-based, how cognitive and how typologically universal they are oriented (see also Chapter 6).

Description of the framework | 13

2.1.1.1 Iconicity The notion of iconicity is related to Peirce’s (1974) well-known typology of signs which distinguishes between indices, icons and symbols. An indexical sign “points to something in its immediate vicinity”, an iconic sign “provides a visual, auditory or any other perceptual image of the thing it stands for”, while a symbolic sign “does not have a natural link between the form and the thing represented” (Dirven/Verspoor 1998, 2) but a conventional link instead. In other words, icons are similar to the things they represent. Traffic signs which warn drivers to look out for crossing pedestrians are iconical in that they picture a pedestrian at a zebra crossing. Peirce (1974) further distinguishes between three subtypes of icons, i.e. images, metaphors and diagrams. A diagrammatic icon is “a systematic arrangement of signs that do not necessarily resemble their referents but whose mutual relations reflect the relations between their referents” (Van Langendonck 2007, 398). The latter subtype, i.e. diagrammatic iconicity, encompasses iconicity in language. The principle of iconicity in language refers to the (possible) similarity between a form and its referent, as in onomatopeia (e.g. cuckoo), and has seen a surge of interest in cognitive linguistics (e.g. Dirven/Verspoor 1998; Delbecque 2002b; García 2009; Haiman 1994; Wierzbicka 1985)5 whose credo is diametrically opposed to autonomous linguistics advocating the arbitrariness of linguistic signs (Van Langendonck 2007, 396). Many cognitive linguists claim, in line with Wierzbicka, that: [i]f we study the correlations between grammatical behavior and conceptualization directly, the apparent arbitrariness of some aspects of grammatical behavior will be greatly reduced and in many cases may vanish altogether. (Wierzbicka 1985, 316)

Iconicity in language is generally associated with three principles related respectively to sequential order, distance and quantity (Dirven/Verspoor 1998, 8– 12, Van Langendonck 2007, 402–413). The principle of sequential order pertains to the linear arrangement of linguistic elements in a construction. The textbook example is the linear word order in Julius Caesar’s historic words Veni, vidi, vici ‘I came, I saw, I conquered’ which reflects the sequence of events. Reversing the order as in ‘Vici, vidi, veni’ would produce nonsense (Dirven/Verspoor 1998, 8). As to BQs, the claim will be made that the QN, whose processing has started before N2 is uttered, impose a schematic-image structure of the mass on N2. In

|| 5 Of direct relevance for this book was the paper by Wierzbicka (1985) which challenges the arbitrariness in the count-mass noun distinction.

14 | Framework, state of the art and methodology

line with Fischer (1999), I will show that the tendency towards persistence is iconically motivated. For the sake of completeness, the principle of distance accounts for the tendency to put together linguistically elements which belong together conceptually while the principle of quantity suggests that more form entails more meaning and is a.o. associated with politeness strategies. These subprinciples are however less relevant for the study of BQs.

2.1.1.2 Construal Construal refers to the human capacity to “conceive and portray the same situation in alternate ways” (Langacker 2007, 435). As a challenge to the truthconditional model, a basic principle in cognitive semantics is that meaning is “not a matter of relationships between language and the world” (Verhagen 2007, 48), but mainly cognitive and emergent in nature (Geeraerts 1993; Paradis 2011).6 Since language provides speakers with a wide variety of linguistic items to describe particular situations or objects, any utterance automatically implies a choice: A speaker who accurately observes the spatial distribution of certain stars can describe them in many distinct fashions: as a constellation, as a cluster of stars, as specks of light in the sky, etc. Such expressions are semantically distinct; they reflect the speaker’s alternate construals of the scene, each compatible with its objectively given properties. (Langacker 1990, 61, quoted in Verhagen 2007, 49)

Crucially, any construal operation involves two entities: the speaker(/hearer) or the subject of conceptualization7 and the scene or object of conceptualization. It is the speaker who decides how specific his choice is and which element of the scene is profiled or foregrounded. In addition, the speaker’s perspective or viewpoint vis-à-vis the object of construal can differ. Many authors have proposed classifications of construal operations (e.g. Langacker 1987; 2007; Talmy 2000a; 2000b; Croft/Cruse 2004), in which perspective, attention and specificity reappaer as constants. However, after providing an overview of existing construal taxonomies, Verhagen (2007, 56) con|| 6 Geeraerts (1993, 259) suggests that cognitive semantics should be primarily concerned with meaning as “a process of sense creation”. Paradis (2011, 65) reserves a prominent role for the discourse context and argues that the meanings of linguistic expressions “are emergent and get their final interpretation in context”. 7 The subject of construal is also called ground in the Langackerian framework. The typical grounding function (cf. Section 2.1.2.1) in nominals is a subtype of construal.

Description of the framework | 15

cludes that “construal operations may vary in so many different respects that attempts at an exhaustive classification necessarily have a considerable degree of arbitrariness”. A more reasonable approach is to establish the set of construal operations in function of the object of analysis. In my view, the conceptualization of BQs involves a decision (i) as to specificity (or schematicity) of the image-schematic structure N1 imposes on N2; (ii) as to focus of attention (or selection of conceptual content) both within the conceptual image of N1 and at the composite structure level; (iii) as to prominence (or salience) of one or more particular conceptual elements in the set of facets that compose the QN’s conceptual image and (iv) as to perspective in terms of degree of subjectivity. The following sections dwell on the construal operations relating to schematicity and subjectivity respectively.

2.1.1.3 Schematicity Schematicity is the converse of specificity and refers to the coarse-grainedness, respectively fine-grainedness of detail with which something is construed. For instance, daughter is more specific than child, which is in its turn more specific than relative. Since schematization or the ability to generalize over details is “one of the most central human cognitive capabilities” (Tuggy 2007, 83; see also Langacker 2008, 56), it does not come as a surprise that schematicity, schemas, abstraction and elaborations are central concepts in cognitive linguistics.8 A workable basic definition of schematicity is provided by Tuggy: [a] schema is a superordinate concept, one which specifies the basic outline common to several, or many, more specific concepts. The specific concepts, which are called elaborations or instantiations or subcases of the schema, fill in that outline in varying, often contrasting ways. (2007, 83)

It bears pointing out that schemas can only be established by virtue of their more specific instances.9 Further, schematicity is a gradable notion: one and the same concept can simultaneously serve both as an elaboration of a particular schema and as a schema relative to more specific concepts. For instance, mammal is schematic to squirrel but specific in relation to animal. Schematicity rela|| 8 In Langacker’s terms, “[t]he linguistic significance of this ability is hard to overstate” (1987, 135). 9 “It does not make sense to call a concept ‘schema’ or say it is ‘schematic’ except in the context of specific cases relative to which it is abstract or whose information it represents at a coarser level of detail.” (Tuggy 2007, 83–84).

16 | Framework, state of the art and methodology

tionships are diagrammatically represented by arrows, “with the schema at the tail and its elaboration at the head of the arrow” (Tuggy 2007, 84), as in ANIMAL → MAMMAL → SQUIRREL. Schematicity is also a transitive notion in that if A → B and B → C, then automatically A → C.10 Interestingly, linguistic expressions and human concepts always involve a certain degree of schematicity. For instance, the concept of tall can apply to both John and Mary even if they are not equally tall. Without the inherent range of variation or imprecision in concepts such as tall, it would hardly be a useful or common concept in daily communication. Two schematicity relationships are usually distinguished (cf. Langacker 1987, 372–373; Tuggy 2007, 86–88): full sanction, or schematicity in the strict sense, and partial sanction, or similarity. The relation between a source concept and a target concept is described as full schematicity when there is full coincidence and “all the standard’s features are preserved in the target” (Tuggy 2007, 86). Yet more common in comparison judgments is partial schematicity, which occurs when there is no full coincidence but rather “omission, contravention, or distortion of the standard’s specifications” (Tuggy 2007, 86). According to Langacker, the comparison between the more schematic entity A and the more specific entity B which gives way to the judgment of partial schematicity automatically establishes the schema C that activates the features A and B have in common. The establishment of schemas is particularly relevant to the study of BQs. Indeed, a central working hypothesis in this volume is that every QN is characterized by a specific conceptual image or image-schematic structure. The QNrelated conceptual image consists of a set of conceptual facets of varying degrees of specificity (full schematicity). When QNs are used within the binominal construction, a particular (subset of) facet(s) match(es) the plural entity profiled by N2 (partial schematicity). The selection of facets that enters the so called ‘schema C’ is highly context-dependent and differs from one instance to another. It goes without saying that, since schematicization is a basic mental ability and schematicity relationships are pervasive in language structure, schematicity is a fertile concept in many domains of language. In cognitive linguistics schematicity is invoked when explaining “such traditional concepts as polyse|| 10 Note that Lakoff’s notion of image schema is related to a more restricted type of schematization (see also the overview by Tuggy 2007, 84–85). In Lakoff’s view (1987, 267–268), image schemas are relatively simple structures which constantly recur and are based in bodily experience, such as CONTAINERS, PATHS, LINKS, FORCES, PART-WHOLE and UP-DOWN orientations, etc. They are directly meaningful, preconceptual and universal structures in human experience.

Description of the framework | 17

my, syntactic categories, rules, analogy, figurative language, headship and valence, and composition, in useful and intuitively satisfying ways” (Tuggy 2007, 82).

2.1.1.4 Subjectivity Basically, subjectivity relates to speaker-involvement in utterances. In a sense, language is inherently subjective in that “it passes through a speaker and is, as such, speaker-related” (De Smet/Verstraete 2006, 370). For instance, any utterance involves the speaker’s lexical choice on how to represent the extralinguistic world. Speaker-involvement has many dimensions and has been described in several frameworks in varied ways.11 In cognitive grammar, subjectivity and its diachronic counterpart subjectification are associated with two prominent exponents, viz. Traugott (1989; 2003a; 2007; 2010) and Langacker (2006; 2007). The competing views are certainly not unrelated nor incompatible (Athanasiadou et al. 2006, 2; Langacker 2006, 17), yet pertain to different aspects of meaning change. While Traugott’s notion of subjectivity pertains to change in meaning, Langacker’s subjectivity concerns change in construal. The differences between both approaches have been formulated as follows: (i) Traugott looks at semantic change from a diachronic perspective while Langacker is concerned with subjectivity from a synchronic point of view; (ii) the former reinforces the importance of pragmatics within the speaker/hearer negotiation of meaning, while the latter focuses on construal of meaning by the speaker, leaving the hearer out of the picture (Athanasiadou et al. 2006, 5); (iii) the former is concerned with the presence/absence of speaker-relatedness while the latter concentrates on how explicitly reference is made to the speaker (De Smet/Verstraete 2006, 369); (iv) Traugottian subjectification is a mechanism in change while Langackerian subjectification is about the relationship between original and extended meanings (Langacker 2006, 29); (v) Traugottian subjectification is unidirectional, while Langackerian subjectification goes hand in hand with objectification; (vi) in Traugott’s view, the speaker is a real and actual person, whereas in Langacker’s view, subjectivity relates to the expression of a point of view, viz. that of a subjective conceptualizer, which is not necessarily

|| 11 For an overview of the history/definitions of subjectivity and subjectification, see Athanasiadou et al. (2006); Brisard (2006); De Cock (2010; 2014); Desmet/Verstraete (2006); Ghesquière (2011).

18 | Framework, state of the art and methodology

the speaker’s (Brisard 2006, 47).12 A major point of convergence between both approaches is the idea that at least part of the new meaning is inherited from the original meaning, which links subjectivity to a central notion in this volume, viz. persistence. In what follows, both views will be presented separately insofar as relevant for the study of BQs. Traugottian subjectivity relates to the expression of the speaker’s perspective. In subjectification – as a diachronic process – meanings become “increasingly based in the speaker’s subjective belief state/attitude toward the proposition” (Traugott 1989, 35) by the process of “invited inferencing”. The notion of invited inferencing is inherently pragmatic and “produces a strengthening of informativeness or informational relevance, whereby the hearer is invited to infer more than what is directly communicated” (Brisard 2006, 44). A crucial distinction is made between subjectivity that happens to accompany a particular use of a form (pragmatic subjectivity) and its development into a coded meaning (Traugott 2010, 35). She therefore refines her initial definition of subjectification as follows: “the reanalysis as coded meanings of pragmatic meanings arising in the context of speaker-hearer negotiation of meaning” (2010, 60). Traugott repeatedly associates subjectivity with unidirectional clines in semantic changes. In her first assessment of subjectification (Traugott 1989, 34– 38), the shifts in meaning follow the pathway (i) from meanings based in the external described situation to internally-based descriptions (e.g. the pejoration involved in boor ‘farmer’ > ‘crude person’); (ii) from externally or internallybased descriptions to textual or metalinguistic meaning (e.g. the development of lexical forms into connectives); (iii) and finally to meanings which are increasingly based in the speaker’s perspective (e.g. the shift of temporal to concessive while). In the last ten years, Traugottian subjectification has become even more rooted in the speaker-hearer negotiation of meaning and echoes the notion of intersubjectification. Intersubjectivity relates to the speaker’s attention to the addressee’s self-image. By contrast, Langackerian subjectivity is above all an important construal operation. Both subjectivity and objectivity pertain to the status of individual conceptual elements within the overall construction, which as a whole, contains simultaneously subjectively and objectively construed elements. In his view, an element is construed objectively when it functions “onstage” as an explicit, focused object of conception. A subjectively construed element on the

|| 12 Langacker’s view largely follows Benveniste’s (1966) distinction between the “sujet d’énoncé” and the “sujet d’énonciation”.

Description of the framework | 19

other hand, remains “offstage”, as an implicit, unselfconscious subject of conception. Figure 1 diagrammatically visualizes the construal arrangement: minimally, the expression comprises the subjectively construed discourse participants, “in their offstage role as the conceptualizers who employ the expression and thereby apprehend its meaning” (2006, 18) as well as the expression’s profile which is objectively construed.

maximal scope of conception

O 

focused object of conception (profile) onstage region (immediate scope) apprehension/construal by S

S

subject of conception (speaker/hearer)

Fig. 1: Basic construal arrangement according to Langacker (2006, 19)

Langackerian subjectification has to be “thought of as a kind of semantic ‘bleaching’ or ‘fading away’” (2006, 21), i.e. shift from being objectively construed to subjectively construed. By way of illustration, Langacker (2006, 23–24) evokes the shift of to be going to from motion into future sense. While in the original motion reading, the spatial movement through time is profiled, the actual motion fades away in the future reading. Nevertheless, this spatial movement continues to be traced mentally by the subject of conception who attempts to situate the process expressed by the following infinitive with regard to the reference time. This fictive or mental motion is thus subjectively construed and remains as a vestige of the objectively construed spatial movement. In other words, the subjectively construed counterpart was there all along, “immanent in the latter (i.e. inherent in its conception). It simply becomes more evident when the objectively construed element is no longer there to mask it” (2006, 21). Generalizing over the points of divergence between Traugott’s and Langacker’s approach, subjectivity (and with it subjectification) are characterized in both models as inherently gradual notions (Langacker 2010, 18; Traugott’s clines of subjectivity 2010, 36). In addition, although subjectification is usually evoked as a mechanism of meaning change in grammaticalization,

20 | Framework, state of the art and methodology

both Traugott and Langacker underline the independence of subjectification with regard to grammaticalization and vice versa. Traugott (2010) further specifies that if subjectification co-occurs with grammaticalization, it is more likely to occur in primary grammaticalization (the shift from lexical/constructional to grammatical) than in secondary grammaticalization (the development of already grammatical material into more grammatical material). This is because primary grammaticalization often requires prior strengthening of pragmatic inferences that arise in very specific linguistics [sic] contexts prior to their semanticization and reanalysis as grammatical elements. (2010, 40)

Although both Langackerian and Traugottian interpretations of subjectivity are involved in the development and functional organization of BQs, subjectivity will mainly be used in its sense of construal operation in this volume. The development of BQs involves Traugottian (pragmatic) subjectification in that the quantifying interpretation arises first as a pragmatic inference and can become encoded with repetition. In una pila de libros ‘a pile of books’, a scalar evaluation can be inferred. Further, the shift towards qualifying readings as in un hatajo de idiotas ‘a bunch (lit. herd) of idiots’ is clearly internally-based and encompasses the speaker’s evaluation of the referents of idiotas. Yet most BQs have not reached the status of intersubjectivity yet. Langackerian subjectification is at stake in the functional organization of BQs. A shift in profile takes place when the QN is reanalyzed as a quantifier. For instance, literal pila profiles a vertically oriented, orderly arranged pile of entities (e.g. una pila de libros ‘a pile of books’), while the constituting entities (the books) remain at the background. This facet ‘order’ is subjectively construed in the quantifying reading (e.g. una pila de años ‘a lot (lit. pile) of years’): while the focus is now on the constituting entities (the years), the succession or sequence of years is subjectively construed. Figure 2 is intended to visualize à la Langacker the shift from objectively construed neatly arranged configuration to subjectively construed successiveness in pila de. The figure is based on Langacker’s conventions: a circle stands for a bounded entity, an ellipse represents a mass, heavy lines indicate profiling, the double line indicates that the two entities (in this case N1pila and N2) are coextensive.

Description of the framework | 21

Immediate scope

Immediate scope

Maximal scope

S

S Maximal scope

Fig. 2: Subjectification involved in the shift from literal to quantifying pila de

2.1.1.5 Construction A construction is a symbolic unit pairing a specific form to a specific meaning, as visualized in Figure 3. The element form refers “to the overt manifestation of language, i.e. a linguistic expression in its material, or perceptible aspects” (Taylor 2002, 20). Since linguistic expressions are prototypically manifested in sound, this element is called phonological structure. The element semantic structure refers to the meaning of expressions, thereby including not only the propositional content, but also the “broader conceptualization that a speaker entertains” (Taylor 2002, 21) as well as the pragmatic aspects (i.e. the expression’s relation to the situational context). The third necessary element of a construction is the symbolic relation linking the form and the meaning of the construction. The arrow in Figure 3 profiles the symbolic link as “a two-way affair”, in that “each pole of the symbolic relation invokes the other” (Taylor 2002, 21).

phonological structure

symbolic relation

semantic structure

Fig. 3: The three elements of a construction or symbolic unit (adapted from Taylor 2002, 21)

Crucially, the strictly constructional definition slightly diverges from the cognitive one in Figure 3 as to the element form. According to Croft (2007), the formpart can contain properties concerning the specific phonological structure, the internal morphological structure of the construction, as well as the interconstructional syntactic properties (cf. Figure 4). In other words, in the construction grammar approach, grammatical form itself symbolizes meaning. Recall that in

22 | Framework, state of the art and methodology

Langacker’s cognitive approach (1987; 2005), by contrast, grammatical form is itself considered to be inherently meaningful. In this study, I subscribe to Langacker’s more radical view in that I claim that the morphosyntactic behavior of the BQ is largely iconic or conceptually motivated (cf. Chapter 8).13

syntactic properties

CONSTRUCTION

morphological properties phonological properties

FORM

symbolic correspondence (link) semantic properties pragmatic properties

(CONVENTIONAL) MEANING

discourse-functional properties

Fig. 4: The symbolic structure of a construction according to Croft (2007, 472)

One of the basic cognitive-constructional principles is the uniform representation-hypothesis, i.e. the idea that generalized constructions constitute the uniform representation of grammatical knowledge. This principle is also referred to as the syntax-lexicon continuum and reflects an essential characteristic of constructions: they can be arranged on a two dimension-continuum – of symbolic complexity and specificity – ranging from atomic to complex and from substantive to schematic (Croft 2007, 471; Langacker 2005, 108). Put differently, according to Langacker (2005), constructions differ in degree, not in kind. Or as in Goldberg’s (2006, 18) slogan: “the network of constructions captures our knowledge of language in toto, i.e. it’s constructions all the way down”. Provided that items constitute symbolic pairings of form and meaning, they can be called constructions. By way of illustration, lexical items or words consist of a || 13 Another major terminological discussion relates to the prerequisite(s) of construction. In Langackerian terms, any form-meaning pairing is a construction. In Goldbergian terms, constructions have to be either unpredictable (i.e. the meaning of the construction is idiomatic and has to be learned separately, Goldberg 1995) or predictable and sufficiently frequent (Goldberg 2006). To the latter usage-based view, García (2009, 9–10) objects that high frequency is not enough for an item to be called a construction. In her view, the usage-based view detracts the attention from the primary requisite to present a signifié.

Description of the framework | 23

phonological form which is mapped on a conceptual content or meaning it evokes. Likewise, highly schematic and complex syntactic constructions, such as the ditransitive construction, pair a specific form (viz. [Subj V Obj1 Obj2]) to a specific semantic content (viz. the transfer of possession, Goldberg 1995, 38). Further, it is commonly assumed in constructional approaches that constructions are organized in a taxonomic network. Each construction containing specific unpredictable or idiosyncratic properties has to be listed as a separate node in the network. A plausible taxonomic network is listed by Croft/Cruse (2004, 263) for the substantive idiomatic phrase kick the habit: [Verb Phrase] | [Verb Obj] | [kick Obj] | [kick [the habit]] In similar networks, constructions are independent but related in terms of schematicity. More precisely, the idiomatic phrase kick the habit is represented as an independent node for conveying a specific idiosyncratic semantic interpretation, while at the same time constituting an instantiation of the more schematic verb-specific construction. Similarly, the wholly schematic construction has to be represented as independent for imposing a particular, though highly schematic, semantic content. The relations of schematicity between different nodes is alternatively referred to as ‘inheritance relations’ (Goldberg 2009b, 99): the highest nodes in the hierarchical network are broad generalizations, captured by constructions which are inherited by many others. However, strict taxonomic hierarchy doesn’t exist: each specific utterance has multiple parents (“multiple inheritance model”, Goldberg 1995, 31). I didn’t sleep, for instance, does not only instantiate the intransitive but also the negative construction as well as constructions marking tense, mood and aspect (Croft/Cruse 2004, 264). As to BQs, I will claim the status of construction for both the concrete, individual BQ-construction (e.g. un montón de N2 ‘a heap of N2’, una pila de N2 ‘a pile of N2’) and the abstract BQ-construction un(a) QN de N2. At different degrees of schematicity, the binominal pattern will be shown to be meaningful. Further, I will search for similarities and differences among the set of BQ-constructions, between BQs and regular quantifiers as well as between BQs and expressive binominal constructions, both paradigmatically and syntagmatically.

24 | Framework, state of the art and methodology

2.1.2 Towards a cognitive-functional model of binominal quantifiers The basic tenet of the strictly functional (Halliday/Matthiesen 2004) as well as the strictly cognitive account (Langacker 1991; Taylor 2002) of nouns and noun phrases is that there is no one-to-one correlation between grammatical categories and semantic functions. For instance, quantity assessment can be achieved by a series of word classes: determiners (e.g. a, some), numerals (e.g. one, seven), quantifiers (e.g. many, a lot), adjectives (e.g. numerous, infinite), nouns (e.g. majority, deal), etc. Reversely, one of the main aims of this book is to shed light on the variety of meanings the BQ-construction is apt to display. This section will zoom in on the cognitive model provided by Langacker (1991) and Taylor (2002), “which is in fact more radically functional than that of Halliday” (Ghesquière 2011, 21) and to which my model of BQs is largely indebted.

2.1.2.1 Nouns and noun phrases Basically, the difference between nouns and noun phrases – nominals in Langacker’s terms (1991, 51) – resides in the distinction between a lexical and syntactic category. While cat is a noun, the cat or my grandmother’s cat are noun phrases. While cat designates a kind or type of thing, comprising countless real and possibly imaginary instances, the cat designates a specific instance of that type (Taylor 2002, 343). By the addition of the or my grandmother’s, the hearer is able to identify the unique cat instance the speaker is referring to. The internal conceptual organization of noun phrases is subdivided into four interrelated functions: type specification, instantiation, quantification and grounding. Type specification can be achieved by the bare noun, which specifies “the basis for identifying various entities as being representatives of the same class” without tying them “to any particular instance of that class” (Langacker 1991, 53). In other words, although a type specification, which pertains in general to the semantic content of a simple noun, provides an “initial delimitation among the potential objects of thought” (Langacker 1991, 53), it results in an open-ended class of conforming entities. Although nouns can be further specified in more detail by adjectives, as in white cat, this subtype is not more specific than cat as to the instance referred to. Instantiation therefore relates to the ability of noun phrases to single out particular instances of the specified type. Further, the instantiation of a type is “essential to the structure of a nominal and prerequisite to both quantification and grounding” (1991, 55).

Description of the framework | 25

Quantification relates to the additional information which is provided as to number, size or amount of the designated instance by items such as many, most, several, etc. In virtue of the indication of quantity, “quantifiers subsume instantiation; the very fact that the speaker singles out an instance of a certain magnitude entails that the instance has become the focus of the speaker’s conscious awareness” (Taylor 2002, 355). Perhaps more important than quantification is the location or grounding of the designated instance from the perspective of the speech event and its participants (also called ground). Grounding thus relates to the difference between definiteness and (specific or unspecific) indefiniteness. In other words, via the nominal grounding function, the designated instance or set of instances becomes uniquely apparent to both the speaker and the hearer (Langacker 1991, 53). In theory, these four functions are iconically realized in a separate layer of the noun phrase. Moreover, the order of description is chronological: whereas type specification is inherent in instantiation, quantification and grounding presuppose instantiation since they do “not pertain to an unanchored type conception but rather to instances of a type” (Langacker 1991, 73). Likewise, quantification is prerequisite to grounding, as the speech act participants generally do not make mental contact with every single instance (in the case of a set of instances) but with a set of instances as a whole. Taylor (2002, 344) visualizes the ‘layered’ conceptual structure as follows: (Grounding (Quantification (Instantiation (Specification (Type) ) ) ) This structure sanctions noun phrases such as these three cats. Crucially, the analysis of noun phrases is not always as straightforward as the model suggests, since the interaction of quantification with grounding and instantiation is especially complex (Taylor 2002; see also Davidse 2004). Grounding may for instance be realized by quantifiers as well as by determiners (which are typically associated with the grounding function), as in I saw three cats. As to the Spanish BQ-constructions, I will show that they can develop quantifier uses, as in Hay un montón de gente ‘There are a lot of people’, where un montón de combines the nominal grounding and quantification. In addition, they can combine type specification, quantification and grounding, as in Recibió un alud de llamadas ‘He received a flood of phone calls’, where un alud de does not only provide information as to the extraordinary number of calls, but also as to their violent, sudden and overwhelming nature.

26 | Framework, state of the art and methodology

2.1.2.2 Quantification and quantifiers in noun phrases A pivotal claim in Langacker’s approach “that every nominal profiles a single instance of some type, which is generally provided by the head noun together with its number specification. The role of a quantifier is therefore not to specify the number of instances (which is always just one) but rather to indicate the size of the profiled instance” (1991, 81). In general, languages provide two devices to indicate quantity, viz. (morphological) number and quantifiers. Since quantifiers “comprise a fairly heterogeneous set of items” (Taylor 2002, 355), a more fine-grained distinction is generally made between absolute and relative quantifiers. Absolute quantifiers are considered to merely specify the size or number of the profiled instance (e.g. three, much, many, several), regardless of the total number of possible instances. Canonical examples are of course numerals (Taylor 2002, 355). Absolute quantifiers are by default indefinite, as in I’d like to have two cats. This principle can however be easily overridden by specific contextual construals, as in I adopted two cats, or by the addition of definite determiners, as in The two cats were playing in my living room. Relative quantifiers designate the size or number of the profiled instance by making implicit reference to a ‘reference mass’ (e.g. all, no, every, each). By way of comparison, three cats gives an exact number while most cats “implies a majority with respect to a larger set” (Taylor 2002, 356). A typical example are partitive constructions, which make explicit reference to (and usually stress) the reference mass, as in most of the cats. In addition to the description in terms of semantic properties, Langacker provides structural evidence for the distinction between both types of quantifier. Since relative quantifiers are by definition grounding items (they evoke a reference mass), they cannot combine with additional quantifiers nor with grounding elements (e.g. *the some dogs, *that every dog). Absolute quantifiers, however, which are by default indefinite, occur regularly with demonstratives, definite articles and even with relative quantifiers (e.g. those three dogs, these few statesmen, any three ballerinas). Further, relative quantifiers cannot occur as the lexical head of the clause (*The politicians who sacrifice their principles for the sake of election are all/most/some/any/every), whereas absolute quantifiers can (e.g The problems we have to deal with are three). Whereas Langacker’s model suggests that only relative quantifiers subsume a grounding function, Taylor (2002) and Davidse (2004) focus on the subtle interplay between grounding and quantification. Taylor argues that “the very fact that the speaker singles out an instance of a certain magnitude entails that the instance has become the focus of the speaker’s conscious awareness” (2002, 355), regardless of the absolute or relative nature of the quantifying element. A

Description of the framework | 27

similar observation has been made by Davidse (2004) who states that “in all primary determiners basic conceptual mechanisms of quantification and identification are inextricably linked” (Davidse 2004, 507). She argues that “the comparison relation between P [KV: profiled mass] and RT [KV: reference mass] which is measured by relative quantifiers is also present in definite identification as the identified part P of RT” (Davidse 2004 , 512). Analogously, “the correspondence relation between T [KV: type specification] and Ti [KV: instantiation of T profiled by the nominal] presupposed by absolute quantification constitutes the central cognitive operation of indefinite identification” (Davidse 2004, 512). The special status of BQs is usually left out of consideration. On the one hand, no additional quantifiers nor determiners can be added to N1 or N2 in (1), and no reference is made to the set of all possible calls, which suggests that BQs are absolute quantifiers. On the other hand, BQs structurally resemble partitive constructions and to some extent evoke a reference image: the number of calls is metaphorically mapped on the conceptual image evoked by alud. (1)

Recibió un alud de llamadas. ‘He received an avalanche of calls.’ Recibió tres aludes de llamadas. ‘?three avalanches of calls.’ *Recibió este un alud de llamadas. ‘*this an avalanche of calls.’ *Recibió un alud de las llamadas. ‘*an avalanche of the calls.’ *Recibió un alud de muchas/quince llamadas. ‘*an avalanche of many/ fifteen calls.’ ?

Langacker (1991) classifies the BQ-construction among the more complex expressions of absolute quantification, similar to an expression such as three of the tables where he considers the numeral to function as the nominal head modified by the PNP. Further, he observes that in expressions like a bunch of carrots, a glass of water, a lot of sharks the “nouns which appear as heads constitute a diverse and open-ended class” (1991, 88). Whereas some of these nouns still have a reading as the container or constitution for some portion of a mass (e.g. a glass of water), most such nouns have developed a quantifying interpretation: we can for instance drink a glass of water. He leaves open the question whether the noun in the of-phrase has to be considered the head of the construction.

2.1.2.3 A cognitive-functional model of binominal quantifiers In two more recent papers, Langacker analyzes the functional organization and the meaning of BQs – of a lot of in particular (Langacker 2011) – as well as the pathway of change which leads from lot as a lexical item to a lot of as a typical quantifier competing strongly with much/many (Langacker 2009a). He starts from the cognitive-constructional assumption that

28 | Framework, state of the art and methodology

one needs to examine the meaning of each component element and relate this to the meaning it displays in other uses. One also needs to consider the component’s grammatical composition in relation to more general patterns, as well as the composite meaning that arises through conceptual integration at each successive level of organization. (2011, 3)

In accordance with this guideline, he first considers the semantic contribution of the preposition of. Instead of evoking a thing, the preposition profiles an intrinsic relationship between two entities (its trajector (tr) and its landmark (lm))14 and has to be followed by a mass noun or a plural noun. As to BQs, two types of intrinsic relationship are distinguished: (1) the trajector can constitute a restricted subpart of the landmark (e.g. a flock of those sheep), as in partitive constructions (Figure 5a); (2) the trajector and the landmark can be coextensive, as in a cup of tea (which is the limiting case of a restricted subpart, Figure 5b). According to Langacker, pseudo-partitive constructions such as a slice of cake are ambiguous (cf. Figure 5c). He further considers the preposition to be the profile determinant: its profile corresponds to the composite structure profile of the prepositional phrase, while the nominal profile corresponds to the preposition’s landmark.15 Therefore, at the first level of conceptual integration “of is the constructional head because the preposition and the composite expression both profile non-processual relationships” (Langacker 2011, 5). (a) restricted subpart

(b) coextension

(c) N of X

 Fig. 5: The semantic contribution of of according to Langacker (2009a, 67)16

|| 14 In Langacker’s model, the construal of a relationship evokes a trajectory/landmark alignment, since varying degrees of prominence are conferred on the participants (in the relationship). “The most prominent participant, called the trajector (tr), is the entity construed as being located, evaluated or described. Impressionistically, it can be characterized as the primary focus within the profiled relationship. Often some other participant is made prominent as a secondary focus. If so, this is called a landmark (lm). Expressions can have the same content, and profile the same relationship, but differ in meaning because they make different choices of trajectory and landmark” (2008, 70), such as the prepositions above and below. 15 Note that the cognitive account attaches greater importance to semantic than to morphosyntactic features in the determination of the nucleus or head of constructions (see Section 2.2.4). 16 According to Langacker’s conventions, a circle stands for a bounded entity, an ellipse represents a mass, heavy lines indicate profiling, the arrow in the subpart-diagram represents

Description of the framework | 29

As far as Spanish is concerned, I will claim that the coextensive relationship between N1 and N2 is preconditional to the quantifier and specifier uses of BQconstructions. Likewise, Langacker argues that the coextension between N1 and N2 in a flock of geese allows for two alternate construals which are metonymically related: If I am looking at a flock of geese, am I looking at the flock or at the geese? Since the two are coextensive in the world, in a sense the point is moot. Still, I can certainly focus primarily on either the unitary or the collective, mass-like aspect of this complex entity. (Langacker 2011, 7)

The coextension relation can be realized in four different ways according to the extent to which the QN exists independently of the N2. If the QN evokes a container, it represents a physical object with a particular shape that is distinct from its content and fulfills a containing function (e.g. box, barrel). Configurations (e.g. pile, stack) and groups (e.g. flock, collection), “may be physical, but they are not distinct from their constitutive entities – they are coextensive with them in the strongest sense. A unit of measure per se is non-physical, consisting in extension along a scale; it is physically manifested only as an aspect of what it measures” (Langacker 2009a, 64, see Figure 6). (a)

(b)

Container

Configuration Group

(c)

Measure

Fig. 6: Alternate degree of coextensiveness according to Langacker (2009a, 65)

In addition, a single QN can have multiple interpretations, according to the presence of the physical container and physical substance, which are in fact matters of degree. The size noun barrel for instance, can be interpreted as the physical container of some mass (e.g. he rolled a barrel of oil down the ramp). The same QN can also be used as a unit of measurement (e.g. we burned a barrel

|| the relationship that a subpart bears to a whole, the double line in the coextension-diagram indicates that these two entities are coextensive.

30 | Framework, state of the art and methodology

of oil that had been sitting in the basement), in which case it does no longer profile a physical object but “rather an abstract entity which is still however identified in terms of a physical barrel” (Langacker 2011, 10). A third way to interpret barrel concerns the cases where barrel simply designates “a certain extension along the scale, with no thought of containers” (Langacker 2011, 10; e.g. A barrel of oil is leaked from the storage tank). Finally, barrel can also be employed as a purely abstract entity (e.g. A barrel of oil is 42 gallons). The four types of coextensive realizations reflect at the same time the evolutionary path followed by nouns that develop measure noun uses: a QN that originally evokes containers, configurations or groups can shift into an abstract unit of quantity via a metonymical extension. As to the example a lot of, Langacker argues that the original count noun lot has taken on a more general and abstract sense. In comparison to other (less grammaticized and less schematic) QNs such as bunch, lot has become very abstract “by virtue of no longer evoking a physical container” and schematic “because it does not pertain to any particular dimension of measurement” (Langacker 2011, 11), to the extent that it can be used quasi-adverbially, whereas other QNs cannot (e.g. It weighs a lot/*a bunch). Moreover, lot makes no distinction between plural and non-plural masses (cf. much vs. many, a bunch of idiots vs. *a bunch of water) and specifies quantity in the vaguest terms. Parallel with quantifiers such as little and few, lot is non-metric, that is, “the extent they profile is characterized only as falling on one side or the other of some norm or neutral value, not at any particular point on a scale” (Langacker 2011, 12). As diagrammed in Figure 7, lot evokes a norm and upward scanning from that point. As to the way the designated quantity is mentally accessed, lot invokes a norm and a scanning away from that neutral value. This conception can only be realized once as it would be “conceptually incoherent to carry out the same scanning operation once more starting from where the original scan ended – once past the norm, one cannot go past the norm again by scanning in the same direction” (Langacker 2011, 13). In the case of bunch, by contrast, the idea of building up to the total quantity in positive increments is still present as a faint vestige of the original cluster-meaning. In other words, while it sounds perfectly fine to eat more than one bunch of cookies, it is not possible to eat more than one lot of cookies. Therefore, bunch can be combined with another, whereas lot does not.17

|| 17 Langacker seems to overlook the structural possibility of eating lots of cookies, viz. the quantifying potential of the plural lots. In Chapter 8, I will illustrate the semantic contribution of the plural morpheme.

Description of the framework | 31

(a) a lot

(b) little

(c) few

 Fig. 7: Upward and downward scanning according to Langacker (2011, 12)

The constructional slot left unspoken until now is the indefinite article a. Since the expression a lot of is well on its way to being reanalyzed as monomorphemic alotta, one could start to doubt its structural necessity and its conceptual contribution. Since lot is considered an abstract entity analogous to numbers, it constitutes a unique instance of its type. Yet the function of the indefinite article precisely resides in indicating that some instance of a type has to be selected or introduced. However, the indefinite article also shows up in other core members of the quantifier category (a thousand, a few, a little). In comparing expressions such as a few and a little to their article-less alternatives, the contribution of the indefinite articles becomes clear: the presence of a induces positive polarity and specifies “a quantity through upward scanning from the scalar origin, i.e. as a positive increment starting from the baseline of zero” (Langacker 2009a, 76). However, as far as Spanish BQs are concerned, I will argue in this book (cf. Chapter 8) that, in spite of the overall preference for the indefinite determiner un(a), QNs which present a rich source semantics combine with a certain degree of variation in the determiner slot in accordance with the thematic or rhematic nature of the QN in the precise discourse context. After discussing the individual constructional elements, Langacker turns to their integration at the level of the composite structure, as visualized in Figure 8. At the intermediate level of organization, the grounded entity profiled by the article and the quantified entity profiled by lot are mapped onto each other.18 Likewise, the mass profiled by sheep is mapped onto the landmark of of and as a whole they profile a coextensive relationship. At the composite structure level, both subparts are “integrated by a correspondence which identifies the mass that lot serves to quantify with the trajector of the prepositional noun phrase” (Langacker 2011, 15). The composite structure as diagrammed in Figure 8 can be

|| 18 G stands for ‘ground’.

32 | Framework, state of the art and methodology

maintained as long as a lot of sheep is analyzable. From the moment alotta is analysed as a single whole, the intermediate level of organization presents an even higher degree of conceptual integration: the meanings of a and of are “wholly subsumed in the meaning of lot, so nothing is lost if it disappears or is no longer recognized as a separate element” (Langacker 2009a, 13) and the preposition’s landmark is mapped onto the entity quantified by lot. a lot of sheep

a lot

of sheep

a

lot

of

sheep

Fig. 8: Composite structure profile according to Langacker (2009a, 73)

By way of conclusion, Langacker zooms in on the delicate matter of the composite structure profile, i.e. which of the two nouns is regarded the head of the construction. In a flock of sheep, the schematic mass delimited by flock is generally put in correspondence with the trajector of the of-phrase and the composite structure profiles a flock consisting of sheep. However, in its coextensivereading, the mass profiled by sheep can come to constitute the expression’s profile via a metonymic shift. This metonymic shift is even obligatory in the case of a lot of sheep, where lot has no independent existence in the physical world and cannot function as the lexical head. Consequently, the profile of the com-

State of the art | 33

posite structure of a lot of sheep does not correspond to the profile of the component structure a lot nor to that of of sheep,19 and none of the nouns can be considered the head of the construction. Therefore the composite construction a lot of sheep is called ‘exocentric’, i.e. it has no head noun so defined (Langacker 2011, 17).

2.2 State of the art The categories quantifying noun, binominal quantifier and pseudopartitive construction pass largely unnoticed in the mainstream Spanish grammar manuals. If examples of BQ-constructions show up at all, they do so in two sections: (i) in the sections on verb agreement of collective nouns (the so-called agreement ad sensum), it is generally agreed upon that the addition of a prepositional phrase which indicates the constituting entities of the collective nouns motivates plural verb agreement; (ii) in the sections on nominal complementation or prepositional phrases headed by de, examples of the BQ-construction figure in the category of prepositional phrases which express quantity assessment.20 BQs as such are not considered as a noteworthy category, however. One exception to this general tendency is the descriptive grammar Gramática descriptiva de la lengua española (henceforth GDLE) directed by Bosque/Demonte (1999) and the renewed version of the reference grammar Nueva gramática de la lengua española of the Real Academia Española (2009, henceforth NGLE) in particular. The account of QNs and the BQ-construction in these grammars will be outlined in Section 2.2.1. The two following sections give an overview of the few synchronic (Section 2.2.2) and diachronic case-studies (Section 2.2.3) performed until now. Sections 2.2.4 and 2.2.5 briefly introduce two continuing problems related to both Spanish and other languages, viz. the identification of the functional (and/or semantic) nucleus of the binominal syntagm on the one hand and the fuzzy boundaries between partitive, pseudopartitive and appositional constructions on the other.

|| 19 As of sheep profiles a relationship rather than a thing, its profile does not correspond to the profile of the composite expression a lot of sheep which profiles a mass (and thus a thing). 20 E.g. Alarcos LLorach (1999; 1995); Alcina Franch/Blecua (2001); Alonso/Henríquez Ureña (1938); De Bruyne (1979); RAE (1973).

34 | Framework, state of the art and methodology

2.2.1 Description of QNs in GDLE (Bosque 1999) and in NGLE (RAE 2009) Both the descriptive grammar directed by Bosque/Demonte (1999) and the new reference grammar of the Real Academia Española include a substantial section on the so-called sustantivos cuantificativos or quantifying nouns and proclaim their status as a grammatical category as such.21 The basic threefold distinction of QNs is outlined in the GDLE. In the NGLE, additional interesting insights are put forward as to the semantic characteristics of the category and QNs are compared to qualifying nouns (sustantivos clasificativos). Three types of QNs are to be distinguished, viz. bounding nouns (sustantivos acotadores), measure nouns (nombres de medida) and group nouns (sustantivos cuantificativos de grupo). The bounding nouns denote a portion of a mass (e.g. un pedazo de papel ‘a piece of paper’, una loncha de queso ‘a slice of cheese’). By default, they combine with singular count nouns and mass nouns. In some cases, both N2 by itself and the binominal construction refer to the same entity: while un pedazo de papel is still un papel, un pedazo de queso ‘a piece of cheese’ is not necessarily the same as un queso. Nouns such as pedazo ‘piece’, trozo ‘piece, slice’ and porción ‘portion’ are considered ‘wild cards’ in that their meaning is adjusted to the bounded entity (N2): un pedazo de papel manifests other dimensions than un pedazo de melón ‘a piece of melon’. The majority of the bounding nouns are selected in function of the consistency and the shape of the bounded mass, however, and some are even restricted to a specific (set of) N2(s) (e.g. un terrón de azúcar/tierra ‘a lump of sugar/earth’). Bosque (GDLE 1999, 20) also indicates that several bounding nouns can be ambiguous between the interpretation as physical object or the one as a quantifying expression, as in un vaso de whisky ‘a glass of whisky’ that can be broken or be drunk). Measure nouns designate measures established in function of the physical characteristics of the measured entity (e.g. volume, weight, height, extension, etc.). Unlike the bounding nouns, measure nouns cannot represent a physical object and they admit plural complements (un kilo de garbanzos ‘a kilo of chickpeas’ versus *una rebanada de panes ‘*a slice of breads’) in addition to singular mass nouns.

|| 21 Quantifying nouns are traditionally considered a subgroup of collective nouns. While a collective designates a group of entities (e.g. orquesta ‘orchestra’, familia ‘family’), QNs can make quantity judgments of a separate group and allow to identify this group as a specific class or type of group (e.g. un montón de regalos ‘a heap of gifts’) (RAE 2009, 823).

State of the art | 35

Group nouns refer to the diverse ways of forming collections or groups. Their quantifying potential inheres in the ambiguity between the interpretation as a collective or as a QN. While ejército ‘army’ as a collective refers to a group of soldiers, it shifts to a quantifying interpretation in un ejército de curiosos ‘an army of onlookers’, which in no way refers to a troupe of curious soldiers. In other words, group nouns need to be complemented, by default by plural count nouns, whereas collectives can stand alone (e.g. Conoce su vecindario ‘He knows his neighbourhood’ vs. *Son un hatajo ‘*They are a bunch’). As is the case for bounding nouns, several group nouns which can stand alone as regular nouns, can give rise to ambiguous expressions. For instance, in una manada de cerdos (‘a herd of pigs’) both the reading as a collective (una manada) followed by a specifying post-modifier (de cerdos) and the reading as a premodifier (una manada de) preceding a head-noun, are possible. The GDLE briefly touches upon the matter of “semantic selection which is sometimes highly constrained)” (Bosque 1999, 25, translation KV):22 the objects quantified by tropel ‘mob’ have to be disorganized or chaotic, the entities quantified by rosario ‘rosary’ form part of a continuous sequence of inopportune, irritating and inappropriate entities, the objects that combine with alud show up in a tumultuous way and hatajo evokes a highly negative and disdainful evaluation of human entities. The GDLE concludes by the observation that the N2s quantified by group nouns function as the head of the construction and trigger verb agreement (as long as the determiner to N1 is indefinite). The NGLE adds various renewed insights and follows a slightly different approach. The most striking difference is that QNs are described in comparison to qualifying (or type) nouns. The threefold distinction postulated in the GDLE is preserved. The NGLE further specifies that lexical items do not necessarily pertain to one specific class of QNs, yet can be construed as such. Therefore several QNs show up in more than one category (e.g. una porción de manteca ‘a portion of fat’ versus una porción de sucesos ‘a bunch of events’) and several QNs have not-quantifying uses outside the pseudopartitive construction (e.g. alud ‘avalanche’, copa ‘glass’). Further, some bounding and group nouns can have more general quantifying uses as in Nos reímos una barbaridad ‘We || 22 In the discussion of both bounding nouns and group nouns, the GDLE refers to the similarity with ‘classifiers’ or ‘classifying morphemes’ (Bosque 1999, 21–26; see also Langacker 2008 on this topic) in many languages. These morphemes affect not only the gender and number (or person) of the objects indicated by the declined noun, but also their shape, consistency, size and visibility. The fact that quantity and quality assessment can be expressed by a single category in unrelated languages challenges, in my view, the strict differentiation between size and type specification. I will come back on this issue in Chapter 6.

36 | Framework, state of the art and methodology

laughed a lot’ or Es muy ágil, es una barbaridad de ágil ‘He is very agile, he is awfully agile’. Although QNs generally show up in the pseudopartitive construction (which means that they are headed by an indefinite determiner and followed by de introducing a noun without proper determiner), they present a “defective syntax” (RAE 2009, 826) for almost exclusively combining with adjectives referring to size, volume or number. Before turning to the semantic restrictions of QNs, the NGLE zooms in on classifying nouns (sustantivos clasificativos) such as clase ‘class’, especie ‘sort’, suerte ‘sort’, tipo ‘type’ or variedad ‘variety’. Four major points of convergence are mentioned. First, classifying nouns also show up in the pseudopartitive construction. Second, they behave similarly as to verb agreement. Further, the first noun functions as a kind of predicate to the second noun, which is considered the head of the syntagm: while decir un montón de disparates ‘to tell a lot of nonsense’ corresponds to decir disparates, pasar un tipo de examen ‘to pass a kind of exam’ implies pasar un examen. Finally, classifying nouns also result in indefinite syntagms. However, classifying and quantifying nouns diverge as to their interpretation. QNs designate not only the way in which the N2-entities are grouped, but also the magnitude of the portion measured. Classifying nouns bound or restrict the type of N2 they introduce, yet in a vague way: they do not specify the nature of N2. In the description of the lexical and semantic aspects of QNs, the NGLE maintains the threefold distinction. The majority of the refined insights concern the group nouns. Two tendencies are to be distinguished, viz. group nouns which do not impose lexical restrictions on N2 and group nouns which present semantic combinatorial restrictions. Group nouns such as cantidad ‘quantity’, conjunto ‘collection’, grupo ‘group’, infinidad ‘infinity’, mogollón ‘lot, mass’, montón ‘heap’, serie ‘series’, sinfín ‘a great many (lit. without-end)’, sinnúmero ‘a great many (lit. without-number)’, etc. pertain to the former class. They allow both singular and plural verb agreement and some can occur without determiner. The latter class of group nouns is quite extensive, in particular the series of group nouns quantifying animals (e.g. banco (de peces) ‘a shoal (of fish)’, un enjambre (de abejas) ‘a swarm (of bees)’). Further, a number of types of group nouns are differentiated for (i) referring to entities which are linked together (e.g. cadena ‘chain’, retahíla ‘string’, serie ‘series’), (ii) profiling a group of entities which suddenly show up in a tumultuous and impetuous way (e.g. alud ‘snowslide’, aluvión ‘flood’, avalancha ‘avalanche’), (iii) designating a group of mixed, disorganized, interwoven and tangled entities (e.g. enredijo ‘jumble’, batiburrillo ‘mishmash, jumble’, revoltijo ‘mess’), (iv) quantifying bundled or fastened entities (e.g. atado ‘bundle’, hato ‘bundle’, haz ‘bundle’), (v) quantify-

State of the art | 37

ing highly numerous N2s (e.g. barbaridad ‘barbarity’, montón ‘heap’, pila ‘pile’) and finally, (vi) evaluating the N2-entities either positively or negatively (e.g. plantel ‘team’ and hatajo ‘bunch’). To conclude, the NGLE observes that the use as group nouns of nouns which are not-quantifying strictly speaking involves a process of grammaticalization in which a particular facet to the original meaning is retained (e.g. the impetus in the case of alud).

2.2.2 Synchronic case-studies Over the last ten years, the topic of Spanish QNs increasingly gained in interest and has been the focus of at least four case-studies, viz. Vos (2002), Bosque (2007), Di Tullio/Kornfeld (2008) and Gutiérrez Rodríguez (2008). Vos (2002) provides a formal and comparative analysis of Dutch and Spanish QNs and establishes a classification of QNs. In function of the lexical23 or functional nature of the first noun, she distinguishes between quantifying nouns (cf. “nombre cuantificador”; e.g. un montón de empleados ‘a heap of employees’), measure nouns (cf. “nombre de medida”; e.g. un litro de vino ‘a litre of wine’), container nouns (cf. “nombre de contenedor”; e.g. un cubo de zarzamoras ‘a bucket of blackberries’), collective nouns (cf. “nombre colectivo”; e.g. un grupo de estudiantes ‘a group of students’), partitive nouns (cf. “nombre de parte”; e.g. un pedazo de pan ‘a piece of bread’) and class nouns (cf. “nombre de clase”; e.g. un tipo de filtros ‘a type of filters’). The main difference between both languages resides in the absence of a linking element in Dutch and the necessary presence of the preposition de in Spanish, which is not considered to have any semantic content in this study (2002, 54). As to the crosslinguistic similarities, the author mentions combinatorial restrictions on N2, the possibly ambiguous interpretation of N1 (literal vs. functional), the impossibility to add proper quantifiers to N2, the general reluctance to add proper determiners to N2 and the possibility to determine the status of N1 by omission or anaphoricity tests. According to her, the classification of QNs is the crucial (and only) factor in determining verb agreement: while quantifying nouns never combine with singular verb agreement, the set of measure, class and collective nouns is compatible with singular verbs and both container and partitive nouns always yield singular verb agreement.

|| 23 By lexical use (cf. “nombre lexical”, Vos 2002, 50), the author refers to nouns which have proper referential value, i.e. which refer to a physical or tangible object in the extra-linguistic world.

38 | Framework, state of the art and methodology

The second case-study is offered by Bosque’s (2007) analysis of the extension of (functional) paradigms via processes of abstraction of lexical items. In addition to the description of the relatively closed paradigm of pseudocopular verbs in Spanish, the author dwells on the more open paradigm of quantifying nouns in pseudopartitive constructions. The starting assumption of this article – which corresponds to the working hypothesis of the present book – is that “the quantifying properties of QNs are obtained via the abstraction (…) of some characteristics present in the literal interpretation” (Bosque 2007, 192, translation KV). By way of illustration, the quantifying interpretation of mar ‘sea’ originates in the facets ‘amplitude’ and ‘unlimited extension’ associated with its literal meaning. Perhaps more important is the claim that new QNs are obtained via a process of grammaticalization and that the grammaticalized QNs maintain a certain relation with the original literal meaning. Further, the author alludes to the tension between lexicalization and grammaticalization, since some new formations are more or less fixed expressions (e.g. mar de dudas ‘sea of doubts’) while other QNs combine with a wide range of N2-collocates, “provided that the abstract notion which is brought about by the QN can be applied” (Bosque 2007, 193). Equally interesting is the claim that the figurative uses of QNs are not necessarily associated with poetic language (Bosque 2007, 194). For instance, within the series of QNs that profile the sudden arrival or impetuous and tumultuous nature of the mass, viz. avalancha ‘avalanche’, alud ‘avalanche’, aluvión ‘flood’, cascada ‘cascade’, oleada ‘wave’ and tromba ‘tornado’, the author zooms in on alud. In its quantifying use, this QN has lost any reference to snow and foregrounds the force and quantity of the N2-entities. These facets are backgrounded in the literal reading of un alud de nieve ‘a snowslide’ and are not even acknowledged by reference dictionaries. As they are immanent in the literal readings, they are “only to be expected” though (2008, 195): in every-day language, there is nothing problematic about the figurative use of QNs. Di Tullio/Kornfeld (2008) analyze the grammaticalized nature of QNs in colloquial Riverplate Spanish. They start from a three-fold differentiation between QNs depending on the function of N1, viz. (a) expressions which by definition evoke the plurality of N2 (e.g. una cantidad de reclamos ‘a quantity of complaints’), (b) expressions which combine quantity assessment with quality evaluation (e.g. una barbaridad de reclamos ‘a barbarity of complaints’) and (c) expressions where the evaluative interpretation relates to quantity only (e.g. una oleada de robos ‘a calamity of thefts’). The majority of the QNs in Riverplate Spanish pertain to the first group (a) whose quantifying interpretation is intrinsically linked to the expression (whether or not via metonymical and metaphorical extensions). As such, QNs designate either containers (e.g. vagón ‘wagon’),

State of the art | 39

entities which consists of relatively little subparts (e.g. rosario ‘rosary’), more or less organized configurations of objects (e.g. pila ‘pile’), collectives (e.g. banda ‘gang’), a quality (e.g. chorrada ‘nonsense, twaddle’) or entities which evoke some kind of sequence or succession (e.g. bombardeo ‘bombardment’, tsunami ‘tsunami’). In some cases, the process of metaphorization is hard to trace however, as in punta ‘tip, end’ or bocha ‘head’. Interestingly, like Bosque (2007), Di Tullio/Kornfeld (2008) also evoke the lexicalized nature of some QNs which are characterized by strict combinatorial restrictions as to N2, without going on to provide underlying semantic or functional motivations for these restrictions or the lack of combinatorial restrictions for other QNs. Further, the study seeks out to determine the degree of grammaticalization of QNs. They distinguish between three levels of GR according to the QN’s behavior in a set of five morphosyntactic tests: (i) the possibility of N1 to combine with premodifying adjectives, (ii) verb agreement with N1 or N2, (iii) anaphoricity tests, (iv) the syntactic scope24 of the QN and (v) the possibility to occur without N1-determiner and for N1 to pluralize. It goes without saying that a higher degree of grammaticalization corresponds to a lower degree of morphosyntactic freedom. The study of Gutiérrez Rodríguez (2008) actually focuses on the grammatical aspects of weak quantifiers (cuantificadores débiles), also called indefinite quantifiers. An important criterion to distinguish strong from weak quantifiers is the latter’s compatibility with impersonal and existential haber (as in hay muchos niños en el jardín ‘there are a lot of children in the garden’). The final chapter of her PhD is dedicated to pseudopartitive constructions. She partially builds on Bosque’s (1999) classification of QNs (see Section 2.2.1) yet adds a fourth type, viz. ‘lexicalized quantifying nouns’ (nombres cuantificativos lexicalizados). This additional category comprises nouns which either have completely lost their etymological meaning and only express quantity, or exclusively present a quantifying reading from the start. The former type of lexicalized quantity notions thus encompasses nouns such as un montón ‘a heap’, la mar ‘the sea’, la tira ‘(lit.) the strip’ or un sinfín ‘lit. an endlessness’, whereas

|| 24 Some QNs can function by themselves as a predicate (a), the object of the clause (b) or as degree modifier (c). (a) Es *una manga. ‘*It is a multitude.’ / Es una chorrada. ‘It is a bunch of crap.’ / Es (una) bocha. ‘It is a whole lot.’ (b) *Tuvo una manga. ‘*He had a multitude.’ / Cuesta un vagón. ‘It costs a lot.’ / Cuesta (una) bocha. ‘It is extremely expensive.’ (c) *Me gustó una manga. ‘*I liked it a multitude.’ / Me gustó un vagón. ‘I liked it a lot.’ / Me gustó (una) bocha. ‘I liked it a whole lot.’

40 | Framework, state of the art and methodology

quantity expressions such as (gran) cantidad ‘(large) quantity’, gran número ‘large number’ or infinitude ‘infinity’ have a quantity-related interpretation by definition (2008, 320–321). Note that ‘lexicalized’ refers to the conventionalized character of the quantifying interpretation – as in ‘part of the lexicon’ – and not as opposed to the preceding GR process she considers them to have been involved in. The distinction of the fourth category is necessary in view of the diverging morphosyntactic behavior in comparison to the other types of QNs. She repeatedly emphasizes, however, that the functional organization of metaphorically used group nouns and lexicalized quantifying nouns largely overlap. Further, she proposes to distinguish between two types of underlying syntactic structures for pseudopartitive constructions according to the ‘literal (or collective) interpretation’ and the ‘quantifying interpretation’. In the literal interpretation, the binominal syntagm contains two lexical heads; in the quantifying interpretation, the binominal contains one lexical noun only (N2) as well as one functional noun (N1) which in combination with its determiners forms a complex quantifier to N2. In both cases, the preposition marks the genitive case and indicates that N2 is a complement of N1 (whereas in partitive constructions the preposition marks the partitive case). To my knowledge, Spanish linguistic literature does not present detailed analyses of the semantic potential of (individual) QNs, nor of the differences in morphosyntactic make-up of the BQ. No underlying pragmatic motivations have been put forward for using QNs at the expense of more ‘regular’ quantifiers such as mucho/a(s) ‘many, much’.

2.2.3 Diachronic case-studies As far as Spanish is concerned, no detailed diachronic analysis of the BQ has been carried out. In the recent historical syntax edited by Company Company (2009), neither the topic of QNs nor the development of (pseudo)partitive constructions has retained the attention of the authors, which is striking in that the volumes contain at least five chapters on closely related topics.25 Two detailed case-studies focus on closely related topics, viz. on the evolution from Latin object and subject genitive to prepositional phrases to deverbal nouns in Spanish (Rodríguez-Espiñeira 2011) and the grammaticalization of Latin genitive in

|| 25 The volumes include chapters on the evolution of the Latin genitive case to the prepositional phrase introduced by de, on the general structure of noun phrases, on (pure) quantifiers, on interrogative and exclamative quantifiers and on nominal complementation.

State of the art | 41

Romance languages and French in particular (Carlier et al. 2013). Both papers start from the well-known assumption that “there was never a general replacement of the genitive, but rather a replacement of specific uses of the genitive by specific prepositions or syntactic functions” (Carlier et al. 2013, 141–142). In the posthumously published Estudios de morfosintaxis histórica del español (2000) of Rafael Lapesa, two chapters touch upon the evolution from Latin appositional genitive to the expressive binominal construction in Spanish. Although the expressive binominal construction, also called ‘emphatic apposition’ (e.g. una maravilla de niña ‘a wonder of a girl’) is not the main concern of the present book, I will briefly outline Lapesa’s findings which corroborate my intuitions on the fuzzy borderlines between BQs and expressive binominals. In the chapter on the evolution of the Latin cases into the Spanish substitutes, Lapesa (2000, 82) argues that the appositional genitive in Latin as in virtus iustitiae ‘the virtue of (the) justice’ corresponds to the prepositional phrase headed by de in Romance and Spanish (e.g. la virtud de la justicia). He dwells on a remarkable variant of the appositional genitive, la atestiguada desde Plauto hasta Cicerón en ejemplos de carácter familiar y afectivo como scelus viri, monstrum mulieris, deliciae pueri, (….). Lo peculiar de este giro consiste en poner de relieve una cualidad o rasgo presentándolos como esencia de su poseedor. Dada su fuerza expresiva, no es de extrañar que construcciones de este tipo hayan perdurado en las lenguas románicas con la habitual sustitución del genitivo por de (…). (Lapesa 2002, 84)26

While in Medieval Spanish documents no attestations have been found of this construction, this is not a sufficient reason to refute the hypothesis. Indeed, on the one hand, similar examples have been observed for Ancient French. On the other hand, given its informal and affective nature, the construction is not susceptible of occurring in literary texts. Also in Present-Day Spanish, N2 usually designates concrete, tangible entities, even if abstract notions are not excluded. In another chapter, Lapesa further zooms in on the discussion of binominal syntagms which consist of an affective quality or ‘interjection’ in the N1-position and a noun or pronoun in the N2-position headed by de (e.g. el diablo del toro ‘the devil of a (lit. the) bull’, el bueno de Minaya ‘the good (lit. of) Minaya’, ¡Ay

|| 26 [T]he one attested since Plautus until Cicero in examples of familiar and affective nature, as in a wickedness of a man, a monster of a woman, a delicacy of a boy, (…). What is remarkable about this formula is that it underlines a quality or characteristic by presenting it as the essence of its holder. Given its expressive power, it does not come as a surprise that constructions of this type persist in Romance languages with the usual substitution of the genitive case by de (…). (translation KV)

42 | Framework, state of the art and methodology

de mí! ‘Ah (lit. of) me’, ¡pobre de Juan! ‘poor (lit. of) John’, etc.), which according to Lapesa have diverging Latin ancestors (e.g. the combination of the demonstrative ille with nominalized adjectives, the exclamative genitive, the appositional genitive, etc.). As to the appositional construction, Lapesa distinguishes two main types which are the only ‘true’ continuations of the Latin appositional genitive. The first pattern is the one where N2 is a common noun without determiner (e.g. una monada de chica ‘a cuty of a girl’, un desastre de viaje ‘a disaster of a journey’) and which are highly frequent meliorative or pejorative formulas in colloquial Present-Day Spanish. Interestingly, Lapesa suggests that the formula may become partitive when N1 expresses a quantity, as in una enormidad de gente ‘a huge amount of people’, un horror de dinero ‘a horrible lot (lit. horror) of money’. In contrast to other Romance languages, the Spanish construction frequently combines with abstract nouns in the N1-slot and is characterized by the possibility to have a singular N1 and a plural N2 (e.g. una maravilla de trenes ‘a wonder of trains’). The second true inheritor of the appositional genitive is the pattern where N2 is a proper name or common noun with a proper determiner, as in the frequently cited medieval example el asno de Sancho ‘the donkey of Sancho’. Lapesa argues that the construction is particularly frequent in Present-Day Spanish (e.g. esa preciosidad de Juanita ‘that beauty of Juanita’, el zorro del cobrador ‘the cunning of the collector’). I will argue for a strict differentiation of both types of apposition in Section 2.2.5 though.

2.2.4 Identification of the head of the construction The interesting descriptive challenge posed by the BQ-construction is to determine the head or nucleus of the two-constituent structure and to simultaneously define the status of the other element. The NGLE (RAE 2009, 1450) argues that pseudopartitive constructions are instances of the more general type “quantifier [de + noun phrase]”. In spite of the semantic and functional similarity between, on the one hand, un poco de in un poco de pan ‘a little of bread’ and, on the other hand, mucho (pan) ‘a lot of (bread)’ and bastante (pan) ‘enough (bread)’, the constituent boundary falls between N1 and the preposition according to the NGLE. Only the underlying structure [[un poco] [de pan]] can explain the coordination of complements (e.g. un poco de pan y de vino ‘a little of bread and of wine’), the ellipse of the complement (e.g. Solo queda un poco Ø. ‘There is only left a little Ø.’), the modification of N1 (e.g. un poco más de pan ‘a little more of bread’) and the occasional

State of the art | 43

insertion of a parenthetical element between un poco and the prepositional phrase (e.g. Te di lo que traía: nada, un poco quizá de mi antigua soberbia. ‘I gave you what I was bearing: nothing, a little may be of my old pride.’). To my knowledge, except for the generative proposal of Guitérrez Rodríguez’ (2008, cf. Section 2.2.2), no fine-grained analysis of headedness within binominal constructions has been worked out for Spanish. It bears pointing out, however, that both the functional and the cognitive model hold strong views on the status of both nominal elements within binominal constructions. Further, some thought-provoking suggestions are presented by case-studies on binominals in other languages. The strictly functional model proposed by Halliday/Matthiessen (2004) associates the noun phrase primarily with the ideational component of the linguistic system, viz. the functional component which is concerned with representational semantics or the expression of content (Halliday/Matthiessen 2004, 29). This “metafunction” is further divided into two components, viz. the experiential and the logical one. The experiential structure of the nominal group refers to its double function of “specifying (i) a class of things, namely trains, and (ii) some category membership within this class” (Halliday/Matthiessen 2004, 312), illustrated by the example those two splendid old electric trains with pantographs. The categorization within a class is generally realized by one or more functional elements within the nominal group, viz. the Deictic element (those), the Numerative element (two), the Epithet (splendid and old) and the Classifier (electric). The class or Thing is expressed by the nominal element and constitutes “the semantic core of the nominal group” (Halliday/Matthiessen 2004, 325). Since the different elements – from Deictic to Thing – fulfill diverging functions with respect to the nominal group as a whole, the experiential structure is considered a multivariate structure. By contrast, the logical structure is univariate in that a single functional relationship, viz. that of subcategorization or modification, is iterated. By way of illustration, when starting from the most general term (the Thing trains) and “[m]oving to the left, we get: (which trains? – ) electric trains; (which electric trains? – ) old electric trains; (which old electric trains? – ) splendid old electric trains and so on” (Halliday/Matthiessen 2004, 329). While the most general term is identified as the Head of the nominal group, the set of remaining and subcategorizing elements is called Modifier (independent of the premodifying or postmodifying position with respect to the head). It goes without saying that, by default, the Head of the logical structure and the Thing of the experiential structure coincide. However, in some functional environments, the Head and Thing are dissociated from one another, especially in measure expressions such as a cup of tea.

44 | Framework, state of the art and methodology

What happens here is that one of the pre-modifying functions is taken on by something that is itself a nominal group, in such a way that the Thing gets embedded in a prepositional phrase with of, which then functions as post-Head Qualifier (…). Of course, the two dimensions of structure are both present throughout; what we are showing here is the way the total meaning is construed by mapping the headhood of cup onto the thinghood of tea. (Halliday/Matthiessen 2004, 332)

The two-level internal structure can be visualized as follows: a EXPERIENTIAL STRUCTURE

Numerative

LOGICAL STRUCTURE

Premodifier

cup

of

Head

Postmodifier

tea Thing

Fig. 9: Internal structure of measure expressions (adapted from Halliday/Matthiesen 2004, 332)

Crucially, the authors specify that when the Head is dissociated from the Thing, it can coincide with any of the premodifying elements: (i) with the Deictic as in some of that chicken, (ii) with the Numerative as in three of those tiles,27 (iii) with the Epithet as in an absolute nightmare of a client and (iv) with the Classifier as in the city of Rome. Measure expressions such as a pack of cards, a slice of cake and a cup of coffee are defined as extended Numeratives, “the Head being a word of measure or type” (Halliday/Matthiessen 2004, 333). In sum, in the functional model, N1 is the core of the logical structure, while N2 is the core of the experiential structure. In other words, N1 is always both the head and the complement, depending on the level of description. This dual analysis results “not only too static, but also inaccurate” (Brems 2007b, 29; 2011, 18) and leaves no room for possible diachronic shifts in the constituency of constructions. In contrast, Langacker’s cognitive model of binominal constructions is more dynamic. In his first assessment of the nominal structure (1991, 88–89), Langacker suggests the possibility of an evolutionary sequence from N1 as the constructional head to N1 as a complex quantifier (as in [[a lot of]QNT sharks]N]NML).28 In a more recent paper (Langacker 2011), the author zooms in on the issue of constituency and evaluates three different analyses, as in (2):

|| 27 Note that loads of money figures in the list of a Head conflated with the Numerative function. 28 QNT stands for quantifier, N for noun, NML for nominal and PP for prepositional phrase.

State of the art | 45

(2)

a. a lot of QNT b. a lot D N c. a lot QNT

geese N of geese PP of geese PP

The first two analyses in (2a) and (2b) correspond to standard assumptions on constituency, “in particular the notion that there has to be a head noun, and that it has to be an immediate constituent of the nominal it heads” (Langacker 2011, 1). In his view, the first analysis is problematic in that it suggests that a lot of is an unanalyzable whole, while he illustrates in his paper that the components are still recognizable and make a proper contribution to the meaning of the construction. The second analysis is problematic in that lot is not the semantic head of the construction: the construction does not profile an instance of lot yet an instance of geese. In Langacker’s view, the major constituent boundary falls between a lot and of geese, since several contexts can be observed where a lot functions as a nominal (e.g. a lot, a whole lot, something of which I have a lot, etc.). Further, he claims that the component structure whose profile corresponds to the composite structure profile can be considered the head of the construction. In his view, of geese profiles the non-processual relationship of coextensiveness between a trajector and the landmark (the landmark being identified with geese). A lot profiles an abstract quantity unit and is identified with the trajector of the preposition. However, according to Langacker (2011) the quantified mass expressed by the preposition consists of geese, but it is not necessarily identified with the geese designated by the prepositional object – it may just be a portion of them (cf. a lot of those geese). The mass profiled by the overall expression is therefore not strictly identical with the entity profiled by any component element. (2011, 17)

Langacker therefore defines the BQ-construction as exocentric, since none of the component elements function as the head of the construction as a whole. Further, headless constructions are not problematic in cognitive grammar: constructions are not required to have a constructional head, as long as the “expressions are fully describable as assemblies of symbolic structures connected by correspondences” (Langacker 2011, 17, cf. Section 2.1.2.3). In addition to the frame-related analyses of headedness, two remarkable suggestions as to constituency of binominal constructions are made by a casestudy on expressive binominal constructions in English (Aarts 1998), respectively a case-study of BQs in French (Benninger 1999). From a rather generative-

46 | Framework, state of the art and methodology

oriented point of view, Aarts (1998) proposes to attribute to a hell of a problem the underlying structure [Det [N1 of a] N2head]. In his view, only this constituency analysis can account for the fact that a premodifier to N1 can scope over N2 (as in another bitchy iceberg of a woman) and that N2 cannot be moved (as in *a monster was delivered of a machine). Further, the entire string [N1 of a] can be considered a complex modifier and replaced by adjectives (as in we saw a hellish movie). Equally intriguing is Benninger’s (1999) proposal to interpret N2 as the head of the construction, mainly because it cannot be omitted. The entire string [determiner + N1 + preposition] is then considered an “inverse prepositional phrase” (1999, 108, translation KV). Within the inverse prepositional phrase, N1 continues to present nominal characteristics (e.g. it can be replaced by a pronoun, presents discourse-motivated determiner variation and combines with adjectives). In her study of the English noun phrases, Keizer (2007) summarizes the set of operational criteria to determine headedness presented in the literature. She distinguishes between semantic and (morpho)syntactic criteria. The former subset of criteria “was originally used to reflect the intuitive idea that within each phrase, one element was somehow more important than the others” (Keizer 2007, 10). Semantic criteria pertain (a) to the distributional equivalence of the head and the composite construction, (b) to the obligatoriness of the head, (c) to the requirement to comply with the selection restriction of the verb (e.g. since to spill requires a liquid subject, the second noun wine is the head in a bottle of wine spilled). The set of (morpho)syntactic criteria commonly applied includes (a) the subject-verb agreement, (b) the determiner-head agreement, (c) the locus of inflectional marks, (d) the tendency to stress the modifier or complement since this element “is more likely to contain focal information” (Keizer 2007, 19) and (e) the anaphoric reference (in that it is the head that will be pronominalized). Crucially, Keizer notes that the battery of tests has to be interpreted cautiously in that it is often inconclusive and that “syntactic evidence may clash with the semantic evidence” (Keizer 2007, 21). She goes on to suggest that headedness is a gradable notion, i.e. a matter of degree, provided that “the more criteria an element complies with, the more prototypical a head it is” (Keizer 2007, 21). In this book I will illustrate that in Spanish BQ-constructions, the ‘semantic’ head tends to coincide with the ‘syntactic’ head. Further, I will differentiate the constituency of literal and grammaticalized uses in line with the working hypothesis put forward by Brems (2011, 46) for the English BQ, viz. that N1 may shift from head status to quantifier status in grammaticalization, as visualized in Figure 10. In sum, the semantic criterion of obligatoriness prevails. Provided

State of the art | 47

that the entire string [Det. N1 de] can be replaced by canonical quantifiers (e.g. mucha gente ‘many people’),29 the constituency boundary moves from the position in between N1 and de in literal uses (cf. [Det. N1][de N2]) to immediately after the preposition in grammaticalized uses (cf. [[Det. N1 de] N2]). Further, it will be claimed that when grammaticalized uses present syntactic evidence against the head-status of N2, the mismatch can usually be motivated in the light of the conceptual image persistence of the QN (see infra).

un montónN1

de librosN2

un montónN1 de

genteN2

Head

Modifier

Quantifier

Head

Fig. 10: Head shift in grammaticalization (cf. Brems 2011)

2.2.5 Gradient boundaries between partitive, pseudopartitive and appositional binominals QNs have been adduced in the literature to illustrate partitive, pseudopartitive and appositional constructions. All three types of construction consist of two nouns and the linking element de. Simplifying largely, partitive constructions can be characterized as expressing the quantity of N2, pseudopartitive constructions as expressing the quantity of the kind of N2 (cf. Koptjevskaja-Tamm 2009, 329) and appositional constructions as expressing the kind of N2. Of crucial importance to the starting assumption of this volume, viz. that BQs are involved in processes of grammaticalization, is the fact that QNs shift to quantifier function only when occurring in pseudopartitive constructions.30 In the remainder of this section, I will draw neat borderlines between partitive and

|| 29 Respectively adjectives in the case of expressive binominal constructions: una maravilla de mujer ‘a wonder of a woman’ can be paraphrased as una mujer maravillosa ‘a wonderful woman’. 30 Since pseudopartitive constructions (like partitive ones) are not restricted to binominal instances but also combine with pronouns (e.g. algo ‘something’ or nada ‘nothing’) or quantifiers (e.g. poco ‘little’ and mucho ‘lot’) in the N1-slot, I will use the notion of binominal quantifier when discussing the development and functional organization of QNs in the remainder of this book.

48 | Framework, state of the art and methodology

pseudopartitive constructions – or should I say construals? – on the one hand and pseudopartitive and appositional constructions on the other.31

2.2.5.1 Partitive and pseudopartitive constructions Strictly speaking, partitive constructions are two-constituent structures “expressing a part-whole relation, of which the first constituent (the quantifier) indicates a subpart or a subset of the second constituent (the reference mass)” (Verveckken 2013). They differ from pseudopartitive constructions “whose reference mass cannot be specified, as it is not grounded” (Verveckken 2013). In addition to bounding or unitizing the entities expressed by the second constituent, pseudopartitive constructions also “designate other characteristics such as shape, size, extent, spatial configuration or kind” (Verveckken 2013). The fuzzy nature of the boundaries between (binominal) partitive and pseudopartitive constructions is probably due to the fact that some (partitive) contexts invite both interpretations. By way of illustration, in (3), this cake can designate both the specific entity of which a slice is a subpart and the kind of entity John ate a slice of (cf. Koptjevskaja-Tamm 2009; Langacker 2011). (3)

John ate a slice of this cake.

The chapter on quantifiers in the descriptive grammar of Spanish (Sánchez López 1999) and the renewed reference grammar by the Real Academia (RAE 2009) also draw this distinction. Partitive constructions are two-constituent structures, consisting of a quantifier and a “partitive complement or coda” (RAE 2009, 1441, translation KV). The first constituent can be an absolute quantifier (e.g. muy pocas de las iniciativas ‘very few of the initiatives’), a fraction noun (e.g. la mitad de los papeles ‘half of the papers’), a measure noun (e.g. un kilo de estas patatas ‘a kilo of those potatoes), a group noun (e.g. un grupo de los senadores ‘a group of the senators’) or a pronoun (e.g. algunas de las iniciativas ‘some of the initiatives’). The second constituent designates a plurality (i.e. a mass noun or a plural count noun) and is necessarily definite. Both constituents have referential or extensional value: in la mayoría de los estudiantes ‘the majority of the students’, both la mayoría and los estudiantes refer to distinct masses in the extralinguistic world, the one designated by mayoría being a subpart of

|| 31 See Rodríguez Espiñeira/Pena (2008) for a detailed discussion on linguistic categorization and (possibly gradient) categorial boundaries. In the case of QNs, the occurrence in either partitive, pseudopartitive or appositional construction is characterized by a slightly different conceptualization, morphosyntactic behavior and combinatorial patterning.

State of the art | 49

the one designated by los estudiantes. The double referential value allows the partitive complement to be separated from the quantifying element, as in (4). Generally, verb agreement can be made with either the first or the second constituent. (4)

De los manifestantes, un grupo se dirigieron al gobierno civil. (Sánchez López 1999, 1052) ‘Of the demonstrators, a group was going to the civilian government.’

Following Brucart (1997), the GDLE distinguishes between “intrinsic partitives” and “not-intrinsic partitives” (Sánchez López 1999, 1049). The former are considered to present a quantifying interpretation by definition, independently of the second constituent, as is the case for totalidad ‘totality’, mayoría ‘majority’, mitad ‘half’, resto ‘rest’, etc. Not-intrinsic partitives are nouns which only acquire a partitive32 reading when not preceded by a definite determiner (cf. the contrast between (5a) and (5b)). Among the not-intrinsic partitives, the author includes measure nouns (e.g. un montón de libros ‘a heap of books’) and configuration nouns (e.g. una pila de discos ‘a pile of disks’). (5)

a. Un grupo de senadores socialistas votaron en contra. ‘A group of socialist senators voted against.’ b. El grupo de senadores socialistas votaron en contra. ‘*The group of socialist senators votedpl against.’ (Sánchez López 1999, 1052)33

The GDLE goes on to distinguish between two types of partitive constructions depending on the definiteness of the second constituent, viz. partitive and pseudo-partitive constructions. Partitive constructions presuppose the existence of two separate masses or groups of entities (as in (6a)). Pseudopartitive constructions only evoke one mass or group of entity since the second noun does not have proper referential value and merely indicates “the properties of the elements that form the set designated by the quantifier” (Sánchez López 1999, 1051), as in (6b)). (6)

a. una gran parte de los senadores ‘a great part of the senators’ b. una gran parte de senadores ‘a great deal of senators’

|| 32 Note that the author seems to equate the notions of ‘quantifying interpretation’ and ‘partitive reading’, which in my view explains why in the GDLE, the pseudopartitive construction is more or less considered a subtype of the partitive construction. 33 Sánchez López considers the example to be ungrammatical. In Chapter 8 it will become clear why I rather consider the sentence as highly marked, yet not ungrammatical strictly speaking.

50 | Framework, state of the art and methodology

The NGLE, however, treats the pseudopartitive construction as a separate construction type and situates it at the same level as partitive constructions. The first element can be a quantifying pronoun (e.g. algo de pan ‘a bit of bread (lit. something of bread’), a quantifier such as poco or mucho (e.g. un poco de amor ‘a bit of love’) or quantifyings nouns (e.g. un gran número de participantes ‘a large number of participants’, la elevada cifra de detenidos ‘the high number of arrested persons’). In contrast to partitive constructions, the first constituent cannot be a cardinal number (e.g. una docena de huevos ‘a dozen of eggs’ or doce de estos huevos ‘twelve of those eggs’ versus *doce de huevos ‘*twelve of eggs’) and the construction with QNs is highly productive. The second constituent can designate a singular mass noun (e.g. un litro de agua ‘a litre of water’) or a plural count noun (e.g. una retahíla de mentiras ‘a whole string/series of lies’) and can refer to both concrete and abstract entities. While the second noun is necessarily indefinite34, the first noun can in some cases show up without determiner (e.g. infinidad de veces ‘infinity of times’) or combine with a definite one (e.g. la tira de niños ‘a bunch (lit. the strip) of kids’). Since the second constituent lacks proper referential value and hence does not refer to a contextually delimited group of entities, the prepositional phrases cannot be separated from the quantifier. The double possibility of verb agreement, viz. with N1 and N2, that characterizes partitive constructions, generally also holds for pseudopartitive constructions, except when the first element is preceded by a definite determiner. It bears pointing out that although identical quantifying elements can be used in partitive and pseudopartitive constructions (e.g. (7)), the conceptualization will be slightly different. Additional evidence for the distinction between both constructions or construals can be found in typological studies: in her crosslinguistic analysis of partitve and pseudopartitive constructions, Koptjevskaja-Tamm (2009) argues that many languages express partitive and pseudopartitive constructions by different and unrelated construction types.

|| 34 Interestingly, the NGLE stipulates that only some of the QNs in pseudopartitive constructions can occur in their plural variant: while plural measure nouns (e.g. dos kilos de azúcar ‘two kilos of sugar’) or group nouns such as pila (e.g. tres pilas de libros ‘three piles of books’) sound fine, group nouns such as sinfín refrain from pluralization (e.g. *dos sinfines de oportunidades ‘*two infinities of opportunities’). Further, only quantifying adjectives can modify N1 in pseudopartitive constructions. These observations will be highly relevant in the discussion of the degree of grammaticalization of QNs.

State of the art | 51

(7)

a. Una gran parte de los senadores votaron en contra. ‘A great deal of the senators voted against.’ b. Una gran parte de senadores votaron en contra. ‘A great deal of senators voted against.’

2.2.5.2 Pseudopartitive and appositional constructions In binominal appositional constructions, two nominal elements are juxtaposed and iconically mirror the existence of a predicative relation, as in la ópera Fidelio ‘the opera Fidelio’.35 Morphosyntactically and semantically close to pseudopartitive constructions are emphatic appositions, of the type A de B (in which the preposition is obligatory and the element A is a predicate to the element B). Three types of emphatic appositions are to be distinguished (cf. Lapesa 2002), which differ as to (i) the lexical nature of the element A, (ii) the type of determiners and quantifiers of both A and B and (iii) the semantic role of B: Type 1 Type 2 Type 3

el asno de Sancho ‘Sancho the donkey (lit. the donkey of Sancho)’ una maravilla de película ‘a wonder of a movie’ un prodigio de vitalidad ‘a prodigy, marvel of vitality’

The first type of appositional construction is usually used to designate human entities and can be paraphrased by B es un A ‘B is an A’. By way of illustration, the predicative relation expressed by el asno de Sancho corresponds to Sancho es un asno ‘Sancho is a donkey’. In view of the predicate-status of N1, the headstatus of N2 does not come as a surprise. The first noun is necessarily definite and can belong to three lexical categories: (i) evaluative nouns (e.g. el fantasma del gerente ‘the boaster of the manager’), (ii) animal nouns which acquire a usually pejorative evaluative interpretation (e.g. la perra de tu mujer ‘that bitch of your wife’) and (iii) adjectives designating an extreme characteristic (e.g. la muy loca de tu prima ‘the very foolish of your niece’). Adjectives or nouns denoting positive qualities are marginal, to the extent that they are interpreted ironically when they occur. In el inocente de su marido ‘the innocent of her husband’ the reading of inocente shifts from ‘innocent’ to ‘naive’. Further, the element A agrees with B in gender and number, as in lasfem pl tontasfem pl de tus amigasfem pl ‘your stupid neighbors (lit. the stupid of your neighbors)’. The second noun is always either definite or a proper noun. || 35 I will mainly follow the train of reasoning put forward in the NGLE, since it offers the most detailed section on binominal appositions. Although the discussion in the GDLE (Suñer Gratacós 1999) includes examples of two types of emphatic appositions, the author does not go on providing a thorough differentiation.

52 | Framework, state of the art and methodology

In the second type, as in un desastre de vacaciones ‘a disaster of a vacation’, the first noun is by default indefinite. Determiner variation is allowed however, when motivated by discourse reasons, as in Quiero hablarles de [la/una] maravilla de película que vi ayer ‘I would like to talk to you about [the/a] wonder of a movie I saw yesterday’ or ¡Qué demonio de hombre! ‘What a devil of a man!’. Further, the first element A is always nominal and does not agree in gender or number with N2. It is usually a quality noun of a restricted set, such as encanto ‘charm’, horror ‘horror’, espanto ‘fright’, etc. The second is nominal as well, but can only be a common noun. A particular “double semantic role” (NGLE 2009, 892) is reserved for N2: in (8), the N2 chica does not only designate the person which is called un encanto but also categorizes the subject of the copulative main clause. (8)

Maricarmen, mi jefa, es un encanto de chica (NGLE 2009, 891). ‘Maricarmen, my chef, is all charming a girl (lit. a charm of girl).’

The third type largely corresponds to the second type of emphatic appositions, but differs as to the role of N2, which is not ‘double’ in this case. The quality of being efficient in (9) is only attributed to the subject of the main clause, yet not to N1. (9)

Es usted un portento de habilidad. (NGLE 2009, 892). ‘You are extremely efficient (lit. a wonder of efficiency).’

From the former discussion, one can infer that the similarity between appositional binominals and pseudopartitive constructions holds for the second type of emphatic appositions (whereas the definiteness of N2 in the first type of emphatic apposition recalls the partitive construal).36 In the first type of appositions, as in el tonto del niño ‘the stupid (lit. of the) child’, the N2 does not only have full referential value, but also designates a topicalized entity, which allows the binominal construction as a whole to fulfill a participant role in the sentence (e.g. (10a)). In the second type of apposition, as in un horror de tío, N2 does not evoke a specific topicalized entity, but a category instead, which explains its primarily predicative value (e.g. (10b) versus (10c)). (10) a. Mamá le habló al tonto del niño. ‘Mum spoke to the stupid child.’ b. *Mamá le habló a un horror de tío. ‘*Mum spoke to a horror of an uncle.’ c. Javier es un horror de tío. ‘Javier is a horror of an uncle.’

|| 36 See also Suñer Gratacós (1999) on the overlap with partitivity.

Methodology | 53

The distinction between the pseudopartitive construction and the emphatic apposition is less neat however than the one claimed between partitive and pseudopartitive construals, especially in the case of plural N2s, as in un horror de mosquitos ‘a horror of mosquitos’. Further, as Koptjevskaja-Tamm (2009, 343) points out, in many languages the pseudopartitive construction type marks both apposition and quantification. Additional arguments will be adduced in the present volumet to illustrate that the difference is one of degree but not of kind: BQ-constructions smoothly shift from one side of the continuum to the other.37

2.3 Methodology This study is corpus-based and focuses on ten QNs, viz. alud ‘avalanche’, aluvión ‘flood’, barbaridad ‘barbarity, atrocity’, hatajo ‘herd’, letanía ‘litany’, mar ‘sea’, montón ‘heap’, mogollón ‘mass’, pila ‘pile’, racimo ‘bunch’. Three points have to be noted concerning the selected items. First, this study aims to analyze the shift from originally lexical items to quantifying uses and to determine which lexical items display the required quantifying potential for such a change. In spite of their categorization as quantifying nouns by the GDLE (1999), three categories have not been taken into account: (a) abstract quantifying notions (e.g. cantidad ‘quantity’, par ‘pair’, docena ‘dozen’); (b) measure nouns (e.g. kilo ‘kilo’, metro ‘meter’) and (c) bounding nouns (e.g. pedazo ‘piece’, loncha ‘slice’). The former quantifying nouns do not present a quantifying potential but rather have an intrinsic quantifying interpretation instead. Several abstract quantifying nouns also developed a more subjectified and schematic reading (as in (11)), in contrast to the original ‘literal’ reading in (12)), but their development is not characterized by chains of subtle metonymic extensions, neither has it resulted in abrupt shifts in combinatorial pattern. To my knowledge, no measure noun has given rise to productive subjectified uses. What is more, since they are judged by international and objective standards, it is highly doubtful that measure nouns will ever develop grammaticalized uses. As to bounding nouns, they usually occur in partitive constructions and by definition evoke a subpart or quantity. By way of illustration, in un pedazo de jamón ‘a piece of ham’, N1 expresses a part of N2 while N2

|| 37 It bears pointing out, however, that the notion of ‘expressive binominal construction’ will be used in the remainder of this volume to refer to the second type of emphatic apposition. In my view, the notion of ‘apposition’ is traditionally associated with a broader interpretation.

54 | Framework, state of the art and methodology

designates the whole. In pseudopartitive constructions and BQ-constructions in particular, N1 expresses a unit or set of N2s while N2 designates the constituting entities, to the extent that the part-whole relation can be considered inverted. (11) Los acontecimientos ocurridos son en general lamentables, o tristes, o sorprendentes, o risibles, o todo ello a la vez. Es indudable que ha habido un buen número de personas que han cumplido mal su tarea. Los acontecimientos que lo prueban han sido numerosos. (1995, press) ‘The events which have happened are generally deplorable, or sad, or surprising, or laughable, or all that at the same time. There is no doubt that there has been a great number of persons who badly carried out their task. The events which give proof of that have been numerous.’ (12) Bakú conocía entonces un dinámico desarrollo y una gran afluencia de población rural, que duplicó su número de habitantes. (1995, press) ‘Bakú then experienced a dynamic development and a huge flow of country population, which duplicated its number of inhabitants.’ A second selection criterion concerns the search for a diversified set. The aim is to include both highly frequent and marginal QNs and to cover multiple N2domains. More precisely, the dataset has to contain QNs which exclusively indicate a set of people, QNs which primarily serve to quantify things, QNs which are limited to a conceptually restricted set of N2s (e.g. to N2s conceptualized as a kind of ‘invader’ with alud) as well as general QNs which combine with any type of N2. The third principle in the selection of QNs concerns the inclusion of sets of closely related QNs. Two factors motivate the analysis of minimal pairs or nearsynonymous QNs. On the one hand, it is generally known that one sees more in comparison. On the other hand, the question whether closely related source constructions necessarily lead to a similar outcome and follow similar pathways of change are interesting theoretical issues. The selection of montón ‘heap’ is based on the working hypothesis that this QN is the most general and most productive QN in Spanish. Pila ‘pile’ can be considered a closely related item, since it designates a similar vertical configuration (at least literally). Mogollón ‘mass, fuss’ can be considered a functional synonym of montón, at least in colloquial Spanish. The minimal pair alud ‘avalanche’ and aluvión ‘flood’ corresponds to a productive lexical field of QNs, viz. uncontrollable nature phenomena. Mar ‘sea’ is included for being the only highfrequency QN next to montón and for its remarkable aptitude to combine with intensifying adjectives as well (e.g. la mar de feliz ‘completely happy’). The

Methodology | 55

remaining QNs are selected in an attempt to cover as many lexical source domains as possible: hatajo ‘herd’ originally designates a set of animals but has acquired a conventionalized pejorative reading; letanía ‘litany’ is the most specific and therefore less canonical QN included; barbaridad ‘barbarity’ is actually primarily related to another category, viz. the category of evaluative or expressive quality nouns. For this set of 10 QNs, exhaustive data-extractions have been carried out from the synchronic and diachronic online corpora of the Real Academia Española, viz. the Corpus de referencia del español actual (CREA) and the Corpus diacrónico del español (CORDE),38 largely recognized as the most representative and extensive corpora for Spanish. They have one major shortcoming, however, viz. the absence of lemmatizing and morphosyntactic tagging. The dataextraction has therefore been based on the query [QN de], inevitably including lots of irrelevant instances (e.g. partitive construals, not nominal second constituents, etc.).39 The samples were inspected manually to ensure that each N2 is a multiplex entity whose quantity is assessed by N1. The precise extension of the datasets, both before and after manually filtering out the irrelevant occurrences, are provided at the beginning of the diachronic (Chapters 4 and 5) respectively synchronic (Chapter 7 and 8) case-studies. In order to avoid mixing up different language systems, the samples were limited to Spain. No restrictions were introduced as to genre. In order to get a better grip on (the glides in) the lexical field of QNs, the diachronic case-study has been completed with a category search in the complementary Corpus del español by Mark Davies. The latter corpus is the only corpus that allows searches containing abstract grammatical categories such as ‘noun’. The query [noun de nounpl] has been used to retrieve all instances of the BQconstruction attested from the 12th century onwards. The extractions shed light on the semantics of the structural slot ‘N1’, i.e. the lexical field of the QN, provided that the irrelevant data are filtered out. However, no statistical conclusions can be drawn from this subcorpus, for four reasons: (i) full data access is denied from the 15th century onwards,40 (ii) the corpus does not allow regiolectal

|| 38 The data extracted by the query montón de in CREA is reduced to a representative sample of the more manageable size of 500 tokens. 39 At the same time, I am aware that this search string a priori excludes the possibility of finding cases with postnominal modification to the QN. However, I have performed various spot-checks from which I can safely conclude that there are hardly any occurrences of this type. 40 From the 15th century onwards, the corpus design obliged me to limit the dataset to items which occur at least three times.

56 | Framework, state of the art and methodology

differentiation between peninsular and Latin-American variants, (iii) when reconsulting a single query, slight discrepancies with regard to token frequencies are observed, (iv) the corpus includes many wrongly-tagged tokens. The BQs observed in CREA and CORDE were coded for number and determiner of both nominal elements, modification patterns, agreement (verbal as well as phoric with N1 or N2) and collocational patterns (N2, count-mass noun distinction, etc.). Semantic criteria pertain to the reading of N1 as head, quantifier or specifier (based on the degree of co-extensiveness between the QN and N2) as well as on syntactic or pragmatic contextual clues at different levels of abstraction), degree of conceptual persistence, reading of N2 (along the distinctions concrete/abstract, human/animate[non-human]/inanimate, literal/figurative), the pragmatic function of N1 (emphasizer, amplifier or neutral (see Quirk et al. 1985)) and the degree of subjectification in the Traugottian sense (1989; 2003b). It is important to note that this analysis set out as a primarily qualitative analysis, which means that frequency tables have auxiliary status only. The more since the subcorpora of individual QNs usually do no present a comparable size and, as such, do not allow sound statistical analyses. Further, quantitative analyses of historical data in particular are to be handled cautiously, since historical data – at least for Spanish – are sparse, incomplete and not representative for language use (Cornillie 2011). However, the label of “bad data” (Labov 1994) does not entail that detailed qualitative historical analyses would be irrelevant. A final methodological note concerns the treatment of examples. The examples are unaltered copies of corpus data, which means that I did not correct language errors. The translations are not glossed except when relevant to illustrate morphological tendencies. The QN and preposition come in bold, both in the original example and in the translation. The font underline marks linguistic or contextual elements which corroborate my interpretation of the example. Unless stated otherwise, the examples are taken from CREA (after 1975) or CORDE (before 1975) and always followed by detailed information as to the year and the genre only (instead of adding full bibliographical details as to the author for instance), for three reasons: first, for reasons of space; second, I have observed that genre distribution did not significantly influence the claims made;41 third, full bibliographical reference can be easily retrieved by introduc-

|| 41 In fact, the diachronic corpus (CORDE) exclusively provides written examples of the BQ. As to the synchronic corpus (CREA), 1231 of the 1371 withdrawn examples pertain to the genres press, books and novels. However, rather than reflecting different tendencies in use according

Methodology | 57

ing any sequence of words from the examples in the search engine of the online corpora, which will provide immediate and full access to the example in question. Nor do I differentiate between distinct discourse traditions (Kabatek 2005)42 as advocated by the recent trend in Spanish historical linguistics, since the case-studies show that the use of QNs is never ‘evoked’ by a particular discourse situation. Instead, QNs fulfill the pragmatic need of the speaker to express hyperbolic quantification (independent of the discourse tradition (s)he is involved in).

|| to the genre, the skewed proportion of oral and written examples of BQs is partially due to the proportion of written and spoken Spanish in the corpora themselves (merely 10% of the CREA is oral Spanish, see also Verveckken/Delbecque 2014, 6–7). 42 Discourse traditions are defined as the repetition of a text or textual form or a particular way of writing or speaking that has acquired the status of a sign in itself (for providing meaning) and that is typically evoked by a particular communicative situation (Kabatek 2005, 158– 159).

3 Binominal quantifiers as a locus of grammaticalization The binominal construction is cross-linguistically considered a locus of grammaticalization.43 However, as far as Spanish is concerned, the literature pays relatively little attention to the development and grammaticalization of the BQconstructions. To my knowledge, no diachronic analysis has been carried out and no final answer is provided to the status of N1, which can be considered lexical, grammatical or in between. The present chapter shows that the Spanish BQ is not a straightforward case of grammaticalization.

3.1 Grammaticalization and grammaticalization theory Literally, grammaticalization (henceforth GR) refers to the process whereby items become grammatical (diachronic dimension) or whereby concepts are expressed by means of grammatical forms (synchronic dimension). The origin of the term is commonly attributed to Meillet (1912) who defined it as “l’attribution du caractère grammatical à un mot jadis autonome” (1912, 131). 44 Although the phenomenon received little interest in structural and generative linguistics, which are strongly synchronic in their approaches and assumptions (Cuyckens 2007), GR continued to be a hotly debated topic among Indo-Europeanists (e.g. Kurylowicz 1964) and typologists (e.g. Givón 1979; Heine/Reh 1984). The topic has seen a new surge of interest in the late nineties – probably boosted by the sudden availability of large diachronic corpora – and gave rise to the so-called grammaticalization theory (henceforth GT). However, in se, GR is a theoretically independent notion.

3.1.1 Definition The definition of GR has been subject to constant revision and further refinements. The focus shifted from lexical items (and by extension also grammatical

|| 43 E.g. Brems (2011); Traugott (2008b); Trousdale (2008b) on the BQ in English; Masini (Fc.) on the BQ in Italian; Mihatsch (2000; Fc.) for a comparative study on Romance languages and Koptjevskaja-Tamm (2009) for a typological overview. 44 The attribution of grammatical nature to an erstwhile autonomous word (translation KV).

Grammaticalization and grammaticalization theory | 59

ones)45 as the input of GR (cf. Himmelmann 2004, 31, “element-based view on grammaticization”) to strings and to the triggering role of the highly specific contexts they occur in (e.g. Hopper/Traugott 2003; Traugott 2003b). In other words, GR is now considered to work on constructions or strings of items, and not on items in isolation. The definition by Traugott (2003b) fits in with the cognitive-functional frame I subscribe to: The process whereby lexical material in highly constrained pragmatic and morphosyntactic contexts is assigned grammatical function, and once grammatical, is assigned increasingly grammatical, operator-like function. (2003b, 645)

The development of the future marker be going to out of the movement verb to go has become somewhat of a textbook example of GR (e.g. Bybee 2003; 2006; Cuyckens 2007; Lamiroy 1983; Traugott 2011). It is commonly assumed that the lexical movement verb to go (1) combined with the pragmatic inference of futurity in purposive directional contexts such as (2): if one is traveling for the purpose of learning Spanish, one intends to learn Spanish in the near future (cf. Cuyckens 2007). With repetition, the string [to be going [to V]] became reanalyzed as [[to be going to] V], as in (3). The reanalysis becomes overt once the pattern extends to non-purposive or non-intentional contexts (as in (4)) and combines with non-human subjects (as in (5)). Note that the pragmatic inference of futurity is not invited with similar movement verbs such as to travel. While I am traveling to learn Spanish automatically evokes physical movement, *It is traveling to rain today is ungrammatical.46 Further, the string has led to phonological erosion into gonna. (1) (2) (3) (4) (5)

I am going to Madrid. I am going to Madrid to learn Spanish. I am going to learn Spanish. My mother is going to be surprised. *My mother is going in order to be surprised. It is going to rain today.

|| 45 E.g. Kurylowicz (1975, 52) defines GR as “the increase of the range of a morpheme advancing from lexical to a grammatical or from a less grammatical to a more grammatical status”. 46 Traugott’s (2011) diachronic constructional account challenges the text-book assumptions on the development of the future market to be going to. She claims that the purposive directional context was untypical and cannot have been the source construction. Instead, the to be going to-future has multiple sources, viz. the progressive be –ing which entails imperfective temporality, the purposive infinitive construction and the passive which demotes agentivity.

60 | Binominal quantifiers as a locus of grammaticalization

3.1.2 Characteristics Generalizing over framework-related assumptions, it is common practice to characterize GR as a gradual, unidirectional process which is mainly diachronic in nature. The gradualness of GR builds on the assumption that diachronic changes involve small and local micro-steps that are discrete and therefore abrupt “in a tiny way” (Traugott/Trousdale 2010, 20). Yet, “because different parts of a construction may undergo changes at different points in time, the change to the construction as a whole may appear to be gradual” (Traugott/ Trousdale 2010, 21).47 It goes without saying that series of small changes can lead to macro-changes (Lightfoot 1979, Traugott 2003b, 626). The succession of changes ties in with the second characteristic generally associatied with GR, viz. its unidirectionality. Minimally, unidirectionality implies that GR works on more lexical (or less functional) items and leads to more functional or grammatical items. Furthermore, several cross-linguistically recurring structural clines form the backbone of many research in GT. In addition to the well-known verbal (Givón 1979) and nominal clines (Lehmann 1985, Heine/Narrog 2010), construction-specific clines (Traugott 2008a) are established: Nominal cline: relational noun > secondary adposition > primary adposition > agglutinative case affix > fusional case affix (Lehmann 1985, 304) Verbal cline: lexical verb > auxiliary > affix (Givón 1979, 220) Binominal cline: partitive/measure phrase > quantifier > degree modifier (Traugott 2008a: 228)

The unidirectionality claim also pertains to the linguistic domain of change and the semantic-pragmatic aspects of GR (cf. Traugott 2003b, 630–631). Givón (1979, 209) introduced pragmatics into GT and suggests that the relative freedom of a linguistic item – at first discourse-motivated – gradually reduces: discourse > syntax > morphology > morphophonemics > zero

His model gave rise to the well-known aphorisms “Today’s morphology is yesterday’s syntax” (Givón 1971, 413) and, more carefully, “Today’s syntax may be

|| 47 Gradualness (both in GR as in change in general) is twofold: on the one hand it relates to the gradual structural spread in the language system, on the other hand is refers to the “socialcontextual” (Traugott/Trousdale 2010, 21) spread of change across groups of speakers.

Grammaticalization and grammaticalization theory | 61

the product of yesterday’s discourse pragmatics” (Faarlund 1989, 71).48 In addition, the assumption of gradualness has recently been extended to language typology. Within a group of languages belonging to the same family, languages can “occupy different stages on a continuous line of development” (De Mulder/ Lamiroy 2012, 201). The following cline indicates that GR takes place at a different speed in three Romance languages: French > Italian > Spanish (De Mulder/Lamiroy 2012, 201)

It bears pointing out however, that GR processes are not deterministic: they do not have to go to completion but can stop halfway (Fischer 2007; Rodríguez Espiñeira/López Meirama 2008). The third characteristic usually attributed to GR – from a cognitivefunctional perspective at least – is its diachronic nature (cf. Cuyckens 2007). Although the notion of GR can be synchronically related to the different crosslinguistic encodings of grammatical functions, GR in the narrow sense refers to the process of change.

3.1.3 Conditions In addition to the above-mentioned characteristics of GR, a number of approaches have focused on the conditions of GR, which relate to variation, frequency and semantics of the source construction (cf. Lamiroy 2001, 96–97). As to the principle of variation, “[i]t has long been recognized that current variation is both the result of and the reason for change” (Traugott/Trousdale 2010, 4). In other words, it is commonly assumed that “changes are always manifested in synchronic variation” (Andersen 2001, 228), which makes Traugott (1989, 31) and Hilpert (2010) claim that gradient synchronic data allow to evaluate the validity of diachronic hypotheses.

|| 48 A number of counter-examples to the unidirectionality claim and clines have been proposed, usually labeled as degrammmaticalization (Campbell 1991; Newmeyer 1998; Norde 2009) or antigrammaticalization (Haspelmath 2004). The reactions to these studies have been quite diverse, however: while only few GR theorists agree that the unidirectionality claim is false, a number of studies suggest that the cline from less to more bonded is the unmarked and highly frequent GR pathway and many hold the opinion that the cases of degrammaticalization can be accounted for by alternative principles (e.g. Heine/Narrog 2010, 403; Fischer 2000; Narrog 2004). Note also that Traugott (2014) has recently suggested that “uniderictionality is a strong claim that has died of its own science”.

62 | Binominal quantifiers as a locus of grammaticalization

The second sine qua non for GR relates to the role of frequency (Bybee 2003), which is generally – and from the usage-based perspective in particular – considered to be both the result of and a primary contributor to the process. The reasoning is as follows: with repetition, a stimulus loses its impact (habituation (Bybee 2003, 605)). Increasing frequency leads to the weakening of semantic force (also bleaching or generalization of meaning) as well as to phonological reduction or erosion. The linguistic (string of) item(s) become(s) gradually more autonomous and the association with other uses of the same item(s) gradually weakens, which leads to a loss of semantic transparency and compositionality, i.e. to semantic change.49 The generalization of meaning resulting from increased frequency ties in with the third condition usually associated with GR. More precisely, GR is considered to work on semantically general or basic and culturally independent concepts (cf. Cuyckens 2007; Himmelmann 2004, 37; Lamiroy 2001). The candidates for GR are items which “represent concrete and basic aspects of human relations with the environment, with a strong emphasis on the spatial environment, including parts of the human body” (Cuyckens 2007, 19). For example, a noun referring to ‘house’ stands more chances to grammaticalize than highly specific refrigerator or pigsty (cf. the French chez ‘with’ from Latin casa ‘house’).

3.1.4 Desemanticization and decategorialization There are two views on the status of the semantic and formal changes in GR: while a number of studies focus on the reduction and increased dependency of the grammaticalizing item (e.g. Lehmann 1985) as a consequence of routinization or habituation (see supra), other GR theorists focus on the expansion or extension to new contexts (Heine/Narrog 2010; Himmelmann 2004). In fact, the process cannot be reduced to either degeneration or extension, but the “GR as expansion” is the logical outcome of “GR as increased dependency” (Traugott 2011, 12): it is the reduction of the signal by routinization which allows the signal to become appropriate in increasingly more contexts. This train of reasoning is referred to as the loss-and-gain model (Heine/Narrog 2010, 405; Hopper/Traugott 2003) and captures formal as well as semantic changes in GR.

|| 49 The crucial role of high-frequency is downtoned in the constructional approach which motivates the grammaticalized behavior of low-frequency items in the light of GR via analogy to more frequent structural relatives (Hoffmann 2004; Traugott 2008a; see also Heine/Narrog 2010, 406).

Grammaticalization and grammaticalization theory | 63

It is common practice to consider decategorialization as the main morphosyntactic change in GR. Decategorialization refers to the fact that grammaticalizing items, as a consequence of routinization and generalization of meaning, start to lose typical features of the grammatical category they originally belong to. Minimally, they lose their combinatorial or syntactic freedom. By way of illustration, the grammaticalized uses of QNs are no longer compatible with any type of adjectives (*medio hatajo de idiotas ‘*half a bunch of idiots’ vs. medio hatajo de ovejas ‘half a herd of sheep’) and the pluralization of the N1 is marked (?durante pilas de años ‘during lots (lit. piles) of years’). Yet, simultaneously, there are also gains or a re-categorialization to a certain degree: the QN acquires properties typically associated with quantifiers. As to the semantic change in GR, a similar loss-and-gain model has been posited recently. Although GR can result in the “loss of concrete specificity (bleaching) there is no loss of semantic complexity” (Traugott 2003b, 623). The bundle of facets associated to a particular item or string is not reduced, but metaphorically transferred to another domain. By way of illustration, the facets related to to be going to are simply transferred from the domain of SPACE to the domain of TIME (cf. Lamiroy 2001). This model of redistribution of meaning lies at the heart of Traugott’s view on subjectification (see Section 2.1.1.4 in Chapter 2) and starts from the assumption that GR “involves pragmatic strengthening (not weakening)” (Traugott 2003b, 633, italics KV).50 It bears pointing out that the role of semantic process in GR tends to be overestimated to the extent that GR risks to be “equated with simple semantic reanalysis or semantic change as such” (Brems 2010, 99; see also Cuyckens 2007). Even though it is quite plausible that the trigger of GR is necessarily rooted in the communicative goals and expressive needs of the speakers (cf. the following quote), Meaning changes and the cognitive strategies that motivate them are central in the early stages of grammaticalization and are crucially linked to expressivity. (Hopper/Traugott 2003, 76)

|| 50 Pragmatic strengthening “arise[s] out of the cognitive and communicative pragmatics of speaker-hearer interaction and discourse practices” (Traugott 2003b, 634). It refers to the fact that speakers can come to use conversational implicatures strategically. When these inferences are repeated by many speakers in many contexts, they can become conventionalized. The invited inferences are “a kind of conceptual metonymy within the speech chain (…) since it is primarily associative in character, being derived from the uses to which interlocutors put linear sequences of utterances and associations in context.” (Traugott 2003b, 634).

64 | Binominal quantifiers as a locus of grammaticalization

it is unacceptable from a cognitive-constructional perspective that semantic change alone corresponds to “the core defining feature of grammaticalization processes”, even though it is suggested as such by Himmelmann (2004, 33).

3.1.5 Mechanisms Although reanalysis and analogy cannot be identified with GR, these processes are generally considered to be the basic mechanisms in GR, to the extent that “[o]ne of the hotly debated topics in recent years has been whether reanalysis or analogy is the dominant mechanism in change” (Traugott/Trousdale 2010, 6). Reanalysis is a structural and initially covert change. It refers to the process whereby a new underlying structure is assigned to a surface sequence without overtly modifying it (De Smet 2011; 2014; Traugott 2011). By consequence, it is itself invisible and operates abruptly: when a pattern ‘X’ is reanalyzed as a pattern ‘Y’, no possible intermediate alternatives exist. Further, reanalysis is said to take place through ambiguity: it is only when a surface pattern allows for (at least) two alternative interpretations that the shift from one analysis to another can occur (cf. Anttila 1989; De Smet 2014; Harris 2003; Timberlake 1977). For instance, the example una pila de libros ‘a pile of books’ is ambiguous and can either refer to the physical heap-like configuration (made up of books) or to a set of books (heaped up). Through ambiguity, the following rebracketing from [[Det. N1] [de N2]] to [[Det. N1 de] N2] can take place. The reanalysis only becomes overt when the re-analyzed structure is extended to new contexts incompatible with the originally underlying structure: in una pila de ãnos ‘a lot of years’, the abstract noun años blocks the interpretation as a vertically oriented configuration. Analogy refers to the process whereby a specific construction is perceived as structurally or semantically near to another construction (Givón 1991; Fischer 2007; but see Delbecque/Verveckken 2014). The mechanism is traditionally considered to operate paradigmatically. For instance, una pila de (libros, gente, años ‘a pile of books, people, years’) can invite the inference of quantification by analogy to the conventionalized quantifying reading of un montón de or by analogy to absolute quantifiers such as mucho/a(s) ‘much, many’. In the identification of the dominant mechanism in GR – and in diachronic change in general – Traugott/Trousdale (2010, 38–39) propose to neatly distinguish the mechanisms of change from the motivations for change. As preconditions or potentials of change, they distinguish between parsing and interpretation on the one hand and analogical thinking or analogy on the other. In both cases, reanalysis is the mechanism of change. While “all analogization [KV: the

Grammaticalization and grammaticalization theory | 65

change based on analogical thinking] is reanalysis (…) not all reanalysis is analogization” (Traugott/Trousdale 2010, 38), which makes the authors suggest that reanalysis is the basic mechanism of GR “since there is no change without reanalysis” (Traugott/Trousdale 2010, 39). Likewise, Heine/Narrog (2010, 402) state that although GR and reanalysis are clearly not identifiable with each other, GR always involves reanalysis. Their view is challenged by researchers who struggle with the abruptness of reanalysis and the precondition of ambiguity (e.g. De Smet 2011; Fischer 2007). Given that analogy is also a principle of synchronic grammatical organization and language use (De Smet 2011, 1735) and can license new interpretations to surface constructions, it fulfills all requirements to be the fundamental mechanism of language change.

3.1.6 Parameters In order to recognize GR and differentiate GR from other types of change, a number of grammaticalization theorists established parameters or diagnostics of GR. The focus shifted from the morphosyntactic reflexes of the process to the semantic changes involved. The first well-known set of parameters is provided by Lehmann (1995) and starts from the assumption that GR corresponds to a decrease in the sign’s autonomy. He measures the autonomy of a sign in terms of it’s (substantial) weight, it’s (low degree of) cohesion and (high degree of) variability. These three aspects can be viewed both paradigmatically (i–iii) and syntagmatically (iv–vi) and thus result in six parameters: (i) attrition or the loss of semantic or phonological substance, (ii) paradigmatization or the increasing integration of the sign in a paradigm, (iii) obligatorification or the loss of paradigmatic variability (it cannot be omitted nor replaced by another sign of the paradigm), (iv) condensation or reduction in scope and complexity of items which the grammaticalizing sign combines with, (v) coalescence or increase in bondedness or dependency on other signs (e.g. the cline from juxtaposition to cliticization), (vi) fixation or the reduction of syntagmatic variability (it usually occupies a fixed syntactic slot). In other words, the structuralist clines all describe a “shift from fuller, freer, more complex structures to shorter, more bonded, simpler ones” (Traugott 2003b, 629). However, the cognitive-functional perspective on GR considered Lehmann’s reductive approach too restrictive and criticized the stepmotherly treatment of semantic changes. Traugott (2003b, 630) concludes that:

66 | Binominal quantifiers as a locus of grammaticalization

although the structural reductions, the condensations, coalescences, and fixations, that Lehmann highlights are strong and viable tendencies in changes that lead to certain new form-function relationships, such as case and tense-aspect-modality, they cannot be generalized to all domains of grammatical function. They should not be used as gatekeepers to exclude from grammaticalization morphosyntactic developments that are similar in other respects to case and temporal markers. (2003b, 630)

Hopper (1991) similarly observes that Lehmann’s diagnostics are “characteristic of grammaticization which has already attained a fairly advanced stage and is unambiguously recognizable as such” (1991, 21). He proposes a complementary set of five parameters which allow to detect incipient and ongoing GR. Layering refers to the fact that within a functional domain, older layers or meanings may remain to coexist and interact with the newly emerged layers. Divergence is understood as a special case of layering “where one and the same autonomous lexical item becomes grammaticized in one context and does not become grammaticized in another” (Hopper 1991, 24). Specialization is the narrowing down of structural (i.e. combinatorial) choices and semantic nuances of an emergent grammatical construction. Persistence means that in GR of a lexical form, “so long as it is grammatically viable some traces of its original lexical meanings tend to adhere to it, and details of its lexical history may be reflected in constraints and grammatical distribution” (Hopper 1991, 22). For instance, the different ways of expressing the future tense – will, shall, be going to – are continuations of their original lexical meaning.51 De-categorialization refers to the loss of morphosyntactic properties of the full or primary category the grammaticalizing item originally belongs to. For instance, in the shift to the complex preposition in view of, view loses typical nominal characteristics such as the possibility to pluralize or to combine with adjectives or determiners, etc. However, as Hopper points out himself, the “heuristic principles” are not workable diagnostics of GR, since they do not “discriminate between processes of change which result in grammaticization and processes of change which do not result in grammaticization” (1991, 32). Nevertheless, Hopper’s principles somehow bridge Lehmann’s reductive view on GR and the more recent view of GR as expansion (cf. Traugott 2014). More precisely, in an attempt to differentiate between GR and lexicalization, Himmelmann (2004) defines GR as context expansion at three different levels. || 51   The criterion persistence is of particular importance in the development and synchronic functional organization of BQs. I will argue in this book that persistence, which has both a formal and a semantic dimension, should not be considered merely a side-effect of GR. As conceptual persistence is closely interwoven with the pragmatic function of BQs, it can also function as the catalysator of GR and ensures the construction’s productivity (see infra).

Grammaticalization and grammaticalization theory | 67

Crucially, he draws on the assumption that “it is never just the grammaticizing element that undergoes grammaticization”, but the “grammaticizing element in its syntagmatic context” (Himmelmann 2004, 31) instead. First, the host-class or class of elements the grammaticalizing item is in construction with is expanded. This construction-internal extension is called host-class expansion. For instance, when grammaticalizing into articles, demonstratives start to co-occur also with proper nouns, viz. nouns they did not combine with before. Further, when the grammaticalizing item occurs in more syntactic environments than before, the change is called syntactic context expansion. Finally, the semantic and pragmatic usage contexts of grammaticalizing items expand. For instance, while adnominal demonstratives only combine with deictic anaphoric or recognitional reference (e.g. this man), articles allow for broader anaphoric and associative anaphoric use (e.g. a wedding – the bride) (Himmelmann 2004, 32– 33) and hence testify to semantic-pragmatic context expansion. Perhaps more interesting is Himmelmann’s suggestion that although the three kinds of change co-occur in GR, “semantic-pragmatic context expansion is the core defining feature of grammaticization processes (…). Often, but not necessarily, it will be possible to show that semantic-pragmatic usage contexts of the construction at hand have been expanded” (Himmelmann 2004, 33). He goes on to argue that “changes on the element-level (in particular erosion and fusion but also paradigm formation) are here considered epiphenomena which, among other things, depend on basic typological features of a given language (…) and the construction type (…).”(Himmelmann 2004, 33). Somewhere in between the extreme ends of decrease in scope and increase in scope, Heine (2003) and Heine/Narrog (2010) propose four parameters which all concern a different aspect of language structure or language use. Their set of criteria builds on the assumption that GR “is based on the interaction of pragmatic, semantic, morphosyntactic, and phonetic factors” (Heine/Narrog 2010, 404). Extension (pragmatics) refers to the rise of new grammatical meanings via context-induced reinterpretation. Desemanticization (semantics) or bleaching refers to a certain loss or generalization in meaning content. Decategorialization (morpho-syntax) is similar to Hopper’s corresponding criterion and erosion (phonetics) concerns loss in phonetic substance. Perhaps more interesting is their claim that the “ordering of these parameters reflects the diachronic sequence in which they typically apply” (Heine/Narrog 2010, 405). In addition, they argue that GR does not only involve structural “degeneration”: “while linguistic items undergoing grammaticalization lose in semantic, morphosyntactic and phonetic substance, they also gain in properties characteristic of their uses in new contexts” (Heine/Narrog 2010, 405; cf. Hopper/ Traugott’s (2003) loss-and-gain model).

68 | Binominal quantifiers as a locus of grammaticalization

3.1.7 Discussion At least three different issues are at the center of debate among grammaticalization theorists. First, different stances have been taken as to the question what triggers GR. While the usage-based account claims that the increase in frequency (Bybee 2003) and the resulting generalization of meaning motivates GR, some approaches privilege the explanatory potential of structural analogy (e.g. Fischer 2009) and still others focus on the cognitive and communicative pragmatics of the speaker-hearer negotiation (Heine/Narrog 2010, 402; Hopper/ Traugott 2003, 76; Traugott 2003, 634). In line with the latter line of research, I hypothesize that GR is triggered by the communicative need (for expressivity in the case of BQs) of the speaker. Meaning changes then are invited by pragmatic inferences which arise in the flow of speech. Second, the relationship between GR and lexicalization is a hotly debated topic as well. Generalizing over theory-specific interpretations,52 lexicalization refers to the emergence of more lexical (strings of) items out of less lexical and more functional (strings of) items. Various recent approaches (e.g. Brinton/ Traugott 2005, Himmelmann 2004) refrain from considering GR and lexicalization as opposite processes yet describe them as complementary. Starting from his definition of GR as a three-dimensional context-expansion, Himmelmann (2004, 36) argues that in lexicalization, the lexicalizing item starts to form a fixed unit with a particular item of its host-class (vs. host-class expansion). For instance, in the lexicalization of forget-me-not, the imperative form of to forget has become fixed to the first person singular pronoun me and the negation. The fixation is reflected by the use of hyphens. Yet, lexicalization can give rise to syntactic and to semantic-pragmatic expansion as well. Further, lexical generality is usually considered the essential difference between lexicalization and GR: while the input of GR is necessarily semantically general and leads to the conventionalization of a pattern, lexicalization can affect a whole sequence and usually results in highly specific strings or items. The third theoretical challenge concerns the status of GR and its relation to other processes of language change. Many of the semantic changes in GR, the triggers, the mechanisms and parameters have been claimed to be GRindependent or common to language change in general (Brems 2011, 91; Campbell/Janda 2001;Heine/Narrog 2010, 403–404; Hopper 1991, 32; Traugott 2011).

|| 52 By lexicalization, Talmy (2000) refers to the process of putting into words a specific meaning, of how to express particular concepts.

Lexical and functional uses | 69

Plausibly, it is not the set of GR-parameters as such which identifies GR, yet the remarkable correlation and interaction between them.

3.2 Lexical and functional uses The present section zooms in on the constructional semantics of the ‘un N1 de N2’-pattern, i.e. on the different readings the BQ is apt to display. In addition to a fairly literal use, two more functional uses are distinguished as well as two types of ambivalent uses. The starting point of the three-layered distinction of uses – or five-layered, including the ambivalent uses – is Brems’ (2010; 2011) model of the English Size Noun-construction.53 In the head-uses, as in (6), N1 is interpreted literally as a physical constellation of grapes fastened closely together. Further, the ofphrase “functions as a postmodifier specifying what the SN consists of, rather than introducing an additional referent into the discourse” (Brems 2010, 91). In the head uses, the functional and semantic head of the binominal syntagm coincide. (6)

A fox, unable to reach a bunch of grapes that hangs too high, decides that they were sour anyway (…). (Brems 2010, 91)

In addition, SN-constructions develop two types of quantifying uses, which are identified as two distinct paths of GR (Brems 2007b, 127). When the more abstract size implications inherent in SNs are foregrounded via pragmatic inferences, and the lexical semantics is backgrounded, SN-constructions develop quantifier-uses, as in (7), and their N2-pattern extends towards new combinations incompatible with the literal use (such as animates, abstract nouns, concrete nouns other than fruits, etc.). The second major not literal use is the valuing quantifier and “pertains to evaluating the N2-referent by a foregrounding of the usually negative semantic prosody of the SNs involved” (Brems 2010, 97). In other words, the N2-referent is evaluated rather than quantified – the quantifying function can in some cases be lost altogether (Brems 2010, 97). The notion of || 53   Brems’ (2011) category of Size Noun-construction corresponds to my notion of binominal quantifier construction. She prefers the notion of size noun (and type noun) over the traditional category measure noun since it allows to explain the functional difference between size and type nouns in line with Langacker’s (1991) distinction between size and type specification. I opted for the framework-independent notions of quantifying noun and binominal quantifier construction, which will be compared to ‘expressive nouns’ and the ‘expressive binominal construction’ (Verhagen 2009).

70 | Binominal quantifiers as a locus of grammaticalization

valuing quantifier refers both to the evaluative meaning of N1 and to the (usually) negative values of the N2 it typically combines with, as in a bunch of morons in (8). (7)

There’s now a whole bunch of studies from different cities that show the same thing. (Brems 2010, 93)

(8)

Your editorial calling for warnings in the brochures of non-ABTA travel agents makes holidaymakers look like a bunch of morons. (Brems 2010, 97).

Interestingly, Brems suggests that the three readings of SN-constructions are “collocationally constrained constructions” (Brems 2010). More precisely, SNconstructions are considered nodes which combine with specific sets of collocates according to their uses as head, quantifier or valuing quantifier. “Prenominally, these co-selection patterns hold between sets of adjectives and determiners, and the SN. Postnominally, systematic co-selection subsumes semantic prosody and the sets of N2s that SNs combine with.” (Brems 2010, 101) As far as Spanish is concerned, it seems convenient to further unravel the notion of valuing quantifier and to distinguish the evaluative potential of quantifier uses from (two-way) specifier uses. In the specifier use of BQ-constructions, the quantity assessment is backgrounded to the benefit of quality assessment. What’s more, the specifier use is generally used to categorize a third discourse entity. By way of illustration, in (9), the QN hatajo continues to single out a group of idiots, but its primary function is to categorize the N2-entities idiotas as a group of stupid and unpleasant people.54 In the given context, the entire BQconstruction is predicatively used to classify the we-participant (implicit in éramos). (9)

Para ella, no éramos más que un hatajo de idiotas. ‘To her, we were nothing more than a bunch (lit. herd) of idiots.’

In line with Di Tullio/Kornfeld (2008), I claim that both the quantifier and the specifier use are characterized by an evaluative potential. According to Di Tullio/Kornfeld, three types of pseudo-partitive constructions are to be distinguished depending on the function of N1. The first group contains expressions which by definition evoke the plurality of N2 such as un montón de dinero ‘a heap of money’, una cantidad de reclamos ‘a quantity of complaints’, un grupo || 54   Since the noun idiotas by definition refers to stupid beings, the specification by un hatajo de might seem partially redundant. In my view, the partial overlap or shared (connotative) meaning between N1 and N2 enhances the co-extensive relationship between both.

Lexical and functional uses | 71

de niños ‘a group of children’. Interestingly, the other types “ãnaden algún tipo de valoración or ponderación” (2008, 2; ‘add some sort of assessment or appraisal’ [translation KV]). The second type oscillates between a quantitative or qualitative assessment by the QNs which is attributed to N2 (like emphatic appositions, e.g. una calamidad de niño ‘a disaster of a child’) and contains examples such as una barbaridad de reclamos ‘a barbarity of complaints’ and una calamidad de robos ‘a calamity of thefts’. By contrast, in the third group which contains examples such as una montaña de dinero ‘a mountain of money’ and una oleada de robos ‘a wave of thefts’, the evaluative interpretation relates to quantity only. In sum, Di Tullio/Kornfeld (2008) consider the evaluative potential as intrinsically linked to quantity and quality assessment by BQ-constructions.55 In sum, I distinguish three major uses of BQ-construction, viz. head, quantifier and (two-way) specifier uses. Both functional uses present an evaluative potential, which is linked with their pragmatic function: in contrast to canonical quantifiers (e.g. mucho/a(s)) or modifying adjectives, the quantifying and specifying uses of BQs serve to express hyperbolic quantity or quality assessment. In addition, ambiguous and indeterminate uses can be distinguished. The following sections treat the different uses in this order and identify a series of syntagmatic and paradigmatic tests to differentiate them.

3.2.1 Head uses of binominal quantifiers In the head-uses, the QN activates its literal meaning as a full content word and the semantic (or functional) and syntactic head coincide in the QN. The QN evokes a physical and concrete configuration, constellation or group while the prepositional phrase heads the constituting entities. For instance, the QNs montón ‘heap’, pila ‘pile’ and racimo ‘bunch’ in (10)–(12) evoke a heap-like or cluster-like configuration which, as a whole, is discernible to the human eye. The de-phrase specifies which type of heap, of pile or of bunch the speaker refers to. The N2-entities are necessarily spatiotemporally contiguous. (10) Imagina un reloj de arena que va acumulando granitos hasta formar una pequeña montaña. Llega un momento en que el montón de arena no aguanta más, y un solo grano es suficiente para que se forme una avalancha. (1997, press) || 55 The first group of partitive constructions they distinguish is not to be equated with my interpretation of head uses though.

72 | Binominal quantifiers as a locus of grammaticalization

‘Imagine an hourglass which accumulates grains of sand to the extent of forming a little mountain. There comes a moment at which the heap of sand can’t stand any longer, and a single grain is enough to produce an avalanche.’ (11) El médico Manuel Muro, bata blanca y mirada franca, me habla desde el otro lado de una pila de radiografías amontonadas. (2000, books) ‘Doctor Manual Muro, white coat and an open expression, talks to me from the other side of a pile of X-rays heaped up.’ (12) Es bien conocida la anécdota con que se demuestra la perfección de un artista que pintó un racimo de uvas al que acudían los pájaros engañados por la exacta semejanza de lo pintado con lo vivo; (…).(2000, books) ‘Well-known is the anecdote used to demonstrate the perfection of an artist having painted a bunch of grapes which fooled birds came to because of the exact similarity between the painted and the real; (…).’ Typically, head uses allow for the substitution of the QN by a near-synonym. In (13a–b), montón and pila can easily be replaced by each other or by amontonamiento ‘stack, pile’ (nominalization of amontonar ‘to heap up’) or rimero ‘pile’ without making substantial changes to the clausal content. The substitution of racimo in (13c) is less straightforward, since ristra usually combines with onions or garlic to the extent of forming a lexicalized or fixed expression. Further, head uses allow the QN to stand alone, as in (14). In other words, it is the QN which provides the essential and necessary information to the clausal content. For the same reason, pronouns which anaphorically refer to the BQ-construction agree with N1, as in (15). The last paradigmatic test relates to the possibility to pluralize for QNs used literally, as in (16). The latter test also shows that N1 is a valid member of the noun-category. (13) a. Llega un momento en que el montón / la pila / el amontonamiento / el rimero de arena no aguanta más, (…) ‘the heap / the pile / the stack / the pile of’ b. (…) desde el otro lado de una pila / un montón / un amontonamiento / rimero de radiografías amontonadas. ‘a pile / a heap / a stack of’ c. (…) que pintó un racimo / ?una ristra de uvas, (…) ‘a bunch / string of grains’

Lexical and functional uses | 73

(14) a. El montón no aguanta más. ‘The pile can’t stand anymore.’ b. Me habla desde el otro lado de la pila. ‘He talks to me from the other side of the pile.’ c. Un artista pintó un racimo al que acudían los pájaros. ‘An artist painted a bunch which the birds came to.’ (15) a. El montón de arena no aguanta más. Dentro de poco, ya no lo veremos. ‘The pile can’t stand any more. Soon we can’t see it anymore.’ b. Me habla desde el otro lado de la pila de radiografías. Después la coloca en otro sitio. ‘He talks to me from the other side of the pile of X-rays. Afterwards he puts it somewhere else.’ c. Un artista pintó un racimo de uvas al que acudían los pájaros. ‘An artist painted a bunch which the birds came to.’ (16) a. Raras veces hay dos montones de arena seguidos. ‘Only rarely there are two successive heaps of sand.’ b. Me habla desde el otro lado de por lo menos tres pilas gigantescas de radiografías. ‘He talks to me from the other side of at least three giant piles of X-rays.’ c. Un artista pintó una serie de racimos de uvas a los que acudían los pájaros. ‘An artist painted a series of bunches of grapes which the birds came to.’ Syntagmatically, it is possible to specify typical spatial or configurational characteristics of the QN (such as height, width, etc., as in (17)), to use a superlative (18), and it makes sense to add to N2 modifiers which highlight those dimensional features (19). (17) Me habla desde el otro lado de la pila de libros de un metro de alto. ‘He talks to me from the other side of the one-meter pile of books.’ (18) Mi colega lleva la pila de libros más alta. ‘My colleague is carrying the highest pile of books.’ (19) Me habla desde el otro lado de la pila de libros apilados/ amontonados / ordenados en orden alfabético. ‘He talks to me from the other side of the pile of books piled up / heaped up / ordered alphabetically.’ The QN is automatically interpreted as a head when it represents a locative component of some motion event (20) or when the configuration of an object is integrated in a topological description (21). In other words, the co-occurrence

74 | Binominal quantifiers as a locus of grammaticalization

with movement verbs or locative prepositions are typical contextual cues indicating the literal interpretation of N1. In addition, if the BQ is the clausal subject, the verb will agree with N1. (20) Se cayó sobre / se movió hacia el montón de paja. ‘He fell on / he moved towards the heap of straw.’ (21) El ancho tejado negruzco baja en pendiente rápida; el alero sombrea el dintel de la puerta. Dentro, el piso está empedrado de menudos guijarros. En un ángulo hay un montón de leña; apoyadas en la pared yacen la horquilla, la escoba y la pala de rabera desmesurada. (1903, books) ‘The wide blackish roof comes down in a steep slope; the eave overshadows the door’s threshold. Inside, the apartment is paved with little pebbles. In one corner there is a woodpile (lit. heap of firewood); the pitchfork, the broom and the disproportionate shovel are leaning against the wall.’ Finally, N1 and N2 are usually, yet not necessarily, coextensive. In other words, in the case of head uses, a larger (reference) mass of all (or at least more) possible N2-instances may be profiled. Recall the possibility of literal QNs to pluralize. Crucially, the coextensive relationship is not profiled in the head use. Yet without additional contextual cues, the coextensiveness between N1 and N2 can give rise to ambiguity: in the utterance Te traigo una pila de libros ‘I am bringing you a pile/lot of books’, the BQ can evoke both a literal pile (which consists of books) or a lot of books (which might simply be put in a shoulder bag, for instance).

3.2.2 Quantifier uses of binominal quantifiers The QN can also fulfill a quantifying function within the binominal syntagm. Via metaphorization, the number or size of the N2-entities is assessed by means of the conceptual image evoked by N1. Like in the case of relative quantifiers (Langacker 1991), where the number of the profiled instance is assessed by implicit reference to a ‘reference mass’, quantity assessment in BQs is achieved by a comparison construal, not to a reference mass including all possible instances of N2, but to the specific configuration or constellation typically associated with N1. In (22), the number of calls is assessed by comparing the number of N2 to the volume or dimensions associated with alud as a lexical item.

Lexical and functional uses | 75

(22) La sala del 091, la Guardia Urbana, los servicios de Protección Civil de Barcelona y las redacciones de los diarios ("¿qué ha pasado?") empezaron a recibir un alud de llamadas de ciudadanos preocupados por el intenso temblor que durante casi cinco segundos notaron bajo sus pies. (1995, press) ‘The 091 call center, the urban guard, the Barcelone Civil Protection services and the newspaper editorial offices (“What happened?”) started receiving a snow slide of calls from citizens worried by the intense quaking they felt under their feet during almost five seconds.’ The quantity assessment of N2 in terms of N1 (e.g. (22) and (23)), is intrinsically linked to the coextensive relationship between both nominal elements. The coextension is prerequisite to the quantifying reading since it allows for the metonymic shift in profile from foregrounding N1 (as in the head reading) to foregrounding N2. N2 indeed constitutes the essential information in terms of clausal content, whereas N1 is nothing but a quantifier and can therefore be omitted or replaced by other quantifiers. (23) La campaña internacional contra el Régimen español aumenta por días. Si coleccionara caricaturas de Franco, con su barriguita y su fajín de general, podría llenar un montón de álbumes. (1986, books) ‘The international campaign against the Spanish government is gaining force every day. If I were to collect caricatures of Franco, with his belly and his general sash, I could fill a heap of albums.’ The most important paradigmatic test revealing a quantifying interpretation is the substitution by other quantifiers such as mucho/a(s) (‘a lot’), numbers or the strictly quantifying and desemanticized gran cantidad de ‘a large number (lit. quantity) of’, as in (24).56 Further, N2 can stand alone, while N1 is necessarily complemented, as in (25). If anaphorical reference is made to the BQ, the anaphoric element agrees with N2, as in (26). Finally, although pluralization is possible with montón de, it is usually highly marked in quantifying uses and evokes a series of distinct sets of N2-entities, as in (27). (24) a. Recibieron muchas / gran cantidad de / miles de llamadas. ‘They received many / a large number / a great lot / thousands of calls.’

|| 56 QNs which present a rich source semantics can also be replaced by near-synonymous QNs. For instance, similar quantity assessment would be achieved by the alternatives alud ‘avalanche’, aluvión ‘flood’ and avalancha ‘avalanche’ in un N1 de llamadas ‘a N1 of calls’.

76 | Binominal quantifiers as a locus of grammaticalization

b. Podría llenar muchos / gran cantidad de / quinientos álbumes. ‘I could fill many / a large number / great deal of / five hundred albums.’ (25) a. Recibieron llamadas. ‘They received calls.’ *Recibieron un alud. ‘*The received a snowslide.’ b. Podría llenar álbumes. ‘I could fill albums.’ *Podría llenar un montón. ‘*I could fill a heap.’ (26) a. Empezaron a recibirlas a partir de las dos. ‘They started receiving them at two o’ clock.’ b. Te los enseñaría con mucho gusto. ‘I would love to show them to you.’ (27) a. ?Recibieron aludes de llamadas. ‘?They received avalanches of calls.’ A lo largo de los años recibieron (varios) aludes de llamadas. ‘Throughout the years, they received (several) avalanches of calls.’ b. Podría llenar montones de álbumes. ‘I could fill heaps of albums.’ Syntagmatically, it is not possible to add adjectives to N1, apart from intensifying ones, as in (28). Further, the BQ easily answers the question ¿cúanto/a(s)? ‘how many?’, as in (29). (28) a. Recibieron un alud *acuoso / *local de llamadas. ‘They received a *watery / *local flood of calls.’ b. Recibieron un gran / verdadero alud de llamadas. ‘They received a huge / real flood of calls.’ (29) – ¿Cuántos dossieres se llevó Roldán? ¿Se han recuperado? R. – Se llevó un buen montón de dossieres. El contenido no lo sé, pero dejó muchas carpetas vacías en la Dirección General de la Guardia Civil. (1996, press) ‘– How many files did Roldán take away? Have they been recovered? R. He took a good heap of files. I am not aware of the content, but he left many empty folders at the Head office of the Civil Guard.’ In addition, several types of contextual cues can testify to a quantifying interpretation, such as verb agreement with N2 (30), the choice of an animate or abstract N2 incompatible with the head reading (31), the combination with contextual elements making implicit reference to a plurality. In addition, if the N2entities are construed as spatiotemporally not contiguous, the head reading is ruled out in advance (32).

Lexical and functional uses | 77

(30) En arte se han justificado un montón de tonterías con el humanismo. (1997, books) ‘In art, a lot of silly things with humanity have been justified.’ (31) Ninguna cara conocida entre el aluvión de gente que deambulaba junto a él. (1984, books) ‘Not a single known face among the flood of people that was wandering around next to him.’ (32) ¿Pero qué he hecho de mi vida? He tenido tres hijos y me quedan dos; he tenido un montón de mujeres y me quedan casi todas; he conseguido una inmensa fortuna... (1990, books) ‘But what have I made from my life? I have had three sons and I have two left, I have had a lot of women and I have almost all of them left, I amassed an immense fortune…’ As the quantifying use initially arises as an invited inference in the context of speaker-hearer negotiation, it does not come as a surprise that BQs have an evaluative potential. In (33), several contextual clues indicate that the quantity judgment in el alud de proyectos activates this potential by evaluating every single project as useless. (33) También se ha hecho notar cómo, gracias a estas reproducciones de calidad, podemos hoy tener constancia de algunas grandes obras cuyos originales desaparecieron sobre todo durante la Segunda Guerra Mundial, pero entre el alud de proyectos culturales con el que nos amenazan los políticos culturizadores – mediatecas, ludotecas, videotecas y otras vaguedades que ndie ha pedido –, no he encontrado ni uno solo que prevea volver a exponer estas colecciones que hace apenas cincuenta años formaban parte del patrimonio artístico de un país. (1998, books) ‘It has also been noticed how, thanks to those high quality reproductions, we now have record of some great works of art whose originals have disappeared, mainly during World War II, but among the avalanche of cultural projects that politicians committed to the diffusion of culture are threatening us with – multimedia centers, toy libraries, video stores and other vague promises no one has asked for –, I did not find a single one that plans to exhibit those collections again which only fifty years ago still formed part of a country’s artistic heritage.’

78 | Binominal quantifiers as a locus of grammaticalization

Although the substitution test suggests that quantifying N1s have entered the paradigm of (absolute) quantifiers, two important differences are to be noted. First, BQs differ from canonical quantifiers in the way they access the N2entities: while muchas llamadas merely designates a plurality of N2 (cf. Figure 1a), un alud de llamadas bonds or unitizes the plurality into a single set (cf. Figure 1b). The unitizing function of BQs does not entail that the N2-referents are necessarily spatiotemporally contiguous (as is the case in the head uses), but allows to conceptualize a set of N2s at once.



(a) absolute quantifier

(b) binominal quantifier

Fig. 1: Quantity assessment by absolute quantifiers (a) and by BQs (b)

The second major difference resides in the addition of a qualifying component to pure quantity assessment in the case of BQs. Since the number of entities is assessed metaphorically by comparing the set of N2 to the typical constellation of N1, it does not come as a surprise that several matching properties between N1 and N2 continue to be highlighted (e.g. the infinite nature of the sea, its watery substance and horizontal extension in (34)–(35)). The latter phenomenon is referred to as (conceptual) persistence in the remainder of this study. In fact, quantifying instances displaying a high degree of persistence might raise the question whether they are not rather to be considered a metaphorical extension of a QN which is used literally. In examples (34)–(35), mar could easily be replaced by océano ‘ocean’ or piélago ‘ocean’. However, N2 constitutes the essential information of the clause or the semantic head of the construction: mar de can easily be left out without losing crucial pieces of information to interpret the sentence, while the question ¿Qué mar? ‘Which sea?’ is odd. (34) Mi corazón palpita todavía al recuerdo de aquellas horas en que flotaba sobre un mar de eternas delicias. (1918, books) ‘My heart still beats by the memory of those hours in which I was floating in/over a sea of eternal delights.’ (35) Y la belleza de este cuadro maravilloso se acrecienta por el mar de luz que por todas partes baña a la ciudad. (1929, books)

Lexical and functional uses | 79

‘The beauty of this painting grows rises by the sea of light that bathes the city from all sides.’

3.2.3 Two-way specifier uses of binomianl quantifiers The QN can also adopt a specifying function: N1 still bonds or singles out a set of N2-entities, but quality assessment now prevails, to the extent that quantity or size assessment remains unprofiled. Instead of merely bounding a set of N2entities, N1 emphasizes their bondedness in terms of a specific quality associated to N1. In other words, in the specifier use as well, the coextension between N1 and N2 involves a comparison construal. Its function can be compared to that of pre-modifying or post-modifying adjectives or of specifying relative clauses. In (36), for instance, the exact number of kids (N2) addressed by the speaker is relegated to the background, while their categorization as a bunch stands out: they are profiled as a group of idiots incapable of making their own decisions. The characteristic of N2 highlighted by N1 is motivated by properties that are typically associated with the literal use of the QN. In contrast to the quantifying use, however, the N1 properties that persist in the specifying use do not necessarily relate to (spatial) configuration: hatajo ‘herd, bunch’ in (36) evokes the gregarious behavior of adolescents, racimo ‘grape, bunch’ in (37) evokes the clustering of the gang, letanía ‘litany’ in (38) evokes endless repetition. (36) – Dos cosas, por lo menos, debíais aprender de este hecho, hatajo de cabritos. La primera [...] (1986, novel) ‘Two things, at least, you ought to learn from this event, bunch of kids (lit.: small goats; euph.: swines). The first one [...]’ (37) No pierdas tiempo en maldiciones. Enfílame a ese racimo de bandidos y pon tu pincho al servicio del rey. (1976, theatre) ‘Don’t waste time in curses. Get me that bunch of crooks and put your gun at the king’s service.’ (38) ¿Por qué no me matas, Fernando, y acabamos de una vez con esta letanía de angustias y de lágrimas? (1990, theatre) ‘Why don’t you kill me, Fernando, and then we end once and for all this litany of fears and of tears?’ Crucially, the specifier use of the BQ-construction generally evokes a third, external entity X. In addition to grouping N2 in terms of N1, the entire binominal string categorizes this external entity. In other words, two type specifications (of N2 and of X) are evoked. I therefore call this functional use the (two-way) speci-

80 | Binominal quantifiers as a locus of grammaticalization

fier use.57 This use, which is typically predicative, fits expressions such as no ser más que un N1 de N2 ‘to be nothing but/just a N1 of N2’, convertir(se) en un N1 de N2 ‘to turn into a N1 of N2’ or ser un N1 de N2 ‘to be a N1 of N2’, e.g. (39)–(43). These copular and pseudocopular expressions indicate that the subject, the external entity X, is fully assimilated to the entity or category profiled by un N1 de N2 and thus coincides with it. As such, the kind of entity profiled by the BQ conveys the very essence of the entity X. In other words, (two-way) specifier uses usually involve two comparison construals: while N2 is specified by means of N1, a third entity X is depicted as an instance of the category evoked by the BQ. By way of illustration, in (39), the essence of a keyboard is reduced to a heap of buttons; in (40), classical music is evaluated as being nothing more nor less than noises grouped together, etc. (39) El teclado es básicamente un montón de botones que se resisten a ser manipulados coordinadamente. (1993, books) ‘The keyboard is basically a heap of buttons that resist to be manipulated in a coordinated manner.’ (40) Cuando en mi casa sacábamos el tema de la música, mi madre decía que lo que llaman música clásica no era más que un montón de ruidos insoportables capaces de volver loca a cualquiera. (1982, novel) ‘When at our place we touched the subject of music, my mother used to say that what is called classical music is nothing more than a heap of unbearable noises which are capable to turn anybody mad.’ (41) En caso contrario, en lugar de mejorar su equipo conseguirá bloquearlo y convertirlo en un simple montón de hardware. (1997, books) ‘In the opposite case, instead of improving his equipment he will get it blocked and turn it into a simple heap of hardware.’ (42)Ahora las escenas de la tarde pierden su condición de simples hechos y se transforman en un aluvión de significaciones. (1987, novel) ‘Now the scenes from the afternoon are losing their status of simple facts and change into a flood of significances.’ (43) Entré en casa hecho un mar de lágrimas y conté a mis tías, sofocado por la ira, el atentado de que acababa de ser víctima. (1921, novel) ‘I came in at home turned into a sea of tears and I told my aunts, stifled by rage, the attack of which I had just been victim.’

|| 57 For reasons of space, two-way will generally be left implicit in the remainder of this volume.

Lexical and functional uses | 81

The modification involved in binominal quantifiers differs from the more habitual modification construal conveyed by adjectives (Figure 2a). If un aluvión de significaciones in (42) would be substituted by significaciones inesperadas ‘unexpected meanings’ or significaciones desordenadas ‘unordered meanings’, the quantity assessment would be lost and only the facet lexicalized in the adjective would be profiled as a characteristic of the N2 significaciones and of the entity-X las escenas de la tarde. In contrast to expressive binominal constructions, e.g. una maravilla de mujer ‘a wonder of a (lit. Ø) woman’ (note that N2 is not a multiplex referent) which are exclusively used for quality assessment (Figure 2c), the (two-way) specifier use of binominal quantifiers combines quantity and quality assessment, minimally as some kind of homogeneization of all N2s in terms of an umbrella-characteristic generally associated with N1. In contrast to regular quality assessment by adjectives, binominal constructions present the quality associated to N2 as its very essence (Figure 2b).

X

X

(a) adjective

(b) binominal quantifier

X

(c) expressive binominal

Fig. 2: Quality assessment by adjectives (a), BQs (b) and expressive binominals (c)58

In order to distinguish the specifying use from the ‘pure’ quantifier uses, several syntagmatic and paradigmatic tests can be used. As to the substitution test, N1 can easily be substituted by either no/nada más que ‘nothing more than’ (44), a nominal syntagm containing N2 and an adjective highlighting the property related to N1 (45) or by a nominal syntagm containing a demonstrative and N2 only (implying at the discourse-pragmatic level that the hearer knows which category of N2 the speaker is referring to, as in (46)). Further, N2 can stand alone while N1 is necessarily complemented, as in (47). If anaphorical reference

|| 58 The illustration is in line with Langacker’s (2011) symbolization of ‘co-extensiveness’, viz. the two lines that relate the circles. The schemas also highlight the linearity principle: in (a), the entity is mentioned first and the adjective is presented as one facet (out of a set of possible facets) that characterizes X; yet by virtue of co-extensiveness, the characteristic is mentioned first in the binominal construction ((b) and (c)) and as such presented as the very essence of X.

82 | Binominal quantifiers as a locus of grammaticalization

is made to the BQ, the anaphoric element agrees with N2, as in (48). Finally, pluralization is highly marked (and not observed in my corpus). (44) a. Sois nada más que cabritos. ‘You are nothing but small goats.’ b. El teclado no es más que botones. ‘The keyboard is nothing more than buttons.’ (45) a. (…), debíais aprender de este hecho, cabritos idiotas. ‘(…), you ought to learn from this event, stupid small goats.’ b. El teclado básicamente son botones pequeñísimos. ‘The keyboard is nothing more than tiny buttons.’ (46) a. Enfílame a esos bandidos. ‘Get me those crooks.’ b. (…) lo que llaman música clásica no era más que esos ruidos. ‘What is called classical music is nothing more than those noises.’ (47) (…) lo que llaman música clásica no era más que ruidos. / * (…) lo que llaman música clásica no era más que un montón. (48) Enfílame a ese racimo de bandidos y sujétales. ‘Get me that bunch of crooks and tie them up.’ Syntagmatically, the specifier use allows both N1 and N2 to combine with adjectives highlighting the relevant facet of N1, as in (49). Further, specifier uses can be paraphrased by means of a (pseudo)copular verb classifying a specific entity as an instance of the category profiled by the BQ-construction, as in (50). (49) a. (…) debíais aprender de este hecho, estúpido hatajo de cabritos / hatajo de cabritos estúpidos. ‘(…) you ought to learn from that, you stupid bunch of swines / you bunch of stupid swines.’ b. Acabamos de una vez con esta cansina letanía de angustias / letanía de angustias cansinas. ‘We end at once this weary litany of fears / litany of weary fears.’ (50) a. Sois un hatajo de cabritos. ‘You are a bunch of swines.’ b. Nuestra relación se convirtió en una letanía de lágrimas. ‘Our relation has turned into a litany of tears.’ As the (two-way) specifier use always involves some kind of categorization (of N2 and another entity), it often additionally yields an evaluation by the speaker. In (36)–(38), it enhances the negative orientation already present in the N2 entities. However, as can be seen in (51), this evaluative potential is not always activated. In view of this categorization, specifier uses frequently show up in the context of exclamative qué, demonstratives (37), as the second, appositional

Lexical and functional uses | 83

element of a reformulation (52), as apostrophes (36), as predicate complement with copulas (39)–(41) and (51), etc. (51) Y Virginia, acto seguido, puso ejemplos, contó anécdotas, citó nombres de amigas y enemigas que habían leído a Martín: era un alud de pompas de jabón que aparecían, brillaban al sol, se deshacían y reaparecían aún más grandes. (1990, novel) ‘And Virginia, immediately afterwards, gave examples, told anecdotes, cited names of friends and enemies who had read Martín: it was an avalanche of soap bubbles appearing, sparkling in the sun, bursting and reappearing even larger.’ (52) Las palabras que siguieron luego o, mejor, el alud de frases deslavazadas [sic] que Eduardo escupió literalmente sobre mi rostro, me adentraron en una realidad sorprendente. (1980, novel) ‘The words that followed later on or rather, the avalanche of incoherent sentences Eduardo literally spat in my face, forced me into an astonishing reality.’

3.2.4 Ambivalent uses The application of the above listed paradigmatic and syntagmatic criteria does not always yield clear-cut distinctions. Among the remaining ambivalent cases I further subdivide between ambiguous and indeterminate uses, building on the distinction introduced by Willemse (2007: 562)59 and applied to binominal constructions in English by Brems (2011). Instances are classified as ambiguous if two distinct readings are equally plausible, “relying on two distinct ways of contextualization” (Brems 2011, 130). By contrast, indeterminate instances “are irreducible blends, combining elements from two functions in metaphorical contexts, but do not allow disambiguation” (Brems 2011, 130). Interestingly, ambiguity and indeterminacy can hold between the head reading and both types of functional readings as well as between both functional readings. By way of illustration, example (53) counts as ambiguous between head and quantifier reading, since from the context it cannot be derived whether the files were neatly piled up, or whether the speaker just meant that he was working on different cases at the same time (with the files spread out on his desk as well as on a cabinet or on the floor). In (54), montón can be interpreted as a head or a

|| 59 Willemse (2007) and Brems (2011) use the notion of ‘vague’ instead of ‘indeterminate’.

84 | Binominal quantifiers as a locus of grammaticalization

specifier, depending on whether the speaker wishes to profile the actual configuration as a literal heap of sand or wishes to stress the idea that the shapeless heap of bones and ashes is all there is left of the dead poet. In (55), the sea of confusions can be interpreted both as an immense number of confusions and as nothing but confusions (which is somehow the limiting case of large quantity). The difference is subtle, however, and already indicates that the borderlines between type specification and size specification by BQs are thin. (53) Una noche estaba en mi oficina de Managua, trabajando en una pila de casos de violación de los derechos humanos y pensé (…). (1996, press) ‘One night I was in my office in Managua, working on a pile of cases of human rights violation and I thought (…).’ (54) Cuando años más tarde se exhumaron los restos de Espronceda para trasladarlos a otro cementerio, yo estaba presente. Los restos eran un montón informe de huesos, cenizas y arrapiezos. (1941, books) ‘When (many) years later the mortal remains of Espronceda were exhumed to move them to another cemetery, I was there. The remains were a shapeless heap of bones, ashes and shreds.’ (55) – Yo no sé qué deciros, ni qué hacer en el trance en que me veo. Lo que a mí me pasa, es para volverme loco. ¡Qué mar de confusiones! (1876, books) ‘I don’t know what to tell to you, nor what to do in the state of trance I find myself in. What is happening to me is driving me crazy. What a sea of confusions.’ By contrast, expressions are referred to as indeterminate when their overall meaning blends two different uses. In (56), the verbal predicates emerge and se amontona support the literal sense of ‘pile’, while the abstract and unspecified nature of the things piled up (lo que nos sobra and lo que nos falta) enhances the quantifying potential of una pila de despojos. In (57), the abstract subject quince años and the past participle convertidos trigger the specifying use of montón, while the relative clause and the verb retiró make implicit reference to a tangible constellation or literal heap of ashes. In (58), the appearance in an appositive reformulation and the demonstrative suggest that aquel mar de silencio is predicatively used to categorize the entity las inmensas vacaciones. At the same time, the adjective inmensas, the complement sin fin and the scenery evoked indicate that the vacation covers a long period.

Lexical and functional uses | 85

(56) Miami, esa ciudad con nombre de cafetería, digo, y de chucho lanudo o niña pija, que viene a ser lo mismo, emerge como una pila de despojos donde se amontona por igual lo que nos sobra y lo que nos falta. (1994, press) ‘Miami, that city with the name of a pub, I would say, and of a hairy dog or a posh girl, which comes down to the same, emerges as a pile/ lot of waste, where what we have left over and what we lack is piled up in equal piles.’ (57) Todo ardió en la chimenea en tan poco tiempo que a Cecilia le recorrió un escalofrío; quince años consumidos en segundos, convertidos en un montón de cenizas que Paulina retiró en un santiamén. (1993, novel) ‘Everything got burned in the chimney in so little time that a shiver ran down Cecilia’s spine; fifteen years consumed in (a few) seconds (only), turned into a heap of ashes that Paulina removed in an instant.’ (58) Las inmensas vacaciones – quince días –, aquel mar de silencio para nosotras, particular, aquella arena sin fin de las horas no muertas: pasadas entre nosotras como si una y otra formásemos las dos probetas de un reloj del tiempo, o como si le dejásemos escurrir entre los dedos y nos gustara así. (1965, novel) ‘The immense vacations – fifteen days –, that sea of silence, extraordinary for us, that sand of not wasted time without borders: spent among us as if together we formed the two tubes of a sandglass, or as if we let it slip through our fingers and liked it that way.’ In the remainder of this book, the cover terms ambiguity and indeterminacy generalize over the threefold (sub)distinction among the ambiguous and indeterminate uses. They are not separately visualized in the frequency tables. One remarkable tendency bears pointing out however: the indeterminate occurrences observed in my corpus data usually correspond to indeterminacy between both functional readings, which suggests that the distinction between quantifier and specifier is less clear-cut than the distinction between functional and lexical uses.

86 | Binominal quantifiers as a locus of grammaticalization

3.2.5 Concluding remarks In the previous sections, a crucial role was attributed to coextension between N1 and N2. I have argued that the head, the quantifier and the (two-way) specifier use all construe this coextensive relationship in a proper way. The illustration in Figure 3 builds on Langacker’s (2011) model of BQs (cf. supra) and visualizes the three realizations of coextensiveness between pila and N2.

a) head









b) quantifier

c) premodifier

X

Fig. 3: Three realizations of coextensiveness between pila and N2

Lexical and functional uses | 87

For the sake of simplicity, I generalize over the individual semantic contribution and the functional status of the determiner and preposition de. As to the headreading in Figure 3a, the N1 pile profiles a specific configuration, while the N2 profiles a plurality of instances of the type expressed by N2. The bonded mass expressed by pile corresponds (dotted lines) to the bonded plurality profiled by de N2. At the composite structure level, the conceptualizations of both nominal elements are mapped onto each other: the mass profiled by pile is coextensive (double line) with the mass profiled by the N2, yet the typical vertically oriented configuration is foregrounded (thick line). As to the functional uses, the illustration is limited to the composite structure level. In the quantifying uses (see Figure 3b), the N2s are bonded or unitized by their coextension with (or in terms of) N1, but their plurality is foregrounded. In the (two-way) specifier use (see Figure 3c), the bonding of the N2s in the light of a specific facet of N1 prevails over the plurality of N2. Crucially, the specifying BQ tends to evoke a specific category and serves to classify a third entity X. Table 1 gives an overview of the paradigmatic and syntagmatic tests differentiating between head, quantifier and specifier uses. Recall that the contextual cues individually are not to be considered as conclusive for a particular use. Tab. 1: Paradigmatic and syntagmatic tests to distinguish head, quantifier and specifier Test

Head

Quantifier

(Two-way) Specifier

Anaphoricity

Reference to N1

Reference to N2

Reference to N2

Pluralization

+

(+/) –

–

Substitution

Near-synonymous N1

Mucho/a(s) (/near-synonymous N1) N1 de

N1 de

Omission Agreement

de N2 N1 only

N2(/N1)

N2(/N1)

Spatiotemp. contiguity of N2

Required

Possible

Possible

Coextension

Unmarked

Required

Required

Movement Vs (e.g. sentarse en ‘to sit on’), locative prepositions (e.g. encima de ‘above), ADJs and complements referring to the configuration (e.g. medio ‘half’, de un metro de alto ‘one-meter’, etc.)

N2s incompatible with head-use; Vs and ADJs inviting for quantifying or intensifying reading of N1 (e.g. tardar ‘to wait’, inmenso ‘immense’, verdadero ‘real’)

Contexts inviting for categorization: reformulations, apostrophes, (pseudo) copular Vs, exclamative ¡qué!, demonstrative to N1, etc. (e.g. estar/ser reducido a ‘to be reduced to’, no ser más que ‘to be no more than’, etc.)

Contextual cues

88 | Binominal quantifiers as a locus of grammaticalization

Table 2 presents the distribution of uses attested in the corpus. In line with Brems, I approach the variety of uses in terms of synchronic layering or as “the co-existing synchronic sediments of diachronic processes of change” (Brems 2011, 2). Tab. 2: Synchronic distribution of uses per QN Head Alud Aluvión Barbaridad Hatajo Letanía Mogollón Montón Pila Racimo

16 0.19 5 0.03 0 0 1 0.06 4 0.015 0 0 114 0.23 47 0.59 25 0.52

Quantifier 62 124 9 6 17 44 311 16 21

0.72 0.83 1 0.38 0.65 0.96 0.62 0.20 0.44

Specifier 3 8 0 7 2 1 29 1 2

0.03 0.05 0 0.44 0.08 0.02 0.06 0.01 0.04

Ambiguous 0 1 0 0 1 0 33 10 0

0 0.01 0 0 0.04 0 0.07 0.13 0

5 12 0 2 2 1 13 6 0

Indet.

Total

0.06 0.08 0 0.13 0.08 0.02 0.03 0.08 0

86 150 9 16 26 46 500 80 48

As to the diachronic processes of change involved, Brems (2007b; 2011) claims that two paths of GR are to be distinguished, one leading to pure quantifying uses and one resulting in valuing quantifier uses. On the basis of the functional analysis of a synchronic dataset of English BQs, she suggests that both pathways of GR have a common starting point, viz. the head-use, since they are defined as “one leading from head status to quantifier status and the other moving from head to a more evaluative quantifier” (2007b, 127). Especially in the case of bunch, quantifying and valuing quantifying uses seem to coexist “rather than constituting two truly separate and chronologically consecutive paths” (2007b, 127). By contrast, in the light of the case-study on small size noun-constructions (2007a) and a diachronic account (2010), she suggests that the development of quantifying uses is a precondition for the valuing quantifier uses to arise. The valuing quantifier is then considered as a highly subjectified quantifier use and operationalized in terms of a re-clustering of the collocational pattern – in the case of bunch of to animate or abstract negative collocates (e.g. morons, lies). Whereas the quantifier uses are primarily characterized by a generalization of the lexical semantics and the extension to increasingly more contexts, the valuing quantifier uses are identified as the reclustering (after the initial extension) of the N2-combinations to negatively evaluated abstract and animate nouns (Brems 2010, 97; Brems 2004, 303). The author strikes a sensitive chord indeed: although the primacy of the quantifier uses seems a sound explanation – both semantically and functionally – this directionality claim cannot be maintained

The grammaticalization of binominal quantifier constructions | 89

in the light of diachronic data – at least as far as Spanish is concerned. I will return to this issue in Chapter 5.60

3.3 The grammaticalization of binominal quantifier constructions BQs are cross-linguistically considered as a locus of (ongoing) GR. While the majority of the case-studies focus on the English BQ (Brems 2007a; 2010; 2011; Traugott 2007; 2008a; 2008b; Trousdale 2008b; 2010), two analyses identify the quantifying uses of Spanish BQ-constructions as the outcome of GR: Bosque (2007) and Di Tullio/Kornfeld (2008). The latter analysis suggests that the GR of BQs results in semi-fixed expressions: Un proceso de gramaticalización particular se advierte en los sustantivos que se integran en construcciones pseudopartitivas, en las que van precedidos de un determinante indefinido y seguidos por la preposición ‘por defecto’ de. (2008, 1)61

However, to my knowledge, no diachronic (corpus) analysis of Spanish BQs has been carried out up until now. Before turning to the case-studies in the remaining chapters, this section outlines the underlying conceptual motivation of GR in BQs (3.3.1), identifies the Spanish BQ-construction as a locus of GR (3.3.2) and formulates some working hypotheses (3.3.3).

3.3.1 The underlying conceptual motivation of GR in BQs In an attempt to explain the evolution of expressions such as a lot of X from the source construction [ [aart lotn] [ofp Xnml] ] into the outcome of GR [alottaqnt Xn],

|| 60 In the case of QNs which present a rich lexical source semantics such as letanía, alud, aluvión, mar, etc., one might even wonder whether it is useful to posit the primary existence of a productive literal use. It is for instance hard to believe that, at first, speakers used to specify the nature of the constituting entities of literal litanies, since the discourse or recited nature of N2 is inherent in the definition of letanía. Likewise, un alud de nieve ‘an avalanche of snow’ and un mar de agua ‘a sea of water’ are highly redundant. It is more plausible that letanía is first used in binominal syntagms when it combines with atypical N2s, viz. when it is used as a quantifier. 61 A particular process of grammaticalization can be noticed with the nouns which are integrated in pseudopartitive constructions. In these constructions they are preceded by an indefinite determiner and followed by the “default” preposition de. (translation KV)

90 | Binominal quantifiers as a locus of grammaticalization

Langacker (2009a) attributes a crucial role to metonymy and the coextensive relation between N1 and N2. As has already been outlined in Section 2.1.2.3, four types of QNs are distinguished depending on the degree of coextension they express. Further, these four types of coextension are argued to reflect the direction taken by the process of GR (Langacker 2009a, 66). Initially, QNs can refer to containers, configurations or groups (cf. Figure 4a): while they may have a typical size, size does not pertain to their central semantic specification. However, as the QN profiles a bounded entity, “it serves a unitizing function with respect to a mass by delimiting the portion that is coextensive with it” (Langacker 2009a, 66), which explains its quantifying potential, “where the size of an instance is measured in terms of the units thus defined” (Langacker 2009a, 66). Furthermore, via a metonymic shift in profile, the QN “might be used for a unit of mass whose association with the container, configuration or group is merely recalled or imagined” (Langacker 2009a, 66, cf. Figure 4b). Another shift in profile induces the measure sense (cf. Figure 4c), “the association with the container, configuration of group receding into the background” (Langacker 2009a, 66). Eventually, this association may even disappear completely (cf. Figure 4d). (a)

(b)

(c)

(d)

 Fig. 4: Chain of metonymic shifts in coextensiveness (according to Langacker 2009a, 66)

The English QN lot seems to approximate this last stage of the measureinterpretation which does not recall the images of a configuration or group any more. After all, lot does not have a referent existing in the physical world. At this point, the role of the coextensive relationship becomes clear. QNs such as lot are coextensive with the N2 as they “are mentally superimposed on this entity for purposes of measurement, and the extent of the measured entity maps onto the extent of the quantifying unit on the relevant scale” (Langacker 2011, 16). This coextension-relationship allows for the obligatory metonymic shift in profile: from individual entity to spatially separated mass or N2-entities. While in a flock of geese, one can focus both on the flock (unit) and on the geese, a lot

The grammaticalization of binominal quantifier constructions | 91

of geese necessarily focuses on N2 (since lot has no real-world referent). At the same time, the coextensive-reading of lot dovetails with the central contribution of the preposition of: at the composite structure level, the relationship of coextension profiled by the preposition and the one profiled by lot “collapse into one: they are two encodings of the same relationship” (Langacker 2009a, 71). The resulting tighter integration and the conceptual overlap, which is characteristic of highly grammaticalized expressions, suggest that the GR process of the expression a lot of X is well on its way to the eventual reanalysis as a monomorphemic alotta. However, as the construction a lot of X still reflects a high degree of analyzability and the final stage of ‘pure’ measure noun has only been reached by lot so far, the English BQs and a lot of in particular cannot be considered as completely grammaticalized yet. A few intermediate steps have to be taken still: the article and the preposition62 must lose their status as separate morphemes; the components a, lot and of, must be considered as a whole or single chunk and finally, the N2 has to be reanalyzed as a simple noun instead of a nominal (Langacker 2009a, 77–80). In short, the construction as a whole should lose its analyzability. Yet ample structural evidence indicates that this process has not yet run its course completely. First of all, a lot can be used anaphorically or adverbially (e.g. We ate a lot/*alotta). Second, the fact that some intensifying adjectives can be put between the article and lot suggests the persistence of the source construction (e.g. an awful lot of, a whole lot of, etc.). Moreover, the possibility of employing the plural lots presupposes the status of count noun (e.g. lots of tons). Further, N2 can be definite (e.g. a lot of them, a lot of that cherry-flowered whisky). Another structural argument concerns the possibility of preposing the prepositional phrase (e.g. Of that I have a lot. That’s something of which I have a lot.// That I have a lot of. That’s something I have a lot of.). Finally, from a semantic perspective, examples can be found where lot seems to retain its original meaning of group or collection (e.g. They’re a sorry lot. We should get rid of the lot of them). For all these reasons, Langacker (2009a, 62) states that the source construction [ [aart lotn] [ofp Xnml] ] is still the primary analysis, or is at least still accessible, while the more grammaticalized construction is emerging.

|| 62 While the article is still slightly meaningful (it designates positive increment but does not fulfil a referential function anymore), the preposition of does not contribute any semantic content that is not already profiled by lot. As a consequence, they are both susceptible to phonetic erosion.

92 | Binominal quantifiers as a locus of grammaticalization

As to the Spanish BQ-construction, I will argue in this volume that precisely this lexical (or rather, conceptual) persistence which characterizes the Spanish construction motivates the formal persistence and prevents the BQs from being morphosyntactically unanalyzable. They correspond to the intermediate stage of GR identified by Langacker, where “a lot of is still internally analyzable (…) but forms a constituent which, as a composite whole, functions as the quantifier” (Langacker 2009a, 77–78).

3.3.2 The Spanish BQ as a locus of GR: towards the case-studies The motivation or potential for change of BQs – as described in the previous section – is not to be equated with change as such. One of the main conclusions of this book is that the development and synchronic organization of Spanish BQ-constructions is characterized by a high degree of formal as well as semantic persistence. Consequently, except for some rare instances of montón, Spanish BQs are not even close to the last stage of the measure-interpretation, as defined by Langacker (2009a). Nevertheless, for the reasons highlighted in the present section, I take the GR of Spanish QNs and the BQ-construction for granted. In line with the loss-and-gain model of GR (Hopper/Traugott 2003; Heine/ Narrog 2010), the development of BQs has clearly resulted in a redistribution of both formal and semantic properties. The de-categorialization concerns the loss of typical noun-features of the QN. Within the (quantifying) binominal context (59), the QN has indeed lost its combinatorial freedom as to determiners (60) and adjectives (61), as well as the possibility to pluralize (62).63 (59) Hay una pila de gente en la puerta. ‘There is a heap of people at the door.’ (60) *Hay la / ?esta pila de gente en la puerta. ‘the / that’ *Hay tres pilas de gente en la puerta. ‘three’ (61) *Hay una pila alta de gente en la puerta. ‘high’ ? Hay una pila grande de gente en la puerta. ‘big’ ? Hay una gran pila de gente en la puerta. ‘big’ (62)

?

Hay pilas de gente en la puerta.

|| 63 Note that the morphosyntactic ‘tests’ to identify quantifying/specifying uses outlined in Section 3.2 present a great overlap with the set of properties determining the membership to a specific grammatical class.

The grammaticalization of binominal quantifier constructions | 93

At the same time, the BQs testify to a certain re-categorialization for having moved towards the grammatical category of quantifiers. BQs lift the combinatorial restriction to N2’s which are compatible with their literal use: while the literal use of pila heads nouns designating easily stackable objects usually made of paper, the host-class expansion extends towards nouns referring to human entities and abstract notions of time when grammaticalizing into quantifying una pila de. The lifting of combinatorial restrictions on N2 is even more drastic in the cases of montón de and mogollón de, which are free to combine with any noun, generalizing over the abstract-concrete and count-mass noun distinctions. In addition, the entire string un(a) QN de can be substituted for by the quantifier mucho/a(s) and N2 can no longer combine with proper quantifiers or determiners. It bears pointing out that the originally more specific QNs, such as alud de or letanía de, testify to a certain degree of formal persistence. In preview of the detailed description of the morphosyntactic behavior of the BQ-constructions in Chapter 8, it can be noted that especially the QNs which are characterized by a rich source semantics can continue to combine with a restricted set of adjectives in the grammaticalized uses, allow for N1-determiner variation when topicalized, impose conceptually motivated restrictions on N2 and can combine with verb agreement with N1. For now it suffices to say that the shift into quantifying readings clearly corresponds to a shift in morphosyntactic and combinatorial behavior. Semantically, the shift to quantifying and/or specifying readings indeed resulted in a loss of concrete meaning or de-semanticization. Quantifying or specifying QNs no longer (necessarily) evoke a concrete, tangible configuration or group of spatiotemporally contiguous N2-entities. When looking at un alud de cartas ‘an avalanche of letters’, one doesn’t see an actual mass of letters loosened from a mountain side and descending swiftly and violently from it. Nor is there any mass of letters simultaneously crushing down on the speaker. Instead, the BQ acquires a quantifying reading and permits to expressively and hyperbolically quantify the group of letters. In other words, the desemanticization is closely intertwined with a re-semanticization. In preview of the detailed overview of the QN-related image-schematic structures in Chapter 7, it is fair to say that there is no loss of semantic complexity. Instead, in close interaction with the specific usage context, the individual QNs allow for a series of different construals of N2. By way of illustration, I will argue that receiving un aluvión de críticas ‘a flood of criticism’ is slightly less face-threatening than un alud de críticas ‘an avalanche of criticism’. Interestingly, De Smet (2014) attributes a ‘hybrid’ status to the BQ-construction (cross-linguistically). Although they have come to express quantity assessment and expanded their structural scope, they continue to be compositional

94 | Binominal quantifiers as a locus of grammaticalization

and analyzable. By way of comparison, no phonological attrition has taken place as is the case of some English constructions (e.g. helluva, (a)lotta). De Smet therefore considers BQs as “hybrids” (2014, 30), i.e. instances that do not resort to a single analysis yet in their surface manifestation present characteristics of both underlying alternative analyses. Even if the operator-like function of BQs cannot be denied, the concept of ‘syntactic reanalysis’ is at odds with the formal persistence in the development of QNs. Yet from a discourse perspective, there seems to be nothing problematic about the structural ambiguity of the construction: whether or not the QN combines with typically nominal features, the speaker will automatically exploit its quantifying potential. I therefore subscribe to De Smet’s (2014) polemical claim that syntactic change or innovation does not need reanalysis strictly speaking. If one allows syntactic categories to have a gradient constituent structure (De Smet 2014, 39), the syntactic indeterminacy resulting from partial decategorialization of BQs is no longer an issue. Instead, the innovative power of analogical thinking, whereby “a construction extends its range of application into the domain of (what used to be) another” (De Smet 2014, 34; see also Fischer 2007) and the importance of persistence will be examined in the following chapters. In addition to the redistribution of formal and semantic properties, support for the GR hypothesis is found in the systematicity and the cross-linguistic recurrence of the BQs. The binominal construction, as far as Spanish is concerned, is a highly productive tool for expressive quantification (and also qualification) of N2. In addition to a whole series of conventionalized QNs, such as montón ‘heap’, pila ‘pile’, racimo ‘bunch’, etc., any N1 with scalar implicatures can give rise to a quantifying BQ without necessarily resulting in a productive type. Further, the binominal pattern appears to be a cross-linguistic locus of GR (Brems 2011; Koptjevskaja-Tamm 2009; Masini Fc.; Mihatsch 2000; Traugott 2008b). In terms of the parameters of GR, the criteria which focus on the semantic changes involved in GR are more useful in the case of BQs than Lehmann’s set of formal parameters. As to the primarily morphosyntactic criteria, the development of BQs is characterized by host-class expansion, decategorialization and syntactic context expansion. First, the class of N2s the QN is in construction with expands to animate and abstract nouns. Second, the shift into quantifying uses gave way to partial decategorialization or syntactic indeterminacy. Since the string [indefinite determiner + QN + de] is the default realization, the GR of BQs also presents some degree of fixation. Finally, the GR of some Spanish BQs definitely involves syntactic context expansion to adjective intensifier uses (in the case of mogollón de (63) and la mar de) and adverb uses (in the case of montón, mogollón (64) and barbaridad). Interestingly, the extension towards

The grammaticalization of binominal quantifier constructions | 95

adjective intensifier and adverb uses is only observed with a subset of the QNs, viz. the QNs which are originally vague as to the internal constitution of the mass.64 (63) Luis Tosar no es guapo, pero es mogollón de atractivo. (2002, press) ‘Luis Tosar is not handsome, but he is quite attractive.’ (64) ¿No sales de marcha con los amigos? Sí, mogollón, sí, o sea, salgo y los viernes y los sábados con los amigos. (year unknown, oral) ‘Don’t you go for a drink with your friends? I do, a lot, indeed, I mean, I go out both on Friday and on Saturday with my friends.’ As to the semantic parameters, the development of BQs gives proof of semanticpragmatic context expansion, persistence, divergence and layering of uses. First, not only do QNs acquire quantifying and specifying readings, they are also endowed with an evaluative potential. Second, the differences in conceptualization between un montón de gente and una pila de gente, for instance, pertain to differences in the original lexical meaning and therefore rely in the contrast between literal un montón de libros ‘a heap of books’ and una pila de libros ‘a pile of books’. As can be derived from Table 2, the GR of Spanish BQs has resulted in a synchronic layering of uses, where the ‘original’ lexical meaning remains next to the ‘newer’ functional uses. Finally, the development of BQs is also characterized by divergence, since the literal QN continues to be used outside the binominal construction and in some cases exploits its quantifying potential in other construction types as well (e.g. a montones ‘in large number’, a mares ‘abundantly’, etc.). The fact that the development of Spanish BQs is more easily operationalized in terms of the semantic parameters might suggest that we are facing a case of incipient GR. Recall that formal changes usually lag behind semantic changes (Traugott 2003a). However, the incipient or less advanced stage of GR is hard to reconcile with the productivity and cross-linguistic nature of the binominal pattern. That is why this study seeks for possible cognitive, functional and pragmatic motivations for the analyzable and compositional nature of grammaticalized items.

|| 64 I thank the native speakers at conferences who pointed out to me that in some regiolects, mazo and puñado combine with these types of syntactic context expansion as well.

96 | Binominal quantifiers as a locus of grammaticalization

3.3.3 Concluding remarks The hypothesis that the Spanish BQ-construction is the locus of GR is the starting assumption of this book. The case-study presents us with an interesting descriptive challenge since Spanish BQs are certainly not a straightforward case of GR: the tendency towards formal as well as semantic persistence clashes with the productivity of the construction. Furthermore, evidence will be adduced against the claim that the persistence is kept up artificially. In preview of the next chapters, it can be noted that the formal persistence is conceptually motivated, while the semantic persistence hinges on the pragmatic function of the BQ. Yet the productivity of the construction is to be handled with care as well. The fact that the GR of BQs gave way to a whole series of conventionalized QNs – in fact any noun with scalar implicatures can occupy the N1-slot – should not overshadow the fact that the correlation between high or increased frequency and GR only holds for un montón de and (la) mar de (cf. Table 2 and infra). The high type frequency of the (schematic) BQ-construction contrasts with the limited token frequency of some individual QNs which nevertheless present a high proportion of grammaticalized uses. The most remarkable example is barbaridad de, which only occurs nine times in the corpus yet always evokes a quantifying reading. Similarly, the proportion of grammaticalized uses of mogollón de (98%), aluvión de (88%), hatajo de (82%), alud de (75%) and letanía de (73%) exceeds the number of grammaticalized uses of montón de (68%). The correlation between high frequency and an advanced stage of GR does hold in the cases of montón de and (la) mar de since both QNs give rise to dramatic syntactic expansion (to adverbial uses and adjective intensifier uses) respectively. However, these extensions are also observed with barbaridad de and mogollón de, which are not particularly frequent items. In sum, the irregular distribution in Table 2 casts doubt on the GR-hypothesis. In this context, the question arises what exactly grammaticalizes when QNs start to express quantity assessment of N2. Can the individual pathways of GR be maintained in the light of the complex frequency pattern? Is there a need to posit a schematic BQ or un N1 de N2-pattern which grammaticalizes? The diachronic case-studies will try to provide a satisfactory answer to this matter. In addition, the high proportion of grammaticalized uses of specific QNs such as alud, aluvión and letanía calls into question the condition of semantic generality as the input of GR. Provided that the more specific the original meaning is, the more chances it stands to remain stable over time, it does not come as a surprise that these QNs tend to combine with a high degree of conceptual persistence. Perhaps more interesting is the fact that the meaning change of less

The grammaticalization of binominal quantifier constructions | 97

specific QNs is not arbitrary either: QNs such as montón, pila and racimo continue to have a meaningful import on the construal of N2. In this context, the question arises why QNs in general are likely to block desemanticization. And if conceptual persistence is to be recognized as a typical property of BQ-construction, one might wonder if and how it interacts with the tendency towards formal persistence. These questions will be the focus of the synchronic case-studies.

| Part 2 Diachronic case-studies – towards a constructional network model

4 The development of montón de, a typical locus of grammaticalization The synchronic layering of uses of the BQ-construction can only be regarded as the outcome of GR process if quantifying and specifying uses are always attested later than the head uses. The next chapters will zoom in on the development of several individual QNs and the BQ-construction in general, in order to verify whether diachronic evidence is found in favor of the GR process. Chapter 3 has touched upon the importance of the triggering ‘context’ of grammaticalizing linguistic items, i.e. the truism in grammaticalization theory that “[e]verything that happens to the meaning of a gram happens because of the contexts in which it is used” (Bybee et al. 1994). The present chapter zooms in on the stepwise evolution of one particular QN, viz. montón, in order to verify whether the semantic glides and changes in the morphosyntactic make-up of the construction qualify for GR. The choice for montón is motivated by the prototype role it appears to fulfill both synchronically and diachronically: it is the first QN to systematically develop quantifying readings, it is the most frequent QN from Classical Spanish onwards, it is nowadays used as a fairly unrestricted quantifier combining with all types of N2 and its syntactic context expansion involves – in contrast to the other QNs – adverbial uses as well (such as Me alegra un montón. ‘It makes me very happy.’). The prototype function of montón implicitly suggests that its development might have attracted, or more carefully affected the GR of the remaining QNs via analogical thinking. As a construction grammarian approach primarily focuses on ‘sets’ of similarly behaving constructions, I also dwell on the evolution of the remaining QNs in Chapter 5. For lack of space, however, I only provide a thumbnail sketch of the latters’ developments, the more since their individual histories reflect similar semantic and syntactic micro-changes. However, analogical thinking alone cannot account for the serious gap in time between the development of montón de and the spectacular rise in frequency of the BQ-construction in Present-Day Spanish. Chapter 6, then, shows that the emergence of the BQs can best be captured in a constructional network model and illustrates how the Spanish case-study can help refine diachronic constructionist approaches. In the present chapter, descriptive analysis prevails over quantitative analysis and frequency tables are only included for the sake of completeness, since the diachronic corpora used unfortunately are not completely statistically reliable. First, historical data are sparse and – at least for Spanish – incomplete and not representative for actual language use (Cornillie 2011). Until Present-Day Spanish, almost exclusively written data or reproduced oral discourse traditions

102 | The development of montón de, a typical locus of grammaticalization

survived. However, the usage-based approach I subscribe to predicts that language change starts in the interactional context, in the speaker-hearer negotiation of the clausal meaning. As nothing but written language is at hand, we have to make the best use out of bad data (Labov’s definition of historical linguistics, 1994, 11). Second, medieval data often include different versions of a single text (such as bible texts copied by monks). The passage given in (1) for instance, occurs 11 times in our corpus, each instance diverging slightly from the remaining ones.65 Finally, apparently the CORDE and Corpus del Español are flawed in terms of editions: for instance, the 13th century data include the V part of the General Estoria (Alfonso X), which is in fact an edition published at the end of the 15th century (Cornillie 2013, p.c.).66 (1)

Et porque fallamos que en dar estos diezmos se fazien muchos engannos, defendemos firmemientre daqui adelantre que ninguno non sea osado de carer nin de medir suo monton de pan que toviere limpio en la era si non desta guisa: que sea primeramientre tanida la campana tres veces a que vengan los terceros o aquellos que deven recabdar los diezmos. (1255, anon., Carta de Alfonso X ) ‘And since we observe that while dividing these tithes many swindles are committed, we firmly defend that from now on no one should dare to price or measure his heap of bread he would have clean on the field if not in this way: first the bell should ring three times before thirds or those in charge of collecting the tithes come.’

Assuming that dictionaries primarily include conventionalized readings of lexical items and usually lag behind actual language use, the present chapter starts by outlining the lexicographic treatment montón received in history. The corpus-based diachronic analysis then tries to tackle the issue of the process of change involved and to answer the question why precisely montón (de) has developed into typical binominal quantifier.

|| 65 The original spelling in the following diachronic examples has been left unaltered, even when the spelling found in the data set is not consistent with the orthographic rules of PresentDay Spanish. 66 Although the diachronic analysis is based on the classification imposed by CORDE, I will specify for each example, in addition to the year, the author and/or title so the valuable information as to edition or discourse tradition can be retrieved (see infra).

Methodological background | 103

4.1 Methodological background The data which lend support to the claims made in this chapter were extracted from the diachronic on-line corpus Corpus diacrónico del español (henceforth CORDE) offered by the Real Academia Española. CORDE is generally considered the most representative diachronic corpus. It is a 237 million words corpus that covers, in so far as possible, all geographic, historical and genre varieties. The search engine also allows for Boolean operators and the universal wildcards ‘*’ and ‘?’. It has one major shortcoming however, viz. the absence of lemmatizing and morphosyntactic tagging. To search for all attested occurrences of the adjective mucho ‘many’ for instance, one has to introduce separately the queries [mucho], [mucha], [muchos] and [muchas] or extract all occurrences of [much] (with or without wildcards) and filter out irrelevant cases, e.g. muchacho ‘boy’. Likewise, category searches specifying nothing but the grammatical category are not possible: to extract all occurrences where montón de is followed by a noun phrase, one has to use the query [montón de] and discard all irrelevant examples manually (such as montón de bien ‘very good’). The data for montón are extracted from CORDE by means of the queries [montón] and [monton], which correspond to 2331 examples. As I am mainly interested in the BQ-construction, I only looked at the other uses of montón insofar as necessary for understanding the precise original usage contexts which gave way to a quantifying interpretation. The binominal uses for montón (de) count 1387 instances in CORDE,67 of which 1298 express coextensiveness. All instances were coded for number and determiner variation of both nominal elements (N1 and N2), modification, agreement and collocational patterns. The parametrization further covers syntactic and pragmatic contextual indices at different levels of abstraction, the degree of conceptual persistence, the reading of N1 and of N2, the pragmatic function of N1 (e.g. emphasizer, downtoner, booster) and the presence of subjectified elements in the reading of N2 (Traugott 1989). The delineation of the four periods in the history of Spanish builds on the general periodization proposed by Eberenz (2009) for diachronic studies of Spanish. In the diachronic analysis, I distinguish four periods in the history of montón de, viz. Medieval Spanish (1250-1450), Classical Spanish (1450-1730), Modern Spanish (1730-1900) and Present-Day Spanish (1900-1975). The data for

|| 67 Although CORDE allows for the query [montón de], I started from all instances of [montón] in order to retrieve adverbial uses and to withhold the occurrences of ‘QN + adjective’.

104 | The development of montón de, a typical locus of grammaticalization

Present-Day Spanish present the artificial cut-off point 1975, as CORDE only includes texts until that year.

4.2 Encyclopaedic information on montón This section summarizes the lexicographical information provided by the encyclopaedic thesauri of Nieto Jiménez/Alvar Esquerra (2007) and the Real Academia Española (2001b) as well as by the medieval Spanish dictionary of Lloyd/Nitti (2002). The thesauri are built to list, per lexical entry, the definitions yielded by dictionary sources from the 15th century onwards. Whereas Jiménez/Esquerra provide the lexical definitions until Autoridades (1734), i.e. the first reference dictionary edited by the Real Academia Española (henceforth RAE), the RAE thesaurus continues the survey until Academia Usual (1992). Although dictionary definitions usually lag behind actual usage – and can therefore not be used to provide the exact chronology of changes – they certainly inform on the conventionalized use of lexemes. In other words, when a quantitative interpretation of montón is mentioned in a particular 18th century dictionary, the change from literal ‘heap’ into quantifying montón de can be considered to have spread to the majority of the language users somewhere before the dictionary’s publication date. Until the 17th century, Spanish reference dictionaries used to define montón as a heap of little things, usually of wood, land or hay. It is generally translated by Latin cumulus ‘heap, pile’ and acervus ‘heap’,68 i.e. two nouns referring to the typical conical configuration. Nevertheless, the quantifying syntagm a montones ‘abundantly’ is often alluded to. The dictionary by Covarrubias in 1611 is the first to mention an extension with regard to the objects the heap is made of: Qualquiera cosa que se ayunta y va haciendo colmo, como montón de trigo, montón de tierra. [emphasis mine]69

The first reference dictionary by the RAE, Autoridades (1734), is the first dictionary to add configurational aspects of the heap by specifying its extensive and chaotic nature:

|| 68 The Latin translations are based on the online Database of Latin Dictionaries (2010). 69 Any thing that is put together and forms a heap, as a heap of wheat, a heap of earth. (translation KV)

Encyclopaedic information on montón | 105

Agregado ò junta de muchas cosas de una misma, ù de diversa especie puestas en algún lugar, confusamente y sin orden, de modo que sobresalgan y se eleven al plano en que están. [emphasis mine]70

This definition indicates that by the beginning of the 18th century, the pragmatic inferences ‘chaotic heap’ and ‘lots of constituting entities’ were considered conventionalized, i.e. amply recognized by the native speakers. Further, the dictionary also mentions that montón was used colloquially for useless and idle persons and includes the idioms en montón or de monton ‘together, jointly’ and a montones ‘excessively, abundantly’. The dictionary adds that, etymologically, montón stems from latin monte ‘mountain’. A century and a half passes by before the first unequivocal quantifying use is illustrated, again by (the new edition of) the reference dictionary by the RAE, viz. Academia Usual (1884), which mentions: (De monte.) m. Agregado (…)|| fig. y fam. Persona inútil (…).|| fig. y fam. Número considerable, en frases como la siguiente: Tengo que decirte un montón de cosas. || Min. Capa ó lecho (…).|| de tierra. (…).|| Á montones. (…).|| De montón, ó en montón. (…). [emphasis mine]71

In other words, the 19th century dictionary acknowledges the quantifying interpretation of the binominal construction as well, providing a chameleon-like right collocate cosas, which may be of any kind. However, the examples yielded by the following dictionaries usually refer to ‘spoken’ things (words, news, etc.). According to Academia Usual (1884) the binominal phrase montón de tierra has lexicalized and expresses ‘very old and weak person, or suffering lots of aches’. Further, cara de montón is an agricultural notion indicating the better part of the wheat harvest. The 1925 edition of Academia Usual additionally mentions the fixed expressions a montón ‘approximately’ and ser uno del montón ‘to be run-of-the-mill or ordinary as a person or regarding the social status’. The definition of montón still remains unaltered in the Present-Day edition of the Diccionario de la Real Academia Española (2001a). Interestingly, the medieval dictionary by Lloyd/Nitti (2002), reflecting the language system during the kingdom of Alfonso X, lists three interpretations for

|| 70 Aggregate or union of many things of the same or different kind put in some place, in a confused way and without order, so that they stand out and reach out at the level at which they were placed. (translation KV) 71 (From monte.) m. Aggregate (…)||fig. and fam. Idle person (…)|| fig. and fam. Considerable number, in phrases such as the following: I have to tell you a lot of things. || Min. layer. (…)|| of earth. (…)|| Abundantly. (…)|| Jointly. (…)’. (translation KV)

106 | The development of montón de, a typical locus of grammaticalization

montón from the 13th century onwards. In addition to the usually attested conical configuration sense, montón is considered to refer to a large number (número considerable) or to the hill construed behind a wall to fight a city or a fortress. Crucially, it includes an authentic example of an unambiguous quantifying reading: (…) Aqui dize agora orosio que nos a contadas muy chicas batallas en muy pocas palabras. & que non departio. tanto quam grandes fueron los fechos de las obras daquellas guerras; como que ayunto y a dezir montones de mesquindades. Ca dize assi. que quien es aquell. que departa la tempestad daquell tiempo Nin las muertes que y fueron fechas quien las podrie contar.” (anno 1280) [emphasis KV]72

In other words, the medieval dictionary suggests that the quantifying reading was already spontaneously associated with (the plural variant of) montón de and commonly exploited in the 13th century, which means six centuries before this use was recorded in the 1884 reference dictionary. In an attempt at getting a better grip on ambiguous information as to the source construction, I propose not to equate the first attested use of grammaticalized readings and the systematic exploitation of particular grammaticalized uses. It bears pointing out that no dictionary makes mention of the particular contexts in which montón was used originally. The next section fills this gap by analyzing the small (but crucial) changes in the contextual and constructional make-up of montón de, which all contributed to its actual usage as the exemplar QN.

4.3 A diachronic corpus analysis of montón In line with the construction grammar axiom that constructions are symbolic units of form and (conventionalized) meaning (Croft 2001; Traugott 2008b) and the cognitive truism that grammar is meaningful (Langacker 2008), both the stepwise semantic changes and the resulting changes in form are discussed separately. As it is a widely held view in grammaticalization theory that semantic change precedes formal change, the subtle shifts in meaning or the possibly conventionalized pragmatic inferences are discussed first. Leaving aside for the || 72 Here he tells now in a glorious way that he has told us a lot of little battles in very little words. And that he has not shown how great the deeds were of those wars; as he gathered (lit. gathered and to say) a lot of (lit. heaps of) mean things. Because he says as follows, that the one who evokes the tumult of that time could not count the deaths that were caused there. (translation KV)

A diachronic corpus analysis of montón | 107

moment whether head, quantifier and specifier use constitute separate ‘constructions’, these uses and their respective usage contexts are discussed separately.

4.3.1 Medieval Spanish: the emergence of quantifying pragmatic inferences The lexical item montón derives from the item monte ‘mountain’ and the augmentative suffix -ón (Corominas/Pascual 1991, 131; Lázaro Mora 1999, 4673), meaning basically ‘a huge mountain’. In Latin, the suffix -(i)o was added to body part nouns in order to designate persons whose oversized version of the respective body part immediately catches the eye. Similarly, in Spanish, narigón stems from nariz ‘nose’ and refers to a ‘man with a big nose’. Given the stem mont- ‘mountain’, the focus on the vertical orientation of montón in its later quantifying uses should not come as a surprise. However, in Medieval Spanish, montón was apparently never used literally to designate an extremely high natural elevation of the earth’s surface in comparison to other ‘smaller’ mountains for instance.73 Its function and productivity reside in two additional conceptual facets that differentiate montón from the usage contexts of monte ‘mount’ –in spite of the latter’s undeniable quantifying potential, it almost exclusively stuck to literal uses up to Present-Day Spanish (cf. Section 4.3.4). In addition to the importance of the vertical orientation of the high, steep and conical configuration, two more conceptual facets are generally associated with montón in Medieval Spanish, viz. the accumulated or gathered nature of the heap and the human intentionality in heaping things up (2). Not surprisingly then, the lexical item montón was generally accompanied by verbs such as allegar ‘to gather’, adjuntar ‘to enclose’, poner ‘to put’, levantar ‘to raise’, hacer ‘to make’, traer ‘to bring’, meter ‘to put’, alzar ‘to raise’, echar ‘to throw’ or their

|| 73 According to De Bruyne (1979, 17), -ón can combine an augmentative with a pejorative value as the distance between both values is not hard to overcome: if something is too big, it often looks ridiculous or ugly. Further, suffixation is often a semantically motivated way of creating a new lexical item. The derivation by –ón leads to the bleaching of the reference to physical or objectively measurable dimensions and the addition of a certain affective load (usually evoking the concept of ‘imperfection’). Consequently, montón can be made to fit hyperbolic contexts. In other words, when adding –ón to monte ‘mountain’, it is fair to say that abstraction is made of the literal reference and that montón is used for heaps of things which are drawing the speaker’s attention, both in a good and bad sense. In the meantime, the combination got so conventionalized that speakers are not aware anymore of the original reference to a large mountain.

108 | The development of montón de, a typical locus of grammaticalization

respective nominalizations. Notwithstanding the apparent redundancy, montón is frequently preceded by the adjective grande ‘big’ (3), which suggests that the literal meaning ‘heap + -big’ had already faded away by the 13th century. (2)

(3)

E, así perquiriéndolos, falló un montón de piedras alto, que bien paresçían manualmente allegadas fuesen, en testimonio de algúnd sepulcro en aquel lugar, segúnd oy se faze (…). (1427, E. de Villena, Traducción y glosas de la Eneida) ‘And, investigating like that, a high heap of stones was found, that seemed put there by hand, as testimony of some kind of sepulcher on that spot, as today is done, (…).’ E por ende publicaron e establescieron las leyes de las diez tablas en público delante los centurias de todo el pueblo, (...); en las quales grant montón de leyes la una sobre la otra fueron ayuntadas, (…). (1400, P. López de Ayala, Traducción de las Décadas de Tito Livio ) ‘And therefore the laws of the tables were published and established in public in front of the centuries of the entire village, (…); in which a big heap of laws were put together one above the other, (…).’

The original usage contexts of montón are rooted in the feudal system that reigned in the West-European Middle Ages (from the 9th until the 15th century). Putting it very simply, feudality came down to a set of social and military customs regulating medieval society. A (land)lord exchanged the holding of a land and military protection to the vassal for loyalty and an annual subpart of the harvest. Similarly, the king or emperor could reward his troops by distributing (a subpart of) the war booty. A common division or tax measure in the feudal system was the tithe: farmers or craftsmen had to offer a tenth of their harvest or production. For an overview of the volume of the profits, the war booty, the harvest or the production were brought together in one point (cf. the plenty of tithe barns along Western Europe) on a heap. Heaps of war booty, of harvest or of manufactured products are by definition vertically oriented, gathered and man-made. In Table 1 the instances of montón as referring to a heap of harvest (4) or a heap of war booty (made up by spoils or dead people (5)) figure among its 5 major usage contexts. The use of montón to refer to a collection of things piled up for distribution (6) or tithe calculation is the most productive context for montón in Medieval Spanish and gave rise to the extension of montón as a measure unit (7). (4)

Qui bien laura su tierra alçara monton de miesses. (1280, Alfonso X, General Estoria IV) ‘The one who farms his land well will raise a heap of corn.’

A diachronic corpus analysis of montón | 109

(5)

(6)

(7)

E aquí se juntaron los peones del Rey con los de Muça, e comiençan su batalla tan brava que antes de una hora tenían fecho entre ambas las batallas el montón de los muertos que no se veían unos a otros; (1430, P., Crónica del rey don Rodrigo, postrimero rey de los godos) ‘And here the King’s soldiers joined Muça’s (soldiers) and they started their battle so fiercely that within an hour they had made between both armies the heap of dead bodies so (high) that they could no longer see one another.’ Otrosy alo que nos pidieron por merçet que aviendo dezmado el monton del pan que cogiese qual quier persona, que despues que dezmado en monton, que non ouiese otro rrediezmo, por que acaesçe que vn labrador que arrendaua su heredat por vn anno o por mas tienpo por çierta quantia de pan, e aquel labrador que tenia la heredat arrendada, dezmado de monton todo el pan que cogia e que sacado todo este diezmo atal enteramente, quel pan quel tal arrendador auia de dar al sennor dela heredat que demanda, despues los arrendadores e los clerigos otra vez queles diezmen dela tal rrenta que dan al sennor dela heredat, seyendo primera mente dezmado de monton. (1386, anon., Cortes de Segovia) ‘In addition to the favor they asked us that having decimated the heap of bread that anyone would take, that after having decimated everything, that there would be no other second tithe, because it happened that a farmer who rented his estate for a year or longer for a certain amount of bread, and that farmer who had leased an estate, decimated all the bread he received and once obtained these tithes entirely, the bread this tenant had to give to the landlord of the estate he took charge of. Afterwards the tenants and clergymen decimate from the rent they give the landlord for the estate, but first it would be decimated from the whole.’ Tomaron a Lope de Mellante un hovete apreçiado en dozientos maravedis, mas un monton de cal apreçiado en çiento maravedis, mas dos toneles de vino del primer año apreçiados en ochoçientos maravedis. (1409, anon., Sentencia) ‘They took from Lope de Mellante a silk fabric estimated at twohundred maravedis [ancient currency unit], as well as a heap of lime estimated at hundred maravedis, as well as two barrels of wine of the first year estimated at eight hundred maravedis.’

110 | The development of montón de, a typical locus of grammaticalization

Equally productive is montón’s reference to a memorial, viz. a heap of stones or earth signaling the tomb of a beloved person (8) or symbolizing a historical fact (9). (8)

(9)

& sobre la fuessa mando el rey ayuntar un grant monton de arena. (1275, Alfonso X, General Estoria II) ‘And the king ordered them to collect a big heap of sand on the grave..’ E la razón porque esto se faze, (…) , es qu'el grand gigante Enchelado, medio quemado del rayo que lo firió su cuerpo en este lugar, e d'él fizo el grand montón e sobr'él por los dioses fue puesto el monte d'Enna. (1427, E. de Villena, Traducción y glosas de la Eneida) ‘And the reason why this is done, (…), is that the great giant Enchelado, half burnt from the lightning that hit his body in this place and that made a big heap out of him and on top of him the gods builded the mount Enna.’

In addition, the heap indicated by montón referred to a woodpile (10) or recalls the literal mountainous configuration in metaphors (11). (10) (…), de los senyores de los dichos sieruos aplegado grant monton de sarmientos fue cremado alli biuo con toda familia. (1376, J. Fernández de Heredia, Traducción de la Historia contra paganos, de Orosio) ‘Together with his family he was burned alive there in a great heap of vine shoots by the lords of those serfs.’ (11) Vn dia mannana quando nascie el sol. Nascio de parte de septentrion un fuego muy grand cuemo monton a manera de otero. E fizo el cielo muy grand sueno. (1270, Alfonso X, Estoria de Espanna que fizo el muy noble rey don Alfonsso) ‘One day in the morning when the sun came up. From the north a fire big as a mountain and similar to a hillock showed up. And the sky made a very loud noise.’ Crucially, the contexts of usage or the image evoked by montón are closely interwoven. In several battle field examples, the spoils or corpses are first heaped up for symbolic value and then burned. Heaps of goods gathered in order to be distributed or divided by ten still consist of harvest heaped up. The ruins of a city may count as a memorial of the historical victory. In other words, the different usage contexts presented in Table 1 should not be regarded as entirely separated readings, but as different activations of interrelated frames. Examples which seem to fall outside the scope of the feudal frame, the woodpile frame and memorial frame, do qualify for a vertically oriented conical configuration of

A diachronic corpus analysis of montón | 111

things brought together by man (12). In other words, the vertical configuration, the accumulated nature and the human intentionality are the meaning components which are minimally evoked. Interestingly, the idiom a montón also originates in the feudal organization of the Middle Ages. It is always preceded by verbs referring to an act of ‘gathering’ (aducir ‘to adduce’, traer ‘to bring’, allegar ‘to gather’, echar ‘to throw’) and invokes the image of a person gathering or throwing a lot of harvest or weapons on a heap (13). (12) Et tomo Jacob vna piedra, e alçola por estatua; [46] e dixo Jacob asus hermanos: coget piedras e cogieron piedras e fizieron monton ; e comieron ende sobre el monte. (1400, anon., Biblia Escorial I-j-4) ‘And Jacob took a rock and lifted it like a statue. And he said to his brothers: take the rocks and they took rocks and made a heap; and then they ate on the hill.’ (13) Mas esto es el derecho que despues que el lugar ffuere ganado que aduga cada vno lo que ganare a monton & den al Rey ssi ffuere y ssu quinto ssegunt dixiemos en esta otra ley. (1260, anon., Espéculo de Alfonso X. BNM 10123) ‘But this is the right; that after the place is conquered, that each one should bring together what he has won and give the king his privilege and the fifth part of what is gained as stated in this other law.’ Insofar as we can project current social norms on medieval data, we can expect people to have been proud of their heap of achievements. Not by coincidence, the immediate context of montón frequently stresses the height of the heap (mainly by the adjective gran(d) 'big' (cf. (3), (14)), by verbs such as crescer ‘to grow’ y alzar ‘to lift’ indicating an upwards direction (4) or verbal expressions such as semejar montaña ‘to look like a mountain’, etc.). The importance the speaker attaches to the scope of his achievement(s), productions or realizations might have been one of the original bridging contexts for enabling the QN to acquire quantifying uses. It also adequately fits the additional information provided by the affectively loaded augmentative suffix and explains why montón de has become a far more productive quantifier than its stem monte. By the 14th century, we find examples wavering between head use and quantifying interpretation, e.g. between a big heap of gold piled up and a lot of gold in (14) or the heap of words in (15) which is metaphorically described as the gathering of words and the outcome of human effort, as well as examples where montón de unambiguously exploits its quantifying potential (16)–(17).

112 | The development of montón de, a typical locus of grammaticalization

(14) (…) por que el ha trobado vn grant monton de oro. (1400, Libro del Tesoro. Girona, Catedral 20a5) ‘(…) because he has found a big heap of gold.’ (15) Pues que assi es, ¿pora que traballare de aplegar grant monton de paraulas en contarte la manzilla de aquesta muyt suzia et pudient amistança et conueniencia de pactos que los romanos la hora firmaron con los sannites? (1376, J. Fernández de Heredia, Traducción de la Historia contra paganos, de Orosio) ‘Because that is how it is, why try so hard to gather a big heap of words while telling you about the dishonor of that very dirty and powerful friendship and the convenience of pacts that the Romans then signed with the Samnites?’ (16) E estos tales son conparados a los ypocritas que, quando ensennan el bien que saben, non lo dizen para hedificar a los que lo oyen mas por que digan que son grandes sabidores e ayan dende grand monton de vanagloria. (1400, anon., Traducción del Soberano bien de San Isidoro) ‘And these ones are compared to the hypocrites who, when they teach everything they know, they don’t tell it to edify those who hear them, but they tell it so they would say of them that they are great wisemen and have a great heap of vain glory.’ (17) E el diablo, enemjgo grande, nunca esta de uagar contra el sieruo de Dios: o le allega grand monton de pensamjentos de tribulaçion en el coraçon, o le despierta dolores en el cuerpo. (1400, anon., Traducción del Soberano bien de San Isidoro) ‘And the devil, great enemy, never leaves the Lord’s slave in peace: either he gathers in his heart a great heap of black thoughts, or he wakes him with physical pains.’ It bears pointing out however that, at first, the quantifying interpretation is always triggered by the premodifying adjective grande ‘big, great’ or by a preceding QN which is used literally (as in gran aplegamiento y montón de… ‘big gathering and heap of…’). In the latter case, using montón literally would be redundant. Further, the feudalism frame has not entirely faded: examples (15)– (17) illustrate the importance of ‘possessing’ the montón de N2, of possessing a mental or physical trait or a particular fortune or richness. Table 1 shows the distribution of frames or usages observed in my corpus.74 || 74 For Medieval Spanish (in contrast to the corpus-study for Classical, Modern and PresentDay Spanish), I analyzed the occurrences of montón as a bare noun exhaustively as well, as I am convinced that the original use as a bare noun influences the further developments of the

A diachronic corpus analysis of montón | 113

Tab. 1: Distribution of frames observed with montón in Medieval Spanish Frames evoked by montón (Huge) mountain Vertical configuration of objects intentionally accumulated and heaped up by human endeavor War booty Heap of harvest Heap of possessions that are to be divided (cf. tithes) Woodpile Memorial stone (or graveyard) The idiomatic expression a montón Heap of ruins after a battle Bridging contexts (between head and quantifier use) Pure quantifier uses

Total

%

12 19

8 13

16 15 26 10 25 4 3 3 8 141

11 11 18 8 18 3 2 2 6 100

With regard to the morphosyntactic make-up of the construction, medieval montón still functions as a full lexical noun. In its head-use, which obviously constitutes the unmarked use (cf. Table 2), montón is unrestricted with regard to the determiner to N2 (18) and the prepositional phrase ‘de N2’ does not necessarily follow montón (19). In other words, in Medieval Spanish, montón de N2 is not processed as a single chunk. The separability of the prepositional phrase is of course linked to the free(er) word order that characterizes Latin and Medieval Spanish. Interestingly, the separability of de N2 does not combine with the quantifying use in my corpus. Neither do the observed quantifying uses in Medieval Spanish combine with determiners to N2.75

|| QN. As the main focus of this study is the binominal construction in itself, I did not analyze how the remaining QNs are used as a single noun. Studying the historical development of individual lexemes which display a quantifying potential would constitute yet another interesting track of research, but one that would deviate me too much from the initial focus. I thus did not take into account montón’s uses outside the binominal context for the remaining periods and leave this issue for further investigation. The query introduced in CORDE generalizes over the purely orthographic distinction between montón and monton ([montón O monton]). 75 I do not wish to jump to conclusions here: corpus based analysis can adequately show what and how constructions are used. The fact that a particular use is not attested does not necessarily mean that it did not exist. In other words, lack of evidence is no strong counterevidence – especially in historical analyses – but may be due to lack of representativeness of the corpus. Historical data are by definition limited and possibly manipulated. In Classical Spanish for instance, quantifying montón still allowed determination of N2. It thus makes no sense to argue that in Medieval Spanish montón had lost this nominal characteristic to regain it a century afterwards.

114 | The development of montón de, a typical locus of grammaticalization

(18) Desí dixo Labam, segund cuenta la Biblia: – Si tú apremiares mis fijas e las dexares e tomares otras mugieres, (…). E este otero e este montón d'estas piedras sea entre nós testigo d'ello. (1275, Alfonso X, General Estoria I) ‘So Labam said, according to what the Bible says: if you would put pressure on my daughters and you would leave them and take other women, (…). And this hillock and this heap of these rocks will be between us as a witness.’ (19) Et allegaron le de suso de piedras & de tierra un monton tamanno que por cosa estranna fue alli tenudo despues toda uia. (1275, Alfonso X, General Estoria II) ‘And from above they threw a heap of stones and earth at him, so much that because of a strange reason it was still kept there afterwards.’ The head uses of montón frequently combine with adjectives denoting configurational dimensions (especially height and volume) of the heap, such as alto ‘high’ and grande ‘big’ or adjectives confirming their accumulated nature, such as hecho (por mano de hombre) ‘manmade’ and traído ‘brought (together)’. Among montón’s quantifying uses, gran is the only attested adjectival combination. Quantifying uses are syntactically more constrained, in the first place since they exclusively appear within the binominal construction. In other words, N2 has to be specified in quantifying uses, whereas in head uses, the extralinguistic referent of N2 is often obvious from socio-cultural customs or the scenery evoked. Where montón designates a woodpile for instance, the prepositional phrase specifying the type of N2 is added in 9 of 10 instances. The nature of the heap of harvest that will be divided and distributed is also specified in 19 of the 26 instances: whether you receive a tithe of a heap of money, of silk, or only of a heap of bread or wheat makes a substantial difference. The remaining typical usage contexts of montón’s literal use – i.e. when montón refers to a war booty, a vertically oriented accumulation of objects, a memorial stone or a heap or ruins – occur as easily in the single noun construction as in the binominal one. When montón designates a huge mountain, however, it appears without prepositional phrase, since it is superfluous to add that the mountain is made of stones, rocks or earth.76 || 76 Of 141 tokens, 83 instances – or a bit more than half of the occurrences – appear in the binominal construction. I schematically mention the total number of binominal realizations per use: 3 of 12 mountain readings, 9 of 19 vertical configuration uses, 7 of 16 occurrences of the

A diachronic corpus analysis of montón | 115

This observation brings us to the description of the combinatorial pattern of montón. The attested uses indeed pattern with a particular type of N2. When the vertical configuration of the heap is profiled, or when montón refers to a memorial or a mountain, the heap is always made of earth or stones. Unsurprisingly, tierra ‘earth’ and piedras ‘stones’ count as particularly frequent right collocates (yielding 7 instances each), in addition to pan ‘bread’, which is, with 11 occurrences (henceforth occ.), the most frequent N2 attested in Medieval Spanish. Needless to say, other senses of montón are hard to combine with a rocky nature. Similarly, pan exclusively combines with the frame of a heap of harvest (which is susceptible of being divided or distributed in the near future). War booties are generally made up of weapons or corpses. Woodpiles consist of (a type of) wood. Heaps of harvest combine with different types of crops. Tab. 2: Layering of uses activated by montón de in Medieval Spanish

1250–1450

# %

Head

Quantifier

Ambiguous

47 (0.76)

8 (0.13)

7 (0.11)

Table 2 gives the exact proportion of the different readings medieval montón de evokes in the binominal syntagm only. Slight deviations from the numbers provided by Table 1 can be explained in the light of the data set used, as Table 1 includes the uses outside the binominal phrase as well.

4.3.2 Classical Spanish: the emergence of the specifier use The development of montón de in Classical Spanish vis-à-vis its use in Medieval Spanish can be easily captured by the notions of continuation and gradual extension. The comparison of frequency tables 2 and 3 leads to the assumption that montón de’s GR set out, given the steady rise in grammaticalized uses (16%). Not only did the quantifier use become more exploited, montón de also adopted a second grammaticalized use. Tab. 3: Layering of uses activated by montón de in Classical Spanish

1450–1730

# %

Head

Quantifier

Specifier

162 (0.65)

44 (0.18)

27 (0.11)

Ambiguous Indeterminate 9 (0.04)

8 (0.03)

|| war booty sense, 9 of 15 heaps of harvest, 19 of 26 heaps of tithes, 9 of 11 woodpiles, 14 of 25 memorial heaps, 2 of 3 ruin heaps, and all of the ambiguous and quantifying uses.

116 | The development of montón de, a typical locus of grammaticalization

Without wishing to address at this point the question whether head, quantifier and specifier readings constitute separate constructions in Classical Spanish, the respective slight changes they underwent are described separately. Since in grammaticalization theory semantic changes are claimed to precede possible morphosyntactic changes (Hopper/Traugott 2003), I will start with the changes in meaning montón de underwent from the late 15th century onwards.

4.3.2.1 Head uses of montón de in Classical Spanish The original usage contexts or frames evoked by the literal interpretation as a heap or typical conical configuration (cf. Table 1) are continued in Classical Spanish, though in different proportions. Most commonly, the vertically oriented shape with a wide surface and tiny edge is profiled. Yet, this heap-like configuration is no longer necessarily high in the way the literal interpretation of the augmentative suffix -ón foresees. In (20), for instance, Aristotle is explaining natural phenomena and pigeons normally do not produce high heaps of bird droppings. In (21), the absolute quantifier muchos ‘many’ is added to N2, in addition to gran ‘big’ premodifying montón which appears to be redundant in combination with the augmentative -ón. This tendency is further corroborated by the fact that the adjective grande ‘big’ is slightly more frequently added in head than in quantifying uses, which thus suggests that the augmentative interpretation has bleached to a certain extent. Frequently, although not necessarily, the contextual frames are of religious nature (e.g. altars), evoke battle fields of market places. Crucially, the conical or vertically oriented configuration and the fact that the N2-entities are somehow gathered – i.e., that the heap of N2 is artificial and not accumulated coincidentally or by nature – are always profiled in the literal uses (22). (20) La manera como se pudren las cosas, dice Aristóteles, es y acontesce cuando el calor del ambiente es mayor que el calor natural de la cosa que se pudre: (…). Especialmente cuenta que un montón de estiércol de palomas se pudrió por darle muchos días el sol, (…). (1575, J. Huarte de San Juan, Examen de ingenios para las ciencias) ‘The way things go bad, says Aristotle, is and happens when the environmental heat is higher than the natural heat of the thing that is rotting: (…) Especially he explains that a heap of dove dung rotted for lying several days in the sun (…).’ (21) Desde allí llegamos á unos pueblos á donde los indios nos tenian prestos un gran monton de muchos plátanos, cañas dulces, cocos, almendras, raices, bizcocho, petates y dos puercos presos: y así de

A diachronic corpus analysis of montón | 117

pueblo en pueblo nos dieron catorce puercos, y de lo demás tanto que no se pudo traer todo. (1605, anon., Historia del descubrimiento de las regiones austriales hecho por el general Pedro Fernández de Quiró) ‘From there we arrived at some villages where the Indians offered us a big heap of many bananas, sweet reeds, coconuts, almonds, roots, cake, bedrolls and two caught pigs: and so from village to village they gave us fourteen pigs, and of the rest so much that we could not carry everything.’ (22) (...) dize que ay una unidad natural & otra graçiosa o otra sobrenatural & sobregraçiosa. La natural es en quatro maneras ca una es llamada unidad por una manera de ayuntamiento de muchas partes diferentes la una de la otra como pareçe en un monton de piedras que llamamos uno por razon que las piedras son yuntamente ayuntadas. (1494, Fray V. de Burgos, Traducción de El Libro de Propietatibus Rerum de B.A.) ‘(…) states that there is one natural unity and another gracious one or another supernatural or super-gracious one. The natural comes in four modalities because something is called a unity for the way in which it brings a lot of different parts together like a heap of stones, which we call one because the stones have been put next to each other.’ Further, two extensions of typical medieval usage contexts are observed. First, in addition to graves remembering beloved deceased persons or memorials symbolizing historical achievements, heaps of stones were frequently used to indicate the road or path to follow. Example (23) adequately illustrates this milestone interpretation by paraphrasing mojón ‘milestone’ by montón. The context of (24), where one interlocutor asks for the use of the heap of stones, equally indicates that montón de piedras serves as a landmark. Both contexts confirm that milestones are by definition made by men intentionally piling up stones along a vertical axe for maximal visibility. (23) (…) tomó la posesión, en nombre de Vuestra Majestad y de sus sucesores y herederos, por Castilla y León, y en señal de posesión real, cortó hierbas y mudó piedras e hizo con sus manos un gran mojón y montón de piedras guijarros o callaos, a lo cual le ayudaron los demás que allí estaban, y luego plantó en él la cruz que traía en el hombro, (…). (1580, P. Sarmiento de Gamboa, Los viajes al estrecho de Magallanes) ‘(…) he took possession of Castile and León, in the name of Your Majesty and your successors and heirs. And as a sign of royal possession, he cut herbs and moved stones and with his hands he made a big boundary stone and heap of pebbles or similar stones. The other peo-

118 | The development of montón de, a typical locus of grammaticalization

ple there helped him with it. After that he planted in it the cross he had brought on his shoulders, (…).’ (24) (...) – ¿Qué montón de piedras es aquel – preguntó Andrenio – que está en medio de las sendas? – Lleguémonos allá – dixo Critilo –, que el índice del numen vial juntamente nos está llamando y dirigiendo. (…) – Pero el montón de piedras ¿a qué propósito? – replicó Andrenio –: ¡estraño despejo del camino, amontonando tropiezos! – Estas piedras – respondió suspirando Critilo – las arrojan aquí los viandantes, que en esso pagan la enseñança: (…). (1651, B. Gracián, El Criticón I) ‘What a heap of stones is that over there – Andrenius asked – in the middle of the paths? – Let’s go there – Critilus said – because the sign of the road numen is calling and leading us. (…) – But what is the use of a heap of stones? – Andrenius replied: What a strange way of clearing the road, piling up obstacles! – These stones – Critilus answered sighing – are thrown down here by passerbies to pay for their education: (…).’ The second extension concerns the frequently exploited metaphor tu vientre es como un montón de trigo ‘your waist is like a heap of wheat’. The image of a heap of wheat evokes the idea of a source of fertility or an outstanding capacity for reproduction and is commonly used to describe biblical mother figures, such as Rebecca (26) or Mary, or young ladies physically looking fertile and who therefore constitute interesting matches ((25), (27)). Although in examples (25)– (27) montón de trigo is used metaphorically (cf. es como ‘is like’, llamar algo montón de trigo ‘to call something a heap of wheat’), these metaphors verge on quantifier and specifier uses. First, a heap of wheat by definition consists of a lot of grains (25). In (26), Rebecca is presented as the mother figures of all Christians, and given that the Roman Catholic Church dominated Western Europe since the Middle Ages, there must have been a lot of them (cf. abundan in (26)). Second, when used in predicative or categorizing contexts, the image evoked by tu vientre es ø un montón de trigo ‘your waist is ø a heap of wheat’ (cf. infra, (42)) has shifted from pragmatic inference to conventionalized metaphor where the images of the woman’s belly and the heap of wheat fully match or overlap. (25) (…), que juzgarán los desatinos que oyeron una religión con dolores de parto preñada y con un vientre, como dice el sposo de su sposa, sicut acervus tritici; como un montón de trigo, que, aunque es uno, son casi infinitos los granos. (1603, San Juan Bautista de la Concepción, Pláticas

A diachronic corpus analysis of montón | 119

a los religiosos) ‘(…) that they will judge the nonsense that followed a religion with pregnancy labor pains and a stomach, as the husband said to his spouse, sicut acervus tritici; like a heap of wheat, that although it is only one (heap), it has infinite grains.’ (26) Rebecca est Ecclesia. El vientre d'ésta es el sacro Baptismo, donde somos regenerados y nacemos de nuevo en el regadío celestial, donde abundan las miesses de Aquél que dize a su Yglesia: "Tu vientre es assí como un montón de trigo cercado de lirios". Cada fiel christiano es un grano de trigo que nace en el regadío inferior de la Yglesia, (…). (1540, F. de Osuna, Quinta parte del Abecedario espiritual) ‘Rebecca est Ecclesia. Her stomach is the holy baptism, where we are reformed and reborn in the celestial irrigation, where the grains are abundant of those who say to their church: “Your stomach is like a heap of wheat enclosed by irises.” Every true Christian is a grain of wheat that is born in the lower irrigation of the Church, (…).’ (27) Tu ombligo fabricó tan bien natura / como una bella luna en redondeza, / y siempre es fuerte, siempre fruto tiene. / Tu vientre, cual conviene, / como un montón de trigo, que en belleza /envuelto, y lirios, viene. (1554, B. Arias Montano, Paráfrasis sobre el Cantar de los cantares de Salomón en modo pastoril) ‘Your navel has produced such a good nature, is like a beautiful round moon, / and it is always strong, / it always bears fruit. / Your waist, as it should be, / is like a heap of wheat, that comes in wallowing in beauty and with lilies.’

4.3.2.2 Quantifying uses of montón de in Classical Spanish In comparison to Medieval Spanish, the quantifier uses in Classical Spanish have abstracted further away from the original frames evoked. When medieval montón de was used to introduce a huge number of inanimate N2 entities, the binominal syntagm evoked the idea of ‘possession’, viz. un montón de N2 constituted somebody’s worthy property. A common image in Medieval Spanish was a heap of remarkable traits, moral or immoral acts someone gathered during his life (cf. (16)). In many religions, a person’s life is thought of as an accumulation of virtues or sins the person is judged for by Heaven’s gatekeeper. In line with the literal reading of montón as a heap of goods that will be measured precisely, various religions or belief systems predict that at Heaven’s door, a balance measures both the weight of the heap of virtues and of the heap of bad moral behavior in order to decide whether the deceased person could pass through.

120 | The development of montón de, a typical locus of grammaticalization

When medieval montón de introduced human N2 entities, the N2s were spatiotemporally contiguous and gathered by an external cause or principle. In (32), different roman characters are brought together in one particular epic poem by Virgil, viz. the Aeneid. (28) Vimos Omero tener en las manos / la dulçe Yliada con el Odosia; /el alto Virgilyo vi que lo seguia / en vno con otro monton de romanos, / tragicos, líricos, elegianos, / comicos, satiros, con eroquistas, / e los escriptores de tantas conquistas / (…). (1424, anon., Cancionero de Juan Fernández de Íxar) ‘We saw Homer holding in his hands /the sweet Iliad and the Odyssey; / the tall Virgil saw he was following him/ with a heap of romans, / who were tragic, lyric, elegiac, / comics, satyrs and epics (lit. heroics) / and the writers of many conquests; (…).’ In Classical Spanish, quantifying uses still combine with the frames of ‘property or possession’ related to wealth or moral attitudes, yet not exclusively. Further, the quantifier uses are no longer mainly triggered by premodifying adjectives such as gran ‘big’: only 6 instances of quantifying montón de combine with gran or grueso ‘thick’. In addition to human N2s or inanimate concrete objects, the combinatorial pattern of quantifying montón de extended as to include all kinds of abstract nouns (instead of remaining restricted to abstract notions of moral behavior which were still metaphorically heaped up), as in (29). The latter extensions testify to a conventionalized association between montón de and quantity assessment in Classical Spanish, since it is hard to imagine a context where abstract nouns such as mentiras ‘lies’ or siglos ‘centuries’ are heaped up literally. (29) (…), pero don Quijote, temeroso que Sancho se descosiese y desbuchase algún montón de maliciosas necedades y tocase en puntos que no le estarían bien a su crédito, (…). (1615, M. de Cervantes Saavedra, Segunda parte del ingenioso caballero don Quijote de la Mancha) ‘(…), but Don Quixote, afraid that Sancho would reveal and blurt out some heap of malicious nonsense and would say things that would discredit him, (…).’ Interestingly, the observed quantifying uses in Classical Spanish often continue to activate one or more facets of the original usage contexts or frames, such as the idea of accumulating good deeds by singing lots of psalms in (30), the image of heaped up possessions in (31) or simply the war booty (32). Yet many examples are also observed which cannot be directly linked to the original frames

A diachronic corpus analysis of montón | 121

evoked by montón de. However, in (33), the N2-entities continue to be somehow gathered. In other words, the conceptual image of montón de may have schematized to the extent of simply expressing a large quantity of spatiotemporally contiguous N2s. (30) Y de aquí viene que piensan otros porque rezan un montón de salmos o manadas de rosarios, otros porque traen un hábito de la Merced, (…), que ya no les falta nada para ser muy buenos cristianos, (…). (1527, A. de Valdés, Diálogo de las cosas acaecidas en Roma) ‘And that is why some think because they pray many (lit. a heap of ) psalms or rosaries, others because they bring a habit from the Mercedarian Order, (…), that they are well on their way to be good Christians, (…)’ (31) (…) Su Majestad os quiere dar el hábito de Santiago, sin que toméis trabajo de hacer la probanza, en recompensa de lo que habéis servido, y por enmienda del daño que recibisteis cuando os pusisteis la Cruz de San Juan, pues en verdad que campeará mal el humilladero sobre el monton de brocado que vuestra merced suele traer. (1550, D. Hurtado de Mendoza, Carta) ‘(…) Her Majesty wants to give you the habit of Santiago, without having to give proof of it, as a reward for what you have served and as an amendment for the pain you suffered when you had to wear the Cross of Saint John. The truth is the shrine will not look very good with the heap of brocade your mercy usually brings.’ (32) No le abastaua que leuaua vn monton de reyes moros consigo / y todo el poder de sus reynos. (1499, G. García de Santa María, Traducción de la Corónica de Aragón de fray G.F. de Vagad) ‘It wasn’t enough that he had brought a heap of North African kings with him / and all the power of their kingdoms.’ (33) (…), está la iglesia con más gente, ve el cortesano muchos conocidos, levántase de allí y vase con ellos. Hácese un montón de hombres, con tal organización que todos se hablan y todos se escuchan, (…). (1654, J. de Zabaleta, El día de fiesta por la mañana) ‘(…), the church is full of people, the courtier sees many familiar faces, he stands up and goes with them. There are a heap of men. They are so well organized that they talk and listen to each other without disturbing, (…).’ It is worth mentioning a recurrent type of N2. In 16 occurrences, N2 denotes speech acts or pieces of discourse. In other words, the N2-entities are of ‘spoken’ nature, e.g. mentiras ‘lies’ (34). The image of a heap of N2-entities ‘pronounced

122 | The development of montón de, a typical locus of grammaticalization

out loud’ recalls the literal facet of a heap of things gathered by human endeavor: N2s ‘produced by men’ are then a metonymical extension of ‘heaped up intentionally’ in which the facet ‘produced by men’ shifts from N1 to N2. (34) (…), después de haber derramado un montón de mentiras y burlerías sobre mí y sobre la doncella, lo dejé. (1587, F. de Mena, Traducción de la Historia etiópica de los amores de Teágenes y Cariclea de Heliodoro) ‘(…), after having told a heap of lies and mockeries about me and about the maiden, I left him.’ The facet ‘gathered’ or ‘accumulated’ is indeed the most frequently activated one in quantifying uses of montón de. When human N2s are quantified by montón de, the conceptualization as ‘being gathered’ often bleaches into mere spatiotemporal contiguity, as in (35). (35) Señor Araujo, dije sepa que después que se acostó han venido un montón de huéspedes, y yo, por la lástima que he tenido desta pobre mesonera y porque no pierda la ganancia, los he hecho las camas y acomodádoselos a todos. (1605, F. López de Úbeda, La pícara Justina) ‘Mister Araujo, I said so you would know that after you went to bed a heap of guests came, and because I felt sorry for the poor landlady and so you wouldn’t lose any income, I made up the beds and accommodated them all.’ The final extension observed for quantifying montón de concerns its interpretation as a ‘mishmash or mess of heterogeneous N2s’. In (36), the coordination of the speaker’s passions, thoughts and affections testifies to a diversified range of N2s which is further considered to disturb the speaker’s heart. Juxtaposing tropel ‘troupe, mob’ and montón equally corroborates the turbulent nature of the N2s, as tropel literally denotes a ‘troupe’ and is (or will later come to be) used for quantifying a confused mass. In (37), the premodifying adjective confuso ‘confused’ intensifies the diverse nature of the entering people, which is further corroborated by describing the scenery as el alboroto de la gente ‘the agitation, commotion of the people’ and the resulting confusión de sus padres ‘the confusion of their parents’. (36) Ca si la gente se llama turba por que perturba alos otros qual ayuntamiento de gentes mas rezia mente perturba mi coraçon que el tropel / y monton de mis passiones pensamientos afecciones. (1456, A. de Cartagena, Apología super salmo "Judica me Deus") ‘Because if people are called a mob because they disturb the others

A diachronic corpus analysis of montón | 123

like a union of people disturbs my heart more stoutly than a mob / and a heap of my passions, thoughts, affections. (37) (…), vieron entrar con ellos un confuso montón de gente, que traían en hombros, sobre una silla sentado, un hombre como muerto, que luego supieron ser el conde que había heredado al enemigo que solía ser de su tío. El alboroto de la gente, la confusión de sus padres, el cuidado de recebir los nuevos huéspedes, las turbó de manera que no sabían a quién acudir ni a quién preguntar la causa de aquel alboroto. (1616, M. de Cervantes Saavedra, Los trabajos de Persiles y Segismunda) ‘(…), they saw that along entered a confused heap of people. On their shoulders these people were carrying an apparently dead man sitting on a chair. Later they got to know he was the count who had inherited the enemy that used to be his uncle’s. The commotion of the people, the confusion of the parents, the care in receiving the new guests, confused the girls in such a way that they didn’t know who to go to nor who to ask about this disturbance.’ Summarizing, the previous paragraphs argue that the conceptual image originally evoked by montón de progressively fades away. In line with both Langacker’s (2006) and Traugott’s (2010) view on subjectification, I do not consider this ‘bleaching’ as a loss of complexity. Instead, Figure 1 illustrates how slight alterations in profiling occur: while some elements which are originally profiled cease to be stressed, others come out of the background.

a. conical configuration of

b. gathered heap or

c. gathered heap or

(nearly) identical N2s heaped

accumulation of

accumulation of

up by human endeavor

homogeneous N2s

heterogeneous N2s

Fig. 1: Gradual schematization of montón de N2

The leftmost Figure 1(a) visualizes a literal heap of goods or achievements heaped up intentionally by human endeavor. The triangle reminds of the typical conical configuration which results from people accumulating things for a spe-

124 | The development of montón de, a typical locus of grammaticalization

cific purpose. Initially, heaps were always made up of a particular type of N2. There was for instance a heap of wheat next to the heap of barley, which were nicely separated for purposes of measurement. Mixing both heaps up would render the measurement and distribution infinitely difficult. The accumulated nature becomes profiled in the first quantifying uses of montón de, as in Figure 1(b): the bold triangle symbolizes the accumulated nature of the homogeneous N2 entities, which are by definition spatiotemporally contiguous. If the facet ‘accumulation’ is the most prominent one, it is does not matter whether the N2s are nearly identical or heterogeneous. In other words, when abstracting further away, montón de may come to profile the accumulated nature or spatiotemporal contiguity of different types of N2 which share a similar purpose (in (36) the passions, thoughts and affections all together disturb the speaker; in (37) the confused jumble of men came in together carrying the count). The idea that a heterogeneous collection of different N2 entities may be quantified by montón de simply for sharing a similar purpose, paves the way for the idea of putting different N2s in one box, i.e. for categorizing or specifying uses, as sketched in Figure 1(c).

4.3.2.3 Specifying uses of montón de in Classical Spanish The subjectification or schematization montón de underwent heralds its final extension in Classical Spanish, viz. the development of the two-way specifier use. The categorizing potential of montón de stems from its original aptitude to denote the heaping up of identical nominal entities (N2). Via a metonymical shift in profile, montón de comes to evoke an accumulation of (possibly different) things which serve a common purpose or share a specific and prominent characteristic. Specifying montón de then serves to profile a specific quality of the set of N2s. This categorizing use continues to single out a set of N2s –which is the basic function of BQs– in that it bounds the N2s under a single denominator. The specifying potential was already obvious in recurrent medieval occurrences where montón de N2 (generally of stones or ruins) denotes the left-over of a city after a battle, as in (38). Example (39) explicitly illustrates the chain of pragmatic inferences: if we see a heap of barley, we see a lot of barley and may equally say that everything we see is barley or that we see nothing but barley. When a city is defeated, there is nothing left but stones. (38) & fablo a tod el pueblo de Juda & dixol. estas cosas dize el sennor de las huestes. Arada sera syon como heredat de pan. & Jherusalem tornada en monton de piedras. (1280, Alfonso X, General Estoria IV)

A diachronic corpus analysis of montón | 125

‘He spoke to all the people of Judah and said, this is what the Lord of the armies has said. Zion will become a wheat field and Jerusalem will turn into a heap of stones.’ (39) Onde, conosçense los pecados mortales en general & nonbranse asi munchas vegadas, mas quando vienen a ser por menudo escodrinnados, non se pueden tan çierto connosçer nin tan çierto nonbrar por mortales, ca se fallan a las vegadas veniales, asi como si vemos un monton de ordio, podemos dezir que todo es ordio, mas si viniesemos a contar los granos por menudo, fallariamos y granos de trigo. (1500, M. Pérez, Libro de las confesiones) ‘Where, the mortal sins in general are known and many sins are named like that, at the most when they happen to be often examined, they can’t really be known nor can they truly be called mortal, because in rare occasions they are considered venial sins, just like when we see a heap of barley, we can say that it is all barley, but if we would count the grains in particular, we would fail and consider them wheat.’ In Classical Spanish, specifier uses of montón de are triggered by the appearance in particular predicative contexts: in clauses such as X no es más que… ‘X is nothing more than’ and no es sino… ‘he is nothing but…’, in apostrophes, after exclamative qué, as the second element of reformulations, as the predicate of copular verbs, etc. Examples (40)–(41) nicely illustrate how, in specifier uses, a particular entity (e.g. the heart of a sinner in (40)) is categorized as un montón de peccados. Literally, the heart of a sinner is nothing but a heap of sins. A heap of sins implies both a lot of sins as well as nothing but sins (instead of also containing some virtues). The pragmatic inference of ‘nothing but N2’ is conventionalized in specifier uses. In other words, in specifier uses, montón de serves to present N2 as the essence of its referent. Similarly, in (41), the entire Koran is identified as a heap of foolishness, as nothing but blunder. The fact that the context usually depicts a certain entity X which can be ‘reduced’ to un montón de N2, motivates the negative connotation that is associated with the specifier uses. (40) Assí que el coraçón del hombre peccador no es sino un montón de peccados, el qual, aunque de fuera no peque, continamente en lo de dentro siempre pecca, pues que está determinado de executar y poner en obra qualquiera de sus malos propósitos. (1540, F. de Osuna, Sexta parte del Abecedario espiritual) ‘So the heart of a sinner is no other than a heap of sins, that even if on the outside it doesn’t sin, on the inside it is continuously and always

126 | The development of montón de, a typical locus of grammaticalization

sinning, as he is determined to execute and carry out every one of his bad intentions.’ (41) (…); y siendo todo el Alcorán un montón de desatinos, sobresalió tanto éste que, con toda su barbaridad, le han conocido los sectarios y no le observan. (…). (1653, J. de Zabaleta, Errores celebrados) ‘(…) and the entire Koran is a heap of nonsense, but this one stood out so much that in its atrocity sectarians got to know it and they don’t notice it. (…)’. Two frequent conceptual images stand out as typical specifier uses in Classical Spanish, viz. un montón de trigo and un montón de tierra. As in (42) the metaphor is explained in the context and reference is made to the number of grains in a heap of wheat, the example may be considered a borderline case. Still, the idea the speaker wants to convey is that ‘your waist is nothing but fertility’. Example (43) demonstrates the origin of the lexicalized idiom montón de tierra which is still used in Present-Day Spanish to denote an extremely old, weak and ailing (and by extension idle) person. The metaphor is used to identify persons as completely devoid of human development or humanity and stems from the biblical Creation story ‘For you are dust, and to dust you shall return’ (Gen.3:19). Originally, a heap of dust or earth is of course worthless in comparison to a heap of achievements. In (43) Death is trying to convince a very old man to join him and die, as he has already become a heap of earth and shows several signs of physical deterioration. In (44), the speaker categorizes himself as an outsider, as an underdog in comparison to the king. This binominal combination thus seems to have lexicalized during the 16th and 17th century.77 It is first mentioned as a colloquial expression by the reference dictionary Autoridades of the RAE in 1734. (42) Alaba, pues, esposo, en el alma santa la fecundidad y abundancia de buenas obras. Tu vientre es un montón de trigo. Partos corporales ha habido numerosos, pero todo esto es poco respecto de la fecundidad espiritual del alma. ¿Cuántos granos hay en el montón de trigo? Pues

|| 77 The categorization of un montón de tierra as a lexalized expression is based on Himmelmann’s (2004, 31) definition of lexicalization as the process whereby a particular node (e.g. montón), begins to form an especially close collocation with a particular item of its hostclass (e.g. tierra), because this phrase is frequently used in an extended metaphorical meaning (e.g. montón de tierra starts to be used for all kinds of idle persons, and not exclusively for the elderly ones any more). The new phrasal unit continues to be used in essentially the same syntactic and pragmatic contexts and may continue to remain transparent (or compositional) for a long time.

A diachronic corpus analysis of montón | 127

más son los hijos espirituales de la buena voluntad; muchos y buenos, como el trigo, que es el mejor fruto de la tierra. (1598, Fray A. de Cabrera, De las consideraciones sobre todos los evangelios de la Cuaresma) ‘So spouse, praise in the holy spirit the fertility and abundance of good deeds. Your belly is a heap of wheat. There have been numerous corporal deliveries, but all of this is little compared to the spiritual fertility of the soul. How many grains are there in the heap of wheat? Well, the spiritual sons of goodwill; there are many of them and they are good as is the wheat. It’s the best fruit on earth.’ (43) Habla la muerte con el viejo en la decrépita edad. Tú, que ya montón de tierra / estás hecho y consumido, /según lo que has ya vivido la vida te haze guerra. / Ya tienes calva la sierra / y aún has del mundo cariños, / dexa el pan para los niños / que ya el mundo te destierra. (1540, S. de Horozco, Cancionero) ‘Death speaks with the man of decrepit age. You, already a heap of earth / are done and consumed, / according to what you have lived life wages war on you. / The mountains are already bald / and still you receive affection from the world, / leave the bread for the children / the world already exiles you.’ (44) (…) vos se los dareis despues en mi nombre, y mirad que la hagais, que tiene buen natural y lo sabra aprouechar el Principe, que digo Principe? el Rey, que yo no soy sino un monton de tierra. (1621, A. de Almansa y Mendoza, Carta que escribió un señor de esta Corte a un amigo) ‘(…) you will give them (the good advices) to him later on my behalf, and make sure you do so, then he is a good person and the prince will know how to make the most of it, what am I saying, the Prince? I mean the King, then I am no more than a heap of earth.’

4.3.2.4 The combinatorial pattern of montón de in Classical Spanish The morphosyntactic make-up of the binominal uses of montón de in Classical Spanish continues to display typical noun features. The separability of the prepositional phrase is observed in no more than 10 instances, however. In 11 occurrences (of which one quantifier use), an additional determiner (generally the definite) precedes N2. Although formal persistence should not surprise us in montón de’s head uses, the progressive conventionalization of the functional uses, i.e. the acquirement of a more operator-like function of montón de, is expected to go hand in hand with formal signs of the shift from noun to quantifi-

128 | The development of montón de, a typical locus of grammaticalization

er/specifier category. Grammaticalized uses of montón de in Classical Spanish still combine with typical noun features, yet formal variation as to the determiners to N1, N1’s adjectival modification and fluctuation in verb agreement is not unrestricted. Discussing the fluctuations in the co-selection pattern regarding montón de from left to right, it is important to note that all three uses show strong preference for the indefinite determiner. In about 60% of its uses, montón de combines with un.78 The head use also frequently combines with the definite determiner (in 33 occ.), in contrast to the quantifier and specifier use (which combine with el in 3 and 1 instances respectively). Further, whereas the head use easily combines with a large range of determiners (including the unspecific indefinite algún ‘some’, the demonstratives aquel ‘that’ and este ‘this’, interrogative qué ‘what’, the postdeterminer otro ‘other’ and possessive su ‘his, her’), quantifying montón de does allow for exclamative qué ‘what’, demonstrative este and algún, yet only sparsely. In no more than 3 occurrences, specifying montón de combines with other determiners, more precisely with el ‘the’, aqueste ‘that’, este ‘this’ and tal ‘such’. As demonstrative determiners may also signal the speaker's endorsement of the categorization (cf. Delbecque 2011), they are not necessarily incompatible with the more operator-like function of specifying montón de. As to the modification of N1 by adjectives, each use is associated with a particular adjectival pattern. Adjectives either immediately precede the QN or immediately follow montón. In its head use, montón de only combines with configurational adjectives, viz. adjectives specifying a dimensional characteristic of montón, such as gran ‘big’ (25 occ.), compuesto ‘composed’, menguado ‘diminished’, rojo ‘red’, redondeado ‘roundish’. The quantifier uses of classical montón de combine 5 times with gran and 1 time with grueso ‘thick’, which then function as intensifiers. Two instances also combine with descriptive adjectives such as ciego ‘blind’ (45) and buen barato ‘good (and) cheap’, which suggests that the morphosyntactic change had not come to completion yet by the 18th century. Specifier uses exclusively combine with qualifying adjectives such as alegre ‘happy’, caduco ‘faded’, ciego ‘blind’, confuso ‘blind’, hermoso ‘beautiful’, quimérico ‘chimeric’ (46) and sucio ‘dirty’, which conceptually seem to scope over N2 (rather than over N1). Since specifying montón de primarily identifies a particular entity as nothing but N2, the addition of a qualifying adjective may reinforce the categorization construal. The adjective may then equally combine with N1 or N2. In other words, head uses of montón de combine with dimension-

|| 78 Un occurs in 90 of 163 head uses (0.55), in 25 of 44 quantifier uses (0.57) and in 15 of 27 specifier uses (0.56).

A diachronic corpus analysis of montón | 129

al adjectives, grammaticalized uses with adjectives intensifying the quantifying or specifying interpretation of the QN. (45) En igual ademan el campo griego / Vió á los fieros verdugos entregada / La bella hija del rey, que el sagaz ruego / De Ulíses dió por víctima sagrada, / Y á la orilla del mar, de un monton ciego / De armas, hácia la selva mas guardada, (…). (1624, B. de Balbuena, El Bernardo) ‘In the same gesture the Greek army saw the king’s beautiful daughter handed over to the fierce tyrants. Ulysses’ shrewd request declared her a holy victim. And at the seaside from a blind heap of weapons to the most hidden forest. (…).’ (46) Grandemente – dixo – os ha contentado este montón quimérico de gustos, este agregado fantástico de bienes, pero advertid que es tan fácil de imaginar quan impossible de conseguir; porque ¿quál de los mortales pudo jamás llegar a esta felicidad soñada? (1657, B. Gracián, El Criticón III) ‘Greatly he said, this chimerical heap of tastes, this fantastic addition of goods has pleased you, but be warned that it is as easy to imagine as difficult to obtain; because which one of the mortals could ever achieve this divine happiness?’ Interestingly, in almost half of the occurrences where quantifying montón de N2 functions as the clausal subject (in 3 of 8 occ.), the verb agrees with N2, as in (47). This tendency gives proof of the operator status of montón de, as the verb is supposed to agree with the head of the binominal syntagm (Keizer 2007; Aarts 1998). Nevertheless, the agreement criterion is of limited applicability, as it only applies when the binominal syntagm appears in the subject position (and agreement with N1 continues to be more frequent). (47) De la simiente mediante la & ccedil;ima la espiga se llevanta la qual segund Isidoro es dicha de spiculo que quiere dezir saetilla porque en la espiga naçen un monton daristas que son agudas como saetas. (1494, Fray V. de Burgos, Traducción de El Libro de Propietatibus Rerum de B.A.) ‘From the seed through the top the ear raises itself, which according to Isidoro is a sharp herb which means arrowhead because from the ear sprout a heap of thorns that are sharp like arrows.’ In contrast to the quantitative study of the use of montón de in Medieval Spanish which analyzes all occurrences of montón exhaustively, the occurrences of montón of Classical Spanish are filtered manually in order to assure that in all occurrences of the final dataset montón is coextensive with N2. Of 625 examples

130 | The development of montón de, a typical locus of grammaticalization

in CORDE, 251 were withheld, which suggests that montón does not prefer the binominal context over its use as a bare noun, rather reversely. The occurrences which were discarded from the quantitative analysis count hundreds of examples of the semi-fixed expressions traer/echar/tomar/tirar a montón ‘to bring/to throw/to take/to throw everything together (lit. on (the) heap)’, (correr) en montón ‘(to run) together’ and sacar/pagar de montón ‘to pay from the heap’ which continued to evoke to feudal frame (as in (48) and (49)). Crucially, the pure quantifying interpretation of montón has not yet expanded towards other categories than the nominal one. (48) Para que pueda hazerse mejor la dicha partición, toda la ganancia deve ser traýda a montón, y, aviéndose d'ella sacado los derechos de Rey y pagado las enchas, y gastos y otras cosas que de suso diximos, deven aver los officiales; (…). (1540, H . de Celso, Repertorio universal de todas las leyes de estos reinos de Castilla) ‘In order to make the distribution easier, all the profits must be brought all together (lit. to heap). After having extracted the king’s rights and having paid the help and costs and other things that are said from above, there must be the official ones.’ (49) Repartiose por cabeças entre los españoles; no se dio todo, sino señalose a cada vno según era. Al de cauallo, doblado que al peón, y a los oficiales y personas de cargo o cuenta se dio ventaja. Pagósele a Cortés de montón lo que le prometieron en la Veracruz; (…). (1553, F. López de Gómara, II parte de la Crónica general de las Indias) ‘It was divided by heads between the Spaniards; not everything was given, but each one was indicated his share. The cavalrymen, twice as many as the infantrymen, the officers and the people of authority or high charge were given advantage. Cortés was paid from the whole amount (lit. from the heap) what they had promised him in Veracruz.’ The rightmost slot of the binominal syntagm, viz. N2, appears to be less constrained than in Medieval Spanish. With regard to head uses, heaps of harvest (67 occ.) are still made of harvested crops and count 53 combinations with trigo. Memorials or milestones (29 occ.) are composed of stones, heap-like configurations out of earth (tierra, 20 occ.), sand or mud. Woodpiles contain wood only, war booties are heaped up corpses or weapons. However, as Classical montón de N2 is characterized by the semantic glide towards ‘heap of things accumulated along a vertical axe’, abstracting away from the initial frames, the combinatorial pattern slightly extends towards all kinds of stackable and concrete objects, such as hojas ‘leafs’, vajilla ‘plates’, cartas ‘letters, basura ‘dirt’. Since the restriction to ‘stackable objects’ is lifted in quantifier uses, the combinatorial

A diachronic corpus analysis of montón | 131

pattern extends so as to systematically include human or animate N2s and also abstract notions. Given that abstract N2s are in no way compatible with the original meaning of montón de, the quantifying interpretation appears to have spread to new contexts incompatible with the head use. In other words, ‘reanalysis’ seems to have occurred or reached the surface structure. Although no particular lexeme is particularly preferred with quantifying montón de, it is possible to trace some recurrent semantic families: discourse entities (16 occ.) such as quejumbres ‘complaints’, people (9 occ.) such as huéspedes ‘guests’ or animals (4 occ.) such as ganado ‘catlle’ and moral behavior (5 occ.) such as virtudes ‘virtues’. In specifier uses, the original collocational pattern equally extended to include abstract notions (2 occ.), such as miserias ‘misfortunes’, people (4 occ.) such as gente ‘people’, discourse entities (5 occ.) such as dichos ‘saying’. As has already been mentioned before, two particular N2-combinations stand out in the specifier uses, viz. tierra and trigo, counting 5 and 2 instances respectively. Against this background, it is fair to say that montón de’s progressive GR set out in Classical Spanish: the quantifier interpretation has become conventionalized and a second grammaticalized use, viz. the specifier use, emerged. Further, fluctuations in the co-selection become increasingly restricted for the grammaticalized readings while the expansion of the right combinatorial pattern gives evidence of the spread of the grammaticalized uses to new contexts.

4.3.3 Modern Spanish: partial continuation and gradual extension Globally speaking, the development of montón de in Modern Spanish consists in a further loss of specificity or schematization, which is illustrated in the following passage from friar Benito Jerónimo Feijóo, a famous Spanish monk and essayist known for his effort to unmask superstitions and popular beliefs: (50) Crít. (…) Porque si preguntáis a varios hombres sobre la cantidad de trigo que es menester para tener la denominación de montón, os responderán con mucha diversidad. Unos os dirán que son menester, pongo por exemplo, quatro hanegas; otros dirán que basta medio celemín; otros ocurrirán a la pregunta distinguiendo montón grande, pequeño y mediano; otros más formales añadirán a estas tres diferencias las dos de mínimo y máximo. Dialéct. No obstante la diversidad que me representáis, creo yo que todo el mundo convendrá en entender por montón de trigo una colección de muchos granos de trigo, pues esta explicación se verifica en el montón grande, en el pequeño, en el mediano, &c. (1739, B. J. Feijoo, Theatro Crítico

132 | The development of montón de, a typical locus of grammaticalization

Universal o discursos varios en todo género de materias…) ‘Crít. (…). Because if you would ask several men about the amount of wheat needed to be called a heap, they would give different answers. Some will say that, I’m giving an example, four bushels are required, others will say half a bushel will do, others will answer the question distinguishing a big, small or medium-sized heap; others, more formal persons, will add to these three differences the distinctions minimum and maximum. Dialéct. In spite of the diversity you present me, I think everyone will agree on the understanding of a big heap of wheat as a collection of many grains of wheat, then this explanation is established in the big, small, medium-sized heap etc.’ The data analysis reveals that the conceptualization of montón de, both in its literal and grammaticalized uses, abstracts further away from its original uses and allows for new extensions. Literal heaps are no longer necessarily high – their surface can extend so that the facet ‘vertically oriented’ does no longer apply. Yet, what persists is the conglomerate configuration or accumulated nature of their components. After having affected the head use already in the preceding period, the pragmatic inference of ‘chaotic heap’ that originates in the lifting of the obligatory facets ‘intentionally accumulated’ – and by extension ‘orderly’, now also reaches the quantifying and specifying uses: montón de now usually refers to a chaotic accumulation of objects, persons or abstract notions. In other words, although montón de’s meaning appears to become more vague or general, the proportion of its uses (cf. Table 4), continues the distribution found in Classical Spanish. The rise in specifying uses catches the eye, which by then exceed the quantifying uses. Further, slightly more ambivalent uses are observed. A rise in ambiguous uses could imply the conventionalization of the quantifying reading of montón de, but as ambiguity may be produced both accidentally as on purpose, it is less significant than a rise in indeterminate uses, where the speaker simultaneously appeals both to the original literal and the newer grammaticalized reading. In (51) and (52), indeterminacy between a literal interpretation on the one hand, and respectively a quantifying or a specifying reading on the other hand is observed. Interestingly, half of the occurrences where montón de simultaneously exploits two readings show indeterminacy between a pure quantifying and a specifying reading (as in (53)), which corroborates the observation that the specifying use continues to ‘bound’ or specify the size of a collection of N2 (yet relegates quantifying information the background) and that quantifying uses do more than merely quantifying a group of N2s, they also add a qualifying component (see Chapters 6 and 7).

A diachronic corpus analysis of montón | 133

Tab. 4: Layering of uses activated by montón de in Modern Spanish

1730–1900

# %

Head

Quantifier

Specifier

217 (0.60)

42 (0.11)

58 (0.16)

Ambiguous Indeterminate 24 (0.07)

26 (0.07)

(51) Se hacen unos croquetones aumentando la sazón, (…). Se sirven muy bien arregladitos en una fuente, en pila simétrica, oculta bajo un montón abundante de patatas fritas, cortadas en ruedas. (1891, Á. Muro, El Practicón. Tratado completo de cocina) ‘Make big croquettes increasing the flavor, (…). Serve them nicely on a dish in symmetrical rows hidden under a big heap of chips cut in wheel shape.’ (52) Las calles, estrechas y tortuosas, estaban obstruidas con los muebles y escaparates que los moros habían roto en su despiadada saña; algunos cadáveres completamente desnudos, asomaban por entre este montón de escombros, y un pueblo loco de alegría, pero andrajoso y repugnante, abalanzándose frenéticamente a nuestros soldados (…). (1860, G. Núñez de Arce, Recuerdos de la campaña de África) ‘The narrow and tortuous streets were obstructed with the furniture and shop windows that the Moorish had broken with merciless brutality; some totally naked bodies stuck out of this heap of rubble while an insanely happy, but ragged and disgusting village was leaping on our soldiers, (…).’ (53) Bajaba, (…), recapitulando lo más sustancial y práctico de lo muchísimo que había cavilado por la noche; contemplaba por última vez, (…), examinaba después el cuadro de sucesos e impresiones que me había traído últimamente a aquellas tan peregrinas andanzas; empeñábame de nuevo en distinguir lo principal de lo accesorio, las causas de los efectos, en el complejo montón de ideas e impresiones que me llenaba la cabeza y el corazón; (…) (1895, J.M. de Pereda, Peñas arriba) ‘I went (…) thinking over the most important and practical things I had been pondering on that night; I contemplated for the last time (…). I then examined the picture of the events and impressions that recently had led me to these wandering adventures. Again I insisted on distinguishing the main thing from the incidental, the cause and the effect in the complex heap of ideas and impressions that tilled my head and heart (…).’

134 | The development of montón de, a typical locus of grammaticalization

4.3.3.1 Head uses of montón de in Modern Spanish With regard to the literal uses in Modern Spanish, the corpus shows that part of the medieval frames continue to be exploited, although to a limited extent. Instances of woodpiles, milestones, mountainous configurations, graveyards and heaps of harvest are found marginally. The war booty sense is activated in about a fifth of the Modern occurrences. More important, however, is the exploitation of montón de’s more general use of a heap of things intentionally accumulated and put one above the other. However, in Modern Spanish, heaps of N2s are only rarely conceptualized as high, which means that montón de progressively gains independence from its original usage contexts, which facilitates its broadening to new contexts. Three particular extensions are observed. First, montón de is commonly used to refer to a pile of paper, as in (54). Further, in addition to the reference to a heap of stones serving as a seat (already in Classical Spanish), montón de (and especially montón de paja) is frequently used to indicate a kind of bed or place to sleep, as in (55). This example highlights a pragmatic inference frequently evoked by montón de in Classical Spanish: Ruperto’s bed is presented as restricted to a heap of straw and thus of poor quality. In line with the loss of the facet ‘high’, montón de acquired the association with inferiority and even poverty: if a heap (of achievements of possessions) is not high, it is less worthy. The latter pragmatic inference brings us to the third extension of montón de to indicate a heap of residues, left-overs and waste. Example (56) literally says that the heap of ashes is what is left from the burned house and its owner, and illustrates that the idea of ‘intentionally accumulated N2s’ may fade into mere ‘spatiotemporal contiguity’ as a result of a transformation process. This final extension of montón de’s head uses verges on the specifier use. Yet in (56), a given entity (the house and its owner) turns into ash and the question ‘what is left in its place’ implies that the ashes literally occupy some space. When used as a specifier, the binominal syntagm would simply categorize the entity as nothing but ashes. (54) Asmát. ¡Y que luego en Madrid nos motejaran de impolíticos! (Sale Don Pedro, con un montón de esquelas.) D. Pedro. (Al Asmático, al entrarse.) Tomad; tomad. (Al Lacayo.) Asmát. Decid de palabra qué mandáis. (1768, R. de la Cruz, Las superfluidades) ‘Asthmatic: And to think that later in Madrid they branded us apolitical. (Pedro leaves with a heap of death announcements.) Pedro: (To the Asthmatic entering): Here, take this. (To the Lackey.) Asthmatic: Speak, what have you brought?’

A diachronic corpus analysis of montón | 135

(55) La barraca, que no tenía más de cincuenta pies, estaba dividida en dos mitades, habitada la una por tres gallinas y un cerdo; habitada la otra por Ruperto y su mujer, cuyo lecho se reducía a un montón de paja. (1876, Á. Grassi, El copo de Nieve) ‘The hut, no bigger than fifty feet, was divided into two parts, one was inhabited by three chicken and a pig, the other by Rupert and his wife, whose bed was no more than (lit. reduced to) a heap of straw.’ (56) (…); pero el huésped me asesinó y puso fuego a mi casa. ¿Qué ha quedado en lugar suyo y de su dueño?, ¡unas gotas de sangre y un montón de cenizas! (1844, E. Gil y Carrasco, El Señor de Bembibre) ‘(…); but the guest murdered me and burned down my house. What was left in its place and in its owner’s place? A few drops of blood and a heap of ashes!’ In order to get a better grip on the broadening of montón de’s head use, Table 5 contains the N2 combinations observed, which can be organized into several semantic families of clusters of N2s. The number immediately following the lexical items corresponds to the number of attestations observed in my corpus. Although the most frequent combination is trigo – which cathes the eye with 17 occurrences – the harvest-frame is not the most prominent one in Modern Spanish. In fact, the conceptual image most frequently evoked is the schematized one, which contains the facets ‘vertically oriented’, ‘conical configuration’ and ‘accumulated (intentionally)’ and does not tie down to a particular frame. This is not to deny that montón’s uses as woodpile, war booty, heap of harvest, heap of ruins or memorial are partially continued in Modern Spanish, but they obviously lose ground to the more schematized image of a vertically oriented and conical accumulation of N2s. Tab. 5: Distribution of N2-collocates to literal montón de in Modern Spanish79 N2-cluster

N2-collocates attested in corpus

‘crops, vegetation’ (47)

trigo ‘wheat’ (17), paja ‘straw’ (12), yerba ‘grass’ (3), heno ‘hay’ (2), arina ‘flour’ (1), arroz ‘rice’ (1), flores ‘flowers’ (1), grano ‘grain’(1), harina ‘flour’ (1), hierba ‘grass’ (1), junco ‘rush, reed’ (1), maíz ‘corn’ (1), paja de cuadra ‘stable straw’ (1), orujo ‘marc’ (1), paja y de hojas ‘straw and (of) leafs’ (1), panes, bollos y tortas ‘breads, rolls and cakes’ (1), tabaco ‘tobacco’ (1)

‘stones’

piedras ‘stones’ (12), grava ‘gravel’ (5), baldosas ‘floor tiles’ (1), cal ‘lime’(1), cal

|| 79 The association of a particular N2 to a particular semantic family is primarily discoursebased, viz. based on its use in context. E.g. proyectos is modified by amontonados and refers to the project descriptions on the speaker’s desk. It therefore shows up in the ‘paper’-cluster.

136 | The development of montón de, a typical locus of grammaticalization

N2-cluster

N2-collocates attested in corpus

(25)

y residuos ‘lime and residues’ (1), cok ‘coks’ (1), escorias ‘slags’ (1), piedra ‘stone’ (1), piedras, casquijo ó ladrillos ‘stones, gravel or bricks’ (1), rocas ‘rocks’ (1)

‘residues’ (25)

escombros ‘rumble’ (6), cenizas ‘ashes’ (5), ruinas ‘ruins’ (5), ceniza ‘ash’ (3), pólvora ‘gunpowder’ (3), desperdicios ‘waste, scraps’ (1), limaduras ‘filings’ (1), polvo ‘dust’ (1)

‘earth’ (22)

tierra ‘earth’ (14), lodo ‘mud’ (3), arena ‘sand’ (2), nieve ‘snow’ (2), barro ‘mud’ (1), tierra y estiércol ‘earth and manure’ (1)

‘manure or waste’ (19)

basura ‘waste’ (7), mantillo ‘leaf mould’ (4), estiércol ‘manure’ (3), estiércol y basuras ‘manure and waste’ (2), basura y paja ‘waste and straw’ (1), cucho ‘fertilizer’ (1), zaramá ‘dirt’ (1)

‘textile, tissues’ (16)

redes ‘nets’ (3), ropa ‘cloths’ (3), colchones ‘mattresses’ (1), esteras ‘mats’ (1), estopas ‘tow, burlaps’ (1), filástica ‘rope’ (1), guiñapos ‘rags’ (1), medias ‘tights, socks’ (1), ropas ‘cloths’ (1), tela ‘fabric’ (1), tela blanca ‘white cloth’ (1), zaleas ‘sheepskins’ (1)

‘corpses’ (14)

cadáveres ‘corpses’ (6), muertos ‘dead men’ (2), cabezas ‘heads’ (1), calaveras ‘skulls’ (1), cráneos ‘skulls’ (1), cuerpos ‘bodies’(1), huesos ‘bones’ (1), moros ‘Moors’ (1)

‘wood’ (13)

leña ‘wood’ (6), madera ‘wood’ (2), vigas ‘joists, beams’ (2), ramas ‘branches’ (1), tablones ‘planks, boards’ (1), tizones ‘charred sticks’ (1)

‘paper’ (11)

libros ‘books’ (2), papeles ‘paper’ (2), cajones ‘boxes’ (1), esquelas ‘death notices’ (1), gacetas ‘gazettes’ (1), legajos ‘files, dossiers’ (1), papel (y polvo) ‘paper and dust’ (1), proyectos ‘projects’ (1), tarjetas ‘cards’ (1)

‘weapons’ (7)

armas ‘weapons’ (3), escopetas ‘shotguns’ (2), trofeos ‘trophies’ (1), vanderas ‘flags’ (1)

‘money’ (5) ‘other: stackable objects’ (13)

dinero ‘money’ (2), oro ‘gold’ (2), doblones ‘doubloons’ (1) objetos ‘objects’ (2), banastas ‘baskets’ (1), bolas ‘balls’ (1), cachivaches ‘stuff’ (1), géneros ‘genres’ (1), gusanos ‘worms’ (1), hilas ‘rows’ (1), palabras ‘words’ (1), platos ‘plates’ (1), plumas ‘feathers’ (1), rebanadas ‘slices’ (1), vidriado ‘glazed pottery’ (1)

Equally interesting is the observation that in Modern Spanish, the clusters of semantically related N2s do not display a one-to-one correspondence to the typical usage contexts literal montón de evoked in Medieval Spanish. Although a heap of corpses and a heap of weapons automatically evoke the war bootyframe and a heap of wood the image of a woodpile, heaps of crops or tissues are no longer exclusively linked to the frames of harvest or tithe-division. In addition, several clusters of semantically similar N2s cannot be directly related to one of the usages listed in Table 1. Yet all heaps in Modern Spanish were accumulated along a vertical axe. In other words, the latter facets are minimally present in all instances. Note that the umbrella-category ‘other’ in Table 5 al-

A diachronic corpus analysis of montón | 137

most exclusively contains stackable objects, although they are not semantically similar strictly speaking.80 By way of illustration, Figure 2 supplies a graphic illustration of how these clusters are interrelated.81 The ovals represent the 12 semantic families described above and point to a slight degree of overlap. The distance on the map reflects the conceptual distance or closeness between the families. The size of the ovals is intended to impressionistically indicate the number of occurrences attested. other (13)

money (5) paper  (11)

corpses  (14) weapons (7)

manure,  (19) waste

residues  (25)

wood  (13) crops (47)



earth (22)

stones  (25)

textile  (16)

Fig. 2: Clusters of N2s in combination with montón de’s literal use

Figure 3 presents a simplified version of the partial match between lexical families of N2s and the typical usage frames evoked by literal montón de. The rectangles represent the usage contexts (e.g. war booty frame). Not unsurprisingly, rectangles and ovals partially overlap. Yet perhaps more interesting is the fact that the schematized image of a vertically oriented configuration is always activated.

|| 80 Since similar conclusions can be drawn for the developments of the other QNs, I will claim in the following chapters that persistence is of conceptual nature and that the only constraint the QN imposes on the N2 is that the N2 can be made to fit the conceptual image of the former. 81 The graph is based on Bybee/Eddington’s (2006) visualization of the exemplar representation of adjectives associated to (pseudo)copular verbs in Spanish.

138 | The development of montón de, a typical locus of grammaticalization

VERTICAL CONFIGURATION +

other (13)

ACCUMULATED

WAR BOOTY

corpses  (14)

 money (5) paper   (11)

weapons (7)

manure,  (19) waste

residues  (25)

wood  (13)

WOODPILE

crops (47)



RUINS

earth (22)

stones  (25) MEMORIAL

textile (16)

HARVEST

Fig. 3: Partial overlap between the clusters of N2s and the frames evoked by montón de

4.3.3.2 Quantifying uses of montón de in Modern Spanish The semantic changes montón de’s quantifying use underwent also come down to partial continuation of older usages and partial broadening resulting in new extensions of the category. On the one hand, heaps of N2s are still somehow accumulated and discourse N2s or entities which refer to oral productions continue to be a frequent combination, as in (57). Further, the use of montón de easily gives way to the pragmatic inference of ‘chaos’ or ‘mishmash’, as in (58). On the other hand, the heap of N2s no longer necessarily evokes the possession frame in which N2 refers to someone’s possessions or richness. (57) Mira, niño, entre paréntesis te digo que no agas caso de mi ortografía, no porque sea muy mala, sino porque como me equivoco siempre en las haches, he determinado suprimirlas, y así no tengo que devanarme los sesos por saber dónde caen y dónde no. El montón de haches que me sobran lo pongo al final, por si quieres enmendarme con ellas la plana. (1899, B. Pérez Galdós, La estafeta romántica) ‘Look, kid, in parenthesis I tell you not to pay attention to my spelling, not because it is very bad, but as I always get the letters H wrong, I have decided to delete them so I won’t have to rack my brains to know where they go and where not. I have put the heap of letters H that are left at the end, in case you want to amend the page.’ (58) Las Pleyades son sumamente perceptibles por el irregular y desordenado monton de Estrellas que las componen. (1738, D. de

A diachronic corpus analysis of montón | 139

Torres Villarroel, Anatomía de todo lo visible e invisible) ‘The Pleiades are very perceptible because of the irregular and disorganized heap of stars that make them up.’ In contrast to the clustered structure of N2 combinations retrieved for the head use in Modern Spanish, only one semantic family is worth mentioning in combination with the quantifying uses, viz. the discourse entities, which counts 7 occurrences. Table 6 presents the N2-clusters observed with the quantifying uses. Interestingly, the N2-clusters of the head-uses are only partially continued (the clusters of nouns referring to manure, earth or money only contain 1 lexical item). Further, the umbrella-category ‘other’ is the dominant one (with 25 instances) and it is not restricted to ‘stackable objects’. On the contrary, Table 6 seems to indicate that the collocational pattern of quantifying montón de has become fairly unrestricted. The N2s observed are not necessarily related and the N2-combinations with montón de are not sensitive to the count-mass distinction, nor to the concrete-abstract distinction. Tab. 6: Distribution of N2-collocates to quantifying montón de in Modern Spanish N2-cluster

N2-collocates attested in corpus

‘discourse’ (7) faltas ‘errors’ (1), gracias ‘acknowledgements’ (1), haches ‘letter H’ (1), latinajos, versos, (…) ‘Latin expressions, lines, (…)’ (1), mentiras ‘lies’ (1), neolgismos ‘neologisms’ (1), psalmos ‘psalms’ (1) ‘vegetation’ (5) flores ‘flowers’ (3), claveles ‘carnations’ (1), nueces ‘walnuts’ (1) ‘corpses’ (2)

cadáveres ‘corpses’ (1), muertos ‘dead men’ (1)

‘manure or waste’ (1)

podre ‘rottenness’ (1)

‘earth’ (1)

nieve ‘snow’ (1)

‘money’(1)

oro ‘gold’ (1)

‘other’ (25)

cosas ‘things’ (2), años ‘years’ (1), botellitas ‘little bottles’ (1), cabecitas ‘little heads’ (1), caseríos ‘hamlets’ (1), conceptos ‘concepts’ (1), ejemplares ‘exemplaries’ (1), especies ‘species’ (1), estrellas ‘stars’ (1), galas ‘adornments’ (1), globulillos ‘blood cells’ (1), llamas ‘flames’ (1), luces ‘lights’ (1), metálico ‘metal’ (1), millas ‘miles’ (1), objetos ‘objects’ (1), olas ‘waves’ (1), planos ‘plans’ (1), quimeras ‘chimeras’ (1), reglas ‘rules’ (1), tábanos ‘horseflies’ (1), tejados ‘roofs’ (1), telarañas ‘spiderwebs’ (1), tiros ‘shots’ (1)

There is however a minor change in the evaluative load of montón de. In 16 occurrences, montón de entails a negative semantic prosody, either because of the context in which it is used (as in (59)), because of the negative connotations associated with N2 (as in (60)), or both. By way of comparison, in Classical Spanish, only 8 occurrences of quantifying montón de were negatively connotated.

140 | The development of montón de, a typical locus of grammaticalization

(59) La gramática, concebida bajo un plan filosófico, con perfecto conocimiento del genio de la lengua, y despejada del monton de reglas que abruman y fastidian á los principiantes, ha tropezado con la resistencia de los talentos rutinarios. (1830, Á. Gil Sanz, Noticias y juicios sobre Francisco Sánchez Barbero) ‘Grammar, conceived in a philosophical way, with a perfect knowledge of the temper of the language, and cleared from the heap of rules that overwhelm and annoy beginners, has stumbled with the resistance of routine talents.’ (60) He tenido que hablar mucho y en la presencia de Dios me encuentro con un montón de faltas, que en particular no puedo decir ninguna, aunque me llenan de pena. (1874, Sor Á. de la Cruz, Papeles de conciencia. Diario espiritual) ‘I have had to speak a lot and in the presence of the Lord I find myself with a heap of errors, of which I cannot tell a particular one, although they fill me with embarassment.’

4.3.3.3 Specifying uses of montón de in Modern Spanish As to the semantic development of the specifier use – which by then had only recently emerged –, the original usages of un montón de N2 that refer to ruins or the very essence of a person’s life or character live through in Modern Spanish. The ruins-frame slightly broadens to all types of remains or left-overs, to the extent that the subdivision made in Table 7 between the N2-clusters ‘ruins’, ‘earth’, ‘stones’ and ‘corpses’ has become artificial. They all evoke a frame in which a certain entity has turned into or has become nothing more than a heap of residues of what it has been or has meant before. Basically, a certain entity X is reduced to its very essence. For instance, the only thing that is left or that remains of a person who dies is his dead body (sometimes described as un montón de tierra, after the Biblical idea that one returns to earth and dust). Likewise, cities get reduced to (a heap of) stones or ruins after a battle, as in (61). The image of an entity which is for some reason reduced to a heap of ‘remains’ is evoked in 42 instances (or 72%) of the specifier uses. Not unsurprisingly then, in those cases, montón de is generally accompanied by contextual clues such as convertirse en un montón de N2 ‘to turn into a heap of N2’, no queda más que un montón de N2 ‘there is nothing more left but a heap of N2’, hacer de X un montón de N2 ‘to turn X into a heap of N2’, no es ya otra cosa que un montón de N2 ‘it is now nothing else but a heap of N2’, reducirse a un montón de N2 ‘to get reduced to a heap of N2’, etc.

A diachronic corpus analysis of montón | 141

(61) Las llamas que durante la ominosa guerra de la Independencia española destruyeron en esta ciudad tantos objetos preciosos y notables edificios, apresuraron la ruina de éste que, abandonado al fin por la comunidad que de tan antiguo lo habitaba, ha quedado reducido a un montón de scombros e informes fragmentos de fábrica. (1857, G. A. Bécquer, Historia de los templos de España) ‘The flames that during the despicable Peninsular War destroyed in this city so many beautiful objects and remarkable buildings, speeded up its ruin. The city was abandoned at last by the community that had always lived there and has remained reduced to a heap of rubble and shapeless factory made fragments.’ When montón de premodifies a group of persons, i.e. when N2 refers to a human noun, no specific frame seems to be systematically invoked. The speaker can indicate that he has no particular (set of) individual(s) in mind yet generalizes over different people, over a heterogeneous group strictly speaking. In view of a particular prominent characteristic they all share and which is stressed by the speaker, the N2-entities are profiled as a homogeneous set of persons. In (62) , the girls are of different kind: some are beautiful, some strong, others slender and elegant. Yet nor their nature nor their identity matter, as they all produce the same sad smell. Further, this smell is so prominent that the speaker considers it to be the very essence of these girls. (62) Aunque había algunas jóvenes limpias, de aquel montón de hijas del trabajo que hace sudar, salía un olor picante, que los habituales transeúntes ni siquiera notaban, pero que era molesto, triste; un olor de miseria perezosa, abandonada. Aquel perfume de harapo lo respiraban muchas mujeres hermosas, unas fuertes, esbeltas, otras delicadas, dulces, pero todas mal vestidas, mal lavadas las más, mal peinadas algunas. (1884, Clarín, La Regenta) ‘Although there were some clean young girls, a spicy smell came from the heap of girls of the sweaty work; a smell that the regular passengers didn’t even notice, but which was annoying, sad, the smell of lazy and abandoned poverty. Many pretty women breathed in that rag perfume, some women were strong and slender, others delicate and sweet but they were all badly dressed, most of the women were badly washed and some of the women’s hair looked a mess.’ The comparison of Table 5, which presents the clustered structure of the N2collocates of literal montón de, with Table 7, which lists the N2-combinations of the specifying uses, reveals that the combinatorial patterns largely overlap.

142 | The development of montón de, a typical locus of grammaticalization

Except for the ‘people’ and ‘discourse’ clusters, the N2-clusters of specifying montón de also show up in combination with the literal uses, although in different proportions. This observation corroborates the claim made previously that the specifier use constitutes a pathway of GR on its own (see also Chapter 3) and does not build on or emerge from quantifying uses.82 This is not to deny that specifying uses continue to provide some quantifying information – the QN still serves to single out a set of N2s which share a specific characteristic – yet the quantity assessment is relegated to the background. Returning to the description of the right collocates, it bears pointing out that the newer N2-clusters of ‘people’ and ‘discourse’ are incompatible with montón de’s original head use, which suggests that the specifier use has shifted form pragmatic inference to conventional status. Tab. 7: Distribution of N2-collocates to specifying montón de in Modern Spanish N2-cluster

N2-collocates attested in corpus

‘residues’ (32) ruinas ‘ruins’ (14), escombros ‘rumble’ (9), cenizas ‘ashes’ (3), basura ‘waste’ (2), despojos ‘remains’ (1), escorias ‘slags’ (1), ruinas y escombros ‘ruins and rumble’ (1), ruinas y destrozos ‘ruins and damage’ (1) ‘people’ (7)

aldeanos ‘villagers’ (1), amigos ‘friends’ (1), emigrados ‘emigrants’ (1), gouffetillos ‘Gouffe adepts’ (1), hijas ‘daughters’ (1), sabios ‘wise men’ (1), turbas ‘mobs’ (1)

‘earth’ (4)

arena ‘sand’ (1), cieno ‘silt, mud’ (1), polvo y ceniza ‘dust and ash’ (1), tierra ‘earth’ (1)

‘stones’ (4)

peñas ‘crags, rocks’ (1), piedras ‘stones’ (1), piedras y musgo ‘stones and moss’ (1), sillares ‘ashlars’ (1)

‘discourse’ (4) mentiras ‘lies’ (1), necedades ‘nonsense’ (1), proposiciones ‘proposals’ (1), trampas ‘traps’ (1) ‘corpses’ (2)

cadáveres ‘corpses’ (1), huesos ‘bones’ (1)

‘other’ (5)

carne ‘meat’ (1), gusanos ‘worms’ (1), humanidad ‘humanity’ (1), hojas ‘leafs’ (2), porquería ‘dirt’ (1), rabos ‘tails’ (1)

While only part of the quantifying uses bear a negative affective load, the specifying occurrences of montón de almost exclusively occur in negatively loaded contexts. In Modern Spanish, this negative semantic prosody is generally prompted by context and not ‘yet’ by the choice of the QN (as is the case in Present-Day Spanish, see Section 4.3.4.3). In other words, the semantic prosody

|| 82 This evolutionary path is indeed suggested by the fact that specifier uses often emerge later than quantifying uses (see Figure 7 in Chapter 5) and by the chains of GR proposed by Traugott (2008a; 2008b) and Brems (2011) for English binominal quantifiers.

A diachronic corpus analysis of montón | 143

merely roots in the source contexts specifying montón de emerged in, more precisely, the reduction of a particular entity to its essence, either a heap of remains or a negatively evaluated characteristic. In 6 occurrences, the N2collocate adds up to the negative affective load of the specifying uses, as in (61) or (63). (63) No, y lo que es razón tenía, es claro; el mundo, bien mirado, era un montón de escorias. Él no podía quejarse, en su vida no había habido desengaños terribles, grandes contrariedades, aparte de la muy considerable de no haber sido cómico; pero en tesis general, el mundo estaba perdido. (1884, Clarín, La Regenta) ‘No, and what was right is obvious; the world, if one looks closely at it, is a heap of trash. He could not complain: in his life, there had not been terrible disappointments, great misfortunes, apart from the very considerable one (misfortune) of not having been a comic; but generally, the world was lost.’

4.3.3.4 The combinatorial pattern of montón de in Modern Spanish The morphosyntactic development of montón de in Modern Spanish continues the tendencies started in Classical Spanish. In other words, although the binominal construction remains analyzable or compositional, the different uses, viz. the head, quantifier and specifier reading, further refine the restrictions – more carefully, their combinatorial preferences – they impose on the determiner to N1, the modification pattern, the verb agreement and the N2. In other words, they gradually develop a specific co-selection pattern. Further, in Modern Spanish, the prepositional phrase is no longer separable and in only two head uses, a separate determiner is added to N2. With regard to the fluctuations in the determiner combinations, the indefinite determiner un is the unmarked option (un appears in 184 of 217 occ.). Although the head uses of montón de easily combine with a whole range of determiners to N1, only topicalized occurrences of quantifying and specifying montón de allow for determiner variation. Interestingly, in addition to its preference for the indefinite determiner, specifying montón de combines in 12 occurrences with the demonstratives aquel ‘that’ (8), este ‘that’ (3) and ese ‘this’ (1). As the spatial distance implied by demonstratives often reflects emotional distance, the abundance of aquel ‘that’ corresponds to the

144 | The development of montón de, a typical locus of grammaticalization

tendency of specifying montón de to introduce entities of reduced or little esteem.83 The adjectival modification pattern to montón de is similar to the combinations observed in Classical Spanish. As to the head uses, only adjectives specifying the shape or other configurational characteristics are observed (e.g. colocado ‘put’, ingente ‘enormous’, grandísimo ‘very high’, sucio ‘dirty’, húmedo ‘humid’, etc.). Interestingly, the proportion of adjectives stressing the height of the heap is less extensive than in Classical Spanish: in Modern Spanish, 13 (or 6%) of 217 occurrences of montón de are preceded by an adjective stressing its huge size, which sharply contrasts with 28 occurrences (or 17%) of 162 in Classical Spanish. As to the quantifying uses, 4 intensifying adjectives (such as desmedido ‘excessive, infinito ‘endless’, etc.) are found in addition to two adjectives – confuso ‘confused’ and irregular y desordenado ‘irregular and untidy’ – which highlight the chaotic nature of the heterogeneous grouping of N2s. As to the specifying uses, 8 occurrences combine with qualifying adjectives (of which 5 refer to the confused or heterogeneous nature of the heap of N2s) and one combination with an intensifying inmenso ‘immense’ is found. With regard to verb agreement, no unexpected verb plurals are found with head or specifying uses. Quantifying uses comprise two instance of plural verb agreement, which corroborates the head status of N2 on a morphosyntactic level. To determine the degree of GR of montón de, it is highly useful to also take into account its development as an uncomplemented noun. Although no strictly quantifying use is observed for montón as a single noun in the sense that (un) montón would have acquired an adverbial function, part of its semantic development runs parallel with the binominal construction type. First, the fixed expression en (confuso) montón ‘all together, lit. in a confused/chaotic heap’ as in (64) and (65), generally preceded by a motion verb such as correr ‘to run’, andar ‘to walk, to go’, unirse ‘to join’, acudir ‘to come, to attend’, etc., shows a spectacular rise in frequency in comparison to its marginal use in Classical Spanish and to the fixed expression (traer, tirar) a montón ‘to bring/throw together on a heap’. The fact that the lexicalized combination en montón is frequently specified by confuso or paraphrased by confusamente corroborates the association with the concept of ‘chaos’ observed for binominal quantifying and specifying uses. Second, the binominal uses of montón – which count 394 occurrences –

|| 83 In Spanish, aquel is typically used for entities outside the current discourse space. Ese refers to entities in the domain of the interlocutor and entities identified by este belong to the speaker’s domain.

A diachronic corpus analysis of montón | 145

starts to become slightly more frequent than its uses as a bare noun – which count 301 uses, of which 79 instances of the lexicalization en montón.84 (64) A lo lejos la vista descubría en montón y confusamente el campo, las empalizadas y las demás obras del castillo; (…). (1834, J. de Espronceda, Sancho Saldaña o El Castellano de Cuéllar) ‘In the distance the sight revealed the land all together (lit. in a heap) and in a confusing way, the palisades and the other parts of the castle;’ (65) (…); por el exceso contrario, los aliados, lanzándose en desorden y montón, se expusieron a perder mucho; y si en vez de hallarse esparcidas las brigadas austriacas, hubieran estado más cerca de la voz del general en jefe, estos revueltos ataques, algo imprudentes y arremolinados, habrían sido funestos. (1862, F. Villamartín, Nociones del arte militar) ‘(…); the allies, from the other extremity, who attacked in disorder and all at once, exposed themselves to losing a lot; and if instead of being scattered, the Austrian brigades would have been closer to the general-in-chief, these messy, a little reckless and crowded attacks would have been fatal.’ In sum, the development of montón de in Modern Spanish partially continues the tendencies observed in Classical Spanish while also extending to new contexts. Crucially, the specifying use acquires conventionalized status by the end of the 19th century, when it has extended to contexts incompatible with montón de’s original frame. The co-selection pattern becomes gradually more refined and is closely linked to the semantic developments.

4.3.4 Present-Day Spanish: advanced grammaticalization stage In the 20th century, the occurrences of montón de testify to an advanced stage of GR, for several reasons. First, the proportion of literal uses (cf. Table 8) keeps diminishing and specific configurational features, especially its height, become faint. Second, its conceptualization has further schematized to the point that it || 84 The total of 394 occurrences CORDE returns by the query [montón de O monton de] – the boolean operator O ‘or’ generalizes over the orthographic variants – differs from the total of 367 instances in Table 4. The discrepancy is due to the fact that the fine-grained analysis reflected in Table 4 considers only those binominals where N1 and N2 are profiled as ‘co-extensive’; 27 instances have thus been filtered out manually for being not co-extensive.

146 | The development of montón de, a typical locus of grammaticalization

counts pure quantifying uses near-synonymous to the absolute quantifier mucho/a(s) ‘many, much’. Third, the changes in the specifying uses show that its categorizing function has become associated with the construction in itself, independently of the right collocate. Finally, syntactically, both grammaticalized uses – or should I say ‘constructions’ (see infra) – show a strict co-selection pattern as to determiner, modifier and N2. In other words, by the 20th century, the operator-like functions of montón de have reached the status of conventionalized pragmatic inference, of changes spread over the entire speech community. Tab. 8: Layering of uses activated by montón de in Present-Day Spanish

1900–1975

# %

Head

Quantifier

Specifier

322 (0.52)

116 (0.19)

71 (0.11)

Ambiguous Indeterminate 70 (0.11)

40 (0.07)

Perhaps even more indicative of the advanced stage of GR is the syntactic context expansion of montón to adverbial uses (such as Me alegra un montón ‘It makes me very happy). Although mogollón and barbaridad present similar uses in the CREA-corpus – yet to a limited extent – montón has obviously been the first to display adverbial uses. In view of the fine-tuned co-selection pattern of the three major uses from the 20th century onwards, the semantic and formal developments of each reading are discussed simultaneously in the following sections.

4.3.4.1 Head uses of montón de in Present-Day Spanish As to montón de’s head uses in Present-Day Spanish, one particular semantic glide enables all subtle changes of literal montón de in the 20th century: heaps are no longer high. This schematization, initiated in Modern Spanish, becomes conventionalized in Present-Day Spanish. Heaps continue to be somehow accumulated, however. The smaller dimension does not entail though that they cannot contain lots of (tiny) N2 entities. In (66), the preceding context refers to a large collection of pictures, yet as the woman is capable of carrying them, it is difficult to imagine an extremely elevated pile.85  Thus, if the speaker wants to || 85 The pile of photographs is not high from the perspective of the speaker or in comparison to the speaker’s proportion. Of course, when comparing to the standard height of (piles of) photographs, the bundle the subject is carrying can be considered relatively high. In other words, the quantifying interpretation of montón de implies a certain subjectification or assessment by the speaker of the scalar implicatures inherent in montón.

A diachronic corpus analysis of montón | 147

convey the idea that the heap is high, its height has to be specified in the context, as in (67). Interestingly, when montón de N2 is used metaphorically, the excessive height of the imaginary heap continues to be stressed, as in (68). (66)  Es más, sé fijamente que no le interesará porque, mire usted, aquí tenemos muchísimas y no se ha tomado siquiera el trabajo de mirarlas una por una. Y, mientras hablaba, le tendió el montón de instantáneas que reproducían uno de los principales incidentes de (…). (1928, M. Cumberland, La esposa indiscreta) ‘What’s more, I know for sure he won’t be interested because, you see, here we have lots of them and he didn’t even take the effort to look at them one by one. And, while he was talking, he showed him the heap of snapshots that reproduced one of the many incidents of (…).’ (67) Colgados en perchas innumerables, sudarios y sábanas blancas rayaban con sus pliegues inmóviles la pared; un montón enorme de cadenas oxidadas se alzaba en un rincón, y una verdadera muchedumbre de espectros iba y venía entre los cachivaches esparcidos por el suelo y los maniquíes (…). (1930, W. Fernández Flórez, Fantasmas) ‘Hung up on innumerable hangers, shrouds and white sheets scratched the wall with their immobile pleats; un enormous heap of rusty chains rose up in a corner, and a true multitude of spectrums came and went between the pieces of junk spread out over the floor and on the mannequins (…).’ (68) Y así vamos a la toma de una afirmación sobre los escombros de la que nos desmoronó la lógica, y se van amontonando los escombros de todas ellas, y un día vencedores, sobre la pingorota de este inmenso montón de afirmaciones desmoronadas, proclamarán los nietos de nuestros nietos la afirmacíón última, (…). (1905, M. de Unamuno, Vida de don Quijote y Sancho) ‘And so we go to the taking of an affirmation on the rubble that made all logic collapse and the rubble of all of them is piling up, and one day victors, the grandchildren of our grandchildren will proclaim the ultimate affirmation on top of this immense heap of collapsed affirmations, (…).’ The altered conceptualization of montón de is particularly clear in the two most productive frames, viz. when un montón de N2 refers to a seat or sleeping place (69) and when it constitutes a sheaf or pile of papers (70). Further, the literal uses of montón de in Present-Day Spanish continue the typical usage contexts observed in Modern Spanish, while adapting to changing society and reality.

148 | The development of montón de, a typical locus of grammaticalization

When N2 refers to a type of wood, montón de N2 is no longer directly associated with altars, sacrifices or stakes, but rather with fire(places), kitchen stoves or burning buildings. Economically, heaps for tithing are replaced by heaps of merchandise (such as textiles or food) at markets or stores and by heaps of money. Further, the heaps of merchandise no longer primarily consist of harvested crops and tissues, but extend to all types of food and clothes (or fashion accessories). In war contexts, heaps of corpses no longer symbolize some honorable achievement but are interpreted as the sad result of force of arms and bombardments which have to be removed yet treated with dignity. (69) En el andén, aldeanas en pie junto a sus cestas o sentadas sobre un montón de traviesas negruzcas esperan la llegada del tren, (…). (1943, W. Fernández Flórez, El bosque animado) ‘On the platform villagers standing next to their baskets or sitting on a heap of blackish crossbeams wait for the train to arrive, (…).’ (70) Y el diputado siguió inclinado sobre su pupitre, en el gabinete de escritura del Congreso, terminando su última carta, añadiendo un sobre más al montón de correspondencia que se apilaba en el extremo de la mesa, junto al bastón y el sombrero de copa. (1900, V. Blasco Ibáñez, Entre naranjos) ‘And the deputy remained leaning on his desk, in the writing cabinet of the Congress, ending his last letter, adding one more envelope to the heap of correspondence that was piled up at the end of the table, next to the stick and the top hat.’ In line with the decrease in size, the association between montón de and the facet ‘less worthy’ is continued. In (71), the heap of dust in the zinc box sharply contrasts with the Saint’s canonical status.86 Further, the pragmatic inference of a chaotic heap, of a mishmash of entities jumbled up disorderly has spread to the literal uses as well, as in (72). However, when un montón de N2 refers to a pile of paper or money, the inference of chaos is not necessarily evoked: coins, notes and sheets of paper are generally profiled as nicely piled up, as in (73). (71) “Bueno, has venido a ver las cenizas del Santo Cuerpo.” “¿Dónde están?” “Ahí, en ese cajón.” Clotilde dio unos pasos y golpeó la caja de zinc con la punta del pie. “¿Ese montón de polvo? ¡Vaya porquería!” “Polvo eres, polvo serás.” “Sí, hijo mío, pero ¡para algo la pobre fue santa!”; (…). (1972, G. Torrente Ballester, La saga/fuga de J. B.) || 86 This example verges on the specifier reading, but as the focus relies on the location of the corpse and as the ashes are collected in a box, I categorized this borderline case as a head use.

A diachronic corpus analysis of montón | 149

‘ “Well, you’ve come to see the ashes of the Most Holy Body.” “Where are they?” “There, in that drawer.” Clotilde took a few steps and hit the box of zinc with the tip of her foot. “This heap of dust? How disgusting!” “Dust you are and dust you shall be.” “Yes, my son, but there’s a reason why the poor girl was a saint!”; (…).’ (72) La sociedad política Estado ha de acomodarse a lo que la sociedad civil es. Y España no es “un montón de gente hacinadas sin unidad ni cohesión, como las arenas en el desierto", ni un conglomerado disconforme de provincias caprichosas y arbitrariamente diferenciadas, ni es tampoco un Estado centralista y absorbente (…). (1932, anon., Síntesis del programa de la Comunión Tradicionalist) ‘The political society of the State needs to adapt to what the civil society is. And Spain isn’t “a bunch of (lit. heap of) people crowded together without any unity nor cohesion, like sand in the desert”, nor is Spain a conglomerate in disagreement of capricious provinces randomly differentiated from each other, nor is it a centralist and demanding State, (…).’ (73) En el infierno, (…) le condenarán a contar ininterrumpidamente sus ganancias, y quizá a mí me destinen a soplarle el montón de billetes cuando los tenga ordenados para que tenga que volver a contarlos, y así una vez y otra, y otra... (1972, A. Zamora Vicente, A traque barraque) ‘In hell, (…), they will condemn him to tell nonstop his profits and maybe I would be assigned to blow away the heap of notes once he has them arranged so that he would have to count his money again, repeating this over and over again…’ The slightly altered conceptual image of montón de is mirrored in the clusters of N2-collocates, which continues the pattern of Modern Spanish (see Table 5), though in different proportions. According to descending token frequency, the following family resemblances are observed: crops (and food in general) (64 occ.), (objects made of) paper (45 occ.), textile, cloths and fashion accessories such as shoes (31 occ.), residues such as ashes (31 occ.), types of wood (29 occ.), kinds of stones (27 occ.), manure or waste (22 occ.), money (18 occ.), earth (12 occ.) and corpses (6 occ.). The umbrella-category ‘other’ hosts the remaining 37 instances. In comparison to Modern Spanish, the family of crops, vegetation and other kinds of food continues to be the largest family lexically speaking. It has to be noted however, that a heap of N2crops in Present-Day Spanish does not only occur in the ‘harvest’ frame. Instead, a heap of wheat or straw more frequently refers to a bed or seat. Other N2-clusters, such as the earth-family and

150 | The development of montón de, a typical locus of grammaticalization

the cluster of stony N2-entities decrease significantly. Further, the weaponcluster no longer exists. In contrast, other semantic N2-families, such as the paper-cluster and the cluster of tissues or cloths, dramatically gain in importance. Tab. 9: Distribution of N2-collocates to literal montón de in Present-Day Spanish87 N2-cluster

N2-collocates attested in corpus

‘crops or food’ paja ‘straw’ (14), trigo ‘wheat’ (13), carne ‘meat’ (4), yerba ‘grass’ (4), patatas ‘potatoes’ (3), cáñamo ‘cannabis’ (2) (64) ‘stones’ (27)

piedras ‘stones’ (21), grava ‘gravel’ (3)

‘residues’ (31)

escombros ‘rumble’ (7), ruinas ‘ruins’ (7), cenizas ‘ashes’ (4), polvo ‘dust’ (2)

‘earth’ (12)

tierra ‘earth’ (9), arena ‘sand’ (4)

‘manure or waste’ (22)

estiércol ‘manure’ (11), basura ‘waste’ (9)

‘textile, tissues’ (31)

ropa ‘cloths’ (6), harapos ‘rags’ (4), trapos ‘rags’ (4), ovillos ‘balls of yarn’ (2)

‘corpses’ (6)

cadáveres ‘corpses’ (3), muertos ‘dead people’ (2)

‘wood’ (29)

leña ‘wood’ (16), brasas ‘embers’ (2), ramas ‘branches’ (2)

‘paper’(45)

papeles ‘paper’ (9), fichas ‘cards’ (8), cartas ‘letters’ (6), cuartillas ‘sheets, manuscript’ (5), periódicos ‘newspapers’ (4), ejemplares ‘copies, samples’ (2), fardos ‘bundles, bales’ (2), legajos ‘files, dossiers’ (2), libros ‘books’ (2)

‘money’ (18)

oro ‘gold’ (5), billetes ‘notes’ (4), dinero ‘money’ (2)

‘other’ (37)

cosas ‘things’ (3)

Morphosyntactically, montón de continues to function as a full noun in the head uses. Literal montón de combines with a large series of determiners, ranging from the indefinite (184 occ.) and definite determiner (109 occ.), over demonstratives, to exclamative ¡qué!, interrogative ¿qué?, indefinite otro ‘other’ and algún ‘a certain, some’, possessives and even absolute quantifiers. Although it primarily combines with configurational adjectives stressing its height – and gran ‘big’ in particular – and adjectives highlighting the unorderly composition of the heap, the literal uses of montón de also occur with descriptive adjectives such as indispensable ‘indispensable’, nauseabundo ‘disgusting’, etc. Further, the only occurrence where verb agreement is visible combines with a singular || 87   In order not to disturb the coherence and presentation of this data description by overloading the text with extensive tables, Table 9 only retains the N2 combinations which occur at least two times with montón de’s head use in Present-Day Spanish. The same comment holds for the parallel Tables 10 and 11, which indicate the distribution of N2 combinations with quantifier and specifier uses respectively..

A diachronic corpus analysis of montón | 151

verb, which provides further evidence for the morphosyntactic head status for N1.

4.3.4.2 Quantifying uses of montón de in Present-Day Spanish Turning to the evolution of montón de’s quantifier use, it bears pointing out that it has abstracted even further away from the possession frame. In addition, the facet ‘accumulated’ is no longer foregrounded, yet continues to be necessarily implied. Interestingly, the requirement of ‘being brought together’, whether consciously and on purpose by an identifiable cause or artificially, easily induces the pragmatic inference of abnormality, of an unexpected course of events. The conditional facet ‘accumulation’ can be activated in many ways. In (74), the scenery depicts a literal collection of savings. In (75) the accumulation is even more abstract and is ensured by the single source or producer of the N2entities: a single speakers produces (and by extension accumulates) the exaggerations. This facet can also fade into mere spatiotemporal contiguity: in (76), the houses all figure in one picture. (74) En un momento se recaudó allí un montón de calderilla y algunas pesetas. Metieron todo este dinero en un pañuelo y se lo entregaron al náufrago. (1921, A. Palacio Valdés, La novela de un novelista) ‘In little time a heap of small change and a few pesetas were collected. They put all the money in a handkerchief and handed it over to the shipwrecked person.’ (75) La cantante la saludó con estrepitosa cordialidad y volcó sobre ella un montón de exageraciones: ¡Me devoran a piropos, amiga mía! Por eso me da rabia pasar por aquí a pie. (1940, J. Díaz Fernández, La Venus mecánica) ‘The female singer greeted her with great cordiality and overwhelmed her with a lot (lit. heap) of exaggerations: They devour me with flattering comments, my friend! That’s why I get irritated when I walk by.’ (76) En estas fotografías aparecían sentados en el pretil de un puente, viéndose en segundo término un río quieto, un montón de casitas viejas y un cielo gris con nubes. (1948, M. Mihura, Mis memorias) ‘In these photos they appeared sitting on the parapet of a bridge, on the background you could see a quiet river, a lot (lit. heap) of old little houses and a grey cloudy sky.’ The conditional facet of ‘accumulation’ – or at least ‘spatiotemporal contiguity’ – entails two further extensions. First, in line with the schematization of meaning illustrated in Figure 2, the reference of un montón de to a heterogeneous

152 | The development of montón de, a typical locus of grammaticalization

accumulation has become entrenched: in (77), montón de gathers diverging personalities in one house; in (78), montón de unites complete strangers in a particular religious community. Interestingly, when montón de quantifies the chameleon-like cosas ‘things’, the N2s are almost automatically interpreted as belonging to different categories (as in (79)). What matters in these occurrences of montón de is the idea that the heterogeneous or opposite N2-referents can be lumped together under a common denominator. The next step in the schematization of montón’s conceptual image is the interpretation of un montón de N2 as ‘all kinds of things’, where the focus does not rely on the extension of the group, but on its extensive diversification: montón de N2 refers to an excessive number of types of N2 rather than to a large number of N2 referents strictly speaking. Crucially, the precise identity of the N2-entities does not matter. In (80), the speaker picked up a lot of stories, but what is more important is that they were so diverse that he now wants to know the truth. In contrast to Modern Spanish when the connotation of ‘chaotic’ heap is frequently associated with montón de’s quantifying use, this connotation is not so much exploited any more in Present-Day Spanish.88 (77) Por lo que se refiere al pueblo bajo que es al fin el que nos interesa porque es el que más necesita redención, no se puede ni siquiera hablar de organización familiar. Domina el macho siempre, sólo porque es macho. Se agrupan en una pieza pequeña y sucia un montón de gentes que hacen la vida más común que es posible imaginar. En esa promiscuidad no es posible extrañar nada, nada nos sorprende. (1935, L. Llach, La influencia del medio en la prostitución) ‘Regarding the ordinary people, who in the end are the ones who interest us the most because they are those most in need of redemption, family organization is out of the question. The male always dominates, merely because he is the male. A heap of people who lead the most ordinary life one can imagine gather in a little and dirty room. In this kind of promiscuity nothing seems strange, nothing surprises us.’ (78) El maremagnum, que a simple vista pudiera parecer la “Sociedad Anónima Agerre”, fue, sencillamente, la reunión cariñosa y animada de un montón de hermanos de diferentes apellidos, pero hijos todos de un hogar, del calor de una misma cocina, que saludaban "aita" y "ama", al actual matrimonio que ocupa Agerre. (1969, I. Linazasoro, La otra Guipúzcoa)

|| 88 It seems to be replaced by the focus on the extensive diversification of N2.

A diachronic corpus analysis of montón | 153

‘The mess that, at first sight, could look like the “Agerre Public Limited Society”, was simply a kind and cheerful reunion of a lot (lit. heap) of brothers with different surnames, but all of them children from the same household, from the warmth of the same kitchen, that greeted the couple which actually occupies “Agerre” calling them mom and dad.’ (79) Pero pasemos a otro asunto. Tengo que consultarte un montón de cosas. 1941, R. León, Cristo en los infiernos) ‘Let’s pass to another subject. I have to ask your advice on a lot (lit. heap) of things. (80) Cuéntame algo sobre él. He oído un montón de historias. Por ejemplo, ¿es verdad que Millán Astray le tiene odio? (1951, A. Barea, La forja de un rebelde) ‘Tell me something about him. I have heard a lot (lit. heap) of histories. For example, is it true that Millán Astray hates him?’ Crucial to the determination of montón de’s progress in GR is the development of the absolute quantifier use in Present-Day Spanish. In many examples, un montón de N2 profiles an excessive number (as in (81)). Its quantifier use is often corroborated by contextual clues such as numbers, the interrogative pronoun ¿cuántos? ‘how many’, etc. The difference with the absolute quantifier mucho/a(s) resides in the facets ‘accumulated’ or ‘contiguity’, which are conditional for the quantifying use of montón de. By extension, montón de indicates that the number of N2 referents is for some reason remarkable or surprising. Interestingly, montón de can be used both from an objective and a subjective perspective. In (82) the speaker is looking for good excuses for being too late: whether he really saw many friends or just two with whom he talked way too long, cannot be derived from the context. In other words, the absolute quantity expressed by montón de is not necessarily enormous, but is at least experienced as such: in (83), the precise number of children probably does not exceed the number of 4 or 5, yet they demand a lot of efforts of the mother.89 (81) Don Antonio, como siempre, está leyendo minuciosamente una carta y emplea en ello un montón de minutos. Finalmente la firma y (…) (1951, A. Barea, La forja de un rebelde) ‘Mr. Antonio is, as always, carefully reading a letter and that takes him a long time (lit. a heap of minutes). Finally he signs it and (…).’

|| 89 According to the socio-economic standards in the 20th century Spanish society.

154 | The development of montón de, a typical locus of grammaticalization

(82) No te sientes, Sagrario, que me voy en el de las seis menos cuarto. Si no hubiese quedado con un montón de amigos... Y amanecerá con medio metro de nieve... (1972, J. García Hortelano, El gran momento de Mary Tribune) ‘Don’t you sit down, Sagrario, as I will leave with the one at a quarter to six. If I hadn’t arranged to meet a lot (lit. heap) of friends… And dawn will be breaking with half a meter of snow…’ (83) Tenía un montón de hijos propios, pero la mujer legal se las apañaba como podía, trabajando de lavandera o de asistenta en las casas pudientes, para sacar a la descendencia adelante. A él no le interesaban los problemas caseros. (1965, J. Escobar, Itinerarios por las cocinas y las bodegas de Castilla) ‘He had a lot (lit. heap) of children of his own, but his legal wife managed as good as she could, working as washerwoman or assistant in wealthy families, in order to give the descendants a good start in life. He was not at all interested in domestic problems.’ One final extension is the shift from a huge quantity into an indeterminate quantity, which adds up to the entrenchment of the quantifying reading. In (84) the precise number of tears each person is susceptible to produce is not known to the speaker, nor does it interest him. What matters is that everybody has a certain portion of tears, that will somehow all come out, regardless of their quantity. (84) Porque, lo que yo digo, quien más quien menos, todo el mundo tiene un montón de lágrimas por derramar en la vida, es como una fabrica, lógico, y si no las echas a tiempo, las echas a destiempo, la cosa no tiene vuelta de hoja. (1966, M. Delibes, Cinco horas con Mario) ‘For, what I’m saying, some more than others, everybody has a lot (lit. heap) of tears to shed in life, it is like a factory, of course, and if you don’t shed them in due time, you will shed them at the wrong time, there’s no doubt about it.’ The entrenchment of the pure quantifying reading of montón goes hand in hand with a further narrowing down of the variation in its co-selection pattern, viz. a further refinement of the morphosyntactic restrictions of the construction. With regard to the N2-collocates (cf. Table 10), at least four clusters are added to the cluster of ‘discourse entities’ (17 occ.): N2s referring to ‘people’ (18 occ.), to abstract notions of ‘time’ (15 occ.), to objects made of ‘paper’ (12 occ.) and, finally, ‘money’ (9 occ.). Yet the remaining N2s do not cluster under a specific semantic family but are compatible with – or can be made to fit – the facet ‘accumula-

A diachronic corpus analysis of montón | 155

tion’. Interestingly, the underspecified noun cosas ‘things’ is the most frequent N2-collocate and evidences the claim I will make in Chapter 9 as to the mutual interaction between the QN and the N2. The vagueness as to internal constituency or magnitude of montón de perfectly matches the indefiniteness of the wild card noun cosas: in both cases the exact number, identity or type does not matter.90 Tab. 10: Distribution of N2-collocates to quantifying montón de in Present-Day Spanish N2-cluster

N2-collocates attested in corpus

‘people’ (18)

hijos ‘children’ (6), gente ‘people’ (2)

‘discourse’ (17) historias ‘stories’ (2), cartas ‘letters’ (2) ‘time’ (15)

años ‘years’ (6), veces ‘times’ (5), tiempo ‘time’ (3)

‘paper’ (12)

cartas ‘letters’ (2)

‘money’ (9)

billetes ‘notes’ (4), calderilla ‘change’ (2), diniero ‘money’ (2)

‘other’ (45)

cosas ‘things’ (10), cadáveres ‘corpses’ (2), nubes ‘clouds’ (2)

The remaining slots in the BQ-construction become further delineated, which suggests that the quantifying use of montón de has developed a stable shape. With regard to the determiner to N1, less variation is attested: in 102 instances, the (expected) indefinite article un precedes montón de. Only when the number of N2-entities is topicalized, the definite article can be used (12 occ.). The demonstrative ese shows up just once under the same condition. Moreover, there is one occurrence without determiner to N1. The modification pattern of N1 is also highly constrained. First, proportionally less modification is observed. The few occurrences – 6 to be precise – of a premodifying adjective to N1 concern quantifying (e.g. enorme ‘enormous’) or intensifying adjectives (e.g. verdadero ‘real’) or adjectives mirroring a facet of montón’s conceptual image (e.g. confuso ‘chaotic’, triste ‘miserable’). One occurrence contains the descriptive adjective espantoso. The stabilization of the co-selection pattern can at first sight appear to be at odds with the N2-pattern, since in Present-Day Spanish, montón de is fairly unrestricted as to the type of noun involved: it combines equally fine with abstract notions as with concrete objects. It slightly prefers plural count nouns over mass nouns or singular nouns. However, as I mentioned before, all N2-

|| 90 The collocational tie between montón de and cosas appears to be particularly bond, as it is two-directional: when searching in CREA for the most frequent immediate left collocate to cosas (in peninsular Spanish at least), montón de is observed to be the preferred option (cf. infra).

156 | The development of montón de, a typical locus of grammaticalization

collocates share the characteristic of fitting the ‘accumulation’-frame. In other words, the ‘constraint’ or ‘restriction’ is primarily conceptual in nature. I will extensively dwell on this observation in Chapters 8 and 9. As to verb agreement, only two instances are found where verb agreement is visible. One singular and one plural verb are attested respectively, which makes it impossible to make any statistically relevant claim (but see Chapter 8). Finally, as to semantic prosody, the development in Present-Day Spanish seems to differ from Modern Spanish. The negative affective load, which frequently combined with the quantifying uses in Modern Spanish, is only attested in 20 occurrences of 116 quantifier uses and is consequently considered the trademark of specifying uses, where it is found in 63 of 71 instances.

4.3.4.3 Specifying uses of montón de in Present-Day Spanish The semantic development of specifying montón de can be characterized by the concepts of partial continuation, entrenchment and gradual extension. Three major usages can be distinguished. First, montón de N2 continues to introduce a set of persons whose precise identity doesn’t matter – given that they all qualify for a specific characteristic (cf. (62)) –, though to a limited extent (only in 9 of 71 occ.). Second, the most frequently exploited frame (with 30 occ.) is when montón de introduces a heap of remains. The specifier use as in ‘X is un montón de N2’ can best be paraphrased by ‘X is a heap of remains of X’s previous status’. This interpretation has in fact become highly entrenched, to the extent that it would be artificial to differentiate in Table 11 between the different semantic families (such as corpses, ruins, etc.) that refer to a specific subtype of ‘residues’. The N2s which refer to a kind of ‘residue’ are no longer primarily corpses or ruins but can designate all kinds of left-overs observed (such as astillas ‘splinters’, serrín ‘sawdust’, trozos de cuero ‘pieces of leather’ (as the remains of a shoe), etc.). More importantly, this interpretation of specifying montón de is not restricted to a specific cluster of N2s but can be evoked with all kinds of N2 (e.g. (85)). The third typical specifier usage (with 20 occ.) concerns person metaphors. Particularly the combination un montón de carne ‘a heap of flesh/meat’, repeatedly followed by rota ‘broken’ or informe ‘shapeless’, is the most frequent combination in this use (86). The metaphorical person descriptions highlight the most essential or striking feature of a specific individual (versus a specific set of individuals). Pragmatically, according to the speaker, the metaphor suffices to describe the individual in question. When un montón de N2 categorizes a specific individual X, the metaphor can be paraphrased as follows: X can be reduced

A diachronic corpus analysis of montón | 157

to one particular feature and to nothing but that feature. In line with the conceptual semantics of this usage, a variety of part-whole relations are exploited, e.g. pellejos ‘skins and bones’ (87). (85) Había encontrado aquella muchacha un seguro asilo. Dejaba la pequeñez de su vida tirada a sus pies como un montón de olvidos y decía, casi sollozando, los versos que ella no sabía escribir. (1970, M.T. León, Memoria de la melancolía) ‘He had found that girl a secure asylum. She let the smallness of her life behind her like a heap of memories and she said, almost sobbing, the verses she didn’t know how to write.’ (86) Parecíame que Helena había querido afrentarme y nada más, que había enamorado a Abel por menosprecio a mí, pero que no podía, montón de carne al espejo, querer a nadie. (1917, M. de Unamuno, Abel Sánchez. Una historia de pasión) ‘It seemed to me that Helena had only wanted to affront me, that she had made Abel fall in love with her because she despised me, but that she wasn’t capable of loving anyone. She was (just) a heap of meat in the mirror.’ (87) Resuelto entró en casa de maestro Pancho. Don Nicolás se le antojó, al verlo, un ridículo montón de pellejos. Era fácil tomarlo por las solapas, tal como lo estaba haciendo, y alzarlo hasta tener las narices a tope con las suyas. Ni siquiera parecía peso lo que levantaba. (1957, E. Nácher, Guanche) ‘Decisively he entered master Pancho’s house. When looking at him, Mr. Nicolás seemed to him like a ridiculous heap of rinds. It was easy to grab him by the lapels, as he was doing, and lift him until his nose was as high as his. It didn’t seem like he was lifting any weight.’ For the sake of completeness, Table 11 lists the N2 combinations observed more than once in our corpus. In contrast to the analysis of the semantic families found in Modern Spanish, Table 11 generalizes over the distinction between particular types of ‘residues’ and considers them as one family. Less productive clusters are N2s referring to human entities (5 occ.) and discourse entities (3 occ.). Since none of these N2s is observed more than once, they are not even included in Table 11. In sum, the collocational pattern adds up to the evidence that the conceptual image of ‘remains’ has become entrenched. Unsurprisingly, the context generally contains predicates such as se reduce a ‘it comes down to, it just is’, ya es ‘it already is’, no ser más que ‘to be no more than’, convertirse en ‘to become’, etc.

158 | The development of montón de, a typical locus of grammaticalization

Tab. 11: Distribution of N2-collocates to specifying montón de in Present-Day Spanish N2-cluster

N2-collocates attested in corpus

‘residues’ (31)

cadáveres ‘corpses’ (4), ruinas ‘ruins’ (4), basura ‘waste’ (3), piedras ‘stones’(3), chatarra ‘scrap’ (2), escombros ‘rumble’ (2)

‘other’ (40)

carne ‘meat’ (5), cosas ‘things’ (3)

Very much like the formal stabilization observed for the quantifier use, the morphosyntactic make-up of the specifier use becomes more fine-tuned along the 20th century as well. More precisely, with regard to the determiner to N1, specifying montón de combines preferably with the indefinite determiner (35 occ.). When montón de N2 introduces a topicalized referent, it also easily combines with definite (7 occ.) or demonstratives (12 occ.) determiners. At the discourse level, the smooth combination with definite or demonstrative determiners does not surprise since, in person descriptions, the speaker usually knows very precisely which person (s)he is talking about. The modification pattern admits emphasizers only, i.e. descriptive adjectives which highlight the characteristic the referent can be reduced to (19 occ.). For instance, ridículo ‘ridiculous’ in (87) profiles the insignificance of the referent: this person was nothing but skin, easy to lift when taking him by his lapels only and not heavy at all. The same holds for (88), where sangriento ‘bloody’ primarily serves to highlight a characteristic inherent in heaps of corpses. Yet, by preposing these facets, ridículo and sangriento are given additional emphasis and profiled as the essential characteristic of the respective heaps of skin or corpses. One final remark concerning the modification pattern is the observation that informe ‘shapeless’ is the most frequent combination (with 4 occ.), next to 3 semantically similar adjectives (simple ‘simple’, deshecho ‘ruined’, confuso ‘confused’): example (89) nicely illustrates a chaotic mass of N2, where all N2-entities seem to give up their individuality. (88) ¿Quién no llora al pensar que muy luego tantas y tan nobles vidas van a convertirse en sangriento montón de cadáveres? (1945, R. Sarabia, ¿Cómo se educan los hijos? Lecciones de pedagogía familia) ‘Who does not cry when thinking about the fact that afterwards so many and so noble lifes will become a bloody heap of corpses?’ (89) Mujeres también bajaban y otras subían por la cuesta, a cuyo fondo se veía la Glorieta con el acostumbrado montón informe de autobuses, tranvías, taxis con una tira roja, carritos de mano, vendedores ambulantes, guardias de tráfico, mendigos y público en general detenido con un oculto designio que nada tenía que ver probablemente ni con la llegada de un próximo tren a la estación allí

A diachronic corpus analysis of montón | 159

yacente, ni con (…). (1961, L. Martín-Santos, Tiempo de silencio) ‘Women went down and up the hill. At the end of the hill you could see the Glorieta with its usual disorganized heap of buses, tramways, taxis with a red strip, baggage carts, peddlers, traffic policemen, beggars and people in general with a hidden plan that probably had nothing to do with the arrival of the next train to station situated there nor with (…).’ As to verb agreement, no marked agreement is observed in my corpus. Again, in 2 occurrences only, the BQ-construction functions as the clausal subject and combines with singular verb agreement. In (90), the singular verb form further strengthens the interpretation of the group of houses as a single mass, in which the houses are completely devoid of individuality. (90) Poco á poco la lechosa claror del horizonte se tiñe en verde pálido. El abigarrado montón de casas va de la obscuridad saliendo lentamente. Largas vetas blanquecinas, anchas, estrechas, rectas, serpenteantes, se entrecruzan sobre el ancho manchón negruzco. (1902, Azorín, La voluntad) ‘Little by little the milky light of the horizon dyes pale green. The multicolor heap of houses slowly comes out of the darkness. Long whitish, wide, narrow, straight, twisty streaks interweave on the wide blackish stain.’ Finally, the tendency in Modern Spanish to bring about negative semantic prosody (or at least to primarily occur in affectively loaded contexts) has become conventionalized in Present-Day Spanish, as 63 of 71 specifier uses testify for negative semantic prosody.

4.3.4.4 Conventionalization of the grammaticalized uses Since both quantifying and specifying montón de combine with N2s which are incompatible with their head or literal use, the reanalysis into quantifier/ specifier – or, more carefully the operator-like function – has become overt. Against the background of the entrenchment of the grammaticalized uses, it is no coincidence that in late Present-Day Spanish only the quantifying reading of un montón (de N2) starts to override the boundaries of the binominal syntagm (in three occurrences). Contexts where the full BQ is provided previously, as in (91) indicating un montón de duros ‘a lot of five-peseta coins’, may have functioned as bridging contexts. One could still argue that un montón refers anaphorically to the immense amount of five-peseta coins and de duros remains

160 | The development of montón de, a typical locus of grammaticalization

implicit for stylistic reasons, thus avoiding redundancy. In examples such as (92), however, the antecedent catalanes is not preceded by un montón but by quantifying mucho/a(s) instead. In other words, while anaphoric reference and redundancy avoidance can still be hypothesized, the latter example does not support the idea of an underlying N2. Two kinds of explanation can be put forward for the fact that the morphosyntactic context expansion of quantifying un montón has not yet reached the status of adverbial uses, which is attested (in CREA) from the last 25 years of the 20th century and the first decade of the 21st century onwards, as in Me alegra un montón ‘It makes me very happy’. Either it could mean that the change still has to take place and has not yet spread to the entire speech community. Or else, which is more probable, the first steps of the change towards adverbial uses do occur before the artificial corpus cut-off point of 1974, yet, since the CORDE corpus does not cover oral discourse,91 the hypothesis is hard to verify empirically. Nevertheless, as I consider semantic change to be stepwise (see also De Smet 2009), an abrupt shift towards adverbial uses from 1974 onwards seems implausible. (91) – (…), yo pienso que me gané un montón de duros. – Considere los tiempos que corren. (…). Se me antoja que con sien pesetas va usted muy bien pagado. Ya es un buen montón, me parese a mí. (1957, E. Nácher, Guanche) ‘(…), I think I earned myself a lot (lit. heap) of money. – Consider the times we live in. (…) I think that with one hundred pesetas you are very well paid. It already seems a whole lot/quite a lot (lit. good heap) to me.’ (92) Ahora hay muchos catalanes producto de la guerra civil: los nacidos del 36 al 39 o al 40 y, antes, los refugiados de Madrid o del sur de Aragón. Los que tenían hasta diez años y empezaron a ir al colegio aquí. Un montón. Bien, pues todos ésos: más catalanes que los ampurdaneses de raíz. (1971, M. Aub, La gallina ciega. Diario español) ‘Now there are many Catalans product of the civil war: those born between ’36 and ’39 or ’40. Before, it were the refugees of Madrid or the south of Aragon. Those who were up to ten years old and started to go to school here. Many people (lit. a heap). Well, all of them are more Catalan than the people from Ampurdan.’ || 91 It is common practice among historical linguistics to consider spoken language as the initial locus of variation and change: “it is in the interactional context where language change first happens” (Cornillie 2011; see also Labov 1994; 2011; Faarlund 1989, 71; Kabatek/Jacob 2002).

A diachronic corpus analysis of montón | 161

A few words more on the ‘uncomplemented’ uses of montón de in Present-Day Spanish. Interestingly, its tendency to occur mainly (in almost 60% of the instances) in binominal contexts is continued in Present-Day Spanish, where montón occurs in almost 70 % of its uses (642 of 964 occ.) in the binominal syntagm.92 The remaining bare noun uses comprise literal uses (241 occ.) as well as at least three lexicalized uses (81 occ.). The lexicalization patterns cover to a lesser extent (echar) a montón ‘to throw together, on a heap’, which implies that all individuality of the constituent entities has been lost, while en montón ‘together, jointly, inseparably’ and particularly ser del montón ‘to be average, runof-the-mill, ordinary’ are frequently attested. Interestingly, the lexicalization patterns equally testify to the persistence of conceptual elements of the original use of the changing lexeme in question. A montón obviously stems from the ancient habit of bringing all achievements to a central place, en montón and ser del montón both testify to the concept of ‘indistinctiveness’ or ‘lack of individualization’ also characterizing the grammaticalized uses of montón de (e.g. (93)– (94)). (93) Hoy se llama bizantina a cualquier cosa que se quiera despreciar como decadente, viciada, de poco momento y de complicación inútil; y en arte, en poesía, en historia, en política, en todo, se juzga en montón, por una palabra y en una palabra, cosas que a veces son excelentes y bien distintas de aquellas con que se las agrupa. (1901, Clarín, Siglo pasado) ‘Nowadays we call byzantine anything we want to scorn as being decadent, marred, ephemeral and uselessly complicated. We judge art, poetry, history, politics, everything as a whole (lit. in heap), by one word and in one word, including things that sometimes are excellent and very different from the things we group them with.’ (94) Pero el hombre es algo más que un grano humano entre los otros granos; no se sitúa en la línea de la cantidad sino en la de la calidad; no interesa como uno más del montón sino como Fulano de Tal, personalísimo. (1968, J.J. López Ibor, El libro de la vida sexual) ‘But man is more than a human grain between other grains; his position lies on the quality line and not so much on the quantity line; he isn’t useful as one more of the mass (the heap), but as Mr. So-and-so, very personal.’

|| 92 Note that only 619 of 642 binominal instances are co-extensive (cf. Table 12).

162 | The development of montón de, a typical locus of grammaticalization

Summarizing, the evolution of montón de in Present-Day Spanish indicates that the three binominal uses of montón, viz. head, quantifier and specifier use, have more or less adopted a stable form and meaning. The fact that the quantifying use has become conventionalized is nicely illustrated in (95). The example has been taken from Introducción a la Lexicografía Moderna: the lexicographer argues that the quantifying interpretation deserves a proper definition line in spite of the persistence of the source image (está viva la imagen que le sirve de base). (95) Y terminaremos con el postulado primero mencionando la coyuntura en que el significado recto conserva toda su vigencia junto a uno traslaticio en que está viva la imagen que le sirve de base. Tal ocurre con las dos acepciones de "montón" que aparecen en las frases siguientes: "la Catedral es un montón de escombros"; "tengo que decirte un montón de cosas". En este caso no podría omitirse la indicación de fig. ante la segunda acepción. Si en lugar de "un montón de cosas" se dijese "un cúmulo de cosas" la solución sería contraria, (…). (1950, J. Casares, Introducción a la Lexicografía Moderna) ‘And we will end with the first postulate, mentioning the conjuncture in which the straight meaning preserves all its validity together with a figurative one in which the image that serves as a base is alive. This is the case with the two meanings of “heap” in the following sentences: “the cathedral is a heap of rubble”; “I have to tell you a heap of things”. In this case the figurative meaning can’t be left out before the second entry. If instead of saying “un montón de cosas” (‘a lot of things’) one would say “un cúmulo de cosas” (a set of things) the solution would be different, (…).’

4.3.5 Intermediary conclusion The case-study of montón de provides support for viewing GR and lexical constructionalization in it (Traugott 2011) as stepwise developments (see also De Smet 2009 and Traugott/Trousdale 2010), and not as the outcome of abrupt reanalyses. Table 12 is added by way of synthesis and presents the distribution of the different uses per time period. The chart in Figure 4 visualizes the changes in (relative) frequency per use. In line with the usage-based approach to GR, the proportions of functional uses gradually rise while the proportion of the literal use gradually goes down. However, in contrast to the common assumptions on reanalysis, the emergence of the grammaticalized uses is not preceded by a productive ambiguous use. In addition to an overall rise in frequency of the grammaticalized uses

A diachronic corpus analysis of montón | 163

(Bybee 2003), the binominal construction montón de in itself has become relatively more frequent over time.93 In addition, as the chart and frequency indicate, the development of new uses did not compete out the original use. Instead, per time period, a synchronic layering (Hopper 1991) is observed. Tab. 12: Layering of uses activated by montón de per time period Head

Quantifier Specifier

Ambiguous Indeterminate Total

1250–1450

# %

47 (0.76)

8 (0.13)

0 (0)

7 (0.11)

0 (0)

62

1450–1730

# %

162 (0.65)

44 (0.18)

27 (0.11)

9 (0.04)

8 (0.03)

250

1730–1900

# %

217 (0.60)

42 (0.11)

58 (0.16)

24 (0.07)

26 (0.07)

367

1900–1975

# %

322 (0.52)

116 (0.19)

71 (0.11)

70 (0.11)

40 (0.07)

619

100%

80% Indeterminate 60%

Ambiguous Specifier

40%

Quantifier 20%

Head

0% 1250–1450

1450–1730

1730–1900

1900–1975

Fig. 4: Gradual extension of the uses for mónton de N2

By way of illustration, Figure 5 visualizes the directionality of the major semantic glides. The dashed arrows represent the pragmatic inferences regularly associated to a particular use. Although they arise in highly specific contexts, they are easily repeated in other uses. This is probably due to the fact that the differ-

|| 93 For Medieval Spanish, CORDE counts 22455652 words, for Classical Spanish 91614592, for Modern Spanish 41672542, for Present-Day Spanish 43206559. Consequently, in terms of frequency per million words, montón de presents a relative rise: 2.72 for Medieval Spanish, 2.74 for Classical Spanish, 8.78 for Modern Spanish and 14.26 for Present-Day Spanish.

164 | The development of montón de, a typical locus of grammaticalization

ent uses were closely intertwined, especially in the early centuries. It also shows that the pragmatic inferences were no more than inferences and therefore do not count yet as change strictly speaking, which would imply the new formmeaning to have spread over the entire speech community. Consequently, the clustered structure and directionality of the pathways as represented in Figure 5 may appear artificial. Yet they provide a useful overview of the cascade of semantic micro-changes and metonymical extensions in the initial meaning of a heap of accumulated harvest or war booty which were essential for the quantifying and specifying uses to become possible at all: – ‘accumulated N2s > unusual or unnatural gathering of N2 > surprising or excessive gathering of N2’; – ‘high > not necessarily high > low and/or inferior’; – ‘high > a lot (of N2s) > diversification of N2s; – ‘high > a lot (of N2s) > indeterminate N2s’; – ‘N2 as achievement > N2 as possession’; – ‘intentionally (heaped up or accumulated) > newly produced’ – ‘intentionally (heaped up or accumulated) > brought together in transformation’ (e.g. a destruction or reduction to ruins in war circumstances) – ‘intentionally gathered > mere spatiotemporal contiguity > unexpected cooccurrence’ – ‘accumulation of homogeneous N2 > accumulation of heterogeneous N2 > nothing but N2 > unidentifiable N2 > insignificant N2’.

Fig. 5: Semantic glides in the development of montón de e.g. & sobre la fuessa mando el rey ayuntar un grant monton de arena. (1275)

e.g. todo el mundo tiene un montón de lágrimas por derramar en la vida. (1966)

e.g. Tengo que cónsultarte un montón de cosas. (1941)

e.g. (…) emplea en ello un montón de minutos. (1951)

A lot of N2

e.g. En un momento se recaudó allí un montón de calderilla y algunas pesetas. (1921)

Excessive number of types of N2

e.g. (…) la reunión cariñosa y animada de un montón de hermanos de diferentes apellidos(…) (1969)

Mishmash of heterogeneous N2

(Unusual,) large number of spatiotemporally contiguous N2

diría! (1500)

e.g. ¿Ese montón de polvo? (1972)

Moderate vertical configuration of (low-quality) N2 intentionally heaped up without order

e.g de aquel montón de hijas del trabajo que hace sudar, salía un olor picante (1884)

Remarkable collection of unidentical N2 that share a single, noticeable feature

e.g. Dejaba la pequeñez de su vida tirada a sus pies como un montón de olvidos (…). (1970)

Insignificant N2 which remain

doesn’t matter

precise identity

+ Pragm.Inf.:

Pragm.Inf.:inferiority

e.g. Jherusalem tornada en monton de piedras. (1400)

e.g. ¡oh qué montón de cosas le

Nothing (left) but N2

temporally contiguous N2 

e.g. por que el ha trobado vn grant monton de oro. (1400)

city after a battle

Bridging context: ruins of a

Large number of spatio-

e.g. por que el ha trobado vn grant monton de oro. (1400)

 literally occupying space

(or  heap).

ber of possessions accumulated

intentionally accumulated N2

A high vertical configuration of

Bridging context: large num-

e.g. cuando de tropel vimos bajar un montón de monigotes de todos trajes(…) (1727)

Mishmash of N2

Indeterminate number of N2



A diachronic corpus analysis of montón | 165

166 | The development of montón de, a typical locus of grammaticalization

The case-study on montón de also reinforces the constructionalist aphorism that the “coevolution of form and meaning should be understood not as simultaneous but as potentially separate form and meaning shifts toward a new construction” (Traugott 2011; see also Bybee et al. 1994). Whereas the first quantifying interpretations of montón de are observed in Medieval Spanish, the shift from lexical noun category to ‘quantifier’ has not even occurred in Present-Day Spanish. In other words, no morphosyntactic reanalysis in the strict sense occurs. Instead of absolute structural decategorizalization (Lehmann 1995), the morphosyntactic development of montón de can best be approached as a gradual increase in combinatorial constraints: over time, the three uses develop specific collocational preferences or a specific collocational shape (which I will claim to be conceptually motivated, cf. infra). For instance, by the end of the 20th century quantifying uses of montón de almost exclusively combine with the indefinite determiner, intensifying or quantifying adjectives and N2s designating discourse entities, people, time indications, objects made of paper or money. Present-Day specifier uses show a strong preference for the indefinite determiner, intensifying adjectives and N2s which can somehow be conceived of as ‘remains’. In other words, by the end of the 20th century, all three uses, viz. head, quantifier and specifier, seem to have adopted a stable form – or more carefully, a stable collocational shape. It is thus fair to say that, by the 20th century, the head use, quantifying use and specifying use have crystallized into three different form-meaning pairings or constructions. Before Present-Day Spanish, I rather consider the head, quantifying and specifying readings as different uses – generally triggered by contextual clues – of one construction, viz. montón de. The latter observation raises the question as to determine what took the grammaticalized uses of montón de so long to become conventionalized. It bears pointing out that gradience, i.e. the fuzzy nature of category boundaries, may be relatively stable over long periods of time (Traugott/ Trousdale 2010, 5). It is also to be mentioned that there is no such dramatic morphosyntactic shift from noun to quantifier, given that even the grammaticalized and constrained uses of un montón de continue to be highly analyzable. Further, montón de N2 appears to have been the ‘only’ productive BQ for a long time (cf. Chapter 5). Largely oversimplifying, Medieval Spanish is when the quantifying use emerges. In Classical Spanish, the original uses are partially continued but also gradually extend via new pragmatic inferences and the specifying use emerges. The uses of montón de observed in Modern Spanish testify to the partial continuation of existent uses, the entrenchment of the quantifying use and new extensions within the specifying use. In Present-Day Spanish, the three major uses, viz. the head, the quantifying and the specifying uses, combine with a more or

Conclusions | 167

less stable collocational shape and the quantity assessment by montón extends to new syntactic contexts, viz. adverbial uses. Finally, when comparing the descriptive analysis of montón de to its lexicographical treatment, my findings suggest that from the 18th century onwards, the dictionary entries are quite accurate and capture recent semantic changes. Before the definition provided by the first reference dictionary Autoridades in 1734, montón was often described as a little heap of things, usually of wood, land and hay. The divergence with the pragmatic tendency to stress the height of one’s heap of achievements as observed in my dataset, may be motivated by the meaning of the suffix –ón94 on the one hand and the fact that the earliest dictionaries do not tend to include authentic examples yet merely start from the morphological analysis of monte+ón. Dictionaries are however late in mentioning the lexicalized or idiomatic expression un montón de tierra for designating old and idle persons and fail to mention the typical specifier use, which in Modern Spanish was even more exploited than its quantifying use.

4.4 Conclusions As this chapter is entitled The development of montón de, a typical locus of grammaticalization, it is timely to consider why montón de constitutes a typical locus of GR. Basically, its development corresponds to the widely accepted view of GR. From a lexical item designating a heap of harvest or war booty, montón de came to indicate excessive quantity, indeterminate quantity and category membership. Generalizing over the criteria listed by several authors (see Lehmann 1995; Hopper 1991; Himmelmann 2004), I consider the quantifier and specifier uses of montón de as grammaticalized instances for involving the following correlated features: – REBRACKETING. From [Det.+montónhead [de + N2]] into [[un montón de] + N2head]. Evidence for this reanalysis is observed when verb agreement is made with N2 or when premodifying adjectives scope over both nominals at

|| 94 According to De Bruyne (1979), the suffix -ón stems from Latin -o, -onis which had a “signification individualisante” (1979, 22, ‘an individualizing meaning’), which could either be augmentative, pejorative or diminutive. For instance, “Naso désigne quelqu’un qui frappe par son nez. Cette anormalité du nez peut consister soit dans sa grandeur inaccoutumée, soit, au contraire, dans sa petitesse.” (1979, 22, ‘Naso designates someone whose nose is noticeable. This nose defect may consist either in its unusual magnitude or, on the contrary, in its tininess.’) De Bruyne further argues that the exact pragmatic value of –ón in Spanish is often to be derived from the context.

168 | The development of montón de, a typical locus of grammaticalization

–

–

–

the same time. Crucially, since the structural rebracketing only took place after the ground had been prepared by the stepwise semantic alterations to montón de’s source semantics and un montón de N2 remains highly analyzable and continues to allow for fluctuations in the co-selection pattern, it is more careful to argue that the reanalysis is primarily semantic in nature. PARTIAL STRUCTURAL DECATEGORIZATION. The loss of typical noun features for montón is to be interpreted as the development strong preferences in its coselection pattern. Quantifying and specifying interpretations are triggered in the binominal construction only and develop tight combinatorial restrictions as to the N1-determiner, adjectives, agreement and right collocate. The N2-combinations per use display a clustered structure: from the moment a particular collocation is conceived of as particularly well-formed by native speakers, they are motivated to keep the use of montón de to contexts perceived as somehow similar. The speakers’ conformity to existent combinations leads to specific combinatorial patterns. Finally, the fixation of the constituents prevents the extraposition of [de + N2]. SEMANTIC SHIFT. Montón shifts from the concrete reading of a high heap or conical configuration of accumulated goods, which literally occupies a certain space, to the less compositional and more abstract readings of assessing (an excessive or indeterminate) quantity or specifying adherence to a particular type. In fact, before turning to quantity or quality assessment, montón de undergoes several discrete micro-changes which are summarized in Figure 5. Crucially, in line with Traugott’s (2003b) model of redistribution of meaning, I argue that GR of montón de resulted in a loss of concrete specificity, but not in a loss of semantic complexity. PARADIGMATIZATION. On the one hand, un montón de remains an analyzable chunk and is not subject to phonological attrition as observed in English alotta or helluva. On the other hand, the binominal syntagm as a whole enters the paradigm of quantifiers and modifiers in its grammaticalized uses. In recibí un montón de dinero ‘I received a lot of money’, un montón de can easily be replaced by the absolute quantifier mucho (dinero) ‘a lot (of money)’. Important meaning components get lost in this replacement though: the fact that the amount of money surprises the speakers, that the money is somehow gathered, etc. Similarly, specifying montón de fulfills a categorizing function similar to adjectives by indicating that no distinction can be made between the N2-entities and that the N2-entities constitute its very essence. It is however hard to substitute montón de by an adjective capturing the very same meaning, as such an adjective is not at hand in Spanish (and again, if similar adjectives would exist, this use of montón de would be redundant).

Conclusions | 169

–

–

–

–

–

CONTEXT EXPANSION AT TWO LEVELS. The change into a grammaticalized construction involves host-class expansion in that the class of right collocates is no longer restricted to N2s designating harvested crops, earth, stones, corpses, ruins, etc. On the basis of the written dataset until 1975, the development of montón de does not give rise to syntactic context expansion: the syntactic context remains the binominal syntagm. The GR of montón de also displays semantic-pragmatic expansion, which is nicely illustrated in Figure 5. In addition, as both the quantifying and specifying reading originate in the conventionalization of pragmatic inferences and thus subjectification, both uses are provided with an evaluative potential. LAYERING AND DIVERGENCE. The emergence of the ‘newer’ functional uses of montón de within the binominal syntagm does not discard the ‘older’ literal reading. Instead, the GR process yields a systematic synchronic polysemy or gradience. Nor does the GR of montón in the binominal syntagm discard its use as a bare noun. SUBJECTIFICATION AND PRAGMATIC STRENGTHENING. The development of montón de involves subjectification, both in Traugottian terms (1989; 2003b; 2010) and in Langackerian terms (2000; 2006). On the one hand, the emergence of the quantifying reading involves ‘textual subjectification’, as the quantity assessment relies on the speaker’s evaluation of the number of N2s on the basis of the pragmatically inferable scale inherent in montón’s literal meaning. On the other hand, the quantity assessment by montón de implies a shift in profile, from objectively construed to subjectively construed: whereas the facet ‘high’ is construed onstage as an explicit, focused object of conceptualization in montón de’s literal use, it becomes subjectively construed as ‘a lot’ in montón de’s quantifying use. In addition, several chains of shifts in profile characterize the subtle semantic changes, for instance ‘accumulated N2s > unusual or unnatural gathering of N2 > surprising or excessive gathering of N2’, ‘high > not necessarily high > low and/or inferior’ (cf. Figure 5). PERSISTENCE. Successive shifts in profile or the conventionalization of invited inferences seem to indicate that the new reading was immanent in the source construction. One particular facet persists in any ‘newer’ exploitation of montón de, viz. the ‘accumulated nature’ of N2. As I will illustrate in Part 3, the discourse context generally co-determines which facets of montón de’s source semantics will be activated in a particular occurrence. RISE IN FREQUENCY. Not only does montón de become relatively more frequent overall (cf. footnote 29), the number of grammaticalized uses gradually in-

170 | The development of montón de, a typical locus of grammaticalization

creases over time (cf. Table 12). Further, the binominal syntagm gradually becomes the preferred morphosyntactic context of montón.95 Moreover, the semantic changes observed in the binominal occurrences partially affected some uncomplemented uses of montón as well. Finally, the GR of montón de did not exclude the later lexicalization of a particularly frequent combination: the nowadays fixed expression montón de tierra, which is listed as such in actual dictionaries and designates weak and idle persons, originates in montón de’s specifier use. Its presence in actual dictionaries sharply contrasts however with its decrease in use with regard to the 19th century. Crucially, as the development of montón de lacks other formal characteristics of GR such as (phonological) attrition, morphological bonding and increase in automaticity, it can only be considered an instance of GR when a redistribution of collocational constraints, viz. a narrowing down of the choices in the combinatorial pattern (involving determiner variation, modification and the hostclass), is considered as a valid change in form. Following Brems (2010), I consider the head, specifier and quantifier use as “collocationally constrained constructions”: within the binominal construction, QNs act as fixed lexical items in the sense of collocational nodes that take different sets of collocates depending on whether they are used as a head, quantifier or valuing quantifier. Hence, in SN-patterns [KV: Size Noun] the predetermination takes the shape of strong node-collocate co-selection related to the various constructional uses of SNs. (Brems 2010, 101)

In addition, the development of montón de hints at ‘hesistant’ GR (see also De Smet/Ghesquière 2010): although the earliest bridging contexts and quantifying inferences are found by the end of the 14th century, no immediate dramatic increase in frequency is observed (cf. Table 12), no immediate correlated change in form is attested, and more importantly, the ‘newer’ grammaticalized use in Present-Day Spanish still coexists with the literal source construction. Further, the development of grammaticalized uses of montón did not exclude the lexicalization of un montón de tierra and the rise of particular inferences, semantic facets or connotations in one use affected the developments in other uses (e.g. the pragmatic inference of ‘inferiority’, invited within the head use when the the facet ‘high’ got bleached, was copied to the specifier use). || 95 In frequency per million words, a relative rise in frequency of the binominal construction is observed: from 2.72 in Medieval Spanish to 14.26 in Present-Day Spanish. By way of comparison, the respective frequencies per million words for montón outside the binominal construction come down to 2.58 and 7.45 respectively.

Conclusions | 171

The present case-study thus asks for two refinements of the widely accepted definition of GR. First, the redistribution of co-selection patterns that corresponds to a shift towards a more grammatical function should qualify for ‘grammaticalization’. Second, in line with the recent trend in GT and diachronic construction grammar, GR should be considered a gradual and stepwise phenomenon (Traugott/Trousdale 2010; De Smet 2009; 2011). The second question raised by the title of this chapter is why particularly montón de engages in GR. Does it imply that the remaining QNs analyzed in this study qualify less for GR according to the parameters listed in grammaticalization theory? On the contrary, the comparison of the frequencies in Table 2 of Chapter 3 rather suggests that several QNs are ‘more grammaticalized’ than montón de, simply for counting relatively more grammaticalized uses, although montón de appears to be the only high-frequency QN. The complex frequency pattern will be motivated in the next chapter: it does not indicate that the GR of the remaining QNs is less advanced, but suggests that montón de has functioned (and continues to do so) as a model attracting the quantifying and specifying uses of other QNs via analogical thinking. Not the proportion of grammaticalized uses but the context-expansion should determine the ‘degree’ of GR (see Chapter 6). Still, one might wonder why particularly montón came to function as the model. Howe/Novell (personal communication) attribute the advanced stage of GR of montón de (and mogollón de) to the augmentative suffix –ón “which in itself represents a particular type of nominal abstraction” and as such, preempts the metonymic shift from ‘size’ to ‘quantity’. Although this explanation sounds plausible synchronically, the hypothesis is hard to verify empirically. When looking at the historical data in CORDE returned by the query ‘monte de’ (viz. the stem of montón when taking the suffix off) it is immediately obvious that the binominal syntagm primarily served to indicate the exact region or place where the mountain was located (e.g. el monte de Sinaí ‘the mount Sinai’, el monte de Gibraltar ‘the mountain of Gibraltar’) or other identifying facets (e.g. el monte de Dios ‘the mount of the Lord’). For Classical and Modern Spanish, the query returns some ‘grammaticalized’ readings, viz. specifying uses of monte de, which seem to copy the typical contexts of appearance of montón de (e.g. convertirse en estéril monte de ruinas ‘to turn into an sterile mountain of ruins’). Similarly, in the few instances where literal monte is coextensive with N2, it generally exploits original usage contexts of montón de, such as a heap of war booty, a heap of ruins, a woodpile. Interestingly, no unambiguous quantifier uses for monte de are attested in CORDE. The same query in CREA indicates that the synchronic variation mirrors monte de’s diachronic development – at least

172 | The development of montón de, a typical locus of grammaticalization

in peninsular Spanish: the binominal construction still primarily serves to identify the mountain, few specifying uses are observed, yet no quantifying uses. In other words, the comparison to the dataset of monte de indeed seems to indicate that montón de is more inclined to systematically develop quantifying uses. Still, Howe/Novell’s hypothesis fail to account for the fact that mogollón nor mogolla are used until the end of the 20th century: if –ón would preempt the metonymic shift from size to quantity in mogollón, we would expect earlier grammaticalized uses for mogollón de as well. On the other hand, native speakers use to mention montón and mogollón in one breath as the most general and frequent QNs. Chapters 5 and 6 will explain that the role of the augmentative can still be considered crucial when rethinking the history of montón de, the development of the remaining QNs and the emergence of the abstract BQ-construction in the light of a constructional network model. Intuitively, I argue that the prototype role or triggering function for montón de in the GR of the remaining QNs is correlated with the vagueness of its source semantics as to the internal constitution of the mass, which results in the smooth lifting of collocational restrictions. In fact, anything can be accumulated in a heap by simply putting the entities one above the other. The diachronic analysis of the other QNs in Chapter 5, that pays particular attention to the mechanisms of reanalysis and analogy, nicely fits in with the constructionist approach, which typically analyzes sets and searches for similarities among them, both paradigmatically and syntagmatically (Traugott 2011). The diachronic case-studies of the remaining QNs will enable me in Chapter 6 to answer the question whether the stepwise evolution from a literal reading comprising a quantifying potential to quantity assessment characterizes the history of montón de only. If the metonymic shift is repeated by other QNs as well, it may be characteristic for the abstract BQ-construction, if such a construction exists at all. And if the formal pattern N1 de N2 is indeed automatically associated with quantity assessment, how did this schematic construction emerge? How does it relate to the individual histories of the QNs? For now, I refer to De Smet (2009) who states that a certain construction can only be analyzed as a particular grammatical category (the ‘auxiliary’ in his study), from the moment a second member arises : Before that time, the ‘auxiliary’ would have been an under-analysed and grammatically isolated chunk of language that had undergone both gradual category-internal change and automation. Only when another chunk developed, language users could perceive a similarity between the two. At that point a category ‘auxiliary’ arises, which, however, entails no more than that perceived similarity. (De Smet 2009, 1751)

5 The role of conceptual persistence and analogy in the development of BQs Chapter 4 discussed the stepwise emergence of quantifying and specifying uses of the high-frequency BQ montón de. The aim of the present chapter is to verify whether other QNs followed a similar developmental path. Somewhat misleadingly, Table 2 in Chapter 3 – which presents the synchronic distribution of meanings observed per QN – creates the impression that the GR process of various other QNs is further advanced than the GR montón de is involved in. Synchronically, for different QNs, e.g. aluvión de, alud de, barbaridad de and letanía de, the tendency to evoke functional readings is more outspoken than for montón de. However, their morphosyntactic behavior does not conform to all hallmarks of traditional GR. More importantly, on the assumption that GR usually boosts the construction’s frequency, the ‘low’ frequency of these QNs sharply contrasts with montón de’s high token frequency. Further, in the case of montón de and la mar de only, the GR led to syntactic context expansion (Himmelmann 2004), viz. to adjective and adverb intensifying uses. The finegrained analysis of montón de’s development led me to consider its diachrony as a typical process of GR. Since the conventionalization of its quantifying and specifying uses goes back to Classical Spanish, it may be speculated that its early GR entailed a cascade of changes in semantically close QNs. On closer scrutiny, however, the explanation of the emergence of quantifying and specifying uses of other QNs in terms of a snowball account does not do full justice to the historical facts. The present chapter zooms in on the development of seven QNs, viz. pila de, aluvión de, letanía de, hatajo de, barbaridad de, mogollón de and mar de, and shows that, although they all give way to quantifying and qualifying instances, the pathways of change followed are rather distinct.96 Generalizing over the individual histories, two main trajectories are observed: while some QNs seem to skip the pre-grammaticalization stage and almost immediately combine with quantifying interpretations in the binominal constructions, other QNs do involve in a gradual process of change. Yet, in contrast to montón de, the latter group of QNs tends to continue the original source semantics of the QN to a high degree. Further, their grammaticalized uses tend

|| 96 According to Traugott (2010) it’s a truism in GT that “each string has its own history, but conforms to general schematic change-types in ways that are partly constrained by the particularities of the original meaning-form relationship.” (2010, 46)

174 | The role of conceptual persistence and analogy in the development of BQs

to combine with negatively loaded contexts. Even more puzzling is the particular development of mar de: this QN is also highly frequent, yet primarily intensifies adjectives while preferably combining with the definite determiner. What is beyond doubt is that montón de was clearly given a head start and that the trajectories of the remaining QNs cannot be explained without recourse to montón de’s exemplar function. Of crucial importance for a good understanding of the development of the other BQs – and the schematic BQ-construction – are the notions of analogy and (conceptual) persistence. Analogy is based on the language user’s recognition of similarities between two structural or functional contexts (Givón 1991, 258) and refers to the mechanism whereby “the earlier structure is restructured to match an existing one, and as a result, the newer one has a new structure” (Traugott/ Trousdale 2010, 36). Although analogy is a precondition for much change it is not to be equated with the discrete change in itself, which Traugott/Trousdale suggest to call analogization. My understanding of persistence is based on Hopper’s (1991, 22) GR criterion of lexical persistence and refers to the tendency of some grammaticalizing items to retain particular features of their original lexical use, which may continue to influence their further developments in various ways. For reasons that will become clear in the remainder of this chapter, I situate the persistence of lexical features at a more conceptual level thereby referring to the frame or image conceptualization (Fillmore 1985; 2003)97 activated by the QNs, which exceeds the definition of a QN given by reference dictionaries. As such, the notion of conceptual image persistence allows for unprofiled elements (in pre-grammaticalization stage) to become foregrounded in the grammaticalized uses, while involving metaphoric and metonymic extension mechanisms. Likewise, the extension towards new N2 combinations will be shown to be frame-based, rather than exemplar-based (Bybee/Eddington 2006).

|| 97 The notion of ‘frame’ is borrowed from Fillmore (1975; 1985; 2003) and redefined by many others (Lakoff 1987; Ungerer/Schmid 1996). It referred initially to “any system of linguistic choices – the easiest being collections of words, but also including choices of grammatical rules or linguistic categories – that can get associated with prototypical instances of scenes” (Fillmore 1975, 124). In later publications (Fillmore 1985, 223), the notion of ‘scene’ received a more cognitive (rather than linguistic) interpretation. In line with Fillmore/Atkins (1992, 75), we use the notion of ‘frame’ as the “cognitive structure”, viz. the experiential knowledge or set of concepts, activated by an item in the speaker’s mind, so that to understand the item one has to relate it to the set or system in question. Interestingly, QNs displaying a high degree of persistence often co-occur with lexical items belonging to the same frame (viz. the frame evoked by N1 as a lexical item), which enhance the degree of persistence.

Methodological background | 175

The present chapter addresses the following questions: does the developmental path of other QNs correspond to the pathway observed for montón de? Are the trajectories followed by other QNs influenced by the early GR of highfrequency montón de? The partially affirmative answer to those questions throws up the issue of a diachronic model which allows for capturing at the same time the individuality of the different pathways and the analogizations observed, which will be extensively commented on in Chapter 6. The structure of the present chapter is as follows. First, Section 5.1 motivates the methodological choices made. Section 5.2 then provides a rough outline of the GR of the remaining BQs. Because the developmental trajectories followed by the QNs are so distinct, they are discussed separately. In Section 5.3, the key importance of the mechanisms analogy and persistence is explained in the light of the extensions towards new N2-combinations and the growth of the lexical domain of QNs. Section 5.4 lines up the peculiarities observed in the GR of the set of QNs in this investigation and summarizes the requirements for a diachronic model which is sufficiently good to capture the less expected developments.

5.1 Methodological background The data which lend support to the claims made in this chapter were extracted from two on-line diachronic corpora, viz. the CORDE offered by the Real Academia Española (see Chapter 4) and the Corpus del español compiled by Mark Davies. The Corpus del Español is a 100 million words corpus that supports not only searches for exact words or syntagms (with or without wildcards), but also for parts of speech, surrounding words and synonyms. Although the corpus covers a wide geographic variety, it is not possible to filter for a particular Spanish variety or genre in its interface. Another weakness is the amount of wronglytagged examples. Further, no statistical conclusions can be drawn on the basis of this corpus since from one search to another slight discrepancies show up with regard to token frequencies. When re-consulting the query [ [nn*] de [nn*] ], the following token frequencies were found: 98636, 98600, 98665, 98614, 98456, 98669, etc. Moreover, when the same query was run for the 20th century, full access to the data was denied and the sample was restricted to N1 de N2 combinations that occur at least three times. It is methodologically crucial to compile a database that is as representative as possible, although representative samples of diachronic data are theoretically impossible (Cornillie 2012, p.c.). The criteria for the choice of the corpus used correlates with the issue under investigation. As this study is primarily con-

176 | The role of conceptual persistence and analogy in the development of BQs

cerned with Peninsular Spanish, I chose to use CORDE when analyzing the subtle semantic glides different QNs underwent and the stepwise extensions of their co-selection pattern. For the QNs pila, aluvión, letanía, hatajo, barbaridad, mogollón and mar, exhaustive samples were extracted from CORDE by the query ‘QN + de’, restricted to the Peninsular variety. The orthographic variants of the QNs were also taken into account.98 Table 1 provides a detailed picture of the number of occurrences attested. Tab. 1: Proportions of raw data and manually filtered binominal constructions

pila aluvión letanía hatajo barbaridad mogollón mar

‘QN + de’ in CORDE

Binominal constructions

261 142 53 85 34 0 2232

61 48 29 46 6 0 960

As this search procedure yields a large number of false (i.e., not coextensive) binominal constructions, these have to be removed manually in order to ensure that each N2 is a noun representing a group of entities whose quantity is assessed by means of the QN (i.e., that each QN is co-extensive with N2). Since the primary focus of this investigation is on the development of the binominal construction, occurrences where for instance the QN is followed by non-nominal elements (as in (1)) or the binominal construction constitutes a compound (as in (2)) are filtered out. I might of course lose crucial examples of syntactic context expansion, which could provide further evidence for the GR process assumed. Yet spot checks in the corpus reveal that except for montón and mar, adverbial uses for the remaining QNs are rare or totally absent. It can therefore be safely assumed that the exclusion of the occurrences where the QN is not followed by a PNP – in the interest of a manageable retrieval procedure – does not invalidate the results.

|| 98 It is common knowledge among historical linguists that diacritics are uncommon in Medieval Spanish and that consonants may have different orthographic representations. Of concern to our investigation is the initial ‘h’ that is not always used, the Present-Day ‘v’ which corresponds to ‘u’, ‘b’ and ‘v’ and the consonants ‘t’ and ‘j’ who correspond to ‘d’ or ‘t’ and ‘x’ respectively.

Methodological background | 177

(1)

Se puede construir la pila de la manera siguiente (…). (1935, R. Yesares Blanco, Industrias para el aficionado. El grabado (…)) ‘The pile can be construed in the following way (…).’

(2)

Detúvose junto a la pila de agua bendita sin poder dar un paso. (1876, B. Pérez Galdós, Gloria) ‘He stopped next to the stoup and could not go on anymore.’

In addition, the Corpus del Español has been used in order to construe the semantic map of possible QNs, i.e. of nouns which may occupy the N1-slot of the binominal construction, since CORDE does not allow for similar part-of-speech searches. Per century, I looked at all data accessible for the query [[nn*] de [nn*]], which means that from the 15th century onwards already, I had to limit the extraction to instances which occur at least two times, and from the 19th century onwards to ‘N1 de N2’ combinations that occur at least three times. Table 2 shows per century the number of ‘hits’ (or types, which refers to number of ‘N1 de N2’ combinations found) and ‘occurrences’ (or tokens, which refers to the total number of binominal constructions, generalizing over all types). Again, the search strategy yields lots of irrelevant tokens which are systematically filtered out (e.g. when the PNP designates the owner of N1 or the person N1 is generally associated to (as in un barco de españoles ‘a boat of Spanish men’), the material N1 is made of (as in un vaso de barro ‘a cup of clay’), etc.). In an attempt to arrive at an overview of nouns with a quantifying potential, I based the inventory primarily on the concordances per ‘type’ listed. In case of doubt, I also checked the full context of the occurrence. Tab. 2: Number of ‘hits’ and ‘occurrences’ per century

13 C 14 C 15 C 16 C 17 C 18 C 19 C 20 C

hits (types)

occurrences (tokens)

7145 3789 4084 8241 6055 6545 7670 8321

34155 11989 31047 45330 27470 29440 60556 98602

178 | The role of conceptual persistence and analogy in the development of BQs

5.2 The development of binominal quantifiers The (development) of BQs in Spanish has so far not received a great deal of attention in the literature. There are, however, numerous studies that examine the emergence of the English binominal construction (Brems 2010; Traugott 2007; 2008a; 2008b; Trousdale 2010), suggesting a gradual change from partitive to free adjunct use. More precisely, Traugott (2008a; 2008b) traces for BQs and complex determiners the following pathway of change, thereby suggesting that abstract constructions may serve as attractor sets for local reanalysis: Step I pre-partitive or (limited partitive)

>

Step II (expanded) partitive

>

Step III degree modifier

>

Step IV free adjunct

Fig. 1: Change schema proposed by Traugott (2008, 228)

In her investigation, the GR chain proposed in Figure 1 accounts for the development of free adjunct uses of (a) sort of NP, (a) lot of NP and (not) a shred of. By way of illustration, the different steps in the chain proposed are applied to montón de: Step I (…) e dixo Jacob asus hermanos: coget piedras e cogieron piedras e fizieron monton ; e comieron ende sobre el monte. (1400, anon., Biblia Escorial I-j-4: Pentateuco) ‘‘(…) and Jacob said to his brothers: take the rocks and they took rocks and made a heap; and then they ate on the hill.’ Step II ¿Qué montón de piedras es aquel – preguntó Andrenio – que está en medio de las sendas? (1651, B. Gracián, El Criticón I) ‘What a heap of stones is that over there – Andrenius asked – in the middle of the paths? Step III Pero pasemos a otro asunto. Tengo que consultarte un montón de cosas. (1941, R. León, Cristo en los infiernos) ‘Let’s pass to another subject. I have to ask your advice on a lot of things.’ Step IV – ¿Cómo ha cambiado? – Un montón. De aquel niño chato... – ¡Chato! (1990, press) ‘– How (much) did he change? – A lot. From that lovely boy… – Lovely!’

The development of binominal quantifiers | 179

At first sight, the chain of functional shifts proposed for English binominals applies to the emergence of the quantifying reading of montón de as well. On closer scrutiny, however, it fails to account for the development of specifying uses and does not do full justice to the role of the specific contexts that gave rise to the pragmatic inferences of quantity. Traugott (2008a, 227) recognizes “the locality of structural and semantic changes” and puts more carefully that “at some point in their histories they all participated in the Partitive Construction and at a later point also in the Degree Modifier Construction”. She also adds that “[c]hanges do not have to ‘go to completion’; they do not even have to occur” (2008a, 235), yet continues to assume that, if they occur, the changes will be in line with the change schema proposed. In the remainder of this chapter, the generalizability – if not universality – of this schema will be verified by extending the dataset of QNs under investigation to pila de, aluvión de, letanía de, hatajo de, barbaridad de and mar de.99 Next to the manageability of the dataset, the guidelines underlying the selection are the following: pila de is chosen in view of its semantic closeness to montón; aluvión de, barbaridad de and letanía de for their relatively high proportion of grammaticalized uses in synchrony; hatajo de for the dominance of specifier uses and mar de for the synchronic idiosyncrasy of preferring adjectives in the N2-slot. As mogollón de is systematically mentioned by native speakers in addition to montón de as the prototypical QN in (spoken) PresentDay Spanish, the corpus search included this QN as well. Interestingly, the CORDE does not return any valid instance. Per QN, the returned occurrences are coded for the semantic-pragmatic and morphosyntactic features which turned out to be essential in the stepwise constructional changes montón de underwent. Semantic-pragmatic criteria pertain to the function of N1, the degree of conceptual persistence, the polarity of the semantic prosody or affective load surrounding the binominal construction. Morphosyntactic criteria concern the co-selection pattern of the QN with regard to the determiner (to N1 and in rare cases to N2), the modification of N1, the verb and phoric agreement and the type of N2.

|| 99 Section 5.3.2 briefly touches upon the development of the combinatorial pattern of racimo de. The case-study on racimo de performed in Verveckken (2012) is based on extractions from the Corpus del español and is thus not limited to peninsular Spanish. For the sake of coherence, no frequency tables on racimo de are included in this chapter.

180 | The role of conceptual persistence and analogy in the development of BQs

5.2.1 Outline of the history of pila de Pila turns out to be a highly polysemic noun (Corominas/Pascual 1991, vol. 4, 542–543). In many binominal uses, the preposition de does not profile a coextensive relation between N1 and N2, yet refers to one of pila’s additional readings. In the majority of the discarded cases, pila de evokes a bowl or sink, as in la pila de agua bendita ‘stoup’ (lit. ‘the basin of holy water’), la pila de bautismo ‘baptismal font’, la pila de bañar/lavar ‘the sink for bathing or washing’, etc. It also frequently refers to a battery, as in la pila de Volta ‘Volt’s battery’. Interestingly, the battery sense originated in pila’s configuration sense (as a pile or stack), as the original batteries constituted una pila de discos ‘a pile of disks’. Although in the ‘bowl’ sense, pila fulfills a containment function, the quantifying uses rather originated in its configuration sense where it refers to a pile of stackable (usually flat and tiny) objects intentionally and nicely piled up. Or put more carefully, the conceptualization of quantifying uses of pila de are usually reminiscent of its configuration reading. Somewhat similar to (one of) montón de’s original context(s), the first attestations of pila de concern piles of products a particular person (generally a farmer or craftsman) nicely piled up for sale, as in (3). Again, the focus on the size of the pile of own productions or creations, as in (4), may have invited the inference of a huge quantity. (3)

Suplicamos a V. M. que para el remedio desto mande que todas las personas que compraren lana para lo tornar a vender en estos reynos sean obligados a vender en suzio sin apartarla ni lavarla ni mezclarla con otra: sino que hayan de vender y vendan la dicha lana ansi como la compraron, cada pila de lana por si: porque desta manera sabran los que las compran lo que les cuesta, y no se podra vender a tan excessivos precios como al presente se vende. (1559, anon., Cortes de Toledo de 1559 (…)) ‘We beg thou, in order to solve that problem, to order that all the people who will buy wool to prepare it for sale in these kingdoms are obliged to sell it untreated, without dividing it, nor wash it nor mix it up with other wool: that, instead, they have to sell and that they sell the wool exactly as they bought it, every pile of wool separately: because in that way the ones who buy the wool will know what it costs them, and it will not be possible to sell at such excessively high prices as it is sold at present.’

(4)

(…) y que si hiciere o hechare a curtir muchos cueros juntos en que haga pila de ellos que sean hasta en cantidad de veinte cueros, sea

The development of binominal quantifiers | 181

obligado a hechar y heche en la dicha pila de cueros hasta ocho o nueve fanegas de cal viva y no menos, segun dicho es, so pena que (…). (1571, anon., Antiguas ordenanzas para la conservación del Monte Castañar (…)) ‘(…) and that the one who one makes or throws to tan many skins together in order to make a pile of them that counts up to twenty leathers, is obliged to throw and throws in that pile of skins up to eight or nine ‘fanegas’ [ancient measure unit] of quicklime and not less than that, as it is said, under penalty of (…).’ Unambiguous quantifying uses, as in (5) do not appear with pila de until the second half of the 20th century. No more than six quantifying readings are observed in CORDE (cf. Table 3). The combinatorial pattern of pila de has suddenly extended to abstract notions of time (e.g. años ‘years’) and mental experiences (e.g. recuerdos ‘memories’), yet crucially, pila de only combines with plural count nouns (see also Section 5.3). The occurrences found seem to lend support to our hypothesis that the concept of ‘order’ is reminiscent of pila’s original meaning, as some continuity is implied in the time periods or mental experiences, they are profiled in chronological order. (5)

No sea idiota; puede pasarse en la cárcel una pila de años por esto. (1971, Á. Palomino, Torremolinos, Gran Hotel) ‘Don’t be stupid; on might spend many (lit. ‘a pile of’) years in jail for that.’

The further development of the head reading indeed reveals that the conceptual facet ‘order’ belongs to pila’s source semantics. Although the contexts in examples (3)–(4) do not primarily focus on this facet, the order is implicitly there: in contrast to un montón de cueros which would designate a heap of skins simply dropped at random the one above the other, una pila de cueros profiles the stack of leather as an impeccable pile of skins handled with care. Parallel to the change in society, a shift in profile is to be observed in the head use of pila de. From the late 19th century onwards, literal pila de is exclusively associated with contexts where ‘order’ is produced. In (6), the shoemaker adjusted the insoles and it is the pile of insoles which is impeccably heaped up. In contrast, in (7) the facet ‘order’ applies to the surrounding area of the pile: while it is almost impossible to ‘nicely’ heap up fallen leaves into a perfectly rectangular configuration or straight pile, the immediate vicinity to the pile has become neatly swept again. In other contexts, a ‘pile’ of clothes are orderly folded in a suitcase and a ‘pile’ of garbage is the result of cleaning up someone’s room, etc.

182 | The role of conceptual persistence and analogy in the development of BQs

(6)

También ha de colocarse en la parte alta de la pila de alzas, cuidadosamente superpuestas sobre un tablero o mejor aún sobre una tapa exterior invertida, cerrándolos con otra bien ajustada. (1951, J. Cabezas, Cartilla del Colmenero) ‘They also have to be put in the upper part of the pile of insoles, carefully put one above the other on a board or even better on an exterior heelpiece turned inside out, fastening them with another very tight one.’

(7)

(…) y por debajo de ella entramos en el soportal, donde un perrazo pinto que se despertaba sobre una pila de hojarasca, me enseñó todos los dientes y contuvo un ladrido, y acaso algo más, por respeto a mi acompañante, que debía serle más conocido que yo. (1895, J.M. de Pereda, Peñas arriba) ‘(…) and passing under it we entered the porch, where a big colored dog that woke up on a pile of fallen leaves showed me all his teeth and bit back his bar, and maybe something else too, out of consideration for my companion, who must be more familiar to him than I was.’

With regard to the development of the literal use of pila de, it also bears pointing out that the usage contexts where the latter QN usually shows up often coincide with the typical usage contexts observed for literal montón de. More precisely, in Classical Spanish pila de is generally associated with high stacks of products manufactured for sale. In Modern Society, it also combines with right collocates (or N2s) referring to corpses, stones and residues. In Present-Day Spanish, pila de may also refer to a pile of clothes or a woodpile. But what is more important, is its restriction to ‘thin (and possibly tiny)’ or at least ‘easily stackable’ objects (e.g. sillas de tijera ‘folding chairs).100 Tab. 3: Distribution of uses per time period for pila de Head

Quantifier

Specifier

Classical Spanish

# %

6 1

0 0

0 0

Ambiguous Indeterminate 0 0

0 0

Modern Spanish

# %

11 0.92

0 0

0 0

1 0.08

0 0

Present-day Spanish

# %

29 0.67

6 0.14

0 0

5 0.12

3 0.07

|| 100 See also Section 5.3 for a detailed description of (the shifts in) the immediately right combinatorial pattern over time.

The development of binominal quantifiers | 183

Table 3 is intended to reflect the distribution of uses of pila de per time interval. Three issues are to be highlighted. First, the overall tendency towards literal uses strikes the eye. Second, quantifier uses first show up in Present-Day Spanish and are not preceded by a long period of ambiguous contexts. Nevertheless, examples such as (8) provide undeniable evidence for the functional shift towards quantifying uses. In the latter example, the quantifying reading gives way to an even more subjectified reading: it is not necessarily (and even rather improbably) the case that Jaime is keeping a large number of coats in his car, but the fact that he has already so often helped the speaker makes him conceptualize this number of coats as remarkably large. Finally, pila de does not give rise to specifier uses before 1975. (8)

Entonces Ena y yo nos refugiábamos bajo el impermeable de Jaime, quien se mojaba tranquilamente... Muchas tardes me he puesto algún chaleco de lana, o un jersey suyo. Él tenía una pila de estas cosas en el automóvil en previsión de la traidora primavera. (1945, C. Laforet, Nada) ‘At that moment, Ena and I, we took refuge under Jaimes raincoat, who was getting wet calmly… Many afternoons I got into a woolen sweater or jersey of his. He had a pile/lot of those things in his car as a precaution against the treacherous spring.’

The unexpected presence of a determiner to N2 in quantifying uses brings us to the formal reflexes of the functional shift to quantifying uses. In a nutshell, the imposition of morphosyntactic constraints on quantifying pila de seems far less stringent than is the case for montón de. This can be due to two reasons. First, as the historical dataset only contains eight instances of quantifying pila de, it is hard to verify empirically whether the shift in function corresponds to a narrowing down of its co-selection pattern (and thus gives way to a GR process strictly speaking). On the other hand, as the number of quantifying uses is still very limited (with regard to the synchronic distribution of uses as shown in Table 2 of Chapter 3), pila de also lends support to the hypothesis that formal changes always lag behind semantic or functional changes (a.o. Traugott 2003a). Very briefly then, the co-selection as to determiner variation, modification patterns and verbal agreement do not reveal unexpected phenomena. As to the head uses, pila de shows minimal determiner-to-N1 variation and the indefinite determiner is the unmarked option from the start. The adjectives generally refer to the external configuration by stressing the height of the pile or are deictic in nature. Only four combinations are found: dicha ‘previously mentioned’, grande ‘big’, levantada ‘raised’ and enorme ‘enormous’. Only five occurrences are found where pila de N2 functions as the clausal subject and as such triggers

184 | The role of conceptual persistence and analogy in the development of BQs

verb agreement. The latter instances are head uses and, the verb is singular and thus agrees with N1 (i.e. the grammatical head of the construction). As to the combinatorial pattern of the quantifying use, it should be pointed out that too little data are at hand to establish real tendencies. With regard to the determiner to N1, pila de combines in four instances with the indefinite determiner yet also admits the demonstrative esa and the possessive in topicalized uses of N1. The combination with a determiner to N2 in example (8) seems more striking but can in fact be discourse-pragmatically motivated: the addition of the demonstrative to the umbrella noun cosas ‘things’ implies that the speakers refers to a particular category (coats in this case). If the demonstratives would have been left out, una pila de cosas – which rather sounds awkward – would refer to a heap of all kind of things, and thus no longer fit the precise context of (8) nor the frame evoked by the use of pila. In only one instance, quantifying pila de is preceded by the intensifying adjective alto ‘big’. Interestingly, the quantifying uses of pila de do not primarily combine with high conceptual image persistence. In one occurrence only, the persistence of conceptual features of pila de’s source semantics is high, giving way to a metaphorical interpretation. In two cases, the persistence is medial and quantifying pila de only retains the concept of ‘order’ from its source semantics. In the remaining three examples, the persistence is neutral. I will come back to this observation in Section 5.2.7.101 In conclusion, the history of pila de shares with the development of montón de the existence of a productive head use and the far-reaching schematization or abstraction of the source semantics. As will become clear in the following paragraphs, it shares with the other QNs the rather late GR.

5.2.2 Outline of the history of aluvión de Initially, the lexical congener aluvión de can refer to a flood as well as to an alluvium (Corominas/Pascual 1991, vol. 2, 496–497). The latter is a metonymically related to aluvión’s first acceptation, as it refers to a deposit of earth, sand or other transported material left by water flowing. Although both readings entail a quantifying potential, the quantifying uses of the binominal construction obviously stem from the flood sense, probably because the latter allows for

|| 101 As a rule of thumb (and oversimplifying), the distinction between high, medial and neutral conceptual image persistence depends on whether the relation of the grammaticalized QN with its source frame is a metaphorical, a metonymic or simply an implicit one.

The development of binominal quantifiers | 185

a dynamic construal of N2. When looking more closely to the original conceptual facets which are profiled in the quantifying reading, only one instance is found which is rather reminiscent of the alluvium sense. More precisely, in (9), several contextual clues conceptualize the N2 as a deposit of the flowing of time. (9)

El presente no nos embarga ya con aquella exuberancia exterior que se nos iba de las manos, impaciente. El tiempo fluye ahora con morosidad y viene cargado de un rico aluvión de imágenes pasadas. Nuestro ser es así más denso, menos inseguro, mientras la esperanza queda como enterrada. (1959, D. Ridruejo, Dentro del tiempo. (…)) ‘The present does not overwhelm us yet with that external exuberance that was slipping through our hands, impatient. Time flows now slowly and comes loaded with a rich flood/alluvium of images gone by. That’s why our existence is more dense, less insecure, while our hope remains as if buried.’

In contrast to the more ‘conventional’ pathways of change as attested for montón de and pila de, where the grammaticalized uses are preceded by head uses allowing for inferences of quantity, the history of aluvión de calls into question the hypothesis of GR. The crucial pregrammaticalization stage of a productive literal or head use appears to be ‘skipped’ in aluvión de’s development (see also Table 4). While no binominal head use shows up in CORDE, the first quantifying use is already attested in 1847. In addition, the grammaticalized uses outcompeted right from the start possible literal uses with regard to token frequency (cf. Table 4) and are not preceded by a large period of ambiguous contexts. Finally, the distribution of uses mirrors the synchronic layering observed for Present-Day Spanish (cf. Table 2 in Chapter 3). Tab. 4: Distribution of uses per time period for aluvión de102 Head

Quantifier

Specifier

Modern Spanish

# %

0 0

14 0.78

3 0.17

Ambiguous Indeterminate 0 0

1 0.05

Present-day Spanish

# %

0 0

19 0.65

6 0.21

1 0.03

3 0.10

|| 102 The data-extraction is based on a query generalizing over the diachronic orthographical variants, i.e. [aluvión de OR aluvion de OR alubión de OR alubion de]. The observed ambiguous and indeterminate uses concern ambiguity respectively indeterminacy between the quantifier and the specifier use.

186 | The role of conceptual persistence and analogy in the development of BQs

Interestingly, in both grammaticalized readings which seem to abruptly emerge, the group of N2-entities (usually palabras ‘words’) are conceptualized as strange and potentially dangerous entities which suddenly invade the domain of a particular person (which for now will be referred to as the victim). For instance, in one occurrence, un aluvión de andaluces seseantes ‘a flood of Andalusians who pronounce the Spanish ‘c’ as [s]’ is portrayed as a group of unwelcome immigrants who populate Latin America. As in (10), the impact on the victim is usually huge (e.g. se quedó aturdido) and the invasion itself is frequently triggered – ironically enough – by a particular action of the victim himself. In contrast to the pathway of change traced by montón de, the conceptualization specifying aluvión de imposes on N2, as in (11), is similar to the quantifying uses. It is further to be highlighted that aluvión de tends to combine with a high degree of conceptual persistence. In 30 of 42 grammaticalized uses, high persistence of original conceptual facets is observed. In addition, in 11 instances, a facet belonging to the original background frame of aluvión is profiled, giving way to medial conceptual persistence. (10) El honrado Fiel Regidor se quedó aturdido, no sabemos si de la pregunta o del aluvión de palabras que se le venía encima. (1855, F. Navarro Villoslada, Doña Toda de Larrea o la madre de la Excelenta) ‘The honest and reliable mayor got confused, we don’t know whether because of the question or because of the flood of words that came onto him.’ (11) Aturdido ya, la cabeza mareada con este aluvión de noticias, que no eran en su mayor parte más que repetición de anteriores informes, D. Juan echó a un lado la carta sin acabar de leerla. (1898, B. Pérez Galdós, Mendizábal) ‘Already bewildered, his head sickened with that flood of news items, which were generally nothing more than repetition of previous reports, D. Juan threw the letter away without having finished reading it.’ The leap-like emergence of aluvión de and the prompt productivity of the grammaticalized uses contrast sharply with the apparent lack of bonding in the determination pattern of the QN. The fact that the quantifying uses of aluvión de combine as easily with the indefinite as with the definite determiner and allow in two instances for an additional determiner to N2 (as in (12)), could be interpreted as counter-evidence to the GR hypothesis. However, definite and demonstrative determiners are only used when the number of N2-entities is topicalized. Likewise, in (12), the N2-entities referred to are specified immediate-

The development of binominal quantifiers | 187

ly afterwards, but it can be derived from the context that the speaker considers the precise identity of the barbarians as common knowledge. It is further assumed in Chapter 8, that the high persistence of conceptual features may enhance the accessibility of the definite determiners. The constructional tightening resides primarily in the other types of combinatorial patterns. With regard to modification to N1, only four intensifying adjectives are attested in combination with a quantifying reading (i.e. enorme ‘enormous’, grande ‘big’, rico ‘rich’ and verdadero ‘real, authentic’). Notwithstanding the functional shift towards quantifier use, quantifying aluvión de is combined in eight instances with a singular verb. Again, I argue that the singular agreement can be motivated in the light of the tendency towards strong conceptual persistence (as in (13)). (12) El Imperio bizantino, infamado con el título de bajo, combate, resiste, se defiende durante otros seis o siete siglos más contra el furioso aluvión y continua avenida de los bárbaros de Oriente y Occidente; contra los persas, los godos, los hunos, los búlgaros, los rusos y los cruzados, y contra el islamismo pujante, (…). (1868, J. Valera, Sobre el concepto que hoy se forma de España) ‘The Byzantine empire, slandered with the title of low, fights, resists, defends itself during six or seven more centuries against the furious flood and continuous avenue of the barbarians of the East and the West; against the Persians, the Goths, the Huns, the Bulgarians, the Russians, the crusaders and against the booming Islamism, (…).’ (13) Cuando el aluvión de novelas extrangeras, generalmente traducidas de una manera lastimosa, ayudado por alguna otra española, tan deplorable por su lenguaje como por sus tendencias, inunda nuestro desgraciado país, y lo desnaturaliza y corrompe, ora introduciendo hábitos y costumbres que nos desfiguran, ora (…). (1849, F. Caballero, La familia de Alvareda) ‘When the flood of foreign novels, generally translated in an appalling way, with the help of some other Spanish (novel), as deplorable in its language as in its leanings, inundates our unhappy country, and denatures and corrupts it, either by introducing habits and customs that disfigure us, (…).’ In spite of aluvión de’s tendency towards high conceptual persistence and the less obvious bonding of its combinatorial pattern, examples such as (14) ask for revisiting the GR hypothesis of BQs. The quantity expressions innumerables and sinfín evidence the functional shift of aluvión de towards the quantifier paradigm. On the other hand, the immediate context is crowded with lexical units

188 | The role of conceptual persistence and analogy in the development of BQs

that belong to aluvión’s frame. Consequently, the choice for aluvión de enhances the isosemy between the QN and its context, to the extent that it may be portrayed as a coherence device: the substitution of aluvión de by any other QN would be perceived as particularly marked. (14) Innumerables familias perecieron arrastradas por las aguas. Fue una especie de parodia del Diluvio Universal, sin arca de Noé, pero con aluvión de suscripciones, rifas, espectáculos, y sinfín de arbitrios que se idearon en toda Europa y en América, para socorrer a los infelices huertanos supervivientes de aquel espantoso cataclismo. (1912, B. Pérez Galdós, Cánovas) ‘Innumerable families perished, swept along by the waters. It was a kind of parody of the Flood, without Noah’s Ark, but with a flood of subscriptions, raffle, shows and a great many judgments that were devised all over Europe and America, to help those unhappy inhabitants of Valencia that survived that horrific cataclysm.’

5.2.3 Outline of the history of letanía de The lexical item letanía comes from Ancient Greek where it referred to ‘supplication’ (Corominas/Pascual 1991, vol. 3, 636). In the history of Western Christianity, a litany was a worship or public prayer performed in services or processions in order to implore the help of a particular Saint to overcome embarrassing situations such as diseases (as in (15)). Generally, it consisted of a series of supplications or intercessions in which the priest leaded and the people responded (originally by “Kyrie”). The entire sequence of a short supplication by the clergy followed by a particular formula of response by the people was repeated successively several times. It is exactly this idea of (long-threaded) repetition which the quantifying potential of letanía de takes up. (15) Tras los psalmos, digan los quatro evangelios y después comiencen devotamente la letanía de los santos, y hagan su processión por la claostra o por el cuerpo de la yglesia. (…) Mientra que esto hazen y dizen los clérigos, la otra gente deve secretamente rezar sus devociones por las horas o por sus cuentas y rosarios, según que supieren, teniendo intención de rogar a Dios a sus santos que, por su misericordia, embíe socorro y ayuda del cielo contra aquella tempestad. (1538, P. Ciruelo, Reprobación de las supersticiones y hechicerías) ‘After the psalms, you should recite the four gospels and afterwards

The development of binominal quantifiers | 189

devoutly begin the litany of the saints, and perform your procession through the monastery and the body of the Church. (…) While you do that and the clergymen recite, the other people have to delicately pray their devotions relying on the hours and their stories and rosaries, as they knew them, with the intention to ask the Lord and his Saints to have mercy and to send help and rescue from Heaven against that storm.’ In contrast to the ‘standard’ pathway of change as observed for montón de and pila de, the development of letanía de is partially similar the history aluvión de: Table 5 shows that the development of letanía de also skipped the pre-grammaticalization stage of productive head uses, let alone ambiguous uses. While letanía de is already frequently used as a quantifier from 1521 onwards, the first binominal head use dates from 1603. Moreover, the head use seems to temporarily ‘die out’ in Modern Spanish but then reappears in Present-Day Spanish, though marginally. Tab. 5: Distribution of uses per time period for letanía de103 Head

Quantifier

Specifier

Classical Spanish

# %

2 0.17

10 0.83

0 0

Ambiguous Indeterminate 0 0

0 0

Modern Spanish

# %

0 0

5 0.73

0 0

0 0

1 0.17

Present-day Spanish

# %

2 0.18

8 0.73

1 0.09

0 0

0 0

When the noun letanía de was used literally in the binominal construction, the N2-slot was generally filled by the name of the Saint whose help was implored, as in la letanía de la Virgen ‘the litany of the blessed Virgin’. Since the function of the PNP in those instances is to single out which particular litany is to be prayed, the construction is not coextensive and therefore not included in our frequency tables. The very few (coextensive) head-uses found concern those constructions where the response formula by the people is mentioned in the PNP, as in (16). The far-fetched character of indicating a particular litany by the recited formula explains its marked nature or unproductivity.

|| 103 The data-extraction is based on a query generalizing over the diachronic orthographical variants, i.e. [letanía de OR ledanía de OR letania de OR ledania de OR litanía de OR litania de]. The observed indeterminate uses blend the two functional uses.

190 | The role of conceptual persistence and analogy in the development of BQs

(16) Y acerca desto se hacen graciosos discursos; porque si alguno andaba hoy próspero, mañana le ven sin zapatos y allá a lo bárbaro, como saben, les cantan el psalmo de Conserva me Domine, juntamente con la letanía de conservare diNeris [sic]; y fuera de sus latines, otros proverbios graciosos (…). (1603, F. de Luque Fajardo, Fiel desengaño contra la ociosidad y los juegos) ‘And about that amusing discourses are made; because if anyone is better-off today, tomorrow they see him without shoes and there barbarously, as they can, they sing for them the psalm of Conserva me Domine, in addition to the litany of conservare dineris; and next to their latin expressions, other funny proverbs (…).’ Interestingly, the quantifying uses are usually highly metaphorical and picture the N2 as coming in a long and dreadful sequence. Crucially, the entities quantified are always of discourse nature (generally insults or enumerations). In other words, the N2s are generally conceived of as pronounced out loud, either by their very nature (e.g. insultas ‘insults’, dicterios ‘insults’, blasfemias ‘blasphemies’, etc., as in (17)) or by contextual requirements (as in (18), where the context pictures a tiresome host reciting a long list of types of wines). Not unsurprisingly then, the immediate linguistic context of letanía de generally contains predicates such as decir ‘to say’, relatar ‘to relate’, insultar ‘to insult’, oír ‘to hear’. Equally frequent are combinations with aspectual predicates indicating the start or the continuation of an action/process, such as empezar ‘to start’, seguir con ‘to continue with’ and iniciar ‘to start’. The latter predicates mirror the concept of a sequence or cycle of repetitions that is usually evoked by letanía. In addition to verbal predicates referring to the source semantics of letanía de, the QN shows a strong preference for high conceptual persistence from the start. Example (19) nicely illustrates the original conceptual facets that generally persist in quantifying uses: (a) the N2-entities are portrayed as a long and boring list of nearly identical small parts of discourse; (b) the sequence is usually triggered by (an action of) one particular person (e.g. the old man who accidently spilled something on the speaker’s suit); (c) there is an audience who is expected to agree or support. (17) Ya supondrá el lector la letanía de preguntas con que abrumaría mi buen Tirabeque al amable jóven nuestro acompañante: le importunaba, le molía, le ostigaba; él sin embargo contestaba á todo con una paciencia y una dulzura admirables: (…). (1842, M. Lafuente, Viajes de fray Gerundio por (…)) ‘The reader will already imagine the litany of questions with which my good Tirabeque would bore the kind young man who accompanied

The development of binominal quantifiers | 191

us: he kept on disturbing, overloading and bothering him; the latter, however, responded to everything with an admirable patience and kindness: (…). (18) (…) me preguntó que con qué me serviría, relatándome la larga letanía de vinos que guardaba en su bien abastecida bodega. (1847, S. Estébanez Calderón, Escenas andaluzas, bizarrías de la tierra, (…)) ‘(…) he asked me with what he would serve me, relating for me the long litany of wines he kept in his well-supplied cellar.’ (19) Desfiló toda la parentela del viejo. Ya estaba lanzado y no pudo parar la lengua, se olvidó incluso de Teresa, y sólo cuando acabó la larga letanía de insultos (mientras el pobre hombre se retiraba refunfuñando, (…) ) y miró a Teresa, descubrió su expresión de reproche. – Qué – dijo él, frotándose la solapa con el pañuelo –. ¿No tengo razón? (1966, J. Marsé, Últimas tardes con Teresa) ‘He passed through the entire clan of the old man. He was already launched into it and could not keep his mouth shut anymore, he even forgot about Teresa, and only after he ended the long litany of insults (while the poor man withdraw grumbling, (…) ) and looked at Teresa, he observed her reproachful look. – ‘What?’, he said, rubbing his lapel with a handkerchief. ‘Am I not right about that?’ Notwithstanding the high degree of conceptual persistence, viz. the highly metaphorical load of the quantifying uses, letanía de has shifted to the category of quantifiers. Examples such as (20) illustrate not only that letanía de has become a full member of the (binominal) quantifier paradigm but also that each BQ imposes its own conceptualization on N2. (20) (…) ; y don Alvaro, fuera de sí, con espumarajos en la boca, grita como un energúmeno: – Ta, ta, ta, ta, ta. Y cada ta, por el tono con que don Alvaro le suelta, parece un centón de blasfemias y una letanía de maldiciones. Doña Inés suele acudir entonces y dice: – ¿Por qué chillas tanto, diantre de hombre? (1895, J. Valera, Juanita la Larga) ‘(…); and Sir Alvaro, besides himself, with foam at the mouth, shouts like a lunatic: – Ta, ta, ta, ta, ta. And every ta, because of the tone which Sir Alvaro utters it with, seems a anthology of blasphemies and a litany of curses. Mrs Inés uses to arrive then and says: – Why do you yell like that, you devil of a man?’ At first sight, the high degree of conceptual persistence observed for quantifying letanía de appears to go hand in hand with a certain degree of formal persis-

192 | The role of conceptual persistence and analogy in the development of BQs

tence as well. While in Classical Spanish the indefinite determiner una was the preferred combination, the determiner variation of the Present-Day quantifying uses strikes the eye. Letanía de appears to easily combine with the definite determiner la, the possessive su or the demonstrative aquella. Though marginally, letanía de can also combine with adjectives. Again, the adjectival pattern modifying letanía de is not unrestricted: the N1 only combines with intensifiers such as toda ‘whole’, verdadera ‘real, authentic’, larga ‘long’ or adjectives evoking the original Christian prayer frame such as lento ‘slow’ and amontonada ‘accumulated’. The restriction of the N2-collocates towards parts of discourse or speech (see also Section 5.3.2) adds up to the evidence for considering the high degree of conceptual persistence as motivating the highly constrained formal variations (and not as mere epiphenomenal of GR).

5.2.4 Outline of the history of (h)atajo de The etymological uncertainty about hatajo somehow heralds the even less straightforward GR process hatajo is involved in. Corominas/Pascual (1991, vol. 3, 326–328) cautiously propose two different origins: while hato ‘cloths; portion of (cattle)’104 has often been mentioned in literature, Corominas/Pascual advance the predicate atajar ‘to cut out, to intercept’ as hatajo’s origin. They further raise the issue of the legitimacy of the initial ‘h’, given that the noun was originally written atajo. However, the first instance attested in our corpus starts with ‘h’ (cf. (21)), which in its further development does not appear to be particularly marked either. As neither of the orthographic variants shows a particular preference for particular contexts or readings, they are discussed together.105 (21) Tenía una piña en la mano * que es fruta muy excelente y comenzaba entonces a darse en esta isla; y, * apregonándose un hatajo de yeguas y otras cosas de muncho [sic] precio, poníanselas en quinientos o mill pesos, decía el comendador mayor: "[A] quien lo pusiere en mill y quinientos le daré esta piña." (1527, Fray B. de las Casas, Historia de las Indias) ‘He was holding a pineapple in his hand * which was an excellent fruit

|| 104 Hato originally referred to ‘cloths’ but later, probably due to interference with Arabic hazz ‘deserved portion, one’s portion by right’, came to refer to the ‘portion of provisions shepherds take along’ and to ‘a portion of cattle’ (Corominas/Pascual 1991, vol. 3, 326–328). 105 The orthography without ‘h’ is overall more frequent: while [atajo de] yields 69 occurrences in CORDE, CORDE returns only 16 instances for the query [hatajo de].

The development of binominal quantifiers | 193

and started to grow in that island; and, * while (they were) advertising the herd of mares and other highly priced things and marking them at five hundred or thousand pesos, the chief commander said: “To the one who puts thousand five hundred, I will give this pineapple”.’ The variant atajo was common in the not coextensive binominal uses, where it referred to a fence (as in (22)) or short cut (as in (23)). In the latter uses, the PNP indicates the material the fence is made of or, with respect to the ‘short cut’ reading, the original path or its destination. (22) Por haser vn atajo de madera en la camara de la ynfante, en la villa de Valladolid, 475 mrs. (1477, anon., Cuentas de Gonzalo de Baeza (…)) ‘To make a fence of wood in the quarter of the Infante, in the city of Valladolid, 475 maravedis [old currency unit].’ (23) (…) aunque muy alegres por yr tan cerca de Babilonia, que no tres jornadas les faltauan al traués que ellos venían vieron caminar mucha gente que a atajo de vn camino para Babilonia proseguían, que llegando más cerca lo vieron (…). (1547, J. Fernández, Belianís de Grecia) ‘And although (they were) very happy to walk so near to Babylon, that they had only less than three days’ march to go they saw a lot of people that followed a shortcut of a road to Babylon, and when they came closer they saw (…).’ The more common interpretation of literal (and coextensive) (h)atajo is where (h)atajo de refers to a group of cattle. In some instances, the PNP merely serves to indicate which kind of animals compose the herd (as in (24)). Yet, interestingly, the literal uses generally show up in contexts of trade or dowry, which further highlight hatajo’s quantifying potential (as in (21) and (25)). Consequently, the particular history of (h)atajo de nicely illustrates the hypothesis underlying this monograph: although many nouns have a quantifying potential in addition to their lexical meaning (simply for referring to a kind of container or collective), the contexts in which the QN is used originally co-determine whether the quantifying reading will be exploited, and if so, to what degree. (24) Toda esta plática o conversación pasó estando este hidalgo y yo echados de pechos sobre el guardalado de la puente Segoviana, mirando hacia la Casa del Campo, por donde vimos asomar un buen atajo de vacas, que nos interrumpió la conversación, y viéndolas le dije: (…)”. (1618, V. Espinel, Vida del escudero Marcos de Obregón) ‘This entire talk or conversation happened while this nobleman and I

194 | The role of conceptual persistence and analogy in the development of BQs

were lying with our chest on the railing of the bridge Segoviana, looking at the Casa del Campo, where we saw a good herd of cows suddenly appearing, who interrupted our conversation, and when I saw them I said: (…).’ (25) Al oír esta respuesta, el viejo reniega de lo revelado por Mahoma, se bautiza, y aun se aviene a guardar un atajo de puercos de la muchacha, para darla este servicio en lugar de dote. (1902, R. Menéndez Pidal, Estudios literarios) ‘When hearing this answer, the old man renounced what has been revealed by Mohammed, got himself baptized, and even agreed to look after the girl’s herd of pigs, in order to give her that service instead of dowry.’ Although the first quantifying uses of (h)atajo seem to be triggered by contextual clues, such as the previous binominal quantity expression ejército de ‘army of’, as in (26), and the superlative form of the adjective grande, as in (27), the combination with abstract nouns such as vicios ‘sins’ in Modern Spanish demonstrate the functional shift towards quantity, as vicios is incompatible with the original ‘herd’ (or ‘short cut’) sense. (26) En esto les alteró un ruido de voces, y era que venía un ejército de tablilleros y un atajo de limeras, * que, en viendo hombre con coche de mujeres, vienen como llamados a juicio. (1646, F. Remiro de Navarra, Los peligros de Madrid) ‘During that, a noise of voices disturbed them, and the thing was that an army of carpenters and a bunch of lime sellers, * who, when they saw a man with a carriage of women came as if called to judge.’ (27) (…), preocupado con la misma melancolía (por haber vivido muchos años entre un grandísimo atajo de bribones y bribonas que hacen trato de la virtud, unos para comer, otros para gobernar, y otros para suponer) saqué mi navaja y corté la pluma. (1729, F. Afán de Ribera, Virtud al uso y mística a la moda) ‘(…), worried with the same sadness (for having lived a lot of years among that very big bunch of rascal men and women who abuse virtue, some to eat, others to govern, and others to count) I got out my razor and sharpened the pen.’ The first specifying use of hatajo de shows up in 1758. Although the quantifying reading emerged first (cf. Table 6), the specifying use immediately overrode the

The development of binominal quantifiers | 195

latter in frequency of occurrence. This skewed proportion is maintained in Present-Day Spanish. Tab. 6: Distribution of uses per time period for (h)atajo de106 Head

Quantifier

Specifier

Classical Spanish

# %

7 0.78

2 0.22

0 0

Ambiguous Indeterminate 0 0

0 0

Modern Spanish

# %

1 0.06

2 0.11

15 0.83

0 0

0 0

Present-day Spanish

# %

6 0.33

1 0.06

10 0.56

0 0

1 0.06

Regarding the conceptualization hatajo de imposes on N2, its quantifying and specifying uses share the idea of a homogeneous group of N2-entities, generally people which the speaker disapproves of for being stupid (or evil), as in (28). The homogeneity of this group of people thus resides in their foolishness, in their inability to think – let alone act – individually and is reminiscent of hatajo’s source semantics (cf. the typical herd instinct), at least in its herd sense. (28) Mucha ciencia alemana, que el demonio que la entienda; mucha teoría oscura y palabrejas ridículas; mucho aire de despreciarnos a todos los españoles como a un hatajo de ignorantes; mucho orgullo, y luego (…). (1878, B. Pérez Galdós, La familia de León Roch) ‘A lot of German science, the devil might understand it; a lot of dark theory and ridiculous words; a lot of air of looking down on us, all the Spaniards, as on a bunch of fools; a lot of proud, and then (…).’ However, examples such as (29), leave room for categorial interference with atajo, derived from atajar. In several specifying instances of (h)atajo de, the previous context todo lo demás es ‘all the rest comes down to/is’ makes clear that the binominal construction primarily serves to single out an exceptional character (e.g. Puigblanch in the example below) and thus implies that someone has been ‘cut out’.107 The interference with atajar and atajo obviously cannot || 106 The data-extraction is based on a query generalizing over the diachronic orthographical variants, i.e. [hatajo de OR atajo de]. The observed indeterminacy always holds between the quantifier and the specifier use. 107 Whether or not by coincidence, the ambiguity is also found in other lexical items of hatajo’s original frame: the polysemic predicate guardar is associated with the image of ‘guarding or defending someone/a herd’ as well as with the idea of ‘keeping’.

196 | The role of conceptual persistence and analogy in the development of BQs

explain the emergence of the specifier use, but it might be a crucial factor in the dramatic rise in frequency of the specifier reading of (h)atajo de from Modern Spanish onwards (see Table 6). (29) En cuestiones filológicas suele acertar Puigblanch, y aun ahondar bastante y adivinar cosas que pocos alcanzaban en su tiempo: (…). Pero todo lo demás es un atajo de desvergüenzas estrafalarias y de especies desparejadas, sin ilación ni método, tal que parece escrito en un manicomio o al salir de una taberna, y eso que el autor era por extremo sobrio; (…). (1880, M. Menéndez Pelayo, Historia de los heterodoxos españoles) ‘In philological questions Puigblanch is usually right, and he even goes quite into detail and also foretells things that only few were able to understand in his time: (…). But all the rest/everything else is a bundle of eccentric impertinences and of odd messages, without cohesion nor method, as if it were written in a madhouse or when leaving a bar, and that while the author was extremely sober; (…).’ The typical contexts hatajo de appears in throw up two further issues. First, although all QNs engage in subjectification processes – they arise out of pragmatic inferences of scalarity (see also Traugott 2003b)–, the immediate contexts of specifying hatajo de particulary demonstrate stance-taking by the speaker. Not only are several instances accompanied by predicates expressing someone’s opinion or evaluation (like sostener que ‘to support, to maintain’ or te aseguro que ‘I assure you that’, as in (30)–(31)), the surrounding context is also overloaded with negative evaluations (such as despreciarnos ‘to look down on us’ in (28), sin ilación ni método, tal que… ‘without cohesion nor method, to the extent that…’ in (29), no vale la pena preocuparse de ellos ‘it’s no use to be concerned with them’ in (31)). The negative emotional load of the context brings us to the second issue: the grammaticalized uses of hatajo yield a negative semantic affect, either by the unpleasant character of N2 (as in (27)–(29) and (31)), either by the context which makes clear that the number of N2 implies inconvenient situations (e.g. the noise in (26)). As this tendency can also be observed for the remaining QNs, I will come back on this issue in Section 5.2.7. For now it suffices to note that the majority of the grammaticalized uses (viz. 25 of 31 occ. or 81%) correspond to negatively loaded contexts. I tentatively argue that this preference for negative semantic prosody can be motivated in the light of the negative connotations of either possible source frame: while a herd entails a group of animals guided by herd instinct only, a portion cut out is generally smaller and possibly of lesser quality.

The development of binominal quantifiers | 197

(30) Poleró, sin conocer el drama, sostenía que era un atajo de inocentadas, y que el mayor favor que se podía hacer al joven manchego era quitarle de la cabeza su idea de ser autor dramático. Cienfuegos no pensaba lo mismo, y veía en Alejandro, (…). (1883, B. Pérez Galdós, El doctor Centeno) ‘Poleró, without knowing the drama, maintained that it was a herd of women who had been acquitted , and that the biggest favour one could do to the young guy from La Mancha was to making him abandon his idea of being a drama author. Cienfuegos didn’t think the same, and he saw in Alejdandro, (…). (31) i es que esta gente que no ha salido de España nunca, te aseguro que son un atajo de cenizos, no vale la pena preocuparse de ellos, te lo digo yo. (1972, A. Zamora Vicente, A traque barraque) ‘And the thing is that those people that have never left Spain, I assure you that they are a bunch (lit. herd) of softies, it is no use to be concerned about them, I can tell you that.’ As to the development of the morphosyntactic make-up of hatajo de, the coselection pattern of hatajo de has, right from the start, been more restricted than the patterns observed for the remaining QNs. They also show similar proportions per pattern for each period, as if the distribution contexts of hatajo de were fixed from the earliest instances on. First, grammaticalized hatajo de shows a clear preference for the indefinite article (in 25 of 31 occ. or 81%), whereas the other QNs more easily combined with definite or demonstrative determiners (in topicalized instances at least). The finding is even more surprising since determiner variation is generally more common in specifier uses, which is the most productive grammaticalized use for hatajo de. Further, in only one quantifying instance, (h)atajo de is preceded by an adjective, viz. grandísimo. Recall that in (27), the intensifier grandísimo can be attributed a triggering role. It is also to be noticed that the N2-combinations are restricted to nouns referring to parts of speech or to people (see also Section 5.3.2). Finally, while the QNs letanía de and aluvión de almost exclusively combined with highly metaphorical interpretations (or high conceptual persistence), hatajo de also combines with conceptualizations which contain facets only metonymically related to hatajo’s source semantics. These findings suggest that in the case of hatajo de, the form-part of the construction conventionalized earlier than the formal restrictions governing the remaining QNs analyzed in this chapter.

198 | The role of conceptual persistence and analogy in the development of BQs

5.2.5 Outline of the history of barbaridad de Equally puzzling is the development of barbaridad de. Its overall low frequency and its incompatibility with head uses question of course the GR involved. Etymologically, barbaridad is derived from the adjective bárbaro, which in Latin and Ancient Greek meant ‘foreign’. Pragmatically speaking, it is only a small leap from ‘foreign’ or ‘barbarian’ to ‘cruel’ and ‘uncultured’. When barbaridad is used literally in the binominal construction, the PNP indicates the person who committed the barbarity (as in (32)) or the nature of the cruelty (as in (33)). (32) (…), le disparó una pistola a los pechos. Sus criados, y otros que le seguían, imitando la barbaridad de su dueño, (…). (1645, F.M. de Melo, Historia de los movimientos, separación y guerra de Cataluña) ‘(…),he shot her with a gun in the heart. His children, and others who were following him, copied the barbarity of their master, (…).’ (33) (…), ya el secretario se olvida de su papel y por leer lo que no está escrito comete la barbaridad de leer lo que está escrito; (…). (1855, E. Castelar, Ernesto: novela original de costumbres) ‘(…) and the secretary already forgets about his role and for (aiming at) reading what is not written he commits the barbarity of reading what has been written; (…).’ The first coextensive use only shows up by the end of the 19th century, where una barbaridad de specifies susto ‘fright, fear’. Although the interval between the first specifying use attested (1891) and the first quantifying use attested (1911) is almost negligible, the chain of inferences that leads from ‘a barbarity’ to ‘a lot’ necessarily passes via the specifying interpretation. In other words, in the case of barbaridad de, the quantifying interpretation seems to build somehow on the specifying use. More precisely, in contrast to the other QNs analyzed, the lexical congener barbaridad de typically does not refer to a container, nor to a collective nor to a unit of measurement. Its quantifying potential resides in the qualitative scalarity barbaridad refers to: a barbarity is in itself an extreme instance of cruelty, it qualifies for ‘extremely cruel’ on a scale of barbarity (see also Paradis 2011; Verhagen 2009; Trousdale 2008b).108

|| 108 Interestingly, the original adjective bárbaro also gave way to quantifying instances in Present-Day Spanish (DRAE) meaning ‘excessive, absolute, big’. See also Paradis (2011) on the development of the degree modifier uses of awful, dreadful, horrible and terrible through

The development of binominal quantifiers | 199

The double reformulation of su susto ‘her fright’ in (34) shows that una barbaridad de susto is even more intense than un susto atroz ‘an awful, terrible fright’ and means ‘a highly awful fright’ (which is of course absolute fright). The example thus illustrates the quantifying potential of the specifying use. On a theoretical level, the barbaridad case suggests not only that the productive quantifying use is no necessary condition for the development of specifying uses, but also that the direction of the schematic chain of inferences may be inverted.109 (34) La duquesa le hizo callar de un abanicazo, y la de López Moreno, llena de satisfacción al verse objeto del interés de todos, continuó el relato de su susto, un susto atroz, una barbaridad de susto... (1891, L. Coloma, Pequeñeces) ‘The duchess silenced him with one wave of the fan, and the duchess of López Moreno, who was fully satisfied when realizing that she was the centre of everybody’s interest, continued the story of her fright, an awful fright, a barbarity of a fright…’ (35) Y nos reíamos mucho, especialmente con un abuelísimo mío que había estado conquistando América y nos decía las mil perrerías que hacía con los indios. Por cierto, que como yo dudase de alguna historia, (…), certificó lo que mi abuelísimo decía y se reía contando más detalles sobre la barbaridad de hombre que fue... (1965, F. García Pavón, Los liberales) ‘And we laughed a lot, especially at a very great-grandfather of mine who had been conquering America and who told us of the thousands of terrible things he did with the Indians. Besides, if I was doubting an anecdote, (…), she [KV: the indian woman] ascertained what my old man used to tell and she laughed while telling us more details of the barbarity of a man that he was….’ However, the overall low frequency of barbaridad de (cf. Table 7) invalidates the previous claims on a statistical level. In Present-Day Spanish, no more than two specifying instances are observed. In both cases, the binominal construction

|| metonymization “whereby the superlative content component of ‘terrible’ is generalized into ‘high degree’” (2011, 79). 109 This observation will be elaborated on in Chapter 6 and interpreted as an additional argument for the hypothesis of a continuum between quality and quantity assessment. For now, it suffices to say that the shift from ‘many N2’ to ‘nothing but N2’ seems to work in both directions.

200 | The role of conceptual persistence and analogy in the development of BQs

barbaridad de serves to categorize people, respectively hombre ‘man’ (in (35)) and moza ‘young girl’. In both instances, contextual cues can be found which picture the N2 as particularly unpleasant, impolite and rude entities (such as mil perrerías ‘a thousand terrible things’ and the entire previous description of the man). Further, the context clearly mentions who perceives or categorizes the N2 as una barbaridad (for instance, both the speaker and the interlocutor evoked in (35)). Tab. 7: Distribution of uses per time period for barbaridad de Head

Quantifier

Specifier

Modern Spanish

# %

0 0

0 0

1 1

Ambiguous Indeterminate 0 0

0 0

Present-day Spanish

# %

0 0

3 0.6

2 0.4

0 0

0 0

With regard to the quantifying use of barbaridad de, only three Present-Day instances are attested. In those cases, barbaridad de quantifies different types of N2s, i.e. tiempo ‘time’, sabios ‘wise men’ and quina ‘fortified wine’ (cf. (36)–(37). In the quina-example, N2 forms part of a fixed expression tragar quina (literally ‘to swallow fortified wine’) which means ‘to have to take it, to put up with it’: está tragando mucha quina, una barbaridad de quina ‘he just has to put up with a lot, with an awful lot’. The functional shift towards the category of quantifiers is made clear by the paraphrasis for mucha quina in the previous example or the juxtaposition with la mar de listos ‘a lot (lit. the sea) of clever people’ in (37): similar examples indicate that barbaridad de is a rightful member of the (binominal) quantifier paradigm. (36) Contaba Lulú que de niña la pegaban para que no comiera el yeso de las paredes y los periódicos. En aquella época había tenido jaquecas, ataques de nervios; pero ya hacía mucho tiempo que no padecía ningún trastorno. Eso sí, era un poco desigual; tan pronto se sentía capaz de estar derecha una barbaridad de tiempo, como se encontraba tan cansada que el menor esfuerzo la rendía. (1911, P. Baroja, El árbol de la ciencia) ‘Lulú told that as a child, they beat her so she wouldn’t eat the plaster of the walls and the newspapers. At that time, she had been subject to severe headaches, nervous attacks; but it was already long ago since she did not suffer any disorder. However, she was a little irregular; she could all of a sudden feel both capable of being up for an awful lot of time as so tired that the smallest effort exhausted her.’

The development of binominal quantifiers | 201

(37) En medicina, en leyes, en mecánica, en qué sé yo cuántas cosas más, la mar de listos, una barbaridad de sabios y sobresalientes; pero en religión, tontos de remate, necios de cal y canto..., suspensos. Por eso, sin duda, ha dado Jesucristo a su Iglesia en estos últimos tiempos el Papa que nos convenía, Pío X. (1945, R. Sarabia, ¿Cómo se educan los hijos? Lecciones de pedagogía familiar) ‘In medical science, in law, in mechanics, in I don’t know how many more things, a lot of (lit. the sea of) clever people, an awful lot of wise men and (people with) excellent scores; but in religion, complete idiots, absolute fools…, people who fail their exams. That’s, without doubt, why Christ lately has given to his Church the Pope we needed, Pius X.’ The quantifying instances of barbaridad de share with the specifying use the explicit markers of stance taking such as Contaba Lulú que ‘Lulú was telling that’ in (36) and sin duda ‘no doubt’ in (37): both indicate that the sequence una barbaridad de N2 corresponds to Lulú’s, respectively the speaker’s, point of view. In other words, the grammaticalized instances of barbaridad de seem to be restricted to contexts where someone’s belief state towards the proposition is explicated, viz. to subjectification in Traugott’s (initial) sense (1989, 35). Whereas the specifying uses obviously occur in particularly unpleasant or negatively loaded contexts, the quantifying uses do not display negative semantic prosody to the same degree. In terms of markers of GR, the low frequency of barbaridad de contrasts sharply with the syntactic context expansion towards adverbial uses in 1905 already (as in (38), while the other QNs remain restricted to the binominal construction). Further, una barbaridad de behaves like a single and fixed chunk with regard to determiner variation (barbaridad de is always preceded by the indefinite una), to modification (no adjective is found in combination with barbaridad de) and to agreement (no explicit verbal nor phoric agreement with N1 or N2 has been observed). Whereas the specifier uses have strong conceptual preference, the quantifying uses are not interpreted metaphorically. (38) Lo cierto es que este mozo sabe muchísimo, sabe una barbaridad, de mi profesión; sabe, teóricamente, lo que los pintores siempre ignoramos. (1905, E. Pardo Bazán, La Quimera) ‘It is sure that this young boy knows a lot, he knows an awful lot (lit. barbarity), of my profession, he knows, theoretically, what we painters always ignore.’

202 | The role of conceptual persistence and analogy in the development of BQs

5.2.6 Outline of the history of mar de The challenge to the GR model caused by the unique histories of the QNs analyzed so far is nothing compared to the enigmatic development of mar de, whose most salient peculiarities are the strikingly high frequency of its coextensive binominal realization, the metaphorical load of the ‘grammaticalized’ uses and the ability to combine with adjectival collocates, such as la mar de simpático (‘very, really sympathetic’).110 In view of the prototype role attributed to montón de, the overall frequency of the lexical item strikes the eye. Generalizing over the types of uses and the historical periods, mar is used far more frequently than montón according to CORDE, displaying 334.72 and 11.72 in frequency per million words respectively. When zooming in on the coextensive binominal uses, the appearance of montón in the binominal construction results to gradually become the preferred combination in Modern Spanish, while mar continues to have a productive lexical use outside the binominal construction.111 Basically, mar refers to (a particular part of) the volume of water that covers part of the earth’s surface. Typically, the PNP identifies which particular subpart of the continuous sea body is meant. In addition to PNPs indicating the land-limits, as in (39), the identification may rely – though marginally – on a noun characterizing the sea, as in (40), where muy salada reflects the quantifying potential of mar de. The latter type of identifying phrases can indeed be considered a bridging context: a particular noun (e.g. sirenas ‘mermaids’, peligros ‘dangers’, ondas revueltas ‘rough waves’, etc.) can only characterize a sea if it is associated with the sea in a sufficiently frequent manner.112

|| 110 The gender difference between un/el mar de N2 and la mar de N2/ADJ/ADV will be discussed later in this section. For now it suffices to say that in the coextensive uses the feminine article is rare and highly marked until the second half of the 19th century. 111 The next table lists the frequencies per million words. The productive use of mar as a lexical item might be the reason why quantifying un mar de hardly desemanticizes (cf. infra). Montón Mar Coext. cxn. Other Total Coext. cxn Other Total 1250–1450 2.72 3.92 6.64 0.18 410.85 411.03 1450–1730 2.74 4.53 7.27 3.25 421.21 424.47 1730–1900 8.78 7.90 16.68 7.39 184.87 192.26 1900–1975 14.26 8.05 22.31 8.1 264.24 272.34 112 Corominas/Pascual (1991, vol. 3, 837–389) mention the Latin origin of mar. With regard to gender preferences, the etymological dictionary mentions that in Medieval Spanish, both genders were equally frequent while in Classical Spanish the masculine variant became the

The development of binominal quantifiers | 203

(39) E passando por la oriella del mar de Galilea, uio a Symon e a Andres so ermano, que metien sus redes en el mar, ca eran pescadores. (1260, anon., El Nuevo Testamento según el manuscrito escurialense I-j-6.) ‘Passing along by the sea of Galilee, he saw Simon and Andrew the brother of Simon casting a net in the sea, for they were fishermen.’ (40) (…), e unos le dizen mar de sal por el agua que es muy salada, e los otros le llaman mar muerto, porque nin se fazen árvoles en la ribera, nin se crían en él peces nin aves nin otra cosa biva ninguna como en las otras aguas. (1275, Alfonso X, General Estoria I) ‘(…), and some call it the sea of salt because of the water that is very salted, and other call it the dead sea, as there grow no trees at its shore, as there live no fish in it, no birds no single living thing as in other waters.’ The identifying source-construction is also particularly manifest in the highly metaphorical uses wavering between head and quantifying uses, as in (41)–(42). As the N2 is incompatible with mar’s head reading and the binominal construction is definitely not coextensive, these examples do not qualify for ambiguous nor indeterminate uses. On the other hand, they invite the inference of ‘a lot of love’ and ‘a lot of grief’, to the extent that the subject is conceptualized as ‘drowning’ in those notions. The metaphoric use indeed almost exclusively combines with predicates such as: hundirse / ahogarse / anegarse en el mar de N2 ‘to sink / to drown / to bath (to be overwhelmed by) in a sea of N2’, navegar por el mar de N2 ‘to sail the sea of N2’, no es navegable la mar de N2 ‘the sea of N2 is impossible to navigate’, etc. Interestingly, this use precedes the coextensive quantifying uses of mar de. Not surprisingly then, mar de combines with grounded N2s as well as with N2s without separate determiner and continues to be productive.113 Whereas in Classical Spanish, the combinations with amor (27 occ., usually in me anego en el mar de amor ‘I am lost in the sea of love’) and llanto (9 occ.) catch the eye, Modern and Present-Day Spanish subjects rather sink in a sea of confusiones ‘confusions’ (9 and 6 occ. respectively, while el mar de amor only occurs 1 and 4 times). Generalizing over the possible shifts in pref|| unmarked option under the influence of Latin. However, in CORDE, the masculine article was the preferred option until Modern Spanish. In Present-Day Spanish, la somehow has become a marker of quantification. 113   With regard to the exact frequencies, 2 instances of the metaphorical, quantifying – yet not coextensive – use are found in Medieval Spanish, 140 in Classical Spanish, 100 in Modern Spanish and 62 in Present-Day Spanish. (The instances with additional determiner to N2 have not been taken into account.) An additional reason not to consider this type of metaphorical use as real quantifying instances is that they do not allow substitution by other QNs.

204 | The role of conceptual persistence and analogy in the development of BQs

erences over time, the not coextensive quantifying uses of mar de combine with N2s referring to a state, situation, virtues/sins or thoughts. In other words, the N2s are nouns inviting for intensification and are usually conceptualized as intangible or infinite qualities. (41) Por el profundo mar de amor nauego sobre la naue de mi pensamiento, donde ay tan altas ondas que me anego porque con mis suspiros crece el viento. (1583, P. Padilla, Romancero) ‘Through the deep see of love I sail / on the ship of my thought, / where there are waves so high that I sink / as with my sighs the wind grows.’ (42) Tengo yo el Ave María clavada en el corazón, y es la estrella que me guía en este mar de aflicción al puerto del alegría. (1615, M. de Cervantes Saavedra, Comedia famosa de los baños de Argel) ‘I have the Ave Maria / fixed in my heart, / and it is the star that guides me / in this sea of sorrow / to the port of happiness.’ The (coextensive) quantifying and specifying uses of mar de N2 continue the high metaphorical load and generally function as a metaphor for life on earth or for someone’s character, as in (43) and (44).114 The metaphor is even more obvious in examples such as (45) and (46) where mar de N2 comes in a series of BQs, which enhances the singular conceptualization each QN imposes on N2. (43) ¿Para qué es desear larga vida, pues nos dilata nuestro destierro y nos detiene en este mar de trabajos, sin poder entrar en el puerto del eterno descanso?, lo cual nosotros no podemos hacer sino por medio de la muerte, que es el desembarcadero en que desembarcamos de esta vida para la otra. (1571, anon., Traducción de la Imagen de la vida cristiana de Fray Héctor Pinto) ‘Why should one long for a long life, since it postpones our exile and detains us in this sea of work, without being able to enter the port of the eternal rest, which we cannot do without dying, which is the landing place in which we disembark from this life for the other one.’

|| 114 In addition to occurrences such as el tempestuoso mar de esta vida (1590) ‘the stormy sea of this life’, tempestades en el mar de su corazón (1562) ‘storms in the sea of his heart’, etc.

The development of binominal quantifiers | 205

(44) No entienden estos quién es Dios, ni quién son ellos, ni qué es lo que le deben, ni tampoco el mar de males i vizios que tienen en su pecho. (1560, J. Pérez, Breve tratado de doctrina) ‘They don’t understand who God is, who they are, what it is they owe him, nor the sea of ills and vices they have in their chest.’ (45) Si es verdadera o si es fingida la tristeza que en mí reyna, no es necessario que lo diga mi lengua, pues claramente lo conoscerás en mi cara; porque los ojos están hechos arroyos de lágrimas y el coraçón hecho un mar de pensamientos. (1529, Fray A. de Guevara, Reloj de príncipes) ‘Whether the sadness that reigns in me is true or fake, it is not necessary that my tongue tells it, as you will clearly recognize it in my face; as my eyes have become streams of tears, and my heart a sea of thoughts.’ (46) ¿Qué hombre se hallará tan abastado de todos estos bienes, y tan exempto de todos estos males, siendo esta vida un mar de continuos desasosiegos y alteraciones, un valle de lágrimas, una cárcel de condenados, donde son muchas más las miserias del hombre que los cabellos de su cabeza, donde son tantas las enfermedades del cuerpo, tantos los apetitos y deseos desordenados del ánima, (…). (1583, Fray L. de Granada, Segunda parte de la Introducción del Símbolo de la Fe) ‘Which man would be that supplied with all those good qualities, and that exempt of all those bad qualities, since this life is a sea of continuous agitations and alterations, a valley of tears, a prison of condemned people, where a man’s miseries are much more numerous than the hairs on his head, where so many are the illnesses of his body, so many the disorganized appetites and lusts of his soul, (…).’ Although the difference in time is negligible, the specifying coextensive use emerged prior to the first quantifying reading, in 1485 (see (47)). Specifying mar de N2 continues to mainly characterize life, a person or his/her situation as (nothing but, an unlimited volume of) N2. The persistence of mar’s source semantics is usually high as the construction continues to profile at least mar’s infiniteness. In contrast to the not-coextensive but highly metaphorical uses, the specifying use is no longer close to literal uses where de N2 fulfills an identifying use (e.g. I am drowning in the sea of love, and not in the sea of joy). In examples (47)–(50), the women described are depicted as infinitely virtuous, incredibly beautiful or stuck in a never-ending crying respectively. The blending of the conceptualizations of the women and the mar de N2 is direct in (48), whereas the not coextensive metaphorical variant would be Leocadía se anega

206 | The role of conceptual persistence and analogy in the development of BQs

en un mar de llanto ‘Leocadia is sinking in a sea of weeping’. In the latter case, one can infer that Leocadia is crying ceaselessly, whereas the specifier construction explicitly characterizes her as such. In Classical Spanish, specifying mar de primarily combines with llanto ‘crying, weeping’ (11 occ., as in (48)), virtudes ‘virtues’ (6 occ., as in (47)) and pensamientos ‘thoughts’ (5 occ.). The construction almost exclusively shows up in apostrophes, epithets or the construction ‘X [es/está/other (pseudo)-copula] un/Ø mar de N2’ and combines with nouns referring to virtues/sins, tears, states and situations only. In Modern Spanish, the phrase hecho un mar de lágrimas ‘to have become a sea of tears’ has more or less become a fixed expression or conventionalized single chunk: it applies to 25 of 39 uses (and to 14 of 31 occ. in Present-Day Spanish). Its context always specifies the underlying motivation of the immense grief, usually the departure or death of a beloved person, as in (50).115 (47) Reyna de alto vuelo o mar de virtudes a ver te enel cielo mis sentidos mudes y a morir me ayudes con amor de llama enla cruz por cama. (1485, Fray A. Montesino, Coplas hechas sobre la pasión) ‘Prestigious queen (lit. high flight) / or sea of virtues, when I see you in the sky, / you move my senses / and you help me dying / with passionate love / on the cross as bed.’ (48) En esto volvió en sí Leocadia, y, abrazada del crucifijo, parecía estar convertida en un mar de llanto. (1613, M. de Cervantes Saavedra, La fuerza de la sangre)

|| 115   See also the quote from Casares (1950) on hecha un mar de lágrimas in the following corpus example: Si en lugar de decir que una viuda “estaba hecha un mar de lágrimas” dijésemos que estaba “como un mar de lágrimas”, no sólo emplearíamos un giro inusitado sino que la expresión habría perdido toda su fuerza. Esa viuda del ejemplo no ofrece semejanza alguna con un inexistente “mar de lágrimas”: es que las derrama en tal abundancia, que imaginamos a la pobre mujer anegada en ellas como en un mar. (1950, Introducción a la Lexicografía moderna) ‘If instead of saying that the widow “has become a sea of tears” we would have said that she was “like a sea of tears”, we would not only use an obsolete expression but the expression would also have lost all its strength. That widow of the example does not show any resemblance to an existing “sea of tears”: the thing is that she spills them so abundantly that we imagine the poor woman as drown in them as in a sea’.

The development of binominal quantifiers | 207

‘At this moment, Leocadia came round and, embracing the crucifix, she seemed to have changed in a sea of crying.’ (49) ¿Qué otra cosa es persuadirle un amante a su dama que es un mar de hermosura? (1627, G. Bocángel, Rimas y prosas) ‘What other thing is a lover convincing his lady that she is all beautiful (lit. a sea of beauty)?’ (50) Al día siguiente llegó la Marquesa hecha un mar de lágrimas, tan trémula y tan horrorizada, que no quiso permanecer allí un momento, y volvió a partir sosteniendo en sus brazos y cubriendo de lágrimas a su hija Constancia, que permanecía en el mismo estado. (1852, F. Caballero, Clemencia) ‘On the next day the Marquise arrived all in tears (lit. made a sea of tears), so trembling and so horrified, that she did not want to stay there any longer, and she turned to leave again while holding in her arms and covering with tears her daughter Constancia, who remained in the same state.’ The first attested quantifier use dates from 1508. In Classical Spanish, the specifying and quantifying uses are equally productive (see Table 8, infra). From Modern Spanish onwards, however, the quantifying uses dramatically rise in frequency and count 124 occurrences (60% of 208 coextensive binominal constructions, in contrast to 39 specifying uses or 19%) to the extent of becoming the unmarked interpretation in Present-Day Spanish (in 72% of the uses). In Classical Spanish, the quantifying uses continue to evolve around the same type of N2s as observed with the not coextensive uses – viz. virtues/sins, states, situations and (confused) thoughts – and marginally allow for animates and nouns referring to money. Note that the N2-collocates are inherently gradable nouns or nouns that invite for intensification. A particular virtue or sin is for instance by definition an unlimited (or unbounded) quality. The reason for considering a distinct use (in addition to the metaphoric, not-coextensive quantifying use) is the direct exploitation of the quantifying potential: in contrast to the many N2s referring to some kind of ‘liquid’ (mainly sangre ‘blood’ and lágrimas ‘tears’), the other right collocates are no longer compatible with the original idea of an infinite volume of water covering the earth’s surface. In addition, several semantic-pragmatic contextual clues can usually be found that point at quantification (cf. the quantifying relative pronoun quanto, the contrast with the absolute quantifiers poco, mucho, ninguna and the predicate bastar in (51); in (52), the predicate no saber elegir implies that there are too many, the adjectives vasto y anchuroso strengthen the extensive variety of opinions; the contrast with

208 | The role of conceptual persistence and analogy in the development of BQs

the (virtual) single drop of blood enhances the quantifying potential of mar de sangre in (53). (51) Con gran magnanimidad competían mi padre y mi madre (…) sobre quál distribuyría los bienes a pobres; y en fin los repartía mi madre, pareciéndole poco quanto tenía para lo mucho que desseava dar. Un mar de riquezas no le bastara para hartarse de hazer limosnas; tanto que le oý muchas vezes desear que ella y sus hijos quedássemos sin ninguna hazienda porque los pobres fuessen socorridos. (1548, G. de Toro, Tesoro de misericordia divina y humana) ‘With great magnanimity my father and my mother (…) quarreled about who would distribute the goods to the poor; and eventually my mother distributed them, and it seemed little what she had for the much she wanted to give. A sea of resources would not suffice to get her tired of giving alms; so many times I heard her wishing that she and her children would remain without any property so the poor would be helped.’ (52) Y así, en la ciencia que yo profesé, como en las demás, se advierten lastimosamente barajados los principios, conque la razón natural del viviente se halla precisada a no saber elegir entre el vasto y anchuroso mar de opiniones. (1725, D. de Torres Villarroel, Correo del otro mundo) ‘And so, in the science that I proclaimed, as in the rest, unfortunately one sees that the principles are being toyed with, by which the natural reason of the living being finds itself obliged to not be able to choose among the vast and wide sea of opinions.’ (53) Es verdad que en España se ha derramado la sangre inglesa; ¿pero dónde sino en Portugal se defiende la Inglaterra, y dónde, dónde el Portugal sino en España? Y para una gota de sangre inglesa que ha podido derramarse, ¿qué mar de sangre española no se ha vertido? (1813, J. Romero Alpuente, Wellington en España, y Ballesteros en Ceuta) ‘It is true that English blood has been shed in Spain: but where but in Portugal does England defend itself, and where, where does Portugal but in Spain? And for one drop of English blood that has been spilled, what a sea of Spanish blood has not flown?’ In Classical and Modern Spanish, the source semantics of mar was vividly present in functional uses. The most prominent facet that shades through in the grammaticalized uses is the concept of infinitude (as in (54)–(55)), in addition to

The development of binominal quantifiers | 209

the horizontally oriented extensiveness (as in (56)).116 Literally, seas reach further than the human eye can grasp and at some point in the horizon simply fade in the sky. The binominal mar de is also used for conceptualizing the N2-entities as in continuous or perpetual movement: in (56), the individual women are somehow conceived of as revolving waves. (54) ¿Por dó comenzaré, bondad inmensa, este mar de mercedes que me diste, pues es el comenzarle hacerte ofensa, siendo infinito lo que en mí hiciste? (1588, Fray P. Malón de Chaide, La conversión de la Magdalena) ‘But where will I start, immense kindness, / this sea of favors that you gave to me, / so that start mentioning one is offending you, / since what you have done for me is infinite?’ (55) Por un lado ve, o más bien siente a su Dios, en ese mar de perfección sin principio ni fin: infinito en sus infinitas perfecciones. Y por otro el abismo de su nada, pero esa nada, nada, nada, ese mar inmenso de nada, que no tiene fin ni principio. (1874, Sor Á. de la Cruz, Papeles de conciencia. Diario espiritual) ‘On the one hand, he sees, or rather feels his God, in this sea of perfection without a beginning or an end: infinite in its infinite perfections. And on the other hand, the abyss of his nothing, but that nothing, nothing, nothing, that immense sea of nothing, that has no end nor beginning.’ (56) Imposible poner orden en aquel mar de cabezas femeninas que se volvían y revolvían a todas partes; imposible conseguir silencio de aquellos millares de lenguas, acostumbradas a estar en perpetuo movimiento. (1945, R. Sarabia, ¿Cómo se educan los hijos? Lecciones de pedagogía familiar) || 116 The first quantifying instance that combines with neutral CIP emerges at the end of the 19th century, viz. in 1885. In the following example, mar de seems desemanticized: “Tienes razón (abrazándola con nervioso frenesí y dándole la mar de besos). Quererse y a vivir. Eres el corazón más grande que existe.” (B. Pérez Galdós, Fortunata y Jacinta) ‘You are right (while embracing her with nervously and passionately and giving her the sea of kisses). Love one another and live. You are the greatest heart that exists.’ However, even in this case, the restlessness and impetuosity inherent in passionate and nervous hugs make (la) mar de a far more suitable and coherent quantifier than for instance muchos or montón de. Note also the sudden appearance of the definite determiner la: in the entire corpus, all occurrences in the corpus that qualify for ‘neutral CIP’ combine with la.

210 | The role of conceptual persistence and analogy in the development of BQs

‘It was impossible to bring order in that sea of feminine heads that were turning around and around in all directions; impossible to silence the thousands of tongues, used to be in perpetual movement.’ When zooming in on the evolution in the N2-pattern (see also Section 5.3.2), the combinatorial pattern appears to dramatically extend in Modern Spanish. Although the N2s lágrimas ‘tears’ (17 occ.), sangre ‘blood’ (9 occ.), dinero ‘money’ (5 occ.) and llanto ‘crying’ (5 occ.) are the most frequent combinations next to the cluster of ‘thoughts’ and ‘virtues/sins’, at least two new clusters emerge, viz. abstract notions of time and human or animate nouns. The proliferation is pursued in Present-Day Spanish, where collocational preferences seem even less pronounced: cabezas ‘heads’ (7 occ.), cosas ‘things’ (5 occ.), gracia ‘grace’ (5 occ.) next to the clusters of human N2s, time indications and a lot of nouns which apparently all have different frames. However, if determiner variation is taken into account, the collocational extension for un/el mar de ‘a/themasc sea of’ is less dramatic or at least gradual: until the end of the 19th century, the N2collocates continue to evolve around the prefab clusters of ‘tears, crying’, ‘virtues/sins’, ‘mental states’, ‘situations’ and by extension human nouns and abstract notions of time. From the second half of the 19th century onwards, quantifying mar de is also frequently preceded by the feminine determiner la.117 With regard to the N2combinations in Modern Spanish, la mar de does not combine with abstract nouns other than notions of time and parts of discourse (e.g. digresiones filosóficas ‘philosophical digressions’, compromisos ‘commitments’). In other words, la mar de does not quantify virtues/sins, states or situations, yet mainly combines with persons, little characters (e.g. angelitos ‘little angels’), animals (e.g. caballos ‘horses’, toros ‘bulls’), small objects (e.g. botones ‘buttons’, flores ‘flowers’) and lots of concrete entities that are not conceptually linked. In Present-Day Spanish, la mar de equally combines with abstract and concrete, animate or inanimate nouns. In fact, the construction with the feminine definite determiner is the preferred one when N2 refers to a human or animate noun: of 20 animate nouns quantified by mar de, 16 instances are preceded by la.

|| 117 The reference dictionary DRAE does not specify the gender of the lexical unit mar. However, the examples (21) where mar is preceded by a feminine determiner or modified by a feminine adjective prevail over the masculine examples (7). Interestingly, although in the final definition ‘an extraordinary abundance of some things’, the example is masculine (i.e. Lloró un mar de lágrimas. ‘He cried a sea of tears.’), only la mar and la mar de are individually listed as fixed expressions for ‘in abundance’. By contrast, the etymological dictionary by Corominas/Pascual (1991) does not mention the quantifying or intensifying use of mar de.

The development of binominal quantifiers | 211

Interestingly, the determiner variation is also correlated with the degree of image persistence: generalizing over Modern and Present-Day Spanish, la mar de only combines 13 times with high conceptual persistence and 15 times with medial persistence. In the remaining 81 instances, the conceptualization imposed on N2 by la mar de does not relate metaphorically nor metonymically to mar’s source semantics. Nevertheless, the instances of la mar de N2 all share some important semantic-pragmatic features. First, the chunk la mar de N2 exclusively shows up in highly subjective contexts where the speaker (or protagonist) manifests his stance explicitly, usually in spoken language (cf. (57)– (61)). Second, they share the pragmatic inference of ‘extravagance’: la mar de N2 usually implies that more or less the totality, or at least all possible instances of N2 are taken into account. In (57), the speaker wants to convey the message that he already has plenty of commitments, to the extent that he cannot take an additional one. In (58), la mar de cosas ‘the sea of things’ actually means ‘everything’ and can best be paraphrased by ‘I am anything, yet not the accomplice’. Third, la mar de N2 usually shows up in a context of ‘surprise’, giving way to the pragmatic inference ‘incredible lots of N2’. In (59) and (60), the speaker feels the need to ascertain the veracity of his enunciation (cf. porque yo te lo digo and asegurando con toda veracidad), as his story is somewhat surprising or seems unbelievable. The same holds for (61), where the obviousness of the claim is contrasted with the ignorance of the interlocutor (cf. ¿No lo sabes?). The facet ‘surprise’ probably originates in the highly metaphorical, quantifying yet notcoextensive uses, where N2s referring to virtues or pleasant situations frequently show up in contexts with negative semantic prosody, such as the predicates hundirse / anegarse / ahogarse en un mar de amor / hermosura ‘to sink, to drown, to bath in a sea of love / beauty’. (57) – No le digo que sí ni que no, don José. Veremos. Tengo la mar de compromisos... Pero ya sabe usted que haré los imposibles por servirle… (1885, B. Pérez Galdós, Fortunata y Jacinta) ‘I am not saying you yes neither no, sir José. We will see. I have plenty (lit. the sea) of commitments… But you well know that I will do the impossible to serve you… ’ (58) ¡Ah! ¿Tú eres el cómplice? Yo no, señor Santi. La mar de cosas me soy menos eso... (1926, F. Romero y G. Fernández-Shaw, El casarío) ‘Ah! Are you the accomplice then? Me, no, sir Santi. I am everything (lit. the sea of things, plenty of things) except that.’

212 | The role of conceptual persistence and analogy in the development of BQs

(59) Lo sé, lo he soñado ahora, cuando me dormí pensando que me moría y que entraba en el cielo escoltada por la mar de angelitos… ¡tan monos…! Créetelo, porque yo te lo digo… (1885, B. Pérez Galdós, Fortunata y Jacinta) ‘I know, I was just dreaming of it, when I fell asleep while thinking that I was dying and entering heaven escorted by little angels everywhere (lit. by the sea of little angels)… How cute! You have to believe it, because I tell you so.’ (60) Completo mi retrato asegurando con toda veracidad que en los días á que me refiero hice la mar de conquistas, como verá el que me leyere. (1910, B. Pérez Galdós, Amadeo I) ‘I finish my story and assure in all veracity that during the days I am referring to I made plenty (lit. the sea of) of conquests as the one who reads me will see.’ (61) Dio con la mano abierta contra el mármol, se rió. – ¿No lo sabes? Tiene una [KV: escandalera] hace la mar de tiempo, y creo que le amenazó con quemarle la cara, como una fiera. (1965, E. Quiroga, Escribo tu nombre) ‘He was resting his open hand on the marber and smiled. – Don’t you know? He is having one [KV: scandal] since ages (lit. the sea of time), and I believe that he threatened with burning her face, like a wild animal.’ Still, in line with the central claim of this study that each QN imposes its own conceptualization on N2, I refrain from considering la mar de as a completely desemanticized chunk. If la mar de would only mean ‘many’, the paraphrase in (62) would be redundant. If la mar de botones is for instance replaced by un montón de botones, the facets ‘extravagance’ and ‘unexptectedness’ were lost and, in terms of absolute frequency, un montón de would count less buttons. (62) ¡Ah!... se me olvidó: el pelo ligeramente empolvado. Los guantes eran claros, de muchísimos botones; eso, eso, la mar de botones. Cuando entré, ya los llevaba puestos. Yo habría deseado que no, para ayudarle en la operación de abrochárselos. (1888, B. Pérez Galdós, La incógnita) ‘Ah… I forgot: her hair slightly powdered. Her gloves were light, with lots of buttons; yeah, yeah, the sea of buttons. When I came in, she had already put them on. I would have wished she had not, to help her in the operation of fastening them.’

The development of binominal quantifiers | 213

The fact that the feminine determiner la goes hand in hand with a highly schematized reading of mar de and from the 20th century onwards starts to be the unmarked variant in quantifying uses (76 of 128 occ. of la mar de N2 vs. 32 occ. of un mar de N2), makes me argue that la has specialized into a marker of quantity.118 The final coextensive use la mar de developed provides further evidence for this claim. The most striking context expansion is indeed the extension towards adjectival (and adverbial) right collocates. From the beginning of the 20th century onwards, la mar de starts combining with adjectives while fulfilling an intensifying function, as in (63). The adjectives that combine with la mar de are all inherently gradable notions for referring to character or behavior (e.g. simpático ‘nice’, contento ‘happy’, presumido ‘arrogant’, descarado ‘shameless’, etc.). The adjectives usually modify nouns denoting human entities and generally refer to positively valued qualities. The construction la mar de ADJ always occurs in combination with (pseudo-)copular verbs (e.g. ser/estar/ponerse la mar de ADJ ‘to becategorization, to belocalization, to become highly (lit. the sea of)’), epithets or postmodifying adjectives which are stressed.119 (63) Su amo no era malo; todo lo de antes habían sido locuras de juventud; pero ahora, al sentar la cabeza, resultaba un señorito de chipén, ¡la mar de simpático!, con gran afición a tratar a las gentes bajas, como si fuesen sus iguales. (1905, V. Blasco Ibáñez, La bodega) ‘His master was not bad; the things from before were all youth follies; but now, finally wise, he turned out to be a marvelous young man, really sympathetic, with great interest in addressing lower class people, as if they were his equals.’ || 118 The fact that quantifying mar de only recently started to abstract away from its source semantics is also culturally motivated. Historically speaking, in the Middle Ages, instances such as el mar de Cáliz ‘the sea of Cáliz, la mar de Espanna ‘the sea of Spain’, etc. and seas in general could evoke wide horizons yet were remote concepts for the large majority of the population. Only those people who lived along the coast could rely on their imagination and travel stories for picturing a sea’s extensiveness. The late desemanticization of mar de N2, somewhere halfway the 19th century, can then be interpreted as conform to changing society: in the second half of the 19th century, the sea has become an established and even daily concept. Until then, the quantifying uses of mar de were highly metaphorical. By way of comparison, in the 21st century, coextensive use of universo ‘universe’ could be a proper candidate for evoking the concepts of endlessness and extensiveness as in Todo un universo de sonidos nos rodea. ‘A whole universe of sounds surrounds us.’ (Google). 119 On a theoretical level, the many instances where the positive affect of N2 is neutralized by the negative semantic prosody of the hosting context implies that the prosody associated with (the context of) N1 scopes over the (possibly opposite) affect of N2.

214 | The role of conceptual persistence and analogy in the development of BQs

Interestingly, the construction la mar de ADJ shares with la mar de N2 precisely those features that distinguish the uses of la mar de N2 and un mar de N2. Not only does la mar de ADJ exclusively show up in subjective contexts and spoken utterances, it also typically evokes some element of ‘surprise’ (as in (64)–(65)), of ‘contrast’ with the actual situation (as in (66)). (64) – Siempre tiene una que cargar con todo. – Mañana me voy, Ana. – No, hijo, si yo estoy la mar de contenta de tenerlo aquí. Pero es que en el colegio ya habrán avisado a don Felipe. Figúrese usted el disgusto… (1962, J.M. Caballero Bonald, Dos días de setiembre) ‘– There is always one who has to shoulder everything. – Tomorrow I leave, Ana. – No, dear child, I am really happy to have you here. But the thing is that in the school, they will already have informed Don Felipe. Imagine the trouble…’ (65) No creas, conozco a papá. Cuando se pone terco... Dijo: – Es más burro. Pobre papá, es la mar de bueno. Pero quiere siempre salirse con la suya. Tiene celos. (1965, E. Quiroga, Escribo tu nombre) ‘Don’t believe it, I know dad. When he gets stubborn... He said: – He is more stupid. Poor dad, he is so/really (lit. the sea of) good. But he always want to be proven right. He’s jealous.’ (66) – Bah, antes me habría puesto la mar de contenta, pero ahora, no sé... Me da lo mismo. (1966, J. Marsé, Últimas tardes con Teresa) ‘– Bah, before it would have made me incredibly happy, but now, I don’t know… I don’t care.’ In other words, from a pragmatic perspective, la mar de N2 and la mar de ADJ can trigger the same inferences and both occur in evaluative contexts. From a morphosyntactic perspective, la mar de ADJ is constrained to attributive contexts. It could be argued then that the highly schematic binominal construction la mar de N2 has functioned as a bridging context for the adjective intensifier use in la mar de ADJ. However, the typical morphosyntactic context la mar de ADJ occurs in – i.e. in predicate contexts – does not host nominal quantifying uses but rather recalls the specifying use, as la mar de ADJ typically serve to characterize a particular person or entity a ‘incredibly, extremely ADJ’. However, binominal specifying uses are not observed with the feminine definite determiner in CORDE. Put more carefully, I tentatively argue that la mar de N2 specialized into a marker of quantification and then easily extended to other

The development of binominal quantifiers | 215

syntactic contexts. Next to the adjective intensifier uses, la mar de is observed seven times with the adverb bien ‘good’, each time within the construction estar la mar de bien ‘to be/feel perfectly (lit. the sea of) well’, as in (67). I have not yet found out, however, why it is precisely the combination with the feminine definite determiner which specialized into a marker of quantity. It is however no historical coincidence that of all QNs, mar de has become a productive adjective intensifier. Historically, mar de has been plainly used, both in not-coextensive and in coextensive uses, in quantifying and specifying constructions, to picture someone’s life, character or (mental or physical) state. The single chunk la mar de ADJ fulfills exactly the same role. In addition, mar is a monosyllabic QN with rather abstract source semantics. (67) – ¿Sabes que a la niña le ha salido novio? – ¿Sí? – Sí. – ¿Y qué tal? – La mar de bien, hija, estoy encantada. (1951, C.J. Cela, La colmena) ‘Do you know that the girl has got a boyfriend? – Really? – Yes. – And how is she like? – She's the most perfect, dear, I am delighted.’ The description of the different uses of mar de is not complete without reference to the head use of mar de N2. The literal coextensive use of mar de N2 is however peripheral and does not precede the grammaticalized uses. The first head use does not show up before 1585. In this use, the (masculine) indefinite determiner to N1 is the unmarked variant and mar de mainly combines with liquids such as water and blood (as in (68)), and also with air, fire (as in (69)) and all types of earth. In other words, N2 simply identifies the material that composes the extensive homogeneous volume. (68) Y subió más alto y creyó divisar a lo lejos las tormentosas nubes, semejantes a un mar de lava, y oyó mugir el trueno a sus pies como muge el océano azotando la roca desde cuya cima le contempla el atónito peregrino. (1862, G.A. Bécquer, Creed en Dios) ‘And he rose higher and thought he could distinguish far away the stormy clouds, similar to a sea of lava, and I heard the thunder bellow at its feet, just like the ocean bellows lashing the rock, from which top the astonished pilgrim contemplates it.’

216 | The role of conceptual persistence and analogy in the development of BQs

(69) – Una estatua erigida sobre unos peñascos... Al ponerse el sol, es un efecto maravilloso: la estatua parece de oro, y la rodea un mar de fuego… Es una aparición. (1881, E. Pardo Bazán, Un viajo de novios) ‘A statue raised on some crags… At sunset, the effect is marvelous: the statue seems made of gold, and a sea of fire surrounds her… It is an apparition.’ It is hard to verify empirically whether the coextensive quantifying uses originate in the head use. Yet, since the latter literal use was unproductive from the start and the development of quantifying un/la mar de N2 lacks a period of ambiguous uses, the head use cannot be the only source construction leading to the GR of mar de. Instead, the interference with the highly productive notcoextensive uses inviting for quantification (e.g. anego en un mar de amor ‘I am drowning in a sea of love’) has been a crucial factor in enabling the quantifying uses of binominal mar de. Table 8 gives a detailed overview of the development and distribution of the different uses. Two tendencies can be observed chronologically. In Classical Spanish, the quantifier and specifier were equally productive and five times as frequent as the marginal head use. From Modern Spanish onwards, the quantifier use beats the specifier and relegates the latter to the (frequency) level of the head use. The ambiguous uses are obviously not sufficiently frequent to trigger the shift from head to quantifying reading independently. Tab. 8: Distribution of uses per time period for mar de Head 1450–1730 1730–1900 1900–1975

# % # % # %

12 0.08 34 0.16 35 0.12

Quantifier N2 62 0.39 124 0.60 128 0.44

Adj. 0 0 0 0 70 0.24

Spec. Adv. 0 0 0 0 7 0.02

65 0.41 39 0.19 31 0.11

Amb. 3 0.02 3 0.01 1 0.003

Indet. 16 0.10 8 0.04 16 0.06

The general tendency to highlight more than one facet of mar’s original meaning in the grammaticalized uses of un/el mar de N2 is reflected in the late and gradual morphosyntactic bonding of the construction. For instance, up to Modern Spanish, it was still common to add an premodifier to N2 in grammaticalized uses as in (70), whereas the presence of adjectives between the preposition and the second noun is highly marked in Present-Day Spanish and does not even occur once with la mar de. Interestingly, the adjectives were generally intensifiers or highlighted a facet of the ‘sea’ frame, such as infinita ‘infinite’ in mar de

The development of binominal quantifiers | 217

infinita riqueza, inmensa ‘immense’ in mar de inmensa suavidad, eterno ‘eternal’ as in mar de eterno llanto ‘sea of eternal crying’, etc. (70) A través del follaje perenne que oír deja rumores extraños, y entre un mar de ondulante verdura, amorosa mansión de los pájaros, * desde mis ventanas veo el templo que quise tanto. (1884, R. de Castro, En las orilla del Sar) ‘Through the eternal foliage / that lets hear strange rumors, / and between a sea of ondulating verdure / mansion full of love of the birds, / from my windows I see the temple that I wanted so much.’ As to determiner variation, the article preferences of grammaticalized uses became more and more pronounced. For quantifying uses, the preference for the (masculine) indefinite determiner was already clear in Classical Spanish, but mar de still allows for a variety of grounding elements in the 18th century. In the second half of the 19th century, la suddenly shows up and is generally used in the quantifying uses which are only weakly reminiscent of mar’s source semantics. In Present-Day Spanish, la has taken over the lead in the quantifying domain: it occurs in 153 of 205 (or 75%) of the quantifying uses. Crucially, it is the only article used in the uses where mar de intensifies adjectives or adverbs. Consequently, in combination with mar de, I consider la as a marker of quantification. Generalizing over all periods, there seems to have been more determiner variation in quantifying uses than in specifying uses. The latter use shifted from preference for no determiner at all to preference for the masculine indefinite article.120 With regard to adjectival modification to mar, it has to be noticed that, in contrast to the not-coextensive and highly metaphorical quantifying uses, the grammaticalized uses only rarely combine with modifiers.121 When they do so, they relate either to configurational characteristics typically associated with seas in general and thus foreground which facet becomes highlighted in the

|| 120 The tendency for specifying mar de to lack a determiner correlates with its main appearance in apostrophes and epithets in poetry, such as Reyna de alto buelo, O mar de virtudes, (…) ‘Prestigious queen (lit. of high flight), oh sea of virtues, (…)’. 121 In terms of exact frequency, in Classical Spanish, 4 of 62 (or 6%) of the quantifying uses combine with a modifier to mar and 4 of 65 (or 6%) of the specifying uses. In Modern Spanish, 9 of 124 (or 7%) of the quantifying uses and 2 or 39 (or 5%) of the specifying uses combine with an adjective to N2; in Present-Day Spanish, 6 of 205 (or 3%) quantifying uses and 1 of 31 (or 2%) of the specifying uses.

218 | The role of conceptual persistence and analogy in the development of BQs

grammaticalized use (e.g. proceloso ‘stormy, tempestuous’, inquieto ‘restless’, insondable ‘bottomless’, etc.), or to size (e.g. gran ‘big’, inmenso ‘immense’, amplísimo ‘very wide’, etc.). The only exception to this tendency observed in our corpus is illustrated in example (71): although sombrío ‘dark, gloomy’ morphosyntactically modifies mar, it semantically scopes over N2 as well (which can be derived from the antonym centelleantes ‘twinkling, sparkling’ added to the N2 ojos ‘eyes’ as a paraphrase. In any case, adjectives to mar which scope over N2 do not contradict the claimed GR, as they provide further evidence for the primarily functional semantic content of mar de. (71) La misma presencia de gente, que antes le intimidaba, ahora le enardecía. Ya no tenía delante un sombrío mar de cabezas, sino un firmamento de ojos centelleantes. (1952, J.L. Sampedro, Congreso en Estocolmo) ‘The same presence of people, that before intimidated him, now aroused him. He had no longer before him a gloomy sea of heads, but a firmament of sparking eyes.’ The development of the N2-combinatorial pattern of mar de reveals several parallels with ‘prototypical’ montón de. Whereas in Medieval Spanish, the grammaticalized uses only combined with the clusters of ‘virtue/sin’, ‘(mental or physical) state’, ‘situation’, ‘liquid’ (and lágrimas and llanto in particular), this clustered structure of the N2-pattern gradually becomes less tight. In addition to the emergence of new clusters (such as ‘humans’), from Modern Spanish onwards quantifying and specifying mar de combine with all kinds of objects as well. Interestingly, la mar de N2 admits a wide variety of N2s from the start. La mar de ADJ/ADV only combines with inherent gradable qualities which are typically foregrounded without limits (see also Section 5.3.2). With respect to verb and phoric agreement, only three instances of quantifying mar de are observed where agreement is explicitly made with N2. Per time period, a minority of the instances present verb or phoric agreement with N1. However, when the degree of image persistence is taken into account, the agreement pattern of mar de supports the idea that agreement with N1 ceases to be marked when several facets of N1’s original frame are highlighted. More precisely, generalizing over all periods, 10 of 17 quantifying instances where agreement with N1 is observed display high conceptual persistence in addition to 6 instances of medial persistence. Syntactically, the behavior of adjective intensifier la mar de is rather noteworthy: la mar de ADJ/ADV does not allow for determiner variation nor adjectives to N1 and only intensifies adjectives with the required semantic property. In other words, in contrast to the remaining BQs, la mar de operates as a single

The development of binominal quantifiers | 219

chunk or semi-fixed expression. In this context, the advanced schematization of mar’s semantic content or lack of compositionality perfectly fits the status of conventionalized expression. Further, examples such as (72) evidence the purely functional content of la mar de: the adjective intensifier intervenes in the plural noun phrase frases felices ‘happy sentences’. Independently of the presence of la mar de, felices displays plural agreement with frases. The underlying structure is thus [frasesN [[la mar de]INT felices]ADJ]]. (72) Y es que el exilio, más o menos doloroso, puede también intentar frases la mar de felices, de una cordura ejemplar como aquella que estampó don Gregorio Marañón (…). (1971, M. Aub, La gallina ciega) ‘And the thing is that the exile, more or less painful, can also yield really happy sentences, of an exemplar good sense as the one that Don Gregorio Marañón printed (…).’ In sum, the development of the high frequency BQ mar de partially resembles the less dramatic and highly constrained development of the majority of the QNs (in the variant un/el mar de) and partially recalls montón de’s advanced stage of GR and primarily functional content (in the variant la mar de). In addition, its history is characterized by an additional reading or intermediate step, viz. the high-frequency not-coextensive metaphoric and quantifying reading which has served as bridging context. Furthermore, it is the only QN with a (productive) adjective intensifier use. Finally, it bears pointing out that the lexical item mar has led to a series of quantifying lexicalized and grammaticalized expressions: DRAE mentions – in addition to the BQ-construction and the adjective intensifier use – a mares ‘lit. on seas’ in llover / llorar / sudar ‘to rain cats and dogs / to cry / sweat streams / abundantly’, la mar ‘abundantly (lit. the sea)’, de mar a mar ‘lit. from sea to sea’ as in Juan iba de mar a mar ‘Juan went excessively adorned’ and hablar de la mar ‘to have to discuss a lot’.

5.2.7 Concluding remarks The previous case-studies have shown that the GR chain proposed by Traugott (2008a; 2008b) for the English binominal construction is not cross-linguistically valid. While the pathway from prepartitive over partitive to degree modifier, as illustrated in Figure 1, applies to a limited extent to the development of montón de and pila de, it does not correspond to the evolution of aluvión de, letanía de, hatajo de, barbaridad de, mogollón de and mar de. The most salient inconsistencies with respect to Traugott’s model are the lack of prepartitive or binominal head uses for aluvión de, letanía de, barbari-

220 | The role of conceptual persistence and analogy in the development of BQs

dad de and mar de and the relatively unproductive quantifying use of (h)atajo de. In other words, the source construction of BQs does not necessarily involve part-whole relations. In addition, except for mar de, the GR processes the QNs are engaged in do not entail a noteworthy rise in frequency. Further, the proposed pathway of change does not account for the emergence of the specifying use, which at least in the case of hatajo de, barbaridad de and mar de has become as important and/or productive as the quantifying use. Nor does the proposed pathway leave room for interference with unrelated categories such as the predicate atajar in the history of (h)atajo de or the highly metaphoric notcoextensive use of mar de which have most probably served as bridging context. Neither can the GR chain motivate the high degree of reminiscence of the QN’s original frame in the grammaticalized uses, nor the complete absence of occurrences for mogollón de in the historical corpus. These historical data throw up some further issues. First, the role of ambiguity and reanalysis in the GR of BQs is problematic. In grammaticalization theory it is commonly accepted that reanalysis – in the history of QNs, the functional shift from (coextensive) head to quantifier – takes place in ambiguous environments: a particular construction is assigned a new morphosyntactic and semantic interpretation, typically via analogy to existing constructions. Historically, the ambiguous sequences of BQs have been too infrequent to trigger the shift towards functional uses. More importantly, they do not always precede the unambiguous quantifying/specifying use, and the head use or source construction may be lacking all together. Finally, the relatively infrequent indeterminate uses observed in our corpus do not blend the head and the quantifying/specifying use but usually reflect a mismatch between the quantifying and the specifying use, as in (73). Instead of prefiguring the locus of change, they blur the distinction between the quantifying and the specifying reading and simultaneously exploit the potential of both grammaticalized uses. (73) Conforme en todo, menos en una cosa; y pues usted me pide franqueza, allá va mi parecer sincero. Todo me ha parecido bien, menos la idea de meternos aquí todos los frailes de la Cristiandad. ¿Para qué queremos aquí tal aluvión y acarreo de regulares? Nosotros los seculares nos bastamos y nos sobramos para todo lo que haya que hacer. Sobre que son en su mayoría un hatajo de gandules que vienen aquí con hambre atrasada, y en poco tiempo consumirían todas las subsistencias de la Nación, querrían mangonear ellos solos y nos reducirían á una servidumbre vergonzosa. (1910, B. Pérez Galdós, Amadeo I) ‘Satisfied with everything, except for one thing; and then you ask me sincerity, there goes my sincere opinion. Everything seemed good to

The development of binominal quantifiers | 221

me, except the idea of all of us friars of the Christianity getting here. What for do we want here such a flood and haulage of regulars? We, seculars, are enough for each other and there are more than enough of us for all that has to be done. About the fact that there are majoritarily a herd of lazybones that come here with delayed hunger, and that in little time would consume all the subsistence of the Nation, they only would want to boss people around alone and would reduce us to shameful servitude. The second issue raised by the corpus study concerns the directionality of the changes. Whereas the literature on English BQs (Brems 2010; Trousdale 2010) suggests that the emergence of quantifying uses precedes (and to some extent triggers) the emergence of evaluative or specifying uses, the prior emergence of the quantifier use is not self-evident in the development of Spanish binominals. As illustrated in Figure 2, four QNs seem to support this directionality: the quantifying use of letanía de and hatajo de precede the specifiying use, in the case of prototypical montón de the difference in time is negligible and pila de does not even pursue the pathway towards specifying uses. On the other hand, although only a small lapse of time separates both uses, the specifier use of aluvión de, barbaridad de and mar de is prior to their quantifier reading. Since the specifying use of barbaridad de seems to bridge the gap from literal to quantifying uses, the directionality of the change is obviously inverted here. The latter set of QNs thus shows that the development of the quantifier reading is not a necessary condition for the specifying use to arise. Recall that the usage contexts of the first specifying uses of montón de have more in common with the typical contexts of the head use than with the quantifying uses. Yet, of greater interest for the diachronic model advocated in Chapter 6, is the observation that the specifier use of montón de has already become an established reading before the other QNs start to take on quantifying readings. It will therefore be argued that the functional shift in the GR of BQs concerns the shift towards ‘coextensiveness between N1 and N2’, which can be either more quantity or more quality oriented.122 Once the link between the concept of coextensiveness and the QN in question arises, speakers may start to exploit both the quantifying or the specifying potential of coextensiveness.

|| 122 For the moment it suffices to say that the specifier use is the extreme realization of quantification: it occupies the positive end of the scale implied in quantification.

Fig. 2: Time line illustrating the emergence of each use per QN

Mar de

Barbaridad de

Hatajo de

Letanía de

Aluvión de

Pila de

Montón de

T

P

QADV

QADJ

QN

H

P

Q

P

Q

H

P

Q

H

P

Q

H

Q

H

P

Q

H 1255

1250

1300

1400

1400

1485

1479

1508

1521

1500

1553

1585

1601

1603

1600

1646

1700

1758

1800

1865

1839

1847

1905

1905

1900

1945

1951

1967

1967

222 | The role of conceptual persistence and analogy in the development of BQs

The development of binominal quantifiers | 223

The third issue touched by the development of the BQs is the role of (negative) semantic prosody. For English BQs, Brems (2007a, 208) suggests that the emotive colouring between various collocates and the N1-node does not merely rely on the meaning or connotation of N1, yet is based on the interaction between the node and its collocates. For instance, the negative semantic prosody associated with the ‘valuing quantifying’ uses of load of does not only arise from the almost exclusive combination with N2s that have a negative affect, but is also triggered by the (negative) implicatures in the source semantics of load, since a load of transported entities implies being a burden. Generalizing over all periods and QNs, Table 9 shows that the sharing of affective values between N1 and N2 is indeed more obvious for specifying uses (which count relatively less neutral occurrences). The distinction made depends on whether the semantic prosody exclusively concerns the relationship between the node and the collocate, the node and the context or both. However, except for hatajo de and letanía de, the prosody is not exclusively negative and a considerable proportion of grammaticalized uses have a positive semantic prosody. It should be pointed out at the same time that the emotive sharing between the context and the node is generally more important than the relationship between the right collocate and the node, at least in the case of negative semantic prosody. Recall that mar de predominantly evokes a negatively coloured prosody in spite of the frequent ‘virtue’-cluster it combines with, as in engolfado en un mar de complacencia ‘swallowed by a sea of indulgence’, which suggests that the scene set by the surrounding context can overrule the semantic affect associated with a particular right collocate. Consequently, I do not believe that the negative semantic prosody that the majority of the specifying (and quantifying) QNs share with the right collocate has triggered the emergence of new readings in Spanish. Nevertheless, the considerable presence of evaluative contextual clues indicates once more that subjectification plays an important role in the GR of BQs. Tab. 9: The correlation between semantic prosody and the grammaticalized readings Negative semantic prosody QN Qadj Qadv P

Coll. 39 6 – 14

Context 99 17 2 58

Both 134 12 – 184

Positive semantic prosody Coll. 6 9 2 7

Context 14 – – 5

Both 24 19 3 23

Neutral 273 7 – 36

# 589 70 7 327

The final question raised by the individual pathways of the QNs concerns the prototype and possibly triggering role of montón de. Its high frequency and

224 | The role of conceptual persistence and analogy in the development of BQs

head start sharply contrast with the development outlined for pila de, letanía de, aluvión de, hatajo de and barbaridad de. On the other hand, the early development of grammaticalized uses and the relatively high frequency of mar questions montón’s prototype role. This need not, however, be problematic if the coselection pattern is taken into account, which continues to be more constrained for mar de. Recall also that the dramatic collocational extension mainly concerns la mar de – whereas other QNs do not occur with the definite article unless topicalized – and la mar de’s adjective intensifier use, while other QNs do not develop this use. Attributing a prototype or triggering role to mar de thus appears a forced interpretation of the historical events. Interestingly, only montón, mar and barbaridad have developed quantifying uses outside the binominal construction as well.123 This observation makes me argue that the quantifying potential of QNs is more important than the precise original contexts in which they are used (which are empirically speaking often hard to reconstruct). The recent importance attributed to the original context of a linguistic item in grammaticalization theory, I believe, misleadingly detracts the attention from the precondition for change.   It is for instance commonly accepted in functional and cognitive diachronic construction grammar theories that in order for change to occur, a specific speaker has to use a specific surface construction with a new meaning and/or morphosyntactic make-up and the interlocutor has to copy the new use in other contexts (Trousdale 2010). The constructional change only counts as change when enough speakers use this innovation (Traugott 2008b, 36; see also Paradis 2011 on ‘intersubjective entrenchment’). However, in my view, this ‘spreading’ does not occur by mere coincidence. The construction grammarian interpretation of ‘spread’ might implausibly suggest that individual creative speakers constitute the starting point of changes. I find it hard to believe, however, that major changes which occur cross-linguistically (e.g. the future marker on the basis of the movement verb ‘to go’ in English, French, Spanish, Dutch, German and many other languages, next to the cross-linguistic association between ‘to come’ and the expression of recent past), exclusively rely on the creativity of one or more particular speakers of those languages who independently of each other start using the verb ‘to go’ ambiguously. Similarly Heine et al. (1994) suggest that this universality stems from the fact that the candidates for GR are notions drawn from human experience. Traugott (2003b, 632) claims that mean-

|| 123 According to DRAE, a montones means ‘abundantly (lit. in heaps)’, ser alguien del montón ‘to be someone of the heap’ applies to average people and the adjective bárbaro can also mean ‘excessive’.

The role of persistence and analogy | 225

ing change is not arbitrary but originates in pragmatic strengthening. In line with Paradis’ (2011) claim that ‘metonymization’ is the key mechanism in semantic change, I argue that linguistic items can only change because of their potential to change or intrinsic multi-layered structure. The characteristics of the original context of usage then determine which items will exploit the potential and to what degree.

5.3 The role of persistence and analogy The crucial role played by analogy in the development from lexical noun to BQ can hardly be denied, as the underlying GR process is characterized by a twodimensional clustering around semantically similar domains. First, lexical nouns with scalar implicatures can be structured into several slightly overlapping domains that typify the lexical field of N1. Second, per QN analyzed, the attested N2s generally cluster around one or more central items. Diachronically, this twofold organization around semantically similar N1s, respectively N2s, suggests that new instances of the construction are created by analogy based on semantic properties. This analysis, then, is similar to that described by Bybee/ Eddington (2006), who showed that the Spanish verbs of ‘becoming’ display an exemplar representation. In a nutshell, in a usage-based exemplar model, each construction is mapped by speakers and hearers onto identical existing representations, i.e. exemplars, strengthening them, or is at least evaluated for its degree of similarity to other exemplars. Novel instances are then formed on the basis of analogy with existing ones. Typically, high-frequency exemplars appear to function as central members of their category, which suggests that in the creation of new instances, the actual token frequency of constructions is as important as similarity to previous uses. This usage-based approach implicitly underlines the importance of the QN’s original usage – as lexical item or as the head of the binominal – in the creation of novel instances: “exemplar representation allows for redundant or marginal features to serve as the model for novel expansions of the category” (Bybee/Eddington 2006, 327). In the following paragraphs, I will first argue that a similar clustering around semantically similar constructions (or a similar exemplar representation) can be hypothesized for all QNs on the one hand (Section 5.3.1) and for all N2s occurring with a particular QN on the other hand (Section 5.3.2).

226 | The role of conceptual persistence and analogy in the development of BQs

5.3.1 Persistence and analogy in the extension towards new N1s In order to identify the nouns that have a quantifying potential in addition to their literal meaning, I looked at all instances of the schematic binominal construction ‘N1 de N2’ in the Corpus del español. These QNs, i.e. the nouns with scalar implicatures, can be grouped into six major domains comprising (1) (human) actions of ‘gathering’ or ‘spreading’; (2) natural phenomena (active or static) which are overwhelming by their scale or sudden appearance; (3) collections (of animals, things and persons, and military collections in particular); (4) configurations (as a heap or a cluster); (5) containers and (6) abstract numbers or quantity notions. Despite some degree of overlap, this basic division of the QN’s lexical field into six functional domains remains unchanged from Medieval until Present-Day Spanish. In the late 19th century, however, an additional subdomain emerged, viz. abstract evaluative nouns expressing an extreme value on a scale such as barbaridad.124 By way of illustration, the lexical field of QNs as attested in the Corpus del Español for the 15th century is presented in Table 10.125 Tab. 10: Lexical field of QNs for the 15th century according to the Corpus del español Lexical field

Lexemes observed in BQ-construction

‘Gathering’ or ‘spreading’

abasto ‘supply’ (2), amontonamiento1 ‘stack (deverbal N from amontonar)’ (4), ayuntamiento ‘assemblage’ (47), carga ‘cargo, load’ (14), golpe ‘knock, blow’ (18), presa ‘prey (deverbal N from prender)’ (2)

Natural phenomena

DYNAMIC: arroyo ‘stream’ (8), abundancia2 ‘wealth’ (from Latin ‘flood’) (22: 3), caudal ‘volume of flow’ (3: 2) , fuente ‘source’ (38) STATIC: laguna ‘lake, lagoon’ (5), monte3 ‘mountain’ (7:5), otero4 ‘hillock’ (1), ramo ‘twig, branch, bouquet’ (10: 4)

|| 124 The diachronic constructional network model proposed in Chapter 6 will show why the (peripheral) category of expressive abstracts nouns is a natural candidate for expressing extreme quantity. 125 Several QNs occur in more than one subdomain, e.g. amontonamiento ‘stack’ (deverbal noun from amontonar ‘to pile up, accumulate’: the action of heaping something up results in the typical conical configuration), monte ‘mountain’ is a natural conical configuration, etc. Those N1s have been underlined and numbered. The exact token frequencies are mentioned between parentheses after each QN. Note that these token frequencies comprise literal, ambiguous as well as some systematic quantifying uses (in bold after colon). It was not possible to perform a statistically reliable frequency analysis on the basis of the category search in the Corpus del español for the reasons exposed in Section 5.1.

The role of persistence and analogy | 227

Lexical field

Lexemes observed in BQ-construction

Collections

ANIMALS: hato ‘flock’ (10) , manada ‘herd’(5) THINGS: aparejo ‘equipment’ (3) HUMAN (NOT MILITARY): partida ‘group’ (11: 3) MILITARY COLLECTIONS: caballería ‘cavalry’ (5) , compañía ‘company’ (24), flota5

‘fleet’ (8), gente ‘people (to quantify a troupe of soldiers)’ (26), hueste ‘host’ (34), legión ‘legion’ (3) , tropel ‘mob’ (9: 3) Configurations HEAP: amontonamiento1 ‘stack’ (4), monte3 ‘mountain’ (7: 5), otero4 ‘hillock’ (1), cerro ‘hill’ (2), montón ‘heap’ (7: 1) CLUSTER: ovillo ‘tangle, jumble’ (4) Containers

flota5 ‘fleet’ (8), manojo ‘bunch (from mano [hand, by extension handful])’ (5)

Quantity notions

abundancia2 ‘wealth’ (22: 3), cantidad ‘quantity’ (25: 3), copia ‘abundance’ (26), infinidad ‘infinity’ (1: 1), millar ‘billion’ (8: 3), número ‘number’ (15), multitud ‘crowd of people, dozens of things’ (91), muchedumbre ‘crowd’ (102), par ‘pair’ (26)

Certainly, categories can still be subdivided into more specific subdomains, especially in later centuries, for which more corpus data are available. For instance, static natural phenomena can be divided into ‘mountain configurations’ (montón ‘heap’, monte ‘mountain’, montaña ‘mountain’, cordillera ‘mountain range’, etc.) and ‘volumes of water’ (mar ‘sea’, océano ‘ocean’, piélago ‘ocean’, etc.). Moreover, subdomains may become more or less productive over time (e.g. the collections of military nouns for indicating a group of people were particularly productive in Medieval and Classical Spanish), which may reflect tradition and society related preferences. These findings seem to indicate that new instances of the BQ-construction are generated by analogy to semantically similar QNs: the lexical field of QNs has the same structure in Present-Day Spanish as in Medieval Spanish, except for the appearance of new QNs, the disappearance of archaic QNs and the emergence of a 7th subdomain. At the same time, the development of BQs seems to challenge Bybee/Eddington’s (2006) view on exemplar representation, as the different domains are not clustered around a particular prefab or lexical item which becomes increasingly more frequent over time and generates semantically related new instances. The process rather seems to be driven by analogy to a conceptual image or frame. Instead of being triggered by one specific highly frequent lexeme for ‘avalanche’, new instances of the binominal construction are triggered whenever speakers feel that a particular frame or concept, such as ‘avalanche’, is useful for expressing quantity in a particular socio-historical context. Thus, we refer to a box or carton of carrots, not because box and carton are near-synonyms and frequent occurrence of the former prompts the emergence of the latter, but because they both function as typical containers of car-

228 | The role of conceptual persistence and analogy in the development of BQs

rots, in addition to basket, bucket and bag (which are by no means synonyms). As the QN’s lexical field, as sketched for the 15th century in Table 10, was similarly structured in Medieval Spanish, I assume that the Spanish language simply continued the ramification already existing in Latin (Bassols de Climent 1992). At the same time, it should be pointed out that for some QNs of Table 10 quantifying uses have been observed in the 15th century already, which questions the ‘prototypical’ status of montón de as well as its role as attracting force. However, as the corpus used generalizes over Peninsular and Latin-American Spanish and the frequency of montón de differs quite substantially from the data returned by CORDE, Table 10 not necessarily invalidates the claims made so far. Further, the overall low frequency of these quantifying uses is note-worthy. Since the QNs mainly belong to the subdomain of natural phenomena (which montón’s lexical congener monte is a member of), to the subdomain of configurations (next to montón de) and of quantity notions (recall that I refrain from considering coextensive uses of abstract notions of quantity as grammaticalized (see Chapter 2), those quantifying uses can be motivated. In a nutshell, montón de might not have been the first QN to develop unambiguous quantifying uses, yet remains the first one to systematically exploit its quantifying use and to dramatically rise in frequency. Finally, the spontaneous quantifying exploitations of other QNs can also be explained in the light of the Latin ancestor of the BQ, viz. the partitive genitive: a binominal construction related to quantity thus somehow already existed.

5.3.2 Persistence and analogy in the extension towards new N2s The analysis of the (diachronic) collocational shape of the individual QNs provides additional support for the claim that analogical thinking is an important mechanism in operationalizing the GR of the BQ. In addition, it shows that the N2-slot particularly illustrates the subtle interplay between analogical thinking and conceptual persistence, in that the clusters of semantically linked N2s are generally motivated by the source semantics of N1. The analysis in the present section builds on Verveckken’s (2012) observation that “for each individual QN, the different uses, viz. as head, quantifier and specifier, tend to combine with clusters of semantically related N2s” (2012, 446). On the basis of the partially filled query [[racimo] de nn*] in the Corpus del español, all instances of racimo in the binominal construction were returned. The in-depth analysis of the right collocates shows that the first quantifier use shows up in the late 19th century but does not manage to compete with the large-

The role of persistence and analogy | 229

ly productive head use in Present-day Spanish.126 Further, when used as a head, racimo was first only combined with types of fruit (grapes, bananas and dates in particular) or gem stones. In the 17th century, its collocational range was extended to celestial bodies or objects usually associated with (falling from) the sky and a typical cluster-like constellation. This extension was triggered by the foregrounding of the typical ‘hanging’ constellation of a literal bunch of grapes. From the 19th century onwards, as the idea of ‘bundle’ (instead of the strict ‘cluster’) became more prominent, literal racimo was also occasionally combined with concrete and abstract nouns which were no longer literally fastened together. Even more crucial is the observation that except for montón de, the right collocates occurring in the quantifying uses of the QNs analyzed always show this clustered structure, which generally (although not necessarily nor exclusively), continues the clusters of semantically related N2s in the original head use. In the case of racimo, for instance, the quantifying use continues the cluster of ‘celestial’ N2s and adds the new cluster of ‘animate creatures’. Although the analysis described in the present section does not alter the main claims defended in Verveckken (2012), viz. the dominance of conceptual images over prefabs and the fundamental role of analogical thinking, some refinements are added. In the following paragraphs, the particular developments of the collocational shape is discussed separately for each QN. Although at first sight, two types of collocational patterns can be distinguished, whereby a difference is made between tight N2-clusters and a more dramatic extension of the collocates – the extreme example being mar de who ends up as adjective intensifier –, the following analysis will show that even the cases of spectacular extension can be conceptually motivated and linked to N1’s source semantics. The spectacular collocational extension as observed with montón de and mar de goes hand in hand with their advanced stage of GR, their relative closeness to absolute quantifiers, the more general source semantics and resulting lack of strong tendency towards high conceptual persistence. Table 11 portrays the collocational development of pila de. The N2-pattern of the head use presents a stable clustered structure over time. The main develop|| 126 For the sake of completeness, I list the total number of uses observed per century: 13 H uses in the 16th century; 24 H uses in the 17th century; 4 H uses in the 18th century; 19 H and 5 Q uses in the 19th century; 30 H, 6 Q and 1 S use in the 20th century. However, as the analysis of racimo is based on another corpus, I do not include the table here as it might lead to misleading conclusions: when entering the query [racimo de] (and its orthographic variants) in CORDE and CREA, the total number of returned instances is 275 (in contrast to the 102 occ. found in Corpus del español, in spite of the lack of filter to Peninsular Spanish.)

230 | The role of conceptual persistence and analogy in the development of BQs

ments are the gradual rise of new clusters, all centered around the adjective ‘stackable’. In Classical Spanish, the right collocates that combine with pila are either a type of earth, a type of textile or another type of natural product. To some extent, clusters may overlap. Some types of textile such as skins, leather or wool also pertain to the faunal side of nature. Their relative frequency and the emergence of ‘manufactured’ textiles from the 20th century onwards constitute the underlying motivations for considering N2s referring to a type of textile as a separate cluster. From Modern Spanish onwards, new clusters arise, which share the concepts of being ‘easily stackable’ (because of their typical shape for instance) or ‘typically stacked or piled’: corpses after a battle are usually heaped up to be burned or transported; lots of mattresses usually come in piles, etc. Note that all N2s used in Classical Spanish are by definition easily stackable or usually stacked for the ease of handling them, yet this facet was less highlighted (since inherent). In Present-Day Spanish, a cluster of nouns is observed that refers to objects or entities which only come to be useful when piled up: the disks, for instance, referred to the vertical pairing of copper and zinc disks the Voltaic pile consists of. In sum, initially, the clusters of N2s literal pila de combined with, referred to products of nature typically or usually stacked, and gradually extended to all types of stackable objects. The collocates that show up in the Present-Day quantifying uses at first sight appear to be unconstrained. An in-depth analysis of the broader context detects however their accumulated or ordered nature. Note also that the data returned by CREA for pila de show a clear preference for notions of time (and años in particular) and persons. The collocational pattern of pila de thus demonstrates that while the more recent clusters do not evolve around near-synonyms or semantically near N2s, they do interact with the lexical specificities of pila de. Tab. 11: Collocation pattern of pila de over time N2-cluster 1450-1730 H EARTH

N2-collocates observed arena ‘sand’, greda ‘clay’

TEXTILE

colambre ‘skins’, cueros ‘leather, skins’, lana ‘wool’

NATURAL

leña ‘firewood’

PRODUCTS

1730-1900 H EARTH

nieve ‘snow’, mineral ‘mineral’

TEXTILE

lana ‘wool’

TYPICALLY

colchones (2) ‘mattresses’, sacrificios (2) ‘sacrified bodies’

STACKED

The role of persistence and analogy | 231

N2-cluster

N2-collocates observed

EASILY

duros ‘five-peseta coins’, volúmenes ‘volumes, books’ ‘RESIDUES’: basura ‘waste’, hojarasca ‘fallen leaves’

STACKABLE

1900–1975 H EARTH

adoquines ‘paving stones’, briqueta ‘briquet(te)’, carbon ‘charcoal’

TEXTILE

almohadas ‘pillows’, fieltros ‘(pieces of) felt’, ropa ‘cloths’

NATURAL

carbas ‘kind of bush’, herba ‘grass’, paja ‘straw’, sal ‘salt’ ‘WOOD’: leña (2) ‘firewood’, madera ‘wood’

PRODUCTS

STACKED

discos ‘disks’, bandejas ‘trays’, láminas de vidrio ‘sheets of glass’, placas de cristal ‘sheets of fine glass, of crystal’

TYPICALLY

platos (3) ‘plates’, alzas ‘insoles’, cadáveres ‘corpses’

FUNCTIONALLY

STACKED EASILY STACKABLE

Q

OTHER

almohadas ‘pillows’, sillas ‘chair’, sillas de tijera ‘folding chair’, trozos ‘pieces’ PAPER: cajas ‘boxes’, cajones ‘large box’, carpetas ‘folders’, pliegos ‘sheets of paper’ años (3) ‘years’, cosas ‘things’, dinero ‘money’, recuerdos ‘memories’

Table 12 presents the collocational pattern for aluvión de. It contains N2-clusters for both functional uses, but lacks any modeling head use. At first sight, the clusters might appear unrelated. Outside the context of aluvión de, the categories ‘invader’ and ‘(part) of discourse’ have nothing in common. Nor are the N2elements within the clusters semantically near. If the “mise en discours”, i.e. discourse integration, of the N2 is taken into account however, it becomes clear that the precise context profiles the N2 as an entity that invades another entity’s domain. In (74), the (vast number of) people are conceptualized as invading the domain of the speaker to the extent that the latter cannot track her companion any more. In other words, the clusters of N2s interact with the frame of aluvión. For the same reason, the conceptualization and usage context of the specific N2s are taken into account in the design of Table 13: the asterisks mark apparently misclassified N2s.127 When aluvión de clusters with human N2s, the latter are conceptualized as ‘invaders’ in a broad sense. The fact that sensations and different types of discourse enter one’s inner self via the senses, makes these clusters natural candidates for combining with aluvión. Further, the combina-

|| 127 In the case of pisadas, the context portrays a scene where a person was shocked by the footsteps of rodents he was hearing in the roofing. The total number of ‘apparent’ misclassifications also hints at the predominance of conceptualization over prefabs.

232 | The role of conceptual persistence and analogy in the development of BQs

tion of aluvión de with all sorts of reactions – usually unpleasant – is motivated by the fact that a literal flood is often triggered by an external cause and generally results in drastic consequences. In other words, the clusters of N2s attested with aluvión de turn out to meet the condition of being semantically compatible with the conceptualization imposed by N1. (74) (…) cuando, llegando a la puerta del teatro, un aluvión de gente, que se atropellaba por salir, lo envolvió y me lo separó, arrastrándolo por no sé dónde, y sin poderlo yo seguir, (…). (1847, S. Estébanez Calderón, Escenas andaluzas, bizarrías de la tierra …)) ‘(…) when, arriving at the door of the theatre, a flood of people, that were squashing eachother to leave, took him up and separated him from me, pulling him away I don’t know to where, and without me being able to follow him, (…).’ This observation also holds for the residual N2 class (under ‘other’). The contextualized bailes refers to a large number of new and foreign dances that suddenly entered the Spanish culture competing with the native forms. The bills metonymically refer to a large number of creditors stalking the indebted person. Similarly, the nouns attested in the Modern Spanish specifying uses actually refer to burning projectiles coming in during a battle, to a series of delicacies the speaker suddenly came to see (or experience) and to new pieces of information driving the addressee crazy. In other words, whereas the collocational behavior obviously distinguishes between the grammaticalized uses and the single literal use observed, the clusters of N2s in quantifying uses are identical to the collocational possibilities of the specifying use. Further, some degree of overlap between distinct clusters is to be noted: the context of aluvión de provincialismos describes the danger implied by the amount of borrowings invading the mother tongue. Tab. 12: Collocation pattern of aluvión de over time N2-cluster 1730–1900 Q INVADER (PART OF)

N2-collocates observed bárbaros ‘Barbarian people’, concurrentes ‘contestants’*, gente ‘people’, viajeros ‘travelers’

DISCOURSE

palabras (3) ‘words’, anónimos ‘anonymous (letters)’, novelas ‘novels’, voces ‘words’

REACTION OR

censuras ‘censure, condemnations’

ANSWER SENSATION

pisadas ‘footsteps’*

OTHER

bailes ‘dances’, cuentas ‘bills’

The role of persistence and analogy | 233

P

N2-cluster

N2-collocates observed

---

fuego ‘fire’, manjares ‘delicacies’, noticias ‘(pieces of) news’

1900–1975 H EARTH, MATERIAL barrio ‘mud’ Q

INVADER

artistas ‘artists’*, excursionistas ‘hikers’, pretendientes ‘pretenders, applicants’*, ratas ‘rats’

(PART OF) DISCOURSE

canciones (extranjeras) ‘(foreign) songs’, dicterios ‘insults’, elogios ‘praises’, escritos ‘writings’, provincialismos ‘provincialisms’

REACTION OR

besos ‘kisses’*, protestas ‘protests’, suscripciones ‘subscriptions’

ANSWER SENSATION OTHER

P

felicidades ‘hapiness’, ideas ‘ideas’, imágenes ‘images’, miserias ‘poverties’, sensaciones ‘sensations’, sorpresas ‘surprises’ cosas ‘things’

INVADER

extranjeros ‘foreigners’, madrugadores ‘people who get up early’*, mujeres ‘women’*, razas ‘races’

(PART OF)

exotismo palabrero ‘verbal exotism, exotic gasbag’, vulgaridades ‘coarse things’

DISCOURSE

On the basis of Table 13, similar conclusions can be drawn for the collocational pattern of letanía de. In its head use, letanía de is only compatible with response formulas, whose nature may of course vary over time. When used as a quantifier, letanía preferably combines with N2s referring to persons, which somehow mirrors the highly frequent not-coextensive use where literal letanía de is immediately followed by the name of the Saint whose help is implored (e.g. la letanía de la Vírgen/de los Santos ‘the litany of the Blessed Virgin/of the Saints’). Interestingly, the human N2s always refer to a long list of names which are enumerated in the context. N2s referring to enumerated items can also be concrete or abstract nouns, such as valentía whose context refers to a long list of someone’s virtues. N2s in quantifying uses can also refer to insults, which is of course semantically near to short formulas which are repeated out loud. In Modern Spanish, human N2s are no longer preferred in quantifying and ‘spoken’ N2s can refer to insults as well as praises128 and also extend to include (a repetition or long list of) ‘questions’ and ‘requests’. Note that the N2-entities under ‘other’ always concern entities pronounced out loud. Both peticiones and preguntas echo letanía’s source semantics: the processions and litany were usually performed to invoke the Saint’s help.

|| 128 Antonyms are by definition semantically near: for all conceptual features, insults and praises have the same value, except for the direction of the evaluation.

234 | The role of conceptual persistence and analogy in the development of BQs

Tab. 13: Collocation pattern of letanía de over time N2-cluster 1450–1730 H FORMULAS OF

Q

N2-collocates observed

RESPONSE

“conservare dineris” ‘to save money’, “epítetos y gracias más” ‘more epithets and graces’

INSULTS

injurias ‘insults’, simplicidades ‘simplicities’

TYPES OF

PEOPLE: autores ‘authors’, erradores ‘wandering people’, gente ‘people’, tiranos ‘dictators’, nombres ‘names’ OTHER: gastos ‘costs’, valentía ‘bravery’, guisados ‘stews’

ENUMERATION

1730–1900 Q INSULTS OR

alabanzas ‘praises’, dicterios ‘insults’, grandezas ‘noble things’

PRAISES TYPES OF

vinos ‘(types of) wine’

ENUMERATION OTHER

1900–1975 H FORMULAS OF RESPONSE

Q

preguntas ‘questions’

“¡hijo míó! ¡hijo míó! ¡hijo míó!” ‘¡Oh dear son! ¡Oh dear son! ¡Oh dear son!’, “pequeños, grandes huesos, cartílagos” ‘little, big bones, cartilages’

PRAISES

agravios ‘affronts, insults’, desvergüenzas ‘impertinences’, improperios ‘insults’, insultos ‘insults’, lamentaciones ‘complaints’, maldiciones ‘curses’

TYPES OF

nombres ‘names’

INSULTS OR

ENUMERATION

P

OTHER

peticiones ‘requests’

Table 14 shows that in the case of (h)atajo de the clustered structure of the N2pattern also remains largely stable over time. In contrast to the previous QNs described, a subtle difference can be noted between the clusters of the quantifiying and the clusters of the specifying uses. The subdivision of the quantifying N2-clusters is somewhat artificial, as it only concerns five instances. It bears pointing out, however, that the N2s are semantically near: while bribones make the speaker feel uncomfortable by definition, the context of limeras makes clear that they inconvenience the speaker by the noise they are making; villains and the sins they are committing are of course metonymically related. The specifier use combines with two more clusters, both centered around the notion of ‘stupidity’: ignorant persons or the stupid things they say. The facet ‘stupid’ is in harmony with hatajo’s source semantics – recall that a herd of sheep are guided by a herd instinct and lack any individuality or independency. The distinction between persons and their actions is also made in the

The role of persistence and analogy | 235

quantifying uses. In sum, the collocational pattern of (h)atajo de also suggests that new clusters or new members of a particular cluster do not primarily emerge because of analogy to N2s which are semantically near. Instead, the classification of the N2s corresponds to the way they are portrayed in the context and interacts with the source semantics of N1. While the association with ‘stupid people’ or ‘stupidities’ is prompted by the concept of ‘herd’, the link between ‘unpleasant people’ or ‘villain’ is far more abstract and correlates with the ‘lack of individuality’: the members of un hatajo de borrachos for instance are conceptualized as a homogeneous group of equally drunk people, which are only related because of the shared drunken state. Tab. 14: Collocation pattern of hatajo de over time N2-cluster 1450–1730 H CATTLE

Q

N2-collocates observed ovejas (2) ‘sheep’, yeguas (2) ‘mares’, elefantes ‘elephants’ ganados ‘cattle’, ‘vacas ‘cows’

VILLAINS

bribones ‘rascal’

UNPLEASANT

limeras ‘lime sellers’*

PEOPLE

1730–1900 H CATTLE

cabras ‘goats’

Q

SINS

minchionerías ‘incongruities, absurdities’, vicios ‘corruption, vice’

P

STUPIDITIES

absurdos ‘absurd things’, blasfemias ‘blasphemies’, desatinos ‘mistakes’, desverguënzas ‘shameless things’, inocentadas ‘practical jokes’, mentiras ‘lies’

STUPID PEOPLE

ignorantes (2) ‘fools’, botarates ‘irresponsible fools, spendthrifts’

VILLAINS

canalla ‘swines, scamp’, pillos ‘rascal’, tunantes ‘rascal’

UNPLEASANT

(infames) aventureros ‘(cruel) adventurers’, piojosos ‘lousy, filthy people’, padres ‘fathers’*

PEOPLE

1900–1975 H CATTLE

cabras (3) ‘goats’, ovejas ‘sheep’, puercos ‘pigs’, toros ‘bulls’

Q

hombres y mujeres ‘men and women’*

UNPLEASANT PEOPLE

P

STUPIDITIES

tonterías ‘silly things’

STUPID PEOPLE

indecentes ‘rude, shameless persons’

UNPLEASANT

asesinos ‘murderers’, borrachos ‘drunk people’, castrados ‘castratos’*, cenizos ‘unlucky persons’, desvergonzados ‘shameless persons’, gandules ‘lazybones’, palurdas ‘lumpish girls’, rameras ‘whores’

PEOPLE

236 | The role of conceptual persistence and analogy in the development of BQs

Table 15 suggests that the collocational shape of barbaridad is rather unconstrained. No prefab or conceptual links between the N2s and barbaridad’s source semantics are observed. Tab. 15: Collocation pattern of hatajo de over time N2-cluster

N2-collocates observed

1730–1900 P EXPERIENCE

susto ‘fright’

1900–1975 Q OTHER

quina ‘fortified wine’, sabios ‘wise men’, tiempo ‘time’

P

hombre ‘fellow, man’, moza ‘young girl’

UNPLEASANT PERSONS

The combinatorial pattern of mar de is remarkable: it partially resembles montón de’s N2-pattern in its apparent ‘unrestrictedness’, it extends towards adjectives and adverbs and the high number of lágrimas ‘tears’ combinations recalls the notion of ‘exemplar representation’. Again, if mar’s source semantics and original contexts of usage are taken into account, the development in the N2-pattern ceases to look arbitrary. Since mar de is a high-frequency QN, it is not instructive to add an exhaustive table of the right collocates observed. Instead, per cluster, some illustrative examples are given in Table 16 and only N2s occurring more than once are listed exhaustively. Tab. 16: Collocation pattern of mar de over time N2-cluster 1450–1730 H LIQUID NATURAL

N2-collocates observed agua (5) ‘water’ fuego (2) ‘fire’, plata (2) ‘silver’

MATERIAL

Q

LIQUID

sangre (4) ‘blood’, lagrimas(2) ‘tears’, tinta ‘ink’, vino ‘wine’

CRYING

llanto (4) ‘crying’, lágrimas (2) ‘tears’

VIRTUES/SINS

gloria (2) ‘glory’, ignorancia ‘ignorance’, mentira ‘lie’, sabiduría ‘wisdom’

PHYSICAL/ MENTAL STATE

hermosura (2) ‘beauty’, amargura (2) ‘bitterness’, afición ‘love’, ansia ‘desire’, furor ‘anger’, hiel ‘bile, bitterness’

CONFUSION

confusión ‘confusion’, contradicción ‘contradiction’

SITUATION

inconvenientes (2) ‘inconveniences’, riquezas (2) ‘richness’, desgracia ‘bad luck’, desventura ‘bad luck’

OTHER

movimientos ‘movements’, prisiones ‘prisons’, rayos ‘rays’, trabajos ‘works’, tribulaciones ‘tribulations’

The role of persistence and analogy | 237

P

N2-cluster

N2-collocates observed

CRYING

llanto (11) ‘crying’, lágrimas (5) ‘tears’

VIRTUES

virtudes (6) ‘virtues’, gracia (2) ‘grace, elegance’, donaire ‘grace’, encantos ‘charms’, piedad ‘mercy’, sabiduría ‘wisdom’

THOUGHTS,

pensamientos (5) ‘thoughts’, engaños ‘deceptions’, misterios ‘mysteries’

CONFUSION PHYSICAL/ MENTAL STATE

amargura (4) ‘bitterness’, amor (2) ‘love’, alegría ‘hapiness’, enojo ‘anger’, hermosura ‘beauty’ GRIEF: dolor ‘pain’, penas ‘pains’

SITUATION

miseria ‘poverty’, peligros ‘dangers’, riqueza ‘richness’

OTHER

alabanzas ‘praises’, flores ‘flowers’

1730–1900 H LIQUID NATURAL

sangre (4) ‘blood’, lava (3) ‘lava’, agua (2) ‘water’ fuego (2) ‘fire’, plata (2) ‘silver’, arena ‘sand’, nieblas ‘fog’, verdura ‘verdure’

MATERIAL

Qun LIQUID CRYING

lágrimas (17) ‘tears’, llanto (6) ‘crying’, lloro (2) ‘crying’

VIRTUES

dulzura ‘sweetness’

STATE

afectos ‘affections’

THOUGHTS,

confusiones (2) ‘confusions’, pensamientos ‘thoughts’, reflexiones ‘thoughts’

CONFUSION SITUATION

agitaciones (2) ‘agitations’, desastres ‘disasters’, luz ‘light’

HUMAN

gente (2) ‘people’, cabezas ‘heads’, pueblo ‘people’

MONEY

calderilla ‘change’, oro ‘gold’

OTHER

fuego (2) ‘fire’, palabras (2) ‘words’, vino (2) ‘wine’, libros ‘books’, lluvia ‘rain’, verdura ‘verdure’

Qla MONEY

P

lágrimas (17) ‘tears’, sangre (8) ‘blood’, sudor (4) ‘sweat’

dinero (4) ‘money’, millones ‘billions’

TIME

tiempo (2) ‘time’

HUMAN

angelitos ‘little angels’, chiquillos ‘kids’, novelistas ‘writers’

OTHER

bancos ‘benches’, besos ‘kisses’, botones ‘buttons’, compromisos ‘commitments’, cosas ‘things’, milagros ‘miracles’, platos ‘dishes’, ropa ‘cloths’

CRYING

lágrimas (25) ‘tears’, llanto (2) ‘crying’

VIRTUES/SINS

donaire ‘charms’, eloquencia ‘eloquence’, gracia ‘grace’, traiciones ‘treasons’

STATE

felicidad ‘hapiness’

OTHER

ciencia ‘science’, hojas secas ‘dry leaves’, puntas de lanzas ‘spear tops’

238 | The role of conceptual persistence and analogy in the development of BQs

N2-cluster 1900–1975 H LIQUID NATURAL MATERIAL

Qun LIQUID

lava (2) ‘lava’, agua ‘water’, lágrimas ‘tears’ fuego (4) ‘fire’, vidrio (4) ‘glass’, lodo (2) ‘mud’, nubes (2) ‘clouds’, plomo (2) ‘metal’ lágrimas (3) ‘tears’, sangre (3) ‘blood’, leche ‘milk’, veneno ‘poison

CRYING

lágrimas (3) ‘tears’

STATE

delicias ‘graces’, hiel ‘bitterness’

HUMAN

cabezas (5) ‘heads’, gente ‘people’, madres ‘mothers’

OTHER

aceite ‘vinagre’, banderas ‘flags’, colores ‘colours’, confusiones ‘confusion’, desdicha ‘misfortune’, encinas ‘holm oaks’, sueño ‘sleep’, tiempo ‘time’

Qla MONEY

P

N2-collocates observed

dinero (2) ‘money’, renta ‘income’, rublos ‘rubles’

TIME

veces (3) ‘times’, tiempo (2) ‘time’, años ‘years’, días ‘days’

HUMAN

gente (3) ‘people’, ateos ‘atheists’, cabezas ‘heads’

OTHER

cosas (5) ‘things’, gracia (5) ‘grace’, ilusiones (3) ‘illusions’, maneras (2) ‘ways’, ajos ‘swearwords’, balcones ‘balconies’, besos ‘kisses’, calefacción ‘heating’, consejos ‘advices’, espejos ‘mirrors’, libros ‘books’

CRYING

lágrimas (14) ‘tears’, llanto (2) ‘crying’

STATE

amargura ‘bitterness’, angustia ‘fear’, soledad ‘loneliness’

SITUATION

calma ‘calm’, oportunidades ‘opportunities’

OTHER

camisas ‘shirts’, encinas ‘holm oaks’

The combinatorial pattern of the head use appears to be stable over time: literal mar de only combines with water, other liquids or other elementary materials such as fire, earth and air. Generalizing over all periods, the quantifying – at least those preceded by the masculine determiner – and specifying uses continue to combine with the same N2-clusters as the highly metaphoric, notcoextensive quantifying uses: as they continue to compare someone’s life or character to un mar de N2, the specifying and quantifying uses continue to appear with the clusters of ‘crying’, ‘virtues/sins’, ‘(mental or physical) states’ and ‘situations’. Interestingly, the ‘crying’ cluster is the most important one for specifier uses in all periods and for the Modern quantifying uses with masculine determiner. Further, this cluster hosts the most frequent N2-combination overall, i.e. un mar de lágrimas, which at the same time coincides with the ‘liquid’ cluster and ‘grief’ subcluster. In addition, the quantifying and Modern specifying uses frequently combine with nouns referring to ‘(unpleasant) thoughts’ or ‘confusion’. Since thoughts are generally pictured as intangible, unbounded

The role of persistence and analogy | 239

and in constant movement, this cluster easily accommodates to mar’s source frame of continuously revolving waves.129 Although the clusters of the Classical quantifying and specifying uses are maintained over time, shifts in preferences occur, clusters become less numerous and new clusters show up. More precisely, the majority of the original clusters count step by step less N2-tokens,130 while the umbrella category ‘other’ proportionally gains more members over time and comes to burst in PresentDay Spanish). Further, new clusters are added, comprising nouns referring to persons, notions of time, money, which cannot be directly related to mar’s source semantics. In other words, the combinatorial pattern of quantifying and specifying un mar de are partially reminiscent of the original uses yet also dramatically extend over time. One final remark concerning mar de’s combinatorial pattern as a BQ regards the distinction between the masculine and the feminine BQ. Depending on the determiner chosen, mar de displays distinct collocational behavior. Basically, la mar de N2 does not occur with the original clusters of ‘crying’, ‘virtues’, ‘states’, ‘situations’ or ‘thoughts’. Instead, la mar de N2 has three minor clusters, viz. ‘time’, ‘money’, ‘people’ and allows for all kind of concrete and abstract nouns in the N2-slot. In its adjective intensifier uses, la mar de exclusively admits inherently gradable notions. The referent of the intensified adjective is human in 48 (or 69%) of 70 instances and the attributed quality is valued positively in 49 of 70 occurrences (or 70%). The valued qualities generally refer to character, looks, intellect and state. The three most important clusters are shown in Table 17. Since (un) mar de was originally used to compare an entity X to un mar de N2 while N2 referred to an abstract quality such as hermosura ‘beauty’, it is conceptually only a small leap from the abstract notion hermosura to the adjective hermoso. In other words, the syntactic context expansion builds on the original uses of mar de.

|| 129 Note that the not-coextensive metaphoric use hundirse/anegarse en un mar de pensamientos/meditaciones/confusión ‘to drown in a sea of thoughts/meditations/confusion’ is also highly productive (even crosslinguistically, e.g. in gedachten verzonken (Dutch), plongé dans les pensées (French), etc.). 130 If no more than one N2-member of a particular cluster is observed, this single N2 is moved to the category ‘other’.

240 | The role of conceptual persistence and analogy in the development of BQs

Tab. 17: Adjectives intensified by la mar de in Present-Day Spanish Adj-cluster

Adjectives observed

NICE

simpático (4) ‘sympathetic’, bueno (2) ‘good’, amable ‘nice’, cariñoso ‘loving, affectionate’

HAPPY

contento (4) ‘happy’, alegre ’happy’, felices ‘happy’

BEAUTIFUL

mono (4) ‘cute’, colorada (2) ‘red’, guapa ‘beautiful’

5.3.3 Concluding remarks The emergence of new QNs as well as the extension in the N2-pattern of individual QNs reveal the dominance of conceptual images over highly frequent prefabs or lexical items. In addition to actual token frequency and similarity to previous uses, the analogy to a conceptual image or frame is as important in the creation of new instances. Since the corpus does not return a particularly frequent ‘N1 de N2’-combination, nor a particularly frequent QN per domain, the sanctioning of new instances rather seems to be driven by analogy to a conceptual image. In this context, it seems important to note that the clusters of N2-collocates and the frame of the respective QNs are typically hard to separate. In the case of pila de, for instance, several clusters are centered around the concept ‘stackable’. In the case of aluvión de, the N2-clusters are completely unrelated unless when taking the source semantics of N1 into account. The same conclusion was drawn for the early stages of montón de’s development (cf. Chapter 4). Moreover, it bears pointing out that the clustering of N2s is particularly obvious in the early binominal uses of QNs. If only the Present-Day data of montón de were taken into account, the data would not invite to think in terms of collocational clustering or analogical thinking with its original frame: its conceptualization has already further schematized than the image imposed by QNs that only started to develop functional uses two centuries ago (and generally present rich source semantics). Interestingly, generalizing over the combinatorial patterns of all QNs, a minimal shared constraint can be observed: a particular QN only combines with those collocates that fit, or that can be made to fit (via contextual cues) the original frame of N1. The latter observation throws up an important theoretical question: does the originally clustered structure of the N2-pattern per QN necessarily lead to synchronic combinatorial restrictions?

The role of persistence and analogy | 241

With regard to the role of co-selection patterns in GR, existing accounts of collocational change have mainly focused on its triggering role.131 Typically, the node-collocate combinations involved in processes of GR are characterized by an increase and (morphosyntactic and/or semantic) diversification of the collocates (Bublitz 1996; Vandewinkel/Davidse 2008). This ‘triggering’ perspective on collocational extension is first applied to BQs by Brems (2010). Recall that she distinguishes two pathways of GR. The first path, leading to ‘pure quantifier’ uses such as a bunch of suits/studies/brain-dead louts, combines with a more or less unrestricted variety of concrete as well as abstract nouns. In contrast, the second path, leading to ‘valuing quantifier’ uses such as a bunch of lies/morons/ idiots, combines the initial extension – do note that the right collocate abstracts away from things growing or fastened together such as grapes – with a collocational ‘re-clustering’ towards nouns with a clearly negative semantic prosody. The newly emerged restriction makes her argue that the valuing uses of a bunch of “thrive on the sharing of affective or emotively coloured values between the nodal SN [KV: Size Noun] and its collocates” (Brems 2010, 97–98). She continues to argue that the shared semantic prosody is not a restriction resulting from GR but rather a “facilitating factor” in that it prompts or promotes emergent uses. In addition to the facilitating role of collocational patterns, Ghesquière (2011, 86) argues for a complementary, yet restrictive role. Building on the casestudies of intensifying adjectives and verb particles, De Smet/Ghesquière (2010) show that collocates can also closely interact with the lexically specific meaning of their nodes. The GR process endowing completeness adjectives, such as complete in the complete establishment, with noun-intensifying functions is restricted to “specific sets of nouns with the required semantic properties, viz. degree nouns inviting open-scale or closed-scale intensification of inherent evaluative or scalar notions” (De Smet/Ghesquière 2010), such as a complete fellow/astonishment/nightmare. In other words, in its noun-intensifying uses, complete engages with the specificities of the noun it modifies. This type of collocational restriction thus constrains generalization and explains why “the new uses remain fairly infrequent despite their abstract meaning” (Ghesquière 2011, 86). In view of the previous case-studies and the development of montón de, I argue for a strict differentiation between the (rather restrictive) role of N2 and the (facilitating) role of the broad co-selection pattern of the BQ, which encom-

|| 131 See Ghesquière (2011, 85–86, 203–208, 223) for an extensive overview.

242 | The role of conceptual persistence and analogy in the development of BQs

passes morphosyntactic, semantic and pragmatic features.132 Intuitively, it seems implausible that a collocational pattern suddenly extends and in doing so triggers a shift in meaning. It is more plausible to reverse the order: QNs can only combine with new N2s, incompatible with the literal use of the QNs, after the ground has been prepared by subtle semantic changes in the constructional node or QN itself.133 Consequently, I situate the facilitating or triggering role rather in the semantic-pragmatic developments of the QN and the context of usage it typically appears in. Recall that for montón de, for instance, its typical reference to heaps of harvest, tithes and war booty in contexts where people explicitly highlighted the size of their heap of achievements foregrounded the quantifying potential of montón de (thereby gradually abstracting away from the original heap of harvest). It is only after language users feel this quantifying information to be the central information in montón de N2 that collocational extensions towards new, initially incompatible N2s can emerge. At the same time, it is important to note that the majority of the QNs analysed in this chapter lack this crucial pre-grammaticalization stage of productive literal use and immediately have a quantifying interpretation when they occur in the binominal construction. In those cases, the notion ‘emergence’ of the collocational shape seems more accurate than its ‘extension’. In preview of the constructional network model developed in the next chapter, it can be noted that the schematic binominal construction itself has become associated with the quantifying interpretation. In other words, in the cases of the QNs without a productive head use, the semantic ground has been prepared at the level of the abstract construction. Whether a particular QN ends up as a productive quantifier then depends on the generalizability of the original usage contexts in which it shows up. The question is then whether there is any role left for the collocational clusters observed in this section. There is indeed a restrictive role for this type of combinatorial pattern. The restriction does not so much hold for the distinction

|| 132 See also Delbecque/Verveckken (2010, 71–72) on the differentiation between three levels of conceptualization: (a) the linguistic item itself, (b) elaboration by the immediate linguistic context or host class and (c) actualization by the “mise en discours”. The semantic and morphosyntactic properties of N2 thus pertain to the level of elaboration. The morphosyntactic, semantic and pragmatic features of the broader linguistic context form part of the actualization. 133 See also De Smet’s (2009) view on reanalysis, which he suggests to break down into three more basic mechanisms: (a) analogy or categorial incursion, (b) gradual category-internal change and (c) automation.

The role of persistence and analogy | 243

between the types of uses however.134 Instead, the collocational clusters show that new N2s are semantico-conceptually constrained in that they have to fit the frame or conceptualization evoked by the QN. Yet in the light of the conceptual persistence of the QN’s frame, the clustered structure of the N2-pattern no longer surprises. Further, the observed clusters are of conceptual nature: within the cluster of ‘cattle’ for hatajo for instance, cabras, vacas, ovejas and puercos are not near-synonyms. Similarly, asesinos, borrachos, gandules, rameras, etc. are clustered under the hyperonymic or umbrella notion of ‘unpleasant people’, without being near-synonymous. The different members of a particular cluster are thus not necessarily semantically near, but they share a particular conceptual feature that can be conceptually linked to N1’s source semantics. In other words, the clusters of N2s observed for each QN are no accidental combinatorial restrictions. Instead, they indicate the part of the QN’s initial frame the right collocates belong to or highlight, and in this sense, the clusters are somewhat predictable. Against this background, the fact that the observed clusters are generally stable over time becomes less surprising, and asks for rethinking the notion of collocational extension, as it appears that the phenomenon is to be mainly situated in incipient GR. The conceptually clustered structure of the N2 co-selection pattern thus provides further evidence for the hypothesis that ‘analogical thinking’ and ‘conceptual persistence’ are the fundamental mechanisms of semantic change in GR: the emergence of new N2-combinations occurs via analogical thinking to a part of the QN’s conceptual image that persists.135 Attributing a facilitating role to conceptual persistence, which is traditionally considered the outcome or an epiphenomenon (Hopper 1991) of GR might seem contradictory. However, I consider the possibility to exploit different parts of the QNs original frame as its creative potential and as the motor of its productivity (cf. infra, Chapters 7 and 9). The collocational clusters show where GR “hesitates” (De Smet/Ghesquière 2010), why many items and constructions engaged in GR generally do not immediately nor necessarily end up as markers of a purely grammatical function.

|| 134 Recall that for aluvión and letanía, the collocational clusters of the quantifying and the specifying use fully overlap, for hatajo de the overlap is large. The major diversification in the collocational pattern thus resides in the difference between the literal use and the functional ones. 135 When operationalizing ‘persistence’ as a ‘mechanism’, it is useful to highlight that at some point ‘analogical thinking’ and ‘persistence’ overlap. The fact that facets of the original frame shade through is a subtype of analogical thinking, while analogical thinking is traditionally characterized as alignment towards a construction that already exists.

244 | The role of conceptual persistence and analogy in the development of BQs

5.4 Conclusion In this chapter, I traced the development of seven QNs, viz. pila de, aluvión de, letanía de, hatajo de, barbaridad de, mogollón de and mar de, and measured to what degree their individual histories correspond to the pathways of change followed by ‘prototypical’ montón de. The short answer to that question is that only pila de and hatajo de follow the standard cline from head to quantifying (and/or specifying) uses. I argued that conceptual persistence and analogy play a crucial role in the GR of BQs and raised them to the level of ‘mechanisms’ of change: both are the underlying motivation and precondition for change. While the tendency of the grammaticalized uses to highlight several facets of the QN’s original frame motivates the morphosyntactic behavior of BQs and the clustered structure of their combinatorial pattern, analogical thinking appears to be the guiding principle in the emergence of new QNs and N2s for specific QNs. Recall that, in general, new right collocates are conceptually near to already existent N2-combinations and that the clustered structure of the N2-pattern is framebased. The role attributed to persistence and analogical thinking in the development of BQs nicely fits the constructional network elaborated in Chapter 6, where both ‘mechanisms’ will be claimed to be the motor of the construction’s productivity. The historical case studies have two major theoretical implications for the GR research. First, in contrast to recent studies (a.o. Heine 2002; Hilpert 2010) who claim that history can be reconstructed from synchronic variation, the development of BQs show that is imperative to include diachronic data for sound historical claims. Although the synchronic layering of the BQs (see Table 2 in Chapter 3) looks like the self-evident outcome of GR via the cline ‘prepartitive > partitive > degree modifier’, it only constitutes a snapshot of the diachronic process. The reconstruction of the QN’s history on the basis of synchronic gradience is only partially true for BQs. Although the synchronic data indeed suggest a prototype role for montón de and show a difference in morphosyntactic bonding between literal and functional uses, these hypotheses do not do full justice to the unique histories of the QNs. The conclusions drawn on synchronic data also overlook the lack of the pre-grammaticalization stage (or coextensive head use) of several QNs, the relative recentness of the GR, the recentness of la mar de’s adjective intensifier use and the importance of analogy and persistence. In other words, the case-study of BQs seems to lend support to Traugott/Trousdale’s claim that micro-changes “need to be theorized as much as macro-changes” (2010, 23), since macro-changes do not occur without prior micro-changes. On these grounds, I will try to develop a diachronic model in

Conclusion | 245

Chapter 6, that – in contrast to the GR cline mentioned above which idealizes over many small changes – can reconcile the shared (macro-)GR with the individual (micro-)histories of the QNs in a more plausible reconstruction. The current findings also reinforce the view that the actualization level or “mise en discours” should also be taken into account in diachronic analysis.136 Studies in GR are generally quite reserved as to the role played by semanticconceptual factors in triggering/constraining functional shifts (cf. Paradis 2011, 71). Since semantic-conceptual criteria appear to be less easy to measure objectively, they are often discarded as less valid. Without further differentiation, meaning shifts in GR are simply categorized as ‘desemanticization’ or ‘subjectification’, and, if any facets of the original lexical item shade through, the notion of ‘lexical persistence’ is invoked.137 However, the case-study on BQs shows once more that morphosyntactic adjustments to the newly acquired function can lag behind the shifts in meaning. If only morphosyntactic criteria would be taken into account, the gradual confinement of the determiner-to-N1 pattern and the restriction to N2s without determiner would be the only signs of increasing structural bonding. The remaining GR criteria (Lehmann 1995; Himmelmann 2004) simply do not (yet) apply. An exclusively morphosyntactic approach would for instance fail to motivate why the context expansion towards adjective intensifier uses of la mar de are restricted to inherently gradable notions which specify looks, character, intellect and state. Nor would the formal account insist on the difference in combinatorial pattern between un mar de N2 and la mar de N2: as the latter is less restricted, the functional shift towards intensifying adjectives becomes less dramatic however. Finally, if the diachronic analysis would generalize over the actualization level and source semantics of BQs, it would ‘misclassify’ lots of N2s as collocational extension whereas, in fact, they continue to be frame-based. For instance, Section 5.3.2 pointed out that the N2s com|| 136 See also García (2009) on the non-arbitrary nature of syntax and the imperative to start linguistic analysis from the “communicative burden of utterances” (2009, 291). In the same vein, Paradis argues that “change proceeds from non-conventionalized mapping between lexical items and their readings, construed through a particular focus of attention that is contextually motivated, i.e., change proceeds through metonymization” [emphasis KV] (2011, 70). See also Boye/Hader’s (2012) new, usage-based definition of GR in terms of shifts in “discourse prominence”: the central idea is that “grammar is constituted by expressions that by linguistic convention are ancillary and as such discursively secondary in relation to other expressions – and that grammaticalization consists in the diachronic change that leads to such expressions.” (2012, 2). 137 See also Paradis (2001, 62) on the dearth of research on the “mechanisms of meaningmaking and change covering the whole continuum from contentful through grammatical to pragmatic meanings”.

246 | The role of conceptual persistence and analogy in the development of BQs

bining with letanía de share the frame-related facet of ‘being pronounced out loud repetitively’. Concrete nouns such as vinos ‘wines’ would traditionally be considered as collocational extension, for not constituting chains of discourse. However, the particular context una letanía de vinos appears in pictures a waiter reciting a long list of types of wine. In sum, provided that the actualization level is taken into account, similar occurrences can be shown to meet the required semantic property.

6 Theoretical reflection: a constructional network model of the development of BQs While Chapter 4 provided a detailed overview of the development of highfrequency montón de as a typical example of GR, Chapter 5 focused on the emergence of several relatively less productive QNs. The fact that they only present us with a partial mirror image of montón de’s history makes the role of structural analogy and the claimed GR process less straightforward. The present chapter will try to reconcile the triggering role claimed for montón de with the individual development of every single QN, thereby showing that the case-study of BQs supports Traugott/Trousdale’s hypothesis that a variationist and constructional approach to gradience and gradualness may be preferable (2010, 39). A constructional network will therefore be posited (see also Verveckken 2012). In a nutshell, the four most prominent lines of theory-building are associated with Fillmore et al. (1988), Goldberg’s Construction Grammar (1995; 2006), Langacker’s Cognitive Grammar (1991) and Croft’s Radical Construction Grammar (2001), respectively. These constructional accounts differ in the focus of their theoretical framework and in the perspective they adopt on language production and change. Depending on how formalist, how usage-based, how cognitive, how general and typologically universal its orientations, each construction grammar advances a specific version of the three basic principles: (1) any symbolic entity or form-meaning pairing is a construction; (2) lexicon and grammar form a continuum of constructions, differing only in their degree of schematicity and complexity; (3) the speaker’s knowledge of a language forms a network of intersecting constructions.138 The synchronic definition of ‘construction’ suggests that a constructional change a priori concerns the “coevolution of form and meaning” (Bybee et al. 1994, 24) and that changes in one particular node, viz. construction, of the network may affect other neighboring ones. Although the case-study of BQs challenges the strict simultaneity of semantic and formal changes (see also Traugott 2008, 221; 2011), the constructional account provides a useful model for reconciling analogy with abstract, schematic constructions on the one hand and language use as the actual locus of change on the other. This is achieved by distinguishing different constructional levels of realignment (Traugott 2008). To count as change, the (local) reanalysis has to

|| 138 See Croft (2007), Hoffmann and Trousdale (2011) and Schönefeld (2006) for more extensive literature reviews on construction grammar.

248 | Theoretical reflection: a constructional network model of the development of BQs

become widespread. Eventually, long strings of analogical extensions may then affect the schematic construction.139 The notions of gradience and gradualness refer to variation as a synchronic and diachronic phenomenon respectively. While gradience concerns the prototypical nature of grammatical categories, and the fuzziness of their boundaries in particular, gradualness presents change as a sequence of discrete micro-steps affecting various aspects of the use and structure of a construction (Traugott/ Trousdale 2010, 22–23), which with repetition, can lead to macro-effects. The gradient and gradual nature of morphosyntactic variation and change is stressed by those linguists who are unwilling to accept that language users abruptly introduce radically new categories or meanings (cf. De Smet 2009, 1751). The ongoing debate justifies the importance attributed to analogical thinking, i.e. the recognition of formal or functional similarities between two or more constructions, and to persistence, as commented on in the previous chapter. Both motivations for change fit the constructional model elaborated in the present chapter. The structure of the present chapter is as follows. Section 6.1 briefly surveys the challenges that the case-study on BQs entails for the common cognitive, constructional and usage-based views on GR. Section 6.2 scrutinizes the explanatory potential of existing theoretical accounts and case-studies, and the diachronic network model in particular. Section 6.3 shows which refinements to the diachronic network model may be posited to arrive at an adequate and complete diachronic account of the development of the individual BQs and the BQ-construction in general. Section 6.4 argues that, in spite of the high degree of formal persistence, the GR claim can be maintained and determines at which constructional level exactly the process or ‘macro-change’ occurs. Section 6.5 winds up the final concluding remarks. As such, the present chapter does not add new empirical data to the diachronic analysis, yet critically examines the generalizability of existing accounts. As one of the arguments in support of the constructional network model is synchronic in nature, this chapter bridges the diachronic and the synchronic

|| 139   Cf. Traugott’s definition of innovation and change: “Speakers and hearers presumably match parts of constructions. By hypothesis, an innovated construct-token may be matched to any one or more layers of a construction-type. If the innovation is replicated and conventionalized by other speakers, stronger integration with the multiple layers of a construction-type may occur. This integration may lead to change: the development of a micro-construction-type that is partially aligned with meso-construction-types and ultimately with hierarchically high-level functions (…).” (2008b, 36).

The challenges for the diachronic account of BQs | 249

parts of this volume. It goes without saying that the constructional network model can only be promoted if it proves to be synchronically valid as well.

6.1 The challenges for the diachronic account of BQs 6.1.1 The complex frequency pattern The case-study on BQs challenges the usage-based account on GR in various ways. The latter approach stipulates that only high frequency items engage in GR (Bybee 2003, 602), that the process of GR leads to increased frequency (Bybee 2003, 607) and that the dramatic rise in frequency in its turn leads to automation and the exemplar based structure of categories (Bybee 2006, 726). The triggering role of high frequency in GR – put more carefully, the restriction towards items displaying relatively high token frequency – relates to the process of desemanticization or generalization of meaning that Bybee considers fundamental in GR. According to the usage-based approach advocated by Bybee (2003, 606), a stimulus loses its impact if it occurs very frequently (i.e. via the mechanism of habituation). The loss of semantic force and specificity correlates with an “increase in the contexts in which the gram may be appropriately used” (2003: 605),140 which recalls Himmelmann’s (2004) set of ‘expansion’ parameters. Although I firmly believe that generalization of meaning precedes the construction’s expansion towards new contexts, the case-study on BQs calls into question the precondition of high token frequency. On the one hand, it is beyond doubt that in the QNs montón and mar, high token frequency and context expansion coincide. While the causal relation between high frequency and generalization of meaning applies to the development of montón de, which continues to have a productive literal use in Present-Day Spanish, the history of mar de seems more subject to the conserving effect of frequency: until the emergence of la mar de in the second half of the 19th century, (un) mar de generally continues to combine with the same type of N2-clusters as if these combinations were accessed as single chunks and their memory representations were strengthened with repetition.141 In other words, the generalization of meaning || 140 Note that Bybee argues that generalization or bleaching means that “specific features of meaning drop off, leaving a semantic core” (2003, 607). 141 Bybee (2006, 714–715) identifies three major frequency effects on processing and storage: (1) the reducing effect, (2) the conserving effect and (3) autonomy. The reducing effect relates to the tendency of “high-frequency words and phrases to undergo phonetic reduction at a faster rate than low- and mid-frequency sequences”. The conserving effect relates to the tendency of

250 | Theoretical reflection: a constructional network model of the development of BQs

appears to be negligible until la mar de shows up, which not only abstracts away from the typical N2-clusters the original, highly metaphoric uses of mar combined with, but also extends the N2-slot to other grammatical categories, i.e. adjectives and adverbs. However, the majority of the QNs analyzed in the previous chapter lack a (productive) head use, which excludes the conventional pathway from highly frequent literal uses over generalized or schematized literal uses to extended, i.e. quantifying, uses. The QNs aluvión, letanía and barbaridad indeed gradually expand to new N2-combinations over time, but their GR is certainly not triggered by generalization of meaning. In sum, the case-study of Spanish BQs confirms the correlation claimed by Bybee between high frequency and advanced GR (in terms of context expansion). However, high frequency is definitely not a precondition for a shift (of low-frequency QNs) towards a quantifying reading. In addition to the ‘triggering’ role frequency might have in GR, the usagebased account stipulates that the “grammaticalizing construction’s frequency of use increases dramatically as it develops” (Bybee 2007, 965), since one source of the rise in frequency is “the increase in the type of contexts in which the new constructions is possible” (2007, 965). In this regard, type and token frequency are equally important. The distinction between token and type frequency reflects the difference between the “frequency of occurrence of a unit” (Bybee 2003, 604, e.g. the total number of occurrences of coextensive [montón de] in CREA is 1294) and the number “of different lexical items with which a construction can be used” (Bybee 2003, 605, e.g. the number of possible N2combinations with coextensive [pila de] in Medieval Spanish is 6). According to the usage-based assumption, standard GR processes thus involve an increase in frequency of both types: A grammaticizing phrase is thus said to increase in generality (…) as the contexts in which it is appropriate move from very specific to more general. A much-noted property of grammaticizing constructions is this increase in type frequency of co-occurring lexical items. As a consequence, the token frequency of units such as going to or gonna also increases dramatically. As important as the increase in type frequency or generality is, it is the high token frequency of grammaticizing phrases which provides the triggering device for many of the changes that occur in the form and function of the grammaticizing construction. (Bybee 2003, 605)

|| “high-frequency sequences to become more entrenched in their morphosyntactic structure and resist restructuring on the basis of productive patterns that might otherwise occur” (2003, 715). Autonomy or automatization refers to the loss of semantic and syntactic transparency of originally morphologically complex strings, since, with repetition, they become autonomous from etymologically related forms and are more easily accessed as a whole.

The challenges for the diachronic account of BQs | 251

However, the case-study of BQs again invites to handle with care the frequency assumptions associated with standard GR processes. Table 1 shows per QN and per time period the absolute token frequency of the coextensive binominal constructions and their frequency per million words. Table 1 reveals that the development of all QNs indeed displays a rise in token frequency. While this increase is only moderate for aluvión de, letanía de and hatajo de, the frequencies of use of montón de and mar de increase dramatically, with a factor of 5 and 4 respectively. The spectacular rises in frequency of pila de (with a factor of 14) and of barbaridad de (with a factor of 6) also strike the eye. However, the qualitative analysis has shown that pila de has only limited quantifying uses – viz. 6 quantifying occ. in Present-Day Spanish – and that barbaridad de is a low-frequency item. It bears pointing out that this section does not include a detailed picture of the possible simultaneous rise in type frequency, since such a frequency table would be misleading. Section 5.3.2 of the previous chapter has revealed that all QNs indeed gradually combine with more types of N2s. However, except for the host class and context expansions observed for montón de and mar de, the ‘new’ N2-combinations are proven to be frame-based or conceptually motivated provided the ‘mise en discours’ is taken into account. The continuation of the QN’s source semantics mitigates the rise in type frequency of N2-combinations. Finally, although available online corpora are not suitable for a detailed frequency analysis of the schematic BQ-construction, the category search in the Corpus del español per century provides sufficient support for the assumption that the GR of Spanish QNs entails an increase in type frequency, without necessarily resulting in increased token frequency for low-frequency QNs. Later in this chapter I will argue that any noun with scalar implicatures can be made to fit the BQ-construction. Note that Martí Girbau (2010, 229) also suggests that quantitative expressions “can easily be created with almost any noun that can have the meaning of a container, (…) even with a noun such as autobús ‘bus’” as in ha llegado un autobus de turistas alemanes ‘a bus of german tourists has arrived’. I attribute the productivity of the construction to the CIP, from which it derives creative potential.142

|| 142 Interestingly, for English and Dutch, two poets (Lipton 1991 and Battus 2003) precisely explored the borders of the BQ-construction, listing combinations such as an exaltation of larks, a string of violinists, a gloss of philologists, a conjunction of grammarians, etc. and een lading elektronen ‘a charge of electrons’, een hap eters ‘a bite of eaters’, een bosje bomen ‘a copse of trees’, een ketel heksen ‘a cauldron of witches’, etc. Those hapax legomena should not be confused, however, with particular BQs which have developed systematic quantifying uses.

252 | Theoretical reflection: a constructional network model of the development of BQs

1250–1450 Medieval Spanish 1450–1730 Classical Spanish 1730–1900 Modern Spanish 1900–1975 Present-Day Spanish

n # n # n # n #

61 2,72 251 2,74 366 8,78 616 14,26

– – 6 0,07 12 0,29 43 1

– – – – 18 0,43 30 0,69

– – 12 0,13 6 0,14 11 0,25

– – 9 0,09 18 0,43 18 0,41

– – – – 1 0,02 5 0,12

Mar de

Barbaridad de

Hatajo de

Letanía de

Aluvión de

Pila de

Montón de

Tab. 1: Absolute token frequency (n) and frequency per million words (#) per QN per period

– – 158 172 208 4,99 288 6,67

Summarizing, the usage-based approach predicts that high frequency leads to generalization of meaning, which might lead to context expansion or GR, which in its turn might lead to increased frequency. In addition to the reducing and conserving effects of repetition on language processing, frequency is claimed to have an impact on the cognitive representation of language (Bybee 2007; Haiman 1994): With repetition, sequences of units that were previously independent come to be processed as a single unit or chunk. This repackaging has two consequences: the identity of the component units is gradually lost, and the whole chunk begins to reduce in form. (Bybee 2007, 969)

Again, the development of BQs questions the GR involved, as it runs counter to the tendency towards automatization generally associated with straightforward cases of GR. Except for the fixed combinations una barbaridad de N2 and la mar de N2/ADJ/ADV, the BQs continue to display (limited) variation in their coselection pattern: in addition to determiner variation, modifying adjectives can be inserted (both immediately preceding N1 or postmodifying N1). Further, no strict reduction in form is noted, although compositionality is partly lost: the sum of ‘herd’ and ‘fools’ in un atajo de ignorantes, does not express the idea of a miserable bunch of stupid people in (1). The observation that it is the entire sequence from determiner over QN to preposition which is replaced by nada más que ‘nothing but’, or by a qualifying adjective such as tontos ‘stupid’, provides further support for the claim that the BQ is analyzed as a whole or as a single chunk quantifying/specifying N2. However, the tendency towards formal

The challenges for the diachronic account of BQs | 253

as well as conceptual persistence characterizing the development of BQs implies that the automatization effect is rather limited. (1)

Digo que todos los que me oyen son un atajo de ignorantes, empezando por ti. (1888, B. Pérez Galdós, Miau) ‘I am saying that all those who are listening to me are a bunch (lit. herd) of fools, starting with you.’ (1)’ Digo que todos los que me oyen no son nada más que ignorantes. (1)’’ Digo que todos los que me oyen son ignorantes tontos.

One final effect of increased frequency in standard GR is the exemplar based structure of categories, claimed by Bybee (2006, 711). The frame-based structure of the N2-clusters, as described in Section 5.3.2, questions once more the process of GR in the development of BQs. In sum, exception made of the high-frequency montón de and mar de whose development conforms to the frequency-effects generally associated with GR, the overall low frequency of the other QNs analyzed which have obviously adopted functional uses questions the nature of the process of change involved.

6.1.2 The formal and semantic persistence The second major challenge for existing accounts on GR consists in reconciling the tendency towards formal and semantic persistence with the functional reading of BQs. The continuation of the string’s morphosyntactic structure and original frame of course violates the automatization principle. In the description of the histories of the individual QNs, the semantic changes were taken as a starting point since they are chronologically prior to the formal changes. This procedure does not imply that GR is to be equated with semantic reanalysis or that semantic change is considered sufficient to detect ongoing GR. Yet, since the particular morphosyntactic make-up associated with the quantifying and specifying readings emerges gradually and rather late, the GR of BQs is less prominent from a structural point of view,143 to the extent that up to Present-Day Spanish, the binominal syntagm remains highly analyzable:

|| 143 As Brems (2010, 99–100) correctly points out, the “availability of attestable formal reflexes of a reanalysis depends on the kind of construction looked at”. In the case of BQs, the specific outlook (i.e. the binominal syntagm) remains largely unaltered. However, since stringent restrictions rely on the combinatorial pattern of the QN, the formal persistence does not necessarily threaten the GR involved.

254 | Theoretical reflection: a constructional network model of the development of BQs

no phonological attrition has occurred (as in English alotta or helluva), and functional uses can still co-occur with adjectival collocates (e.g. un inmenso racimo de casitas blancas ‘an immense bunch of little white houses’) or are attested in the plural (e.g. montones de gente llenaron las calles ‘heaps of people filled the streets’). This formal persistence or typical noun features shining through in the grammaticalized use raise fundamental questions about the headness of N2. However, the noun-ness of N1 is only partial, for three reasons at least.144 First, only intensifying adjectives can modify a quantifying or specifying QN. The grammaticalized instances of aluvión de, for instance, exclusively combine with grande ‘big’, verdadero ‘authentic’, enorme ‘enormous’ and rico ‘rich’ in CORDE. In addition to referring to the size-related configurational characteristics of N1, intensifying adjectives in functional uses can highlight a particular facet of the QNs original frame, such as proceloso ‘stormy’ and inquieto ‘restless’ in combination with mar de. Particularly recurring combinations with all QNs are grande ‘big’ and verdadero ‘real’: whereas the former intensifies the quantity assessment by the QN, the latter ironically stresses the authenticity of the QN, as in Recibí un verdadero aluvión de anónimos (1876) ‘I received a real flood of anonymous letters’. Second, generalizing over all QNs, the combination with the indefinite determiner is the unmarked option, unless determiner variation is motivated for discourse purposes (e.g. definite determiners only combine with topicalized instances). Quantifying or specifying uses allow determiner variation as long as the same determiner would be used when the QN is replaced by mucho/a(s) ‘many’ or simply omitted, as in (2). Finally, pluralization of the QN is marked and only montón de appears to have a productive plural alternative (cf. Section 7.2.4, infra).145 As examples (3) and (4) illustrate, the plural morpheme yields its own conceptualization. In fact, the existence of different || 144 According to RAE (2009, 826), quantifying nouns present a defective syntax. 145 Although mar de is not included in the synchronic dataset, its high frequency status urges to verify whether, similarly to montones de, mares de has quantifier uses in Present-Day Spanish. Nine occurrences are observed where mares de is used as a Q, of which 5 combine with tinta ‘ink’. In all nine, the plural morpheme yields a proper conceptualization: in the following, the value of the plural morpheme is aspectual and relates to the fact that in history (in contrast to hoy), various people at various moments have spilled ink on this matter. Sobre la primacía de teoría sobre la práctica y viceversa, se han gastado mares de tinta. Hoy, al contrario de otras épocas, prevalece la exageración practicista, con lo que se obtienen buenos obreros, pero rutinarios y mediocres maestros. (unknown, oral) ‘Seas of ink have been spilled on the primacy of theory over practice and vice versa. At present, in contrast to other eras, the practical exaggeration prevails. Consequently good workmen are obtained, but (only) routine and mediocre masters’.

The challenges for the diachronic account of BQs | 255

groupings of N2s is presupposed and the groups of N2s need not be spatiotemporally contiguous: in (3), e.g. the problems faced vary according to the groups that pose them (socios versus aficionados); in (4), successive arrivals of foreigners are envisaged (japoneses and estadounidenses). Using a singular form instead of a plural (un montón de extranjeros ‘a heap of strangers’) would suggest that they all arrive at once. (2)

Los celos le parecían tan estúpidos como injustificados. No dejaban, sin embargo, de halagarle; y optó por callar, discretamente, ante el aluvión de dicterios que se le venía encima. (1933, H. Rodríguez, ¿Quo vadis, burguesía?) The jealousy seemed to him as stupid as unjustified. He couldn’t help being flattered, however; and he decided to keep silent, discretely, in front of the flood of insults that came upon him.’ (2)’ Optó por callar, discretamente, ante los dicterios que se le venían encima. (2)’’ Optó por callar, discretamente, ante los ?muchos/numerosos dicterios que se le venían encima.

(3)

– Con este tema he tenido montones de problemas. Los socios del Atlético me han llamado de todo, los aficionados me han increpado. Lo he pasado bastante mal. (1982, press) ‘I have had heaps/lots of problems with that theme. The members of Atlético have called me a lot of things, the fans have reprimanded me severely. I went through a rough time.’

(4)

“Nos llegan montones de extranjeros, sobre todo japoneses y estadounidenses. Todos sienten muchísima curiosidad por ver si quien está dentro del ataúd es realmente Doña Josefa y no un muñeco”, señaló. (1989, press) ‘ “Heaps/lots of strangers arrive here, mainly Japanese and Americans. All are very curious to see whether the one who is lying in the coffin really is Doña Josefa and not a doll”, he commented.’

In sum, the formal features which persist in the quantifier and specifier uses are highly constrained. The fact that no additional absolute quantifier can be added after de and that no qualifying adjective can be added to N1 (5), provides further evidence that un/a QN de is analyzed as a single chunk. More importantly, as has been repeatedly observed in the previous chapter, the morphosyntactic behavior of the BQs appears to be conceptually or discourse pragmatically motivated (see also Chapter 8).

256 | Theoretical reflection: a constructional network model of the development of BQs

(5)

Recibí un alud de llamadas. ‘I received an avalanche of calls.’ Recibí un alud de *tres llamadas. ‘(…) an avalanche of *three calls.’ Recibí un alud *local de llamadas. ‘(…) a *local avalanche of calls.’

The latter observation brings us to the tendency towards conceptual persistence that clashes with the expected generalization or bleaching of meaning generally associated with processes of GR. The diachronic analysis has revealed the crucial role of conceptual persistence in the syntagmatic as well as paradigmatic expansion of the BQ: the extension towards new QNs in the N1-slot is conceptually driven and the clustered structure of the N2-combinations is frame-based. Synchronically, the impact of conceptual persistence is twofold. On the one hand, conceptual persistence guarantees the distinctiveness of the conceptualizations evoked by the individual QNs. Iconically, since the QN comes first and already yields a typical configuration of a mass or group of entities, N2 simply adjusts its conceptualization to the image activated by N1. In (6) for instance, alud de ‘avalanche of’ obviously does more than only quantify the hypocrisies: it suggests that the speaker feels overwhelmed, as if snowed under by a literal avalanche. (6)

El griterío es ensordecedor. Me siento sepultado y cautivo en un alud de hipocresías, estrategias, tácticas, vilezas y bajas pasiones. (1994, press) ‘The shouting is deafening. I feel submerged and captive in a lot/ avalanche of hypocrisies, strategies, tactics, vile acts and low passions.’

As (part of) the original image may continue (or come) to be highlighted, each QN thus imposes its own conceptualization on N2. As a consequence, the N2 gente ‘people’ repeated from (7) to (10) is conceived of differently each time: whereas in (7) the people keep on moving in a confused, noisy yet commanded way, they are conceptualized as simply brought together without any structuring principle in (8). In (9) they have been orderly recorded in an appointment diary, whereas in (10) they are presented as vivid persons suddenly appearing and invading the house. (7)

El PRÍNCIPE se halla acostado y duerme. Arde la chimenea. Súbitamente irrumpe en su cámara un tropel de gente capitaneado por el propio REY, que se cubre la cabeza con un casco y lleva debajo del jubón de raso negro una coraza; (…). (1980, theatre) ‘The PRINCE is lying down and asleep. The chimney is burning. Suddenly a mob of people burst into his room, commanded by the KING

The challenges for the diachronic account of BQs | 257

himself who is covering his head with a helmet and wearing a cuirass under his doublet of black satin; (…).’ (8)

Usted mismo ha mencionado antes los liberales, después está la ultra derecha, la ultra izquierda, los anarquistas, un montón de gente. (unknown, oral) ‘You yourself mentioned the liberals previously; then there is the extreme right-wing, the extreme left-wing, the anarchists, a lot of people.’

(9)

Aproveché para mirar mi agenda, cosa que rara vez hago y, como siempre que la examino, me di cuenta de que había un montón de cosas que debía haber hecho esa semana que no hice y una pila de gente a la que debía haber llamado y no llamé, con lo que he quedado fatal para siempre. (1990, novel) ‘I took the opportunity to look at my diary, something I rarely do and, as always when I check it, I realized that there were a lot of things that I should have done this week which I hadn’t done, and a lot of (lit. pile of) people that I should have called (but) who I didn’t call, to the extent that I made a terrible impression forever.’

(10) A partir de las 9.30 de la noche un aluvión de gente importante comienza a invadir la casa. (1988, books) ‘From 9:30 p.m. onwards a flood of important people started to enter the house.’ On the other hand, the conceptual persistence prevents QNs from being mutually interchangeable without (at least slightly) altering the conceptualization of N2. For instance, in (7), tropel de can be replaced by aluvión de (profiling an even larger number of people that continuously enter the room) and montón de (profiling a large, possibly heterogeneous and chaotic mass of people), yet the modification capitaneado por el propio Rey makes tropel de the only coherent choice. For the same reason, except for the general intensifying adjectives such as grande ‘big’ or verdadero ‘real’, QN-specific intensifiers such as incontenible ‘uncontrollable’ for alud or proceloso ‘stormy’ for mar are not susceptible of combining with other QNs. It bears pointing out, however, that the correlation between conceptual persistence and morphosyntactic behavior confronts us with a chicken-and-egg problem: is the phenomenon of conceptual persistence responsible for the collocational restrictions or do Present-Day instances of the BQs display a certain degree of lexical persistence as they systematically combine with a particular set of N2s? A major argument supporting the primary role of conceptual

258 | Theoretical reflection: a constructional network model of the development of BQs

persistence is iconicity (García 2009, 14; Van Langendonck 2007, 411; Wierzbicka 1985, 334): N1 is mentioned first in the flow of speech and thus processed before N2 is even uttered. In addition, conceptual persistence can motivate the formal persistence observed in the corpus data. Still, the tendency towards persistence throws up the question whether the meaning shift towards quantity assessment involves a process of GR. The ‘grammaticalized’ BQ is certainly morphosyntactically constrained, yet not completely devoid of noun-ness. In preview of the extensive argumentation on this issue in Section 6.4, I agree with Bybee that: A model of grammaticalization in which the only change is that lexical meaning is lost or bleached cannot account for all the changes that are documented. Clear cases exist in which meaning is added into grammaticalizing constructions through pragmatic inferencing. (2007, 976)

Although change may result in bleaching, bleaching cannot be the mechanism in itself. Following Bybee (2007), Traugott (1991) and Paradis (2011), I argue that pragmatic inferencing is a key mechanism in creating semantic change in GR. In concrete communication, i.e. the speaker-initiated process of strategic choices, the speakers can decide “to say less than he or she means” (Bybee 2007, 976; also Grice 1975) and to leave part of the meaning to be inferred by the addressee.146 When the association of the inference with a particular item or construction is frequent enough, the inference becomes conventionalized as part of the meaning of the construction. Situating innovation in the context of speakerhearer negotiation of meaning reserves a major role for the (discourse) context. The mechanism of pragmatic inferencing, which is usually characterized as a metonymic extension implies that, at least initially, part of the original meaning of the QN is continued. I would tentatively argue that the conceptual persistence inherent in the mechanism of inferencing motivates the productivity of the BQ-construction. Precisely the facets of meaning that shade through in the quantifying readings of BQs is what distinguishes this construction from genuine (absolute) quantifiers (e.g. mucho/a(s) ‘many, much’, tres ‘three’, etc.). Since BQs yield an individ|| 146 In the argumentation elaborated in Chapter 7, I will take up the idea put forward by Traugott (2003b) that ‘subjectification’ is to be seen as a subtype of ‘pragmatic strengthening’: “Invited inferencing is a kind of conceptual metonymy within the speech chain (…), since it is primarily associative in character, being derived from the other uses to which interlocutors put linear sequences of utterances and associations in contexts. In the case of subjectification, the new polysemies are those in which the speaker’s perspective is an essential element.” (2003b, 634).

The challenges for the diachronic account of BQs | 259

ual conceptualization of the group of entities expressed by N2, the BQ-construction provides the Spanish native speaker with a useful tool for expressive quantification. The fact that the pragmatic or communicative function of BQs hinges on conceptual persistence, leads to the hypothesis that QNs are not likely to become completely bleached quantifiers like montón de.147

6.1.3 The exceptional role of (un) montón de N2 Although the previous chapter insists on the individuality of the history of all QNs analyzed, the course of the development of three QNs, viz. montón de, (la) mar de and barbaridad de catches the eye. Starting from the cognitive grammar assumption that grammatical categories are prototypical categories (Langacker 2008, 103; Tuggy 2007, 90; Vázquez Rozas 2007, 17), montón de can be considered to be the most typical member and to function as a model for the other members of the QN-category.148 Synchronically, montón de’s high frequency stands out. Table 1 shows that in Present-Day Spanish, the coextensive uses are about two times as frequent as those of mar de, i.e. those of the other high-frequency item of the QN-paradigm, and fourteen times as frequent as the most productive low-frequency QN, i.e. pila de.149 However, in line with Geeraerts (1988, 211), I argue that frequency is primarily a result of prototypicality:

|| 147 See also Van de Velde F. (2011, 407) who argues that the entrenchment of a construction can lead to variety or complexification of the construction, rather than to fixation. 148 According to Geeraerts (1994, 45), prototypicality is itself a prototypical concept. Four characteristics of prototypical categories are frequently mentioned (Geeraerts 1994, 45–56; Lewandowska-Tomaszczyk 2007, 145; Rosch 1975): (a) absence of classical definitions (i.e., a single set of necessary and sufficient attributes); (b) clustering of overlapping senses (i.e., a family resemblance structure); (c) degrees of representativity; (d) absence of clear boundaries. Note that these criterial features of prototypicality need not co-occur. 149 Accordingly, montón is the most frequent QN (as a lexical item) of the ‘heap’ cluster (cf. Section 5.3.1), which is corroborated by the category-query in the Corpus del español. It could therefore be argued that montón constitutes a counter-example to the tendency of clustering around a conceptual image and that the other QNs of the ‘heap’ cluster are attracted by analogy to a prefab. Even if this were the case, the reasoning still fails to explain why the ‘heap’ cluster is not the most productive one in creating new instances of the BQ-construction (see Bybee/Eddington (2006) on exemplar representation and ‘prefabs’). Consequently, I continue to consider the QN-category as a prototypical rather than exemplar construction that is framebased.

260 | Theoretical reflection: a constructional network model of the development of BQs

We can use linguistic frequencies to determine what instances of a concept are prototypical (…), but explaining prototypicality on the basis of linguistic frequency is putting the cart before the horse. Some kinds of usage are not prototypical because they are more frequent; they are more frequent because they are prototypical.

It goes without saying that two more reasons can be invoked for characterizing montón de as the most typical member of the BQ category. First, it is the only chameleon-like QN, whose semantic content is so low that it can easily substitute for all the other QNs, although the reverse is not the case. The examples in (11) reformulate the examples (7)–(10). While the substitution of the respective QNs by montón de only leads to a slightly altered conceptualization in (11b), the other reformulations sound rather awkward since the conceptualization activated by montón de clashes with several contextual cues: in (11a), the chaotic nature of montón de conflicts with the sudden appearance of and the hierarchy within the group of people; in (11c), the static nature of montón de is incongruous with the inchoative aspect and the dynamic characterization of the entities. Similarly, in view of the heterogeneity and disordered conglomeration of people in (11d), which is echoed in montón de, tropel de and pila de are not quite appropriate, aluvión de is inappropriate for the dynamicity it implies. (11) a. Súbitamente irrumpe en su cámara un montón de gente capitaneado por el propio REY. b. Aproveché para mirar mi agenda, me di cuenta de que había un montón de cosas que debía haber hecho esa semana que no hice y una montón de gente a la que debía haber llamado y no llamé. c. A partir de las 9.30 de la noche un montón de gente importante comienza a invadir la casa. ? d. Usted mismo ha mencionado antes los liberales, después está la ultra derecha, la ultra izquierda, los anarquistas, un(a) tropel de / aluvión de / pila de gente. The second major argument supporting montón de’s prototype-nature relies in its pioneer role: it was the first QN to systematically exploit its quantifying potential. At the beginning of the 15th century already, unambiguous quantifying uses of montón de are found, whereas the development of quantifying uses of the majority of the QNs does not take off before the second half of the 16th century (see Figure 2 of the previous chapter). However, the first quantifying use of mar de also shows up rather early – in 1508 – and the time interval between the first specifying uses attested for montón de and mar de is negligible (in 1479 and 1485 respectively). Since in the Classical and Modern occurrences of mar de, several facets of mar’s source semantics generally shade through and mar de

The challenges for the diachronic account of BQs | 261

more or less continues to combine with the same type of right collocates, the generalization of montón de’s meaning is further advanced – at least until the string la mar de shows up in the second half of the 19th century. It can therefore be assumed that montón de is a more typical member than mar de. The pioneer role and prototype status of montón de warrants the assumption that the development of montón de might have served as an attracting force, as a model for the GR of the other QNs.150 The lack of semantic specificity of montón de in Present-Day Spanish adds up to the evidence for considering the GR process of montón de as completed to a great extent. With regard to the context expansion parameter proposed by Himmelmann (2004), montón de indeed appears to be the most grammaticalized QN. By the 15th century montón de had already turned into a largely unrestricted quantifier through host-class expansion: it combines with concrete (12a) as well as abstract nouns (12b) and with count (bound, [12c]) as well as mass nouns (unbound, [12d]). (12) a. un montón de {libros/pan} ‘a lot of {books/bread}’ b. un montón de {problemas/impresiones} ‘a lot of {problems/impressions}’ c. un montón de {personas/finalidades} ‘a lot of {persons/aims}’ d. un montón de {gente/basura} ‘a lot of {people/rubbish}’ Its syntactic context got extended so as to include adverbial uses as well, as is illustrated in (13) (where substitution by another QN sounds awkward). In fact, even when used as an uncomplemented noun, montón frequently reminds of the quantifying reading either by instantiating the idiomatic expression ser del montón (14) or by serving as a general measure expression (15). (13) Éramos amigos y nos queríamos un montón. (1994, press) ‘We were friends and loved each other a lot.’ Nos queríamos *un aluvión/*un alud/*un hatajo/*una letanía/*la mar/ ? una barbaridad…. (14) Eres una mujer del montón, sin nada extraordinario y bastante boba, (…) (1990, theatre) ‘You are an ordinary/average/run-of-the-mill wife (lit. of the heap, mass), with nothing extraordinary and quite silly, (…)’

|| 150 Several native speakers have commented (during discussions of earlier drafts of this manuscript at international conferences) that they intuitively consider montón to be the most general and typical QN, in addition to mogollón which is confined to informal communication.

262 | Theoretical reflection: a constructional network model of the development of BQs

(15) He aquí algunos de los textos enarbolados durante la marcha: “Sin agua, sin exención, pero ratas un montón”; (…). (1979, press) ‘I list some of the texts which were displayed during the march: “Without water, without exemption, but a lot of rats [lit. rats a lot]”; (...)’ In contrast, the other QNs do not systematically activate a quantifying interpretation when used without PNP. The case of barbaridad deserves special mention however, in the first place since it has undergone similar syntactic context expansion: in (13), una barbaridad could technically speaking function as degree modifier and replace un montón, though its source semantics clashes with the predicate quererse; but in (16) and (17), the astonishment of the speakers makes una barbaridad the best fit. Further, una barbaridad de combines low token frequency, morphosyntactic fixedness and the restriction to quantifying and specifying readings. Equally remarkable is the chunk la mar de, which exclusively combines with quantity assessment: its N2-pattern seems unconstrained and la mar de has developed adjective and adverb intensifier uses. However, since not a single QN – of my dataset at least – has copied the preference for the definite determiner, nor the extension towards adjective/adverb intensifying uses, it would be hard to maintain a triggering role for (la) mar de. (16) ¡Dieciocho horas! ¡Casi dieciocho horas dormido! – Musitó: “Qué barbaridad”. (1980, novel) ‘Eighteen hours! You have slept for almost eighteen hours! – “What an incredible lot/What a(n) barbarity, atrocity”, she whispered.’ (17) A mí me extrañó una barbaridad. (1975, novel) ‘That amazed me a lot.’ The extreme ease with which montón de seems to expand its host class as well as its syntactic and semantic-pragmatic context (in contrast to the other QNs) is at first sight incongruent with the tendencies which can be derived from Table 2, viz. the distribution of uses per time period. In terms of relative frequency, the other QNs systematically present a higher proportion grammaticalized uses. Again, I am not inclined to consider this observation as a threat to the triggering role claimed for montón de. Attributing an advanced stage of GR to those QNs on the basis of the relatively high frequency of their quantifying/specifying uses disregards the absence of increased token frequency and advanced context expansion. A more plausible explanation resides in the original source semantics of the QNs: whereas aluvión de, letanía de, (h)atajo de and barbaridad de are characterized by a rich lexical content which is hard to desemanticize (cf. the degree of conceptual image persistence observed), the literal meaning of

The challenges for the diachronic account of BQs | 263

montón de does not go beyond ‘a group/collection of things put one above another without order’ (according to DRAE). In other words, when used literally, the N2-complement is generally necessary in order for the hearer to know what (kind of) heap the speaker is talking about. For instance, if a speaker mentions last week’s alud, the hearer automatically understands that reference is made to a recent snow slide. In contrast, in (18), without additional contextual clues, the hearer does not know whether the speaker means a heap of paper, a heap of books, a heap of litter, a heap of work, etc. The semantic specificity of the remaining QNs might also motivate, diachronically, the absence – or at least scarcity – of head uses in their development. In other words, the main observation should be that the other QNs have relatively less literal uses than montón. (18) Tengo un montón en mi mesa. ‘I have a heap on my desk.’ Tab. 2: Relative frequency (in %)151 of functional uses per QN and time period

Montón de Pila de Aluvión de Letanía de (H)atajo de Barbaridad de Mar de

1250–1450

1450–1730

1730–1900

1900–1975

Q 0.13 – – – – – –

Q 0.18 – – 0.83 0.20 – 0.39

Q 0.12 – 0.78 0.83 0.11 – 0.60

Q 0.19 0.14 0.63 0.73 0.06 0.60 0.71

S – – – – – – –

S 0.11 – – – – – 0.41

S 0.14 – 0.17 – 0.83 1 0.19

S 0.11 – 0.20 0.09 0.56 0.40 0.11

In sum, several arguments can be adduced for attributing a triggering or model role to montón de. However, the interaction between analogical thinking and conceptual persistence repeatedly alluded to in the previous chapter fails to explain how the other, highly specific, QNs can change by analogy to montón de if the latter prototype displays only limited conceptual persistence. In Section 6.3, I will argue that a diachronic constructional approach is apt to account for the complex frequency pattern, the apparent lack of automatization of lowfrequency QNs and the exceptional role of montón de.

|| 151 100% corresponds to the total number of uses per QN per time period: e.g. in Medieval Spanish, only 13% of the occurrences of montón de were quantifying uses, in Classical Spanish, 83% of the instances observed for letanía de were quantifying uses, etc.

264 | Theoretical reflection: a constructional network model of the development of BQs

6.2 The explanatory scope of existing accounts The knotty issue of partial productivity of a construction has been theorized both in in research on GR and in constructional approaches. Starting from the assumption that schematic syntactic patterns are inherently meaningful (Langacker 2008; Delbecque/Cornillie 2007; Goldberg 1995; 2006), synchronic constructional approaches correlate low-frequency of constructions with the concept of ‘coercion’. Coercion refers to the construal whereby the constructional meaning overrides the lexical meaning, as in She sneezed the napkin from the table, where the inherent intransitive reading of the lexical item to sneeze conforms to the meaning of the caused-motion structure in which it is embedded (Goldberg 2006; 2009a; Michaelis 2004; Rice 1987). In terms of coercion, montón de is a typical member of the BQ-construction, and the quantifying/specifying reading of a low-frequency QN can be invited by coercion. That position has much in its favor, but it requires that the low-frequency QNs be considered untypical members of the QN-category, which is obviously not the case: Present-Day Spanish has a wide variety of lowfrequency BQs. In addition, the assumption that the binominal construction merely coerces a quantifying reading might overshadow the impact of the original use and source semantics of N1. To account for the GR of low-frequency complex prepositions such as by dint of, Hoffmann (2004) distinguishes between pure frequency of occurrence or ‘lexical frequency’ and ‘conceptual frequency’ or ‘saliency’. Conceptual frequency is “the frequency with which a particular concept or speech event occurs” (2004, 191), which is more difficult to assess empirically since a single concept can be realized “in a great number of different and formally unrelated ways” (2004, 191). In line with the usage-based view on language storage, he argues that: [i]f a particular form covers the majority of realizations of a particular concept, it is the preferred choice of expression, even if its absolute frequency is relatively low. (2004, 191)

It might thus well be that the unproductivity of low-frequency QNs such as letanía de simply mirrors the low conceptual frequency of the meaning intended. Hoffmann observes that the complex preposition-construction, i.e. preposition-noun-preposition sequence such as in spite of, generally behaves as a single word preposition and thus displays a great level of ‘cohesiveness’ (2004, 199). Interestingly, the low-frequency complex prepositions are also characterized by an overall dearth of adjectival modification and the absence of a determiner before the nominal element. In an attempt to reconcile the discrepancy between relatively frequent complex (e.g. in front of, in view of) and lowfrequency complex prepositions (e.g. by dint of, in distinction to, by analogy to)

The explanatory scope of existing accounts | 265

with the abstract, text-organizing property they all share, Hoffmann suggests that “low-frequency combinations grammaticalize by analogy to their more frequent ‘structural relatives’” (2004, 195), which do seem to have developed gradually. Interestingly, this proposal minimizes the relevance of the specific original context in which GR occurs, thereby challenging the current cognitive grammar view on GR. Instead, he claims that GR results in “the establishment of constructional schemas whose slots can be filled with suitable lexical items” (2004, 195). In other words, on the basis of the structural and semantic similarities between the high-frequency complex prepositions, language users distill an abstract complex preposition, whose structural slots can be creatively exploited. Regarding English small size noun constructions (henceforth SSN-constructions), Brems (2007a) invites to cautiousness in describing what served as analogical source for what. With the exception of a bit of, SSNs are uncommon, both within and outside the construction and they “individually display very specific polarity restrictions as well as collocational restrictions and semantic prosodies, i.e. preferences for either positive or negative collocates” (Brems 2007a, 295).152 While whiff of and smidgen of show strong preference for positive polarity, (not) a skerrick of, (not) a jot of, (not) a scintilla of, (not) a shred of, (not) a scrap of, (not) a flicker of are restricted to negative polarity.153 The fact the highly frequent bit of is polarity-independent and does not incorporate specific collocational properties nor semantic prosody leads her to argue that mere structural analogy to a frequent model that did grammaticalize gradually cannot account for the development of SSNs into quantifiers. She suggests that more complex analogies with quantifiers incorporating polarity sensitivity such as some/any are at stake. However, SSNs differ from the fairly unrestricted some/any by specific collocational and pragmatic constraints. In an attempt to combine the distant modeling with partially polarity-dependent quantifiers and the distant modeling with the polarity independent schematized quantifier construction, Brems distinguishes an analogical model at an intermediate level of schematicity (i.e. Traugott’s meso-level, cf. infra) that is polarity-dependent.154

|| 152 Another major distributional difference between Hoffmann’s complex prepositions and Brems’ SSNs (and the Spanish BQ by extension) is the fact that several high-frequency complex preposition can coexist while SSN-constructions and binominal constructions have only one or two highly-frequent models (a bit of and montón de(/mar de) respectively). 153 Brems’ interpretation of polarity sensitivity builds on Israel’s definition of polarity items as “a class of constructions which do not themselves express negation or affirmation, but which are restricted to sentences of one or the other polarity” (Israel 2004, 701–702). 154 From the context, it can be derived that ‘distant modeling’ refers to the fact that the SSNs only partially copy the morphosyntactic and/or semantic characteristics of both models.

266 | Theoretical reflection: a constructional network model of the development of BQs

In line with Hopper/Traugott’s (2003) claim that GR is non-deterministic, she argues that these infrequent SSN-constructions “will not go on to fully macroconstructional status, precisely because of the very specific expressive purpose they serve as quantifying expression”. In sum, she partially supports Hoffmann’s claims on analogy and frequency yet goes on to argue that “grammaticalization as such directly works on and results in (at least partially) substantive constructions, rather than schematic ones” (2007a, 293). Brems’ proposal to distinguish multiple source-constructions and different levels of schematicity in the development of low-frequency SSN-constructions is inspired by Traugott’s constructional network model which frames the development from partitive constructions into degree modifiers in terms of analogical thinking at different constructional levels. Typically, high-frequency items or schematic constructions are claimed to attract local reanalyses of morphosyntactically similar constructions by analogy. In order to account for similarly behaving subgroups of low-frequency items, intermediate constructional levels are posited. In line with the distinction made in grammaticalization theory between (1) schemas or pathways of GR, (2) generalized change-types, (3) specific change-types and (4) the empirically attested tokens, Traugott (2007; 2008a; 2008b) distinguishes between the constructional levels of (1) macro-constructions, (2) meso-constructions, (3) micro-constructions and (4) constructs.155 In her approach to the development of degree modifiers in English, the notion of macro-construction encompasses high-level schemas and is “the highest level relevant for the discussion at hand” (e.g. partitive construction, degree modifier construction). The meso-construction level refers to “sets of similarly-behaving constructions” (e.g. the set a bit/lot [of], the set (a) kind/sort of, etc.). Micro-constructions are “individual construction-types” such as a lot of and a bit of which are instantiated by constructs, i.e. “empirically attested tokens of micro-constructions” (Traugott 2007, 525).156 Crucially, the construct-level is the level of actual || 155 Note that in her analysis of the development of degree modifiers in English, Traugott (2008b, 31) is primarily concerned with reconciling the partially shared pathway ‘prepartitive > partitive > degree modifier/quantifier (> degree adverb > free adverb)’ by (a) kind of, a bit (of) and (not) a shred of with their individual histories and diverging source construction. In addition, her account provides a suitable model for (synchronically) limited productivity of constructions. 156 Traugott uses the constructional levels both in a synchronic and diachronic sense. With the term macro-constructional level, for instance, she refers both to the cline from ‘prepartitive’ to ‘free adverb’ and to the separate steps of this cline (e.g. the superordinate partitive construction or degree modifier construction, etc.). Since she does not feel the need to explicitly comment on the synchronic or diachronic nature of the different levels, the constructional network model seems a natural candidate to account simultaneously for synchronic and diachronic variation.

The explanatory scope of existing accounts | 267

language use – hence the locus of innovation and change – and tokens or instances of a particular construction are “pragmatically rich” (Trousdale 2010, 5). Traugott’s hypothesis on the development from partitives into binominal degree modifier constructions is “that the grammaticalization occurred initially at the micro-level, individual micro-construction by individual micro-construction; as the new micro-constructions came to be aligned with each other new meso-constructions arose, in turn affecting the constituency of the macro-construction, and that all these changes share characteristics associated with grammaticalization, especially increased entrenchment or routinization” (Traugott 2007, 544). In other words, the sequence of micro-changes followed by the individual partitive constructions can be summarized as follows: a. They came to have their own characteristics as micro-constructions, e.g. a shred of came to co-occur mainly with unbounded nouns, a bit/lot of with both bounded and unbounded nouns. b. Some came to be unified with meso-degree modifier constructions, e.g. a bit with very, kind of, sort of with respect to adjective collocations (this is selective analogy since we find a bit hungry, but not *a lot hungry, only a lot hungrier). c. Together with quite and all with nominal heads, they became models to which other degree modifiers can be analogized (very fun). d. The macro-degree modifier construction was therefore expanded (this is reanalysis of the macro-construction, allowing analogical developments at lower levels. (Traugott 2007, 540–541)

Trousdale’s (2008b; 2010) analysis of some non-standard English binominal constructions explicitly builds on Traugott’s constructional model. Apparently, hell of (Californian English) and gobs of (informal American English) differ from the conventionalized pathway of GR form partitive into degree modifier construction, which is followed by lots of for instance. Hell of – also helluva – is particularly interesting because of the unusual source construction, viz. the evaluative construction (e.g. a hell of ugly devils (1591)). Through pragmatic inferencing and analogical thinking associated with hell’s scalar implicatures – note that hell scores extremely high on a scale of badness –, the micro-construction a hell of a N2 has become “aligned with other ex-binominals (e.g. a lot of, a shred of) as instances of the degree modifier construction” (Trousdale 2010, 14) and syntactic reanalysis occurred. The development of gobs of is remarkable for its limited regional spread, in spite of its advanced stage of GR (or loss in compositionality; e.g. digesting great gobs of statistics, they make googobs more money), both from a semantic and morphosyntactic point of view. In line with Traugott’s model (2007), he distinguishes four constructional levels and adds a distinctive feature to each: macro-constructions can be distin-

268 | Theoretical reflection: a constructional network model of the development of BQs

guished from each other by semantic-pragmatic features; micro-constructions differ from each other by formal features and the in-between meso-constructions are characterized by partially shared meaning and form (Trousdale 2010, 2). Following Traugott, he argues that the very first quantifier deriving from a binominal structure (in Middle English) modeled on the (abstract) absolute quantifiers. In other words, when a deal of NP was first reanalyzed as ‘much NP’ (instead of ‘a portion of NP’), the mismatch was resolved by associating the BQ to “an abstract Quantifier Construction” (Trousdale 2010, 6). With successful repetition of a deal of expressing ‘much’, “what was initially a very loose connection between a construct and a very general macro-construction comes to be crystallized as an entrenched micro-construction” (Trousdale 2010, 6). His analysis of two regional BQs sheds light on two issues which are particularly relevant for the constructional network of Spanish BQs. First, he addresses the question of why some partitive constructions spread in some dialects but not in others. The analysis of hell of suggests that this micro-construction does not spread since the “the evaluative macro-construction is too idiosyncratic a construction to serve as a source for the development of further grammatical constructions” (Trousdale 2010, 19), while much theorizing on GR argues that the input to GR is rather general (see also Brinton/Traugott 2005, 99). In addition, not all micro-constructions “invite the necessary inferences for realignment” (Trousdale 2010, 20). Second, his analysis focuses on the question why not all (potential) partitives expand their contexts to the same extent. As some partitive constructions develop new collocational restrictions, “the associations between elements of the constructions become too tight” (Trousdale 2010, 19) and consequently block further growth. For instance, bunch is not susceptible to appear in the construction that was a bit/lot/*bunch better, since this QN has become restricted (after the initial broadening of its N2-pattern) to “collocations of the kind what + a load/bunch of + [negatively evaluated noun, esp. nonsense, rubbish, etc.]” (Trousdale 2010, 19). He goes on the argue that similar collocations are ‘hybrids’ between lexical and grammatical constructions: while load is more ‘lexicalized’ in what a load of rubbish, it is more ‘grammaticalized’ in it’s certainly loads better. I will come back on this observation in Section 6.4. In sum, the constructional network model stands the best chance to account for the case-study of BQs. In view of the almost systematic tendency towards formal and semantic persistence in the development of the BQs analyzed, it cannot be the case that the binominal construction would simply ‘coerce’ a quantifying reading (the more since the binominal construction has also developed a wide variety of not coextensive readings, e.g. the partitive construction in la rebanada de pan ‘the slice of bread’, the possessive construction in el libro de Juan ‘the book of Juan’, the expressive binominal construction in un desastre

The constructional network model revisited | 269

de película ‘a disaster of a movie’, etc. (cf. Traugott 2008a and Massini Fc. on the English and Italian binominals respectively)). Further, the abstract construction distilled from highly frequent complex prepositions whose structure is simply copied by low-frequency constructions, as suggested by Hoffmann, contrasts with the individual and rather different histories of the BQs analyzed, which generally gradually extend their collocational pattern. The present case-study corroborates Brems’ caution in determining the model for realignment. While she suggests multiple sources in the development of SSN-constructions, viz. negative polarity items such as some and any in addition to the highly frequent polarity-independent a bit of, Spanish BQs do not have similar polarity restrictions. I will argue in the next section that a slightly refined version of the constructional network model proposed by Traugott (2007) can assess the development of Spanish binominals.

6.3 The constructional network model revisited In line with Traugott’s (2007, 541) constructional network model, and the distinction between four constructional levels in particular, the constructional changes which take place when BQs arise out of QNs can be summarized as follows: Step 1: pragmatic quantitative implicatures arising from individual QNs (or constructs) in the flow of speech in certain linguistic (binominal) contexts – via analogical thinking to an abstract quantifier construction simply meaning ‘much’, or at least, to existing absolute quantifiers – allow for the semantic change into quantifiers from the moment enough people replicate the construal in enough contexts; Step 2: when the semantic shift into quantifier is replicated micro-construction by micro-construction via pragmatic inferencing and analogical thinking, morphosyntactically similarly behaving micro-constructions become aligned with each other as instances of a meso-construction (e.g. the avalanche-set of similarly behaving alud de/aluvión de/avalancha de); Step 3: the binominal syntagm undergoes syntactic reanalysis (from [det. N1 [de N2]] into [[det. N1 de] N2]); Step 4: with repetition, sets of meso-constructions first give rise to a schematic BQ-construction at the macro-level and eventually strengthen the abstract ‘N1 de N2’ pattern or affect its internal structure. This layered model permits us to reconcile the partially shared history of QNs (which all shifted from head to quantifier/specifier) with the variety of devel-

270 | Theoretical reflection: a constructional network model of the development of BQs

opmental paths followed by individual QNs as micro-constructions. It further allows low-frequency items to grammaticalize, or at least to become re-aligned as instances of the BQ-construction, by analogy to a set of similarly behaving micro-constructions, i.e. to a more abstract meso-construction. However, the case study on the development of BQs in Spanish requires some further refinements of this constructional network model, particularly in view of the prototype-function of montón de, the high conceptual image persistence of BQs and the productivity of the quantifying uses of expressive binominal constructions. Figure 1 visualizes the application of the constructional network to the development of Spanish QNs. The symbol used for constructions is borrowed from Croft (2001): the construction internal symbolic link ties formal properties (upper rectangle) to conventional semantic properties (lower rectangle). The dashed arrows symbolize the sanctioning of a concrete instance by a more abstract construction; the dotted lines sketch the alignment, with repetition, of concrete occurrences into a more schematic, overarching construction. The abbreviations ‘Q’ and ‘CIP’ refer to ‘quantity’ and ‘conceptual image persistence’ respectively.

macroPartitive QN1-construct(1)

Construction

(det) QN1 de N2

QN1-construct(2) Q + CIPQN1 QN1-construct(x) micro-QN1

QN2-construct(1)

(det) QN2 de N2

(det) QN de N2

Q + CIPQN2

Q + CIPQN

QN2-construct(2) QN2-construct(x) micro-QN2

QNn-construct(1)

meso-QN

(det) QNn de N2

QNn-construct(2) Q + CIPQNn QNn-construct(x)

micro-QNn

macroAbsolute Quantifier

Fig. 3: The constructional network of Spanish QNs according to Traugott’s model (2007)

The constructional network model revisited | 271

6.3.1 Revisiting the analogical model(s) The previous chapters have provided plenty of evidence supporting the advanced stage of the GR process montón de is involved in as well as its head start in comparison to the other QNs analyzed. Since montón de was the first QN to exploit its quantifying inferences, there must have been a mismatch between the binominal structure of un montón de N2 and the invited quantifying reading ‘many N2’ (instead of ‘a heap of N2’). According to Trousdale (2010, 6), the mismatch is resolved by pattern matching to an abstract Quantifier Construction, since “[t]he best the hearer can do is to create a link to align the meaning of the construct with the meaning of other, more abstract, extant constructions in the network” (Trousdale 2010, 11). However, there is no need to invoke an abstract quantifier construction to resolve a mismatch: in examples such as (19) – (17) in Chapter 3, repeated here for convenience –, the pragmatic inference of quantity is triggered by several contextual clues, viz. the specifying grant, the fact that the speaker questions the use of informing the hearer, etc. Consequently, I think it is somewhat of an overstatement to argue that the Medieval hearer, whether consciously or not, aligns monton de parauvlas to an abstract quantifier construction, let alone specific absolute quantifiers such as muchas ‘many’, when the quantifying reading associated to monton de is already inherent in and prompted by the (discourse) context. (19) Pues que assi es, ¿pora que traballare de aplegar grant monton de paraulas en contarte la manzilla de aquesta muyt suzia et pudient amistança et conueniencia de pactos que los romanos la hora firmaron con los sannites? (1376, J. Fernández de Heredia, Traducción de la Historia contra paganos, de Orosio.) ‘Because that is how it is, why try so hard to gather a big heap of words while telling you about the dishonor of that very dirty and powerful friendship and the convenience of pacts that the Romans then signed with the Samnites?’ Likewise, both Traugott (2008a; 2008b) and Brems (2010) allow structurally unrelated constructions to be the target construction in the reanalysis from binominal structure into quantifier. While Traugott claims that a shred of and a bit of become unified with very, kind of and sort of at meso-constructional level and, with repetition, imply the expansion of the macro-degree modifier construction, Brems suggest that the low-frequency polarity-dependent SSNs are unified with a polarity-independent meso-construction which is aligned to both negative-polarity quantifiers such as some and any and the high-frequency

272 | Theoretical reflection: a constructional network model of the development of BQs

structural relative a bit of. To account for the emergence of the first Spanish BQ, however no alignment to abstract macro-quantifiers has to be posited. With regard to the development of the Spanish BQ-construction, I propose the following sequence of micro-changes: Step 1: The quantifying potential of montón de is highlighted in specific (discourse) contexts. Since the association with the quantifying reading becomes frequently exploited from the 15th century onwards, a new formmeaning pairing arises, i.e. the association of montón de with the meaning ‘a lot of, many’ at micro-constructional level. As there are no other QNs yet which systematically exploit their quantifying potential, there is no reason to posit an abstract construction.157 Step 2: Although the dramatic increase in functional uses of the highly frequent montón de N2 in combination with its early loss of compositionality might allow montón de to function as a model for other less frequent QNs (e.g. letanía de) from the beginning of the 16th century, the first functional uses of the other QNs are always contextually embedded, as if triggered independently of montón de.158 Step 3: When the quantifying inferences of the other QNs are sufficiently repeated, they become crystallized as separate micro-constructions as well. Since new QNs are generally conceptually similar to existing BQs (cf. Section 5.3.1), they are rather aligned to existing micro-constructions than to a highly abstract ‘N1 de N2’ macro-construction. Yet, every new quantifying binominal micro-construction strengthens this spontaneous association between the binominal structure and the expression of a large quantity and smoothens the emergence of new QNs in the binominal construction. Step 4: From the second half of the 19th century onwards, several QNs, viz. aluvión de, barbaridad de and la mar de in particular, simultaneously develop quantifying uses. In addition, the emergence of the specifying uses of pila de, letanía de, aluvión de and barbaridad de, also dates from the 19th century. On these grounds, it can be plausibly argued that the 19th century is

|| 157 See also De Smet’s (2009, 1751) quote concluding Chapter 4 who argues that the first instance of a grammatical category cannot be analysed as a member of that category before there is a second member. 158 It is empirically impossible to verify whether the GR of montón de initially attracted the GR of the other QNs. However, in view of montón de’s high frequency and general source semantics, it is plausible that early crystallization of un montón de N2 into a quantifying micro-construction at least speeded up the GR of other QNs: thanks to the entrenchment of un montón de N2, the association between the binominal structure and quantity assessment was already less abrupt or dramatic.

The constructional network model revisited | 273

when the different binominal micro-constructions expressing quantity are unified in an abstract BQ-construction (at macro-constructional level), to the extent that in Present-Day Spanish any noun with scalar implicatures may fill in the N1-slot of the BQ to express quantity. In other words, the conventionalization of the form-meaning association between N1 de N2 and quantity assessment, i.e. the emergence of the abstract quantifier construction, does not take place before the end of the 19th century. Because of the high-frequency of quantifying montón de as well as montón’s semantic non-specificity, the early development of montón de has probably speeded up the quantifying exploitations of the other QNs, especially those QNs that lack a productive head-use. However, it is important to bear in mind that the pragmatic inferences of quantity assessment were at first always contextually embedded. In other words, I argue that – at least until the 19th century – the emergence of quantifying readings of BQs originate in concrete usage and is not primarily attracted by some abstract analogical model. This is not to deny that abstract constructions may function as analogical models. On the contrary, from the 19th century onwards, the N1 de N2-structure can ‘coerce’ quantifying/specifying readings of new low-frequency items. Yet, it is not historically correct to posit for the earliest quantifying uses of montón de or other genuine BQs realignment either to absolute quantifiers or to some abstract BQ-construction linked to quantity assessment. Finally, the constructional network model developed so far should not overshadow the fact that, according to the category-query performed on the Corpus del español, montón de was not the first nor the only QN in Medieval Spanish. Given the fact that the binominal construction already existed in Latin as the partitive genitive (e.g. montes auri ‘mountains of gold’, montis argenti ‘a mountain of money’ (Bassols de Climent 1992)), the existence of several quantifying N1 de N2 combinations should not come as a surprise. Still, the majority of these binominal syntagms were used literally and were largely unproductive. Moreover, if a quantifying interpretation was activated, it was triggered by contextual clues (20) or by metaphorical use of N1 (21). As Vergara Wilson (2009, 293) demonstrated that categories may mutate over time – by clusters becoming more centralized or by a shift in the central member within a particular cluster –, these observations should not contradict our hypothesis.159

|| 159 For the same reason, the fact that the diachronic corpora used are flawed in terms of editions (Cornillie 2012, pc.), and consequently that the ‘first’ H, Q or S use of a specific QN is

274 | Theoretical reflection: a constructional network model of the development of BQs

(20) & uio en un campo grand companna de diablos que estauan contando a su princep los males que fizieran por las tierras. & entre todos los otros estaua y uno muy negro a manera de Ethiopiano. (1270, Alfonso X, Estoria de Espanna que fizo el muy noble rey don Alfonsso (…)) ‘He saw in the country a large company of devils who were telling their leader the bad things they did on earth. Among all the others there was one very black [devil], [as black] as an Ethiopian.’ (21) & te sacare de las piedras & te echare en monte de fuego que te quemara. (1280, Alfonso X, General Estoria IV) ‘And I will take you out of the stones and throw you in a mountain of fire which will burn you.’

6.3.2 The explanatory scope of the redefined analogical model The slightly modified form of Traugott’s constructional network offers three major advantages. First, it allows constructs or tokens of low-frequency QNs to be the locus of language change while also recognizing the strength of the link between the binominal pattern and the concept of quantity at a more abstract level. This model differs from approaches that consider the locus of change as an abstract, schematic construction, thus dealing with the knotty issue of lowfrequency items undergoing GR (Hoffmann 2004). It also differs from models focusing on the level of substantial constructions only (see also Noël 2007; Traugott 2003b for a detailed discussion), thus addressing the questions of why some QNs undergo a process of GR while others do not, why some QNs are only used as quantifiers with a particular set of N2s and so on. In contrast to these two approaches, the constructional network model tackles both issues at the same time. In spite of the crystallization of clusters of semantically related N1s at meso-constructional level, the change occurs at the level of the individual constructs, i.e. in the speaker-hearer interaction. The GR of QNs is shown to be initially triggered by the pragmatic inference of quantity by the hearer, and not all QNs invite this inference required for realignment (Trousdale 2010, 10). This observation ties in with the general constructionist claim that the motivation underlying a construction does not equal its prediction (Goldberg 2006, 11): whether a specific QN engages in GR processes and ends up as a pure quantifier || possibly incorrectly considered the first attested one, is not problematic for the constructional network model proposed in this Chapter.

The constructional network model revisited | 275

depends on the highly specific contexts in which it was originally used. Although with repeated use, the binominal construction may have started to impose a quantifying interpretation on single instances from the 19th century onwards, grammaticalized instances of the ‘N1 de N2’-constructions – especially the low-frequency ones – may be constrained by the original head use from which they derive (cf. the combinatorial restrictions imposed on N2 by each N1, the singular concord with N1 even when the construction is clearly used as a quantifier, the constrained modification pattern for N1, etc.). As Traugott’s puts it: “the attractive force of the meso- or micro-construction may be counterbalanced long-term by the backward pull of earlier uses of a micro-construction” (2008, 34). A second advantage of the constructional network model is that the mere existence of a schematic ‘N1 de N2’ quantifier construction justifies the productivity and creative use of the construction (Cano Aguilar et al. 2000, 143): every noun with scalar implicatures, directly or via intermediate pragmatic inferencing, can fill the QN-slot in the BQ-construction, even if the combination is not repeated and therefore thus does not count as change strictly speaking. The combinations made up by Battus (2003) and Lipton (1991, cf. footnote 5) or a random Spanish example I constructed and checked in Google (e.g. (22)) remain at the level of hapax legomena: they do not count as grammaticalized instances of the respective lexical items in QN-position, but they do provide additional support for the GR of the BQ-construction, or at least for the existence of an abstract ‘N1 de N2’ pattern linked to the expression of excessive quantity. (22) El azul intenso del cielo de Madrid, el verde de sus jardines, el colorido de su gente, forman un alegre arco iris de colores que se entremezclan en cada momento y alimentan todos los sentidos con energía. (extracted from Google [“arco iris de colores” site:es]) ‘The intense blue colour of the Madrid sky, the green colour of its gardens, the brightness of its people, form a bright rainbow of colours which mix at any time and feed all the senses with energy.’ In sum, the complex frequency pattern of the development of BQs is not problematic from a constructional point of view. The high-frequency of montón de and its prototype features relate to its general source semantics which might explain its pioneer role in extending to quantifying uses. The partial productivity of QNs such as pila de, aluvión de, letanía de, hatajo de and barbaridad de originates in their specific source semantics which is harder to desemanticize and to combine with new collocations in an unconstrained manner. In addition, once there is a model to align to, partial productivity or low frequency is no longer an issue in construction grammar. While high frequency typically

276 | Theoretical reflection: a constructional network model of the development of BQs

strengthens schematic constructions, low frequency is attributed to the original usage contexts whose specific design blocks the expansion to new contexts. In addition, the lack or marginal nature of literal coextensive uses for QNs such as aluvión de, letanía de and barbaridad de is no longer enigmatic in the constructional model. Finally, the distinction between several degrees of schematicity can shed light on the affinity between BQs and another binominal construction consisting of two noun phrases in a “subject-predicate” relationship (Aarts 1998), e.g. una maravilla de professor ‘a wonder of a professor’, and the quantifying uses of barbaridad de in particular.160 The construction has been given different labels in the literature, ranging from “emphatic apposition” (RAE 2009) or “emotionally charged epithet” (Traugott 2008b) to “evaluative binominals” (Trousdale 2010) and “expressive binominal construction” (Verhagen 2009). In the present study, I will use the latter term, as this type of binominal construction is not restricted to apposition or epithet functions. In the expressive binominal construction, N1 denotes a type of N2 and, consequently, the binominal syntagm as a whole denotes a special kind of entity indicated by N2. Interestingly, this morphosyntactically similar string gives rise to pragmatic inferences of quantification in examples such as un horror de mosquitos ‘a horror of mosquitoes’. Even with more general type nouns (Brems 2010; Mihatsch Fc.), such as una especie de hijo político ‘a sort of son-in-law’ (23), una especie de seems to evaluate the degree to which the referent fits the ‘son-inlaw’-category. Without positing a schematic BQ-construction, or at least, an association between the binominal pattern and the expression of quantity, the linguistic similarities between BQs and expressive binominal constructions would be hard to capture. (23) Tuvo un lugarteniente que era Sila, una especie de hijo político, su hombre de confianza, su brazo derecho; persona de muchas lecturas, gusto refinado, con más relaciones y más mundo que su patrón Mario. (2001, books) ‘He had a deputy who was Sila, a sort of son-in-law, his confidential advisor, his right-hand man; a widely-read person, with refined taste, with more relationships and greater experience than his master Mario.’

|| 160 See also Di Tullio/Kornfeld (2008, 2) who distinguish three types of QNs: (1) intrinsically quantifying notions (e.g. un montón de dinero ‘a lot of money’), (2) qualifying Ns (e.g. una barbaridad de reclamos ‘a barbarity of complaints’, un disparate de dinero ‘a madness of money’, una calamidad de robos y asesinatos ‘a calamity of thefts and murders’), (3) Ns which acquire a quantifying reading via metaphorical and metonymical steps (e.g. un montaña de dinero ‘a mountain of money’, una caterva de locos ‘a bunch of fools’).

The constructional network model revisited | 277

6.3.3 Revisiting the macro-constructional level The constructional account of the development of BQs developed so far has evidently not done full justice to the historical course of events, in that the quantifying (/specifying) uses of BQs tend to exploit their conceptual persistence to a high degree. The latter tendency cannot be copied from the analogical model montón de, since this QN is conceptually underspecified from the start. Nor can the conceptual persistence be explained by the backward pull of earlier uses, as it seems closely tied to the pragmatic motivation of the construction. BQs are pragmatically useful precisely because they provide the speaker with expressive tools for quantification, by distinguishing between different types of quantities.161 As a final refinement to the constructional network model, I claim that the abstract BQ-construction (at meso- or macro-constructional level) that comes to unify the micro-constructions un montón de N2, una letanía de N2 and many others, does not associate the binominal structure un(a) QN de N2 with the expression of quantity, but with a coextensive relationship between both nominal elements. This ‘coextensiveness’ may be more quality or more quantity oriented and builds on the conceptual continuum between quantity and quality assessment. The brief digression in Section 6.3.3.1 and 6.3.3.2 justifies the hypothesized continuum between quantification and type specification as profiled by binominal constructions (and previews the central hypothesis in the synchronic analysis, i.e. the unique conceptual profile each QN conveys). If the underlying conceptualization processes are brought into the picture, BQs and expressive binominal constructions indeed represent two sides of the same coin, which can be dubbed comparison.162 In Langackerian terms (2009a), the comparison of N2 with N1 means that when the speaker maps his/her conceptualization of N2 on his/her prior conceptualization of N1, both conceptualizations fully coincide.

6.3.3.1 Binominal quantity and quality assessment as comparison construals From a cognitive-linguistic perspective, assessing the quantity of a mass, regardless of whether it consists of discrete entities, always involves a process of comparing.163 The canonical method of counting is to compare a group of enti|| 161 Cf. the different groupings of people in examples (7)–(10), according to whether they are bounded by tropel de, montón de, pila de or aluvión de. 162 Cf. Huelva Unternbäumen (2007) on the comparison construal in binominal constructions. 163 Cf. the definition of quantity by Radden/Dirven (2007, 117): “We may understand a quantity of something either relative to a set or relative to a norm on a scale.” [emphasis mine]

278 | Theoretical reflection: a constructional network model of the development of BQs

ties to the well-established unit ‘one’ (cf. Langacker’s (1991) absolute quantification). Yet quantity/size assessment can also be achieved by comparing a mass of entities to a well-established scale (generally linked to a norm, e.g. [too] much/ many, few, some…) or to the conceptualization (in terms of volume/size/ dimensions/configuration/containment...) of another entity (cf. Langacker’s (1991) relative quantification). Interestingly, BQs appear to be the canonical expressions for the latter type of quantity assessment. (24) A las 9 horas, tres médicos del cuerpo de sanidad militar comenzaron a atender las seis líneas habilitadas del teléfono 91-395-54-85. Sin embargo, el aluvión de llamadas desbordó todas las previsiones, por lo que Defensa reforzó el servicio con otros siete facultativos más, (…). (2001, press) ‘At 9 o’clock, three doctors from the military health force started to operate the six telephone lines supplied for 91-395-54-85. Nevertheless, the flood of phone calls exceeded all expectations. That’s why the Department of Defense reinforced the service with another seven physicians, (…).’ In order to assess the size or quantity of the group of entities denoted by N2, the speaker compares the size of N2 to the volume or configuration typically associated with N1 as a lexical item164. In (24), the number of phone calls is conceived of as a literal flood. Important properties of a flood that also correspond to phone calls include its ‘overwhelming’ and ‘unstoppable’ nature. Exactly the same construal can be posited for the assignment of quality to N2 in expressive binominal constructions. In (25), the woman described is categorized as una maravilla ‘a wonder’ for marvelously embodying the ideal woman, in contrast to average women. Crucially, in both examples, N2 deviates from the habitual norm: if the number of phone calls had corresponded to the expected number and if the woman had only been moderately wonderful, the speaker would not feel the need to expressively categorize the calls as a ‘flood’ and the woman as a ‘wonder’ (Leboráns 1999, 2373).165 (25) Hijo, después de lo del aeropuerto, cuando esa fan me ha pedido un autógrafo, esto es lo mejor que me ha pasado desde que he llegado. ¡Esa maravilla de mujer se desnuda cada noche con mi canción! (1995, novel) || 164 See also Doetjes/Rooryck (2003), Den Dikken (2006) on the implicit comparison construal. 165 See also Leonetti (1999: 847) and Leboráns (1999) on the emphatic use of the indefinite article, especially in attributive contexts.

The constructional network model revisited | 279

‘Man, after what happened at the airport, when that fan asked me an autograph, that is the best thing that has happened to me since I arrived. That wonder of a (lit. Ø) woman takes off her clothes every night to my song!’ Whether this mental comparison takes place is hard to verify empirically. I would even argue that, especially in entrenched or conventionalized binominal strings, speakers do not consciously compare two entities. The least we can say is that, while producing/hearing the utterance in (24), the speaker/hearer first processes N1 and keeps this conceptualization of a literal flood in mind. He then goes on to map the group of N2s or ‘phone calls’ onto the literal concept of ‘flood’, thereby highlighting the matching properties (see Figure 2 for an oversimplified sketch). In other words, the conceptualization of N2 is adjusted until it fits the configuration expressed by N1, and conversely, the conceptualization of a literal flood is slightly adapted into a metaphoric interpretation. This mutual accommodation, whether conscious or not, provides a clue to the mechanism underlying the convergence of the BQ and the expressive binominal construction. The actual rationale for bringing both constructions together under one single conceptualization process resides in the ‘coextensiveness’ they share.

t un aluvión

de

llamadas

Fig. 2: Sequence of construal operations involved in un aluvión de llamadas166

|| 166   Figure 2 visualizes the sequence of construal operations by un aluvión de llamadas on a time line. The dashed squares symbolize the boundaries of the profiled masses or groups of

280 | Theoretical reflection: a constructional network model of the development of BQs

An additional argument for considering the two types of binominal constructions as fundamentally intertwined is that they are commonly discussed together in the literature. Quirk (1985, 249–251), for instance, uses partitive as a cover term for both “partition in respect of quantity” and “partition in respect of quality” as instantiated by ‘a bit of N2’ or ‘a sort of N2’ respectively. Similarly, in the recent Spanish reference grammar, the “sustantivos cuantificativos” (or quantifying nouns) and the “sustantivos cualificativos” (or qualifying nouns), which may both occur in the pseudo-partitive construction, are mentioned in the same paragraph (RAE 2009, 823). Lapesa (Cano Aguilar 2000, 143) names una enormidad de gente ‘a vast number (lit. enormity) of people’ an instance of the expressive binominal construction. Further, both Trousdale (2010) and Verhagen underline the requirement for N1 to express “an extreme value on some scale (relevant to the dimension in which the referent is being evaluated)” (Verhagen 2009: 141) when it occupies the N1-slot in the expressive binominal construction. In other words, expressive binominal constructions also convey scalar implicatures. More importantly, the GR cline ‘prepartitive > partitive > degree modifier/quantifier (> degree adverb > free adverb)’ Traugott (2008a) proposes is built on the developmental overlap between kind of, sort of, bit of, shred of: no differentiation between primary quantity or quality assessment is made for the source construction. Finally, the overlap between quantity and quality assessment can also be observed outside the binominal construction. External evidence supporting the continuum between quantification and type specification is found in some evaluative morphemes, which, being originally augmentative, have developed pejorative uses, as in tripón ‘glutton’ (from tripa ‘belly’), barrigón ‘a bit of a paunch’ (from barriga ‘stomach, belly’), santurrón ‘over-pious person’ (from santo ‘holy’), dulzón ‘sickly sweet’ (from dulce ‘sweet’), buenazo ‘kindhearted, softie’ (from bueno ‘good’), etc (Leboráns 1999, 2374). In addition, Paradis (2011) shows that the contentful adjectives awful, dreadful, horrible and terrible – whose expressive content is “based at the extreme end of a scalar structure” (2011, 78), as in a terrible dragon – can give rise to degree reading by foregrounding the scale configuration, as in a terrible bore or a terrible nonsense. Interestingly, the directionality is inverted here: from quality assessment in a terrible dragon to quantity assessment in a terrible bore. Finally, many unrelated African, Asian and Australian languages, can add so-called “classifiers” to nouns, viz. meaningful morphemes that denote a salient characteristic of the

|| entities. The equation symbol mirrors the co-extensive relationship between the mass profiled by alud and the mass profiled by llamadas.

The constructional network model revisited | 281

entity the noun refers to (Allan 1977, 285): interestingly, those classifiers can inform about shape, consistency, size and number – typically ‘measurable’ notions – as well as about gender, visibility, location, possession, etc. (Allan 1977, 297).

6.3.3.2 Co-extensiveness as the motivation of the comparison construal The notion of ‘co-extensiveness’ is borrowed from Langacker (2009a, 64) and refers to the relationship between two entities(/groups/masses) which fully coincide. With regard to the binominal construction, co-extensiveness implies that only one entity or group of entities is profiled by ‘N1 de N2’, as neither N1 nor N2 have full referential value. The difference between the partitive and pseudo-partitive construction (or BQ) lies in the co-extensiveness of N1 and N2 in the latter. In un grupo de los lingüistas ‘lit. a group of the linguists’, both nouns have referential value, as both are preceded by a determiner. (Un) grupo refers to a subset only of the set of linguists actually referred to by los lingüistas. If no determiner strictly precedes N2, as in un grupo de lingüistas ‘a group of linguists’, the binominal string merely invokes the linguists which constitute the group referred to by un grupo, without necessarily conceptualizing a larger encompassing group in the extralinguistic reality. In other words, in un grupo de lingüistas, the mass profiled by grupo and the one profiled by lingüistas are co-extensive with each other (cf. Figure 3, which visualises the contrast between partially overlapping and coextensive N1 and N2).167 Similarly, the expressive binominal construction is said to foreground a quality or feature of N2 “by presenting it as the essence of its holder” (Cano Aguilar 2000: 142, translation mine): in (25), every single aspect of the woman reminds the speaker of the concept ‘wonder’.168 In other words, in the expressive binominal construction, N1 and N2 are considered to fully coin-

|| 167 By way of comparison, generative approaches argue that the difference between partitive and pseudo-partitive constructions relies in the fact that the partitive constructions, as in a slice of John’s pie, is a complex noun, while the pseudopartitive, as in a slice of pie, is a single nominal projection (Stickney 2007). 168 See also Koptjevskaja-Tamm (2003, 539) on the difference in strength of the categorization between a qualitative genitive construction and the adjective compound construction in Swedish: “Thus, one and the same day can be called both en sorgens dag ‘a day of grief’ and en sorglig (ADJ) dag ‘a sad day’ and one and the same person can be called en pliktens man ‘a man of duty’ and en pliktmänniska (‘a duty.man’, COMPOUND), but the former expression in each pair implies a higher degree of unhappiness or duty and raises them to the most salient feature of the day or the man in question.”

282 | Theoretical reflection: a constructional network model of the development of BQs

cide, to be co-extensive: the woman can be categorized as the incarnation of a wonder as the speaker not only considers any minimal and necessary feature of the concept maravilla reflected in the woman, but also considers the woman as the prototypical or most representative example of the maravilla-category (cf. Leboráns 1999, 2374).169 Interestingly, when two entities (or groups of entities) are co-extensive, the focus may be either on the bounded group as one unity or on the constituting entities (Langacker 2011), which is why co-extensiveness may trigger the quantifying interpretation of binominal constructions and the specifying use of BQs by providing the appropriate bridging context. (a)

(b)

grupo

lingüista

grupo

lingüistas

Fig. 3: Partitive (a) vs. pseudo-partitive (b) construal of N1 ‘grupo’ and N2 ‘lingüistas’

Further evidence for the hypothesis that the BQ-construction and the expressive binominal construction share the expression of co-extensiveness can be found in their common Latin ancestor: i.e. the genitive case (Väänänen 1981). Various Latin grammars struggle with the typology of the genitive functions when it comes to distinguishing the partitive genitive from the epexegetic or explanatory genitive (Bassols de Climent 1992). Generally, the partitive genitive introduces the larger set of which the nominal head forms part (26). The explanatory genitive is used to explain or specify the generic notion expressed by the nominal head (27). Interestingly, while the examples in (28) are generally considered instances of the explanatory genitive, they are categorized by Bassols de Climent as a subtype of the partitive genitive (1992, 55). In contrast, Lapesa (2000, 143) considers the Latin explanatory genitive as the source construction of una enormidad de gente ‘a vast number (lit. enormity) of people’.170 (26) multitudeN1 puerorumN2 genitive ‘a crowd of boys’, parsN1 finitimi agriN2 genitive ‘a part of the neighboring territory’ || 169 In Langacker’s (2009a, 6) enumeration of binominal syntagms expressing coextensiveness between N1 and N2, the example that idiot of a president figures among a bar of soap, a mound of dirty clothes, a herd of cattle, etc. 170 See Carlier et al. (2013) on the replacement of the Latin genitive in Romance languages and French in particular by the prepositional phrase headed by de.

The constructional network model revisited | 283

(27) Quis genusN1 AeneadumN2 genitive, quis TroiaeN2 genitive nesciat urbemN1? ‘Who would not know the familyN1 of AeneasN2’ descendants, who would not know the cityN1 of TroyN2?’ (28) flagitiumN1 hominisN2 genitive ‘a shameN1 of a manN2’, scelusN1 viriN2 genitive ‘a crime/sinN1 of a manN2’ The fact that both types of binominal syntagms under analysis can be traced back to a single morphosyntactic marker in Latin, viz. the genitive case, and the continuing effort among Latin grammarians to avoid overlap in the typology of the genitive case suggest that quantity assessment and quality assessment expressed by Spanish binominals originate in one single conceptualization process. More precisely, in both types of binominal, de introduces a relation of coextensiveness between N1 and N2, which can be either quantity or quality oriented. Finally, the fact that at least two typological studies link quantity and quality assessment in the binominal construction can be taken as additional crosslinguistic evidence. First, Koptjevskaja-Tamm (2009, 343) associates the pseudopartitive construction crosslinguistically with marking both apposition and quantification. Likewise, Beckwith (2007) introduces the universal category of ‘phoronyms’, i.e. classifiers in a pseudopartitive construction. The category necessarily encompasses ‘mensural classifiers’ (or measure words) and ‘sortal classifiers’ which express noun classification (see also Lyons 1977 and Langacker 2009, 340). Further, many generative studies on binominal constructions in other languages mention the parallelism between the qualitative and pseudopartitive constructions (see Martí I Girbau 2010, 255 for an overview; Doetjes/Rooryck 2003). In addition, in several languages in Central African (e.g. the Bantu family), the binominal string is found to be the default strategy for adnominal qualification (Mark Van de Velde 2011) and is called the dependency reversal construction. Crucially, this construction also existed in Latin (Malchukov 2000). In the light of these findings, Trousdale’s (2010, 3) observation that evaluative nouns are typically excluded from the restricted set of partitive nouns that can give rise to quantifier and degree modifier uses, needs to be toned down. He considers BQs such as helluva an exception, marginal both in terms of structure and spread, while the Spanish functional equivalent is at least partially productive and can be appropriately motivated in the constructional network model.

284 | Theoretical reflection: a constructional network model of the development of BQs

6.3.4 The explanatory scope of the revisited macro-constructional level The second adjustment to the constructional network model concerns the association between the binominal structure and the expression of a coextensive relationship between N1 and N2. As such, the revised model overcomes the artificial distinction between BQs and expressive binominal constructions. In other words, BQs which originate in genuine QNs are not aligned to a sort of meso-construction pairing the N1 de N2 with quantity assessment, and the quantifying uses of originally expressive nouns such as barbaridad or horror do not derive from a sort of meso-construction pairing the N1 de N2 with quantity assessment. On the contrary, if an abstract construction is to be posited, it is the entrenchment of the association between N1 de N2 and ‘coextensiveness’, which can (temporarily) be either more quality or more quantity oriented depending on the source semantics of the precise noun in the N1-slot and its frequency of use.171 The refined constructional network is visualized in Figure 4, where ‘CE_Q’ refers to the concept of quantity oriented coextensiveness. The constructional network model can now adequately cope with the tendency towards conceptual persistence of grammaticalizing QNs. Although the schematic binominal construction may have become more quantity-oriented because of the preference for quantifying uses of high-frequency montón de and the primarily quantifying dimension of pila de, letanía de, aluvión de and mar de, the quantifying interpretation expressed by binominal syntagms cannot be entirely separated from the initial comparison construal or co-extensiveness relation. Especially in the case of low-frequency BQs, the comparison construal prompted by the backward pull of the source construction generally results in high conceptual persistence, as in (29). The context makes clear that the questions raised by Arrabal are conceptualized as ‘pronounced out loud’ according to a certain cadence or rhythm as in a literal litany. (29) ¿Se trata de un hombre y una muñeca? ¿De un pelele y una mujer? – escribe Arrabal en su cadenciosa letanía de interrogantes – ... ¿Es una || 171 Israel (1996) comes to a similar conclusion when analyzing the development of the Modern way-construction [to Vmovement NP’s way PP] as in The wounded soldiers limped their way across the field. The way-construction is said to originate in multiple sources, i.e. a cluster of ‘manner’ verbs and a cluster of ‘means’ verbs that allowed for increasing schematization and, consequently, far-flung extensions. In modern uses of the way-constructions it turns out to be difficult to establish a ‘manner’ or ‘means’ source, since both pathways can be equally wellmotivated. Yet, “there is no reason we should have to choose. By the time such usages begin to emerge, the two threads are already so entangled that it is often difficult to decide for a particular novel extension whether it should count as means or manner” (1996, 225).

The constructional network model revisited | 285

ligadura simbólica o mágica? (1996, press) ‘Is it about a man and a doll? About a wimp and a woman? – Arrabal writes in his rhythmic litany of questions – … Is it a symbolic or a magic bond?’

montón-construct(1) (det) QN1 de N2 montón-construct(2) montón-construct(x)

CE_Q + CIPQN1 micro-montón

QN2-construct(1)

(det) QN2 de N2

(det) QN de N2

QN2-construct(2) QN2-construct(x)

CE_Q + CIPQN2

CE_Q + CIPQN

micro-QN2

macro-QN

QN3-construct(1) (det) QN3 de N2 QN3-construct(2) QN3-construct(x)

CE_Q + CIPQN3 micro-QN3

QNn-construct(1)

(det) QN de N2 CE_Q + CIPQN meso-QN23

(det) QNn de N2

QNn-construct(2) QNn-construct(x)

CE_Q + CIPQNn micro-QNn

Fig. 4: The constructional network model refined

The revocation of the artificial distinction between quantity and quality assessment also explains why the ambivalent uses observed in the corpus typically relate to indeterminacy or ambiguity between specifier and quantifier uses, rather than involving ambivalence between a literal use and a functional use. In (30), the strange pronunciation of the Spanish [ϴ] can be equally provoked by the huge number of Andalusian colonists (viz. the quantifying reading) as by the oppressive attitude of the Andalusian invaders (viz. the specifying reading). Example (31) blends the quantifying and specifying reading: the disgraces are numerous and the binominal construction pictures Pascuala’s discourse as particularly hurtful and astonishing. Provided that the GR of QNs ties up with the expression of coextensiveness between the two nominal elements, ambiguous and indeterminate uses should no longer be discarded.

286 | Theoretical reflection: a constructional network model of the development of BQs

(30) Lo más corriente es creer que el seseo ha sido traído del Viejo Mundo desde el Descubrimiento por, el aluvión de andaluces seseantes. (1953, A. Alonso, Estudios lingüísticos. Temas hispanoamericanos) ‘The most usual is to believe that the pronunciation of the Spanish [ϴ] as [s] has been brought from the Old World since its discovery by, the flood of lisping (lit. [s]-pronouncing) Andalusians.’ (31) Al oír las primeras palabras de Pascuala, una nube de ira cegó sus ojos; sintió ímpetus de arrojarse sobre ella y clavarle las uñas en el cuello; después tuvo un estremecimiento nervioso y se quedó pálida, demudada, sin querer escuchar lo que le decían y aguantando aquel aluvión de infamias, que iban a clavarse en su corazón como agujas envenenadas. (1884, J.O. Picón, La hijastra del amor) ‘While hearing the first words of Pascuala, a cloud of anger blinded her eyes; she felt the urge to throw herself on her and to dug her nails in her neck; afterwards she was shivering nervously and remained pale, distorted. She did not wish to hear what they told her while standing the flood of disgraces, which were going to stick into her heart like poisoned needles.’ Further, the constructional network model can accommodate the apparently inverted pathways of GR, which form a substantial threat to the directionality claim in traditional theorizing about GR. Chapter 5 has revealed that the specifier use of some QNs, i.e. aluvión de, barbaridad de and mar de, precedes the emergence of quantifying uses. Within the constructional account advocated here, the anteriority of the specifying use of some QNs is no longer noteworthy: the GR of the QNs entails the foregrounding of the coextensive relationship between N1 and N2, which can be either related to size or to type specification. The constructional network hypothesis may also explain the apparent divergence of barbaridad de (with only 7 tokens in the corpus, its syntactic context expansion to adverb uses and its exclusive preference for ‘grammaticalized’ or functional uses) and motivates the membership of ‘expressive nouns’ in the category of BQs. Traditionally, barbaridad de would be considered an instance of the expressive binominal construction, similar to maravilla de ‘wonder of’, horror de ‘horror of’, desastre de ‘disaster of’, demonio de ‘demon de’, encanto de ‘charming man of’, espanto de ‘fright of, ghost of’, estupidez de ‘stupidity, nonsense of’, prodigio de ‘marvel, wonder of’, portento de ‘genius of’ (examples listed by RAE 2009, 891) and many others. Trousdale (2010, 5) and Verhagen (2009, 141) both observe that, in order to occupy the N1-slot, expressive N1s have to display a “scalar extremity”: barbaridad for instance rates extremely high on a scale of badness, which may give rise to the invited inference of a

The constructional network model revisited | 287

large quantity when the scale becomes highlighted.172 In (32) and (33), the respective N2s are not categorized as instances of a barbarity or a horror. Though a single mosquito can still be experienced as a horrible situation, a time span in itself cannot be a barbarity. It is because the mosquitoes are so numerous (cf. the reference to a plague of locusts) and the time span is so extensive (cf. her talking lasted ‘precisely until the door of her house’) that they are compared to a barbarity and a horror. (32) Estuve hablando una barbaridad de tiempo sin que ella cambiase de actitud, exactamente hasta la puerta de su casa. (1981, novel) ‘I was talking to her an awfully long (lit. barbarity of) time without her changing her attitude, exactly until the door of her house.’ (33) – ¿Qué sucede allí? Dicen que plagas de langosta. – También. Pero además un horror de mosquitos y de tábanos. (1991, theatre) ‘– What is happening there? They say a plague of locusts. – Also. But also a horror of mosquitoes and horseflies.’ Interestingly, the pilot (synchronic) investigation of expressive nouns suggests that only a small proportion of the expressive nouns exploits its quantifying potential on a regular basis in binominal syntagms. Crucial to the constructional model elaborated is the considerable number of indeterminate uses which blend the specifying and quantifying readings: in the case of expressive N1s, the specifier and quantifier uses appear thus even more entangled. The observation holds in particular for dechado ‘paragon’, maravilla ‘wonder’, portento ‘genius’ and prodigio ‘marvel’, which are also conceptually near. The specifying uses are easily recognized since they are by definition followed by a concrete N2, as in (34). The binominal syntagm denotes a type of N2 whose most remarkable quality is N1, as in (25), where esa maravilla de mujer qualifies the woman as particularly marvelous and particularly representative for the category of wonderful women. Since dechado, maravilla, portento and prodigio all somehow picture a prototypical member of a category by definition, indeterminateness shows up when N1 is followed by an abstract noun (usually ‘beauty’ or a sort of (moral) quality such as musicalidad ‘musicality’ or elegancia ‘elegance’): in un dechado de belleza ‘a paragon of beauty’, the entity referred to may be rewarded either for being extremely beautiful (and thus displaying a lot || 172 See also Milner (1978) syntactic and semantic description of quality nouns or “noms de qualité”, which have an affective connotation (usually a pejorative one) and lack autonomous reference (and thus depends on the noun of the PNP and the context).

288 | Theoretical reflection: a constructional network model of the development of BQs

of beauty) or for being a representative member of the category of beautiful people. In the examples (35)–(37) the quantifying potential is obviously highlighted, yet the expressive binominals occur in predicative constructions, typical of specifying uses. The quantifying uses of expressive binominal constructions almost exclusively combine with concrete and plural nouns, as in (38). (34) Seguimos a Palma, donde estuve con mis hijas que hacían un curso de dos semanas en esa maravilla de escuela de vela que es Calanova. (1993, books) ‘We proceed to Palma, where I stayed (a while) with my daughters who were taking a two-weeks course in that wonder of a sailing school that Calanova is.’ (35) Mientras el favorito Bithell se mostró nervioso en su Beethoven, quedándose "enganchado" en el primer tiempo poco antes de la cadencia y tocando sin demasiada convicción el resto, por supuesto dentro de un buen nivel, la Kolodin un Concierto en la menor de Schumann, que fue todo un prodigio de musicalidad y precioso sonido. (1977, press) ‘While the favorite Bithell gave a nervous impression of himself in his Beethoven, for having hesitated during the first movement just before the cadenza and for playing not too convincingly during the rest, but of course, at high-quality level, the Kolodin a concerto in A-minor, which was a complete prodigy of musicality and precious sound.’ (36) Hay de todo, pero te hablo de las que conozco, que son bastantes, arriesgan mucho, son más trabajadoras que nosotros, se aprenden primero los textos. Hay compañeras que son una maravilla de belleza exterior e interior también, trabajar con Geles es un lujo, es una persona adorable dentro y fuera del trabajo. ‘There are all kinds [of colleagues], but I talk to you about the ones I know, which are quite a few, they risk a lot, they work harder than we do, they are the first to learn the texts by heart. There are colleagues that are a wonder of external, and also internal, beauty, to work with Geles is a privilege, she is an adorable person during and after work.’ (2002, press) (37) Yo era feo y retraído, un horror de fealdad, como todavía puedes ver, así que no me quedó otra salida que refugiarme en la constancia. (1989, novel) ‘I was ugly and solitary, a horror of ugliness, as you can still see, so

The constructional network model revisited | 289

for me there was no other way out than to take refuge in perseverance.’ (38) Innumerables han sido las definiciones que del mismo se han dado e innumerables también los requisitos considerados necesarios para justificar el título. Pero de todas las definiciones, la más completa y la que, en teoría, mejor encaja con el concepto que antaño se tenía de él, es la dada a últimos de siglo por el famoso cardenal Newman. Desgraciadamente, la debilidad humana es demasiado grande para que produzca el dechado de virtudes y perfecciones que aquél exigía. (1994, press) ‘The definitions that have been given to the very same concept [gentleman, KV] have been innumerable, and innumerable have been the requirements considered necessary to justify the title. But of all definitions, the most complete one and the one that, in theory, most likely fits the concept one had of it in the old days, is the one provided by the famous cardinal Newman at the end of the century. Unfortunately, the human weakness is too big to produce the paragon of virtue and perfection that the former requires.’ The fact that the quantifying and specifying uses of expressive nouns are closely intertwined and that expressive nouns are less susceptible of productive quantity assessment (cf. Table 3), is probably partially due to their strong evaluative load. Since even their literal use is marked, their specific semantic content is hard to desemanticize. In addition, while the scalar implicatures of QNs are inferred directly, the strengthening of the quantifying inferences of expressive nouns requires an intermediate metonymic step.

Quantifier Specifier174

7 1

0 22

2 15

1 20

0 1

0 0

3 1

2 28

3 22

Portento de

Prodigio de

Maravilla de

Horror de

Estupidez de

Espanto de

Encanto de

Desastre de

Dechado de

Demonio de

Tab. 3: Token frequency of the expressive binominal constructions in CREA173

0 2

|| 173 I thank María José Rodríguez Espiñeira for her suggestion to add dechado de ‘model, paragon’ to the list of expressive nouns (with a quantifying potential) provided by RAE (2009). 174 The instances where N2 is grounded, such as la maravilla de Ana / mi hermana ‘lit. the wonder of Ana / of my sister’, are excluded from this count since they belong to another construction type (see Section 2.2.5.2).

Indeterminate TOTAL

45 53

0 69

1 169

1 258

0 57

0 57

1 193

7 115

46 160

Portento de

Prodigio de

Maravilla de

Horror de

Estupidez de

Espanto de

Encanto de

Desastre de

Demonio de

Dechado de

290 | Theoretical reflection: a constructional network model of the development of BQs

3 16

On the other hand, several expressive nouns gave rise to conventionalized quantifying uses in more than one structure: as a noun in fixed expressions and/or as an etymologically related adjective. More precisely, as illustrated in Table 4, DRAE lists quantifying, or at least intensifying, semi-fixed expressions for demonio, espanto, horror and barbaridad, while only the high-frequency QNs montón and mar appear to exploit their quantifying potential outside the binominal construction. The question then arises whether expressive nouns involve in GR processes or undergo lexicalization, since the development of idiomatic expressions is the typical outcome of lexicalization. Tab. 4: Conventionalized quantifying constructions according to DRAE Expressive N / QN

Expression

Description and example

Demonio

como un demonio

Like a devil; e.g. Es feo como un demonio. ‘He is extremely ugly.’ Of the devil; e.g. Hace un frío del demonio. ‘It is extremely cold.’ Of fright; e.g. Tengo un hambre de espanto. ‘I am extremely hungry.’ Horrific, appaling; e.g. Hace un frío espantoso. ‘It is terribly cold.’ Horrors; e.g. Se divierten horrores. ‘They have a very good time.’ Horrible; e.g. Nos dio un susto horrible. ‘It really frightened us.’ Brute; e.g. Le hizo un desplante bárbaro. ‘He really snarled at her.’ In heaps. Abundantemente, sobrada, excesivamente. ‘plentiful, more than enough, excessively.’ In heap. Abundantemente, sobrada, excesivamente. ‘plentiful, more than enough, excessively.’ To seas; e.g. Llorar, llover, sudar a mares. ‘to cry, to rain, to sweat a lot.’ From sea to sea; e.g. Juan iba de mar a mar. ‘Juan went excessively adorned.’ To have to discuss a lot (lit. ‘to talk about the sea’). The sea (of). Con abundancia. ‘Plentiful.’

del demonio Espanto

de espanto espantoso

Horror

horrores horrible

Barbaridad

bárbaro

Montón

a montones a montón

Mar

a mares de mar a mar hablar de la mar la mar (de)

The grammaticalization of the BQ-construction | 291

To avoid possible misunderstandings, it should be pointed out that the present analysis makes no claims on the emergence of the expressive binominal construction. Nor does the constructional network hypothesis suggest that the tendency towards a high CIP displayed by changing BQs should be considered as a consequence of the existence of expressive binominal constructions. Both types of binominals originate in the foregrounding of the co-extensive relationship between N1 and N2. This coextension may be more quantity oriented or more quality oriented. The construction’s high CIP may be strengthened by the repeated activation of the quality-pole by expressive binominal constructions, but crucially remains the result of the initial comparison construal. The question as to know when the first expressive binominal construction showed up and whether it underwent GR are interesting ones, but fall outside the scope of this analysis.

6.4 The grammaticalization of the BQ-construction Another important question remains the issue of whether there is a meaningful role left for GR in the development of low-frequency BQs. Since the answer to that question depends on the definition of GR adopted, the present section will first dwell on my understanding of GR (Section 6.4.1). Only then will I be able to determine at which constructional level GR occurs (Section 6.4.2).

6.4.1 BQs as the locus of grammaticalization The fact that frequency tables throughout this volume generally reflect differences between head, quantifier and specifier uses, might create the impression that I implicitly equate GR with semantic change. In this section I will show that BQs acquired a more operator-like or functional role, in spite of the high degree of formal and semantic persistence that characterizes their GR. In a nutshell, I agree with Traugott (2011) that the constructional hypothesis on ‘coevolution of form and meaning’ should not be interpreted as strictly simultaneous, yet as “potentially separate form and meaning shifts toward a new Cxn [construction, KV]” (Traugott 2011, 13). Yet at the same time it strikes me that both in the usage-based theorizing on GR as well as in diachronic construction grammar, priority is ironically given to (shifts in) form while within the same framework changes in meaning are claimed to be equally important. A construction is usually defined as a form-meaning pairing, a pairing of form and meaning, and never alluded to onomasiologically, as a meaning-form pairing.

292 | Theoretical reflection: a constructional network model of the development of BQs

Since Lehmann’s groundbreaking formulation (1982; 1995), processes of GR are typically associated with his set of parameters. Starting from the assumptions that an autonomous sign has a certain weight; is not closely linked to another sign nor to its environment; that it displays some variability, both paradigmatically and syntagmatically; and, finally, that GR entails a reduction of the construction’s autonomy, GR is generally measured in terms of attrition, paradigmaticization, obligatorification, condensation, coalescence and fixation (Lehmann 1995, 164). However, Hopper (1991, 22) suggests that Lehmann’s criteria are primarily characteristic of completed processes of GR. Ongoing GR primarily combines with layering (of original and new uses), divergence, specialization, persistence and de-categorialization. In refinement of their view of GR as ‘reduction’ or increased dependency (see Traugott 2011, 11), Himmelmann (2004) defines GR as a process of context ‘expansion’ at three levels. Interestingly, Traugott (2011) argues that context-expansion is not a contradiction to GR as reduction, since reduction favors routinization and routinization in its turn leads to increased generality or bleaching, to increased schematicity (i.e. the creation of slots in paradigms) and to less syntactic and semantic compositionality. In other words, GR as context-expansion is the logical outcome of GR as reduction (Traugott 2011). Since the morphosyntactic outlook of the BQ remains largely unaltered, strict adherence to Lehmann’s parameters would not be informative for the process Spanish QNs are involved in. Instead of considering Lehmann’s criteria as ‘gatekeepers’ to exclude many morphosyntactic developments from GR, Traugott suggests that GR typically involves (1) structural decategorialization, (2) unidirectionality, i.e. shift from membership in a relatively open set to membership in a relatively closed one in the context of a particular construction, (3) bonding within a particular construction and (4) subjectification, i.e. the semantic or pragmatic shift from more to less referential meaning via pragmatic strengthening (Traugott 2003, 643–644). The GR of Spanish BQs is shown to be conceptually driven. Specific quantifying or specifying uses can combine with specific clusters of conceptually linked clusters of N2, such as the ‘invader’-cluster in the grammaticalized uses of aluvión de or the ‘unpleasant people’-cluster that combines with letanía de. The constrained and semi-productive nature of N2-combinations leads Trousdale (2010) to argue that lexicalization is at stake: while load is more ‘grammaticalized’ in the quantifying uses of [loads + ADJ], as in loads better, the same item is more ‘lexicalized’ in the idiomatic expression what a load of rubbish, since [what a load of NP] is restricted to negatively evaluated nouns, especially nonsense and rubbish (Trousdale 2010, 19–20). In his view (Trousdale 2008, 58), lexicalization leads to idiom-like constructions, while GR leads to more schematic constructions. Collocational restrictions indeed remind of

The grammaticalization of the BQ-construction | 293

Himmelmann’s definition of lexicalization as a kind of host-class specialization, where a particular node-collocate combination is “singled out” (2004, 36) and becomes the preferred combination, as in hablar de la mar (in contrast to *discutir/charlar/contar de la mar). Likewise, the tendency (or inherent constructional possibility) to display a high degree of conceptual persistence reminds of Brinton/Traugott’s (2005) definition of lexicalization as a change whereby “in certain linguistic contexts speakers use a syntactic construction or word formation as a new contentful form with formal and semantic properties that are not completely derivable or predictable from the constituents of the construction or the word formation pattern” (2005, 96, italics KV). However, the semantic change in lexicalization entails that “a specific semantic component is added, so that the new lexical meaning differs from the former compositional meaning” (Wischer 2000, 364), which does not exactly correspond to the conceptual persistence in low-frequency BQs. The conceptual persistence refers to the semantic components that live on in the grammaticalized quantifier next to the more functional reading which is foregrounded. The development of BQs fails to have another characteristic of lexicalization, i.e. a decrease in pattern productivity. Their development also involves a battery of correlated features generally associated with grammaticalization (cf. Brinton/Traugott 2005; Cuyckens 2007; Traugott 2008a) and definitely leads to – at least partial – loss of compositionality: Rebracketing The shift from [Det. + QNhead + [de + N2]] to [[Det. + QN + de] + N2head].175 Structural decategorialization || 175 Note that the underlying structure [[un N1 de ] N2] is problematic in the light of (1) the possibility to repeat the preposition with co-ordinated N2’s (as in un aluvión de informaciones y de opinions ‘a flood of information and opinions’) and (2) in view of the omission of de in adverbial uses (as in ¿Cuántos discos has oído este verano? – Un montón. ‘How many discs did you listen to this summer? – A lot.’) – I thank Ignacio Bosque for this insightful suggestion. The former characteristics have lead many linguists within the generative framework to maintain the boundary between N1 and the following PNP and to assign distinct functions to the preposition de depending on the partitive or pseudopartitive interpretation of N1 (see Gutiérrez Rodríguez 2008 for an overview). From a functional point of view I suggest that the rare cases of repetition of de (in functional uses!) might either induce a proper conceptualization (see Chapter 8) or be related to the fact the structural reanalysis is rather recent. As to the underlying structure of the BQ when de is omitted in adverbial uses – which is only attested with highly grammaticalized montón, mogollón and barbaridad, I propose to consider the entire syntagm [(un) QN] as a quantifier and therefore argue that strictly speaking, de is not even omitted since it has never been made explicit in adverbial uses nor necessarily mentioned anaphorically in the context either.

294 | Theoretical reflection: a constructional network model of the development of BQs

Partial loss of noun-ness of the QN: stringent combinatorial restrictions as to determiner variation, modification and right collocate. Semantic shift QN shifts from literal, concrete meaning (for instance a literal herd in un hatajo de ovejas) to the less compositional yet equally complex, more abstract interpretation of assessing an excessive quantity (as in un aluvión de gente ‘a flood of people’). Paradigmaticization and increase in productivity Emergence of the QN-paradigm or partially productive construction where the QN-slot is occupied by a noun with scalar implicatures. Context expansion at three levels The process involves (1) host-class expansion, i.e. in the quantifying(/specifying) interpretation, the class of possible right collocates comprises all nouns which can be made to fit the individual construal of the QN; (2) syntactic expansion, as montón and barbaridad can also be used adverbially with verbs and adjectives to express a high degree, as in Me alegra un montón. ‘It makes me very happy’ and la mar de has turned into a productive adjective intensifier; (3) semantic-pragmatic expansion as N1 shifts from more to less referential meaning via pragmatic inferences or changes from a lexical noun into a quantifier or a modifier displaying an evaluative potential. Layering and divergence The emergence of the ‘newer’ quantifying and specifying use of N1 within the binominal syntagm does not discard the ‘older’ literal reading. Instead, GR yields a systematic synchronic polysemy. In line with Brems’ (2010) suggestion to operationalize semantic change in GR as changes in the combinatorial behavior of constructions, I consider BQs as collocationally constrained constructions: within the binominal construction, specific QNs “act as fixed lexical items in the sense of collocational nodes that take different sets of collocates depending on whether they are used as a head, quantifier or valuing quantifier. Hence, in SN-patterns [KV: Size Noun] the predetermination takes the shape of strong node-collocate co-selection related to the various constructional uses of SNs” (Brems 2010, 101). In addition, BQs may develop a strong preferences for a particular use: hatajo de, for instance, is tied to the specifier use, aluvión de preferably combines with the quantifying reading, etc. Provided that redistribution of combinatorial patterns and context of appearance is accepted as ‘formal’ change, quantifying and specifying BQs can be considered as a true locus of GR.

The grammaticalization of the BQ-construction | 295

Trousdale (2010) prefers the notion of constructionalization when outlining the developments in the English binominal network. He suggests that the creation of the degree modifier construction helluva far from the constitutive construction a hell of ugly devils involves two stages: “preconstructionalization, which involves a new polysemy of a construct (an instance of an existing construction), and constructionalization (…)” (Trousdale 2010, 2) or the creation itself of a new grammatical construction. In an attempt to overcome the terminological erosion of grammaticalization, Traugott (2011; 2014, in collaboration with Trousdale) proposes a new terminological and constructional apparatus of GR. They frame the debate in terms of constructional change and (grammatical or lexical) constructionalization. While constructional change can be any language change, for instance GR, semantic change alone, etc., constructionalization is the subset of constructional changes in which “morphosyntactic formnew-meaningnew combinations of signs are created through a sequence of small steps in which form and meaning are reanalyzed” (2011, 6). The diachronic sequence of events in which Spanish QNs are involved can be typified as a locus of grammatical constructionalization, since the new form-meaning pairing serves a procedural function, i.e. the purposes of the grammatical pole of the language system. However, even the innovative definitions of GR and grammatical constructionalization do not provide a water-tight distinction between unconstrained formal change (more ‘grammaticalized’ in Trousdale’s terms) and semantically constrained formal change (more ‘lexicalized’ in Trousdale’s terms). The notion of constructionalization was first used by Noël (2006; 2007), who wishes to distinguish the initial formation of a construction from possible succeeding shifts of the construction towards a more procedural function. The prior association between a specific meaning and morphosyntactic structure is called schematization or constructionalization, while the notion of GR is reserved for the latter type of change (2007, 195). Both types of change need to be kept apart, since GR presupposes constructionalization, whereas the reverse is not necessarily true. However, in practice, “it might not be possible to determine where (or rather when) primary schematization ends and where grammaticalization begins” (2007, 195). The question then arises whether the development of BQs involves constructionalization or GR only. I tentatively argue that it involves both types of change. When un montón de N2 and other pioneer QNs came to develop quantifying and specifying uses, the morphosyntactic outlook did not change dramatically and the specific binominal constructions were the locus of GR. When a series of similarly behaving QNs gave rise to the macrobinominal construction associated with quantity or quality oriented coextensiveness, and afterwards, the individual GR of new QNs further en-

296 | Theoretical reflection: a constructional network model of the development of BQs

trenched the macro-construction, constructionalization has occurred (yet only at macro-constructional level).

6.4.2 GR of the binominal quantifier macro-construction A key problem that remains unsolved is the identification of the level of GR, in particular for low-frequency BQs such as letanía de: is the quantifying use in examples such as (39) an instance of the grammaticalized micro-construction letanía de or merely an instantiation of the grammaticalized abstract BQ-construction at the meso- or even macro-level? (39) El zapatero, con más deseos de hacer reír a la gente que de insultar a la Jerónima, fue diciéndole una verdadera letanía de desvergüenzas: Cállate, penca del diablo, pata de afilador, albarda, zurupeta, tía chamusca, estropajo. (1953, R.J. Seder, Réquiem por un campesino español) ‘The shoemaker, who felt more like making the people laugh than insulting Jerónima, was telling her a real litany of obscenities: Shut up, you devil’s lash, fusspot, packsaddle, piece of shit, shriveled woman, idler.’ The answer crucially depends on the framework chosen and on the definition of GR adopted. I would therefore argue that: (i) the first quantifying/specifying use of low-frequency items is always contextually embedded and arises via pragmatic strengthening of scalar implicatures (recall that not all QNs invite this inference). (ii) from the moment un montón de N2 became an entrenched micro-construction expressing quantity (and/or quality) assessment, the emergence of new partially productive constructions is felt as less abrupt. (iii) corpus data suggest that from the 19th century onwards, the binominal construction is automatically associated with the expression of quantity (and/or quality) assessment to the extent that in Present-Day Spanish, the binomial construction can ‘coerce’ the coextensive relation between N1 and N2 in creative language use, i.e. enable local reanalysis via analogical thinking. In other words, if the change from literal letanía into polysemic letanía de (used both literally and as a quantifier) is not widespread, examples like (39) remain at the level of metaphoric language use or invited inference possibly triggered by analogical pressure from high-frequency montón de and later the entrenched

The grammaticalization of the BQ-construction | 297

link between ‘N1 de N2’ and quantity expression. If the quantifying use of letanía de undergoes routinization and even generalization (as in (40) and (41)), the change into a quantifier function has affected the micro-constructional level. Whereas in (40), the catholic frame as well as the concept of ‘repetition’ (cada día ‘each day’) are recalled, letanía de has abstracted further away from its source semantics in (41). In sum, the shift to functional uses of low-frequency QNs can be regarded as an instance of GR if it gives way to context-expansion and if one allows GR to be prompted by analogical thinking. (40) La letanía de tragedias que causan cada día los malos tratos domésticos ha metido a la Iglesia católica en un nuevo jardín de contradicciones y críticas. (2002, press) ‘The litany of tragedies which are daily caused by domestic abuse again entangled the Catholic Church in contradictions and criticism.’ (41) Durante siete años la guerra de Argelia había sido como una letanía de sangre sobre nuestros ojos y nuestros oídos. (1995, books) ‘For seven years, the Algerian War had been like a litany of blood on our eyes and ears.’ The importance attributed here to the discourse context of grammaticalizing BQs might misleadingly create the impression that there is no need anymore to posit an abstract macro-binominal quantifier: the quantifying uses of lowfrequency QNs might also simply result from the speaker’s spontaneous exploitations of the QN’s scalar implicatures. Several main arguments can be raised against the latter reasoning. First, hypothesizing a process of GR that would only occur at the micro-constructional level cannot account for the productivity of the construction, especially form the 19th century onwards. Second, it is no coincidence that the beginning of the 19th century as a cut-off point coincides with dramatic increase in type-frequency. Third, numerous nouns have scalar implicatures, but only those nouns which can be semantically modeled on already existing QNs develop systematic quantifying uses. One open question remains, however: does the process of GR only occur at micro-constructional level or does it affect the abstract N1 de N2-pattern as well? I propose that the GR exclusively concerns the individual micro-constructions, since it is obviously hard to find empirical evidence for morphosyntactic and semantic micro-changes of abstract constructions, which questions their validity as input for GR. On the question how to relate the abstract partitive construction to the abstract BQ, I would answer that the matter is irrelevant, since the diachronic corpus analysis has shown that the partitive construction cannot be considered a common starting point of the QNs. The relationship between the

298 | Theoretical reflection: a constructional network model of the development of BQs

abstract partitive construction and the abstract BQ is thus not straightforward. Instead of a change at macro-constructional level from partitive construction into a contextually constrained quantifier construction, I propose that the individual development of several QNs led to the emergence of a new construction. In addition, one might wonder whether the quantifying and specifying readings of QNs are to be considered two separate constructions or two uses of a single macro-construction expressing coextensiveness. The fact that both readings usually show up in quite different contexts favors their status as distinct constructions. However, the corpus analysis has shown that it is often hard to decide where the quantifying interpretation ends and the specifying one begins. In addition, although the precise discourse context is claimed – additional arguments will be provided in the following chapters – to interact with the reading of the QN, contexts are not exclusive for particular readings (cf. the number of indeterminate uses).

6.5 Conclusion In this chapter, I have argued that a slightly modified form of Traugott’s constructional model is needed to accommodate the diachronic gradualness and the synchronic gradience in the development of Spanish BQs. More precisely, the frame-based combinatorial constraints of low-frequency BQs; the partial productivity of literal binominal uses for some QNs; the serious gap in time between the GR of montón de and the sudden rise in type frequency of the abstract construction in the 19th century; the common tendency towards conceptual persistence; the ability of la mar de to intensify adjectives; the categorial interferences or multiple sources (especially for (h)atajo de); and, finally, the inclusion of expressive nouns such as barbaridad in the lexical field of QNs are shown to be compatible with GR provided several constructional levels in GR are distinguished and conceptual persistence is allowed to be more than just a side effect of GR. I have argued that conceptual persistence is plausibly a crucial factor in the productivity (and as a consequence, further entrenchment) of the construction. In the next chapters, I will refine the common acceptation of Hopper’s notion of persistence and illustrate why I believe the morphosyntactic behavior of QNs to be conceptually driven. Within this model, a major role continues to be reserved for persistence and analogical thinking. While the role of persistence hinges on coextensiveness between both nominal elements and the hypothesized comparison construal, analogical thinking is crucial in the rise and conventionalization of the abstract BQ as well as in the expansion towards new N1- and N2-clusters. It is important

Conclusion | 299

to note however, that the eagerness to define an analogical model for local reanalysis to realign to, might misleadingly detract the attention from the precise original context which has to invite the necessary inference for realignment. Following Traugott (2011; 2014), I propose to take the re- off in reanalysis and realignment, since the latter notions might insinuate an ‘attracting force’ for high-frequency montón de and the schematic binominal construction. Yet in spite of the modeling function these ‘attracting’ constructions fulfill, innovation begins at the level of concrete usage, where the precise discourse context has to invite the pragmatic inference. The next chapters will show that analogical thinking is not only a precondition for language change (see also Givón 1991; Fischer 2007; Traugott/Trousdale 2010), but also a crucial principle in the interaction between particular constructions and the discourse context in order to arrive at coherent communication.

| Part 3 Synchronic case-studies – towards a constructional network model

7 QN-related schematization and N2-profiling In the previous chapters a crucial role was attributed to persistence and analogy in the development of BQs. The suggestion was put forward to raise ‘persistence’ from epiphenomenal to GR to the status of mechanism. In the synchronic analysis, the role of conceptual persistence is analyzed in the light of the construction’s semantic, pragmatic (the present chapter) and morphosyntactic behavior (Chapter 8). The claim will be made that, on the basis of the notion of conceptual persistence, a proper image-schematic structure can be posited for each QN. This notion of conceptual (image) persistence is derived from Hopper's (1991, 22) parameter of lexical persistence and first applied to BQs by Brems (2011). Lexical persistence refers to the tendency of some grammaticalizing items to retain particular features of their original lexical use, which may continue to influence their further developments in various ways. Hopper considers lexical persistence as a diagnostic for uncompleted processes of GR. I prefer the notion of conceptual (image) persistence (CIP) over lexical persistence (see Verveckken/Delbecque Fc. (2014)) mainly because the notion of CIP fits better the cognitive framework I subscribe to. More importantly, the lexicographic treatment of the Spanish QNs presents quite severe gaps. For instance, most definitions encountered for alud ‘avalanche’, evoke the image of a snow crash involving violence and loudness, but fail to mention that it is irrepressible, unstoppable and life-threatening. In my view, this property is however particularly prominent in the conceptualization of an avalanche and influences the quantifier use of alud, as shown in (1). (1)

Pero lo que de verdad hubiera afectado a la estabilidad del país, habría sido el imparable alud de reproches que habría recaído sobre González. (1995, books) ‘But what would really have affected the country’s stability, would have been the unstoppable flood of reproaches that would have fallen down over González.’

With the notion of CIP, I refer to the frame (Fillmore 1985) or the image schema (Rhee 2002; Oakley 2007) activated by the QNs, which exceeds the definition of a QN given by reference dictionaries and is (partially) preserved in GR. The notion of CIP allows for unprofiled elements (in pre-grammaticalization stage) to become foregrounded in the grammaticalized uses, while involving metaphoric and metonymic extensions.

304 | QN-related schematization and N2-profiling

Interestingly, in his study on (relatively) open paradigms, Bosque (2007) argues that the GR of BQs goes hand in hand with a process of abstraction or schematization of the original lexical meaning. He also claims that the persistence of the original lexical meaning constrains the combinatorial pattern of the QN as to the N2 (2007, 196). The question of how much exactly remains of the source meaning has not yet been fully addressed in the literature, however: me parece que no se ha estudiado con detalle en estos procesos de gramaticalización la cuestión de cuánto queda exactamente del sentido original de tales sustantivos en sus nuevos usos gramaticalizados. (Bosque 2007, 192)176

The present chapter partially fills this gap. The basic tenet is twofold and starts off from the cognitive grammar assumption that a difference in form entails a difference in meaning. It predicts on the one hand that each QN yields its own conceptualization of the N2-entities, and, on the other hand, that this individual conceptualization is a faint vestige of its source semantics, viz. the result of CIP. Recall that the examples (7)–(10) of Chapter 6 illustrate that the conceptualization of the N2 gente differs depending on the QN – tropel, montón, pila or aluvión – that precedes. Given that the only constructional slot subject to alternation is the N1 slot, the difference in construal stems from the semantics of the QN. Likewise, in (2), the reformulation of retahíla ‘string, series’ by letanía ‘litany’ suggests that letanía de agravios not only refers to a long succession of insults, yet also to the inclination of the Cuca-women to react, as in a literal litany where the priest leads and people respond by fixed formulae. (2)

Todas las Cucas se sabían esta retahíla o letanía de agravios y fiel de fechos y en las entrañas sentían la comezón de responder; pero ¿cómo? (1927, E. Noel, Las siete cucas) ‘All the Cuca women already knew that string or litany of insults and offenses and inside, they felt the impulse to respond; but how?’

The structure of this chapter will be as follows. Section 7.1 motivates the methodological choices made as to corpus compilation and parametrization. Section 7.2 provides evidence for the hypothesis that each QN yields its own conceptualization of N2 and presents a fine-grained analysis of the meaning construction per QN. Section 7.3 will line up the observed tendencies and show that the tradi-

|| 176 It seems to me that in those processes of GR the question has not been studied in detail of how much exactly remains of the original meaning of those nouns in their new grammaticalized uses. (translation KV)

Methodological background | 305

tional conception of the notion of persistence might benefit from further refinements.

7.1 Methodological background The empirical data described in Part 3 of this volume are extracted from the synchronic online-corpus Corpus de referencia del español actual (henceforth CREA) patronized by Real Academia Española. CREA is considered the most representative synchronic corpus. It is a 160 million words corpus that covers the time interval from 1975 to 2004. It comprises a variety of written and oral texts (respectively 90% of written data and 10% of oral data) and encompasses several geographical varieties. More precisely, 50% of the material concerns Peninsular Spanish, while the remaining 50% contains Latin American texts (including the traditionally distinguished linguistic zones of Caribbean, Mexican, Central, Andean, Chilean and Río de la Plata Spanish). Like for CORDE, the search engine of CREA does not allow for lemmatizing nor morphosyntactically tagging. For the 9 QNs selected (cf. Chapter 2), exhaustive samples – of the plural as well as of the singular forms – were extracted from CREA by the query ‘QN + de’. I filtered out irrelevant examples manually (e.g. when the PNP is not quantity related (3)177 or not used nominally yet for instance adverbially (4)). In order to assure that each instance contained a (coextensive) BQ. The filtering yielded the totals listed in the second column of Table 1. In order to arrive at representative and comparable samples, I decided beforehand to analyze exhaustively the QNs that returned no more than 250 coextensive examples. If the filtering procedure yielded more than 500 instances of a particular QN, that sample would be reduced to a representative sample of 500 instances.178 The underlying idea was that, the more productive the QN is, the more subtle and varied will be its semantic potential and the more data are needed for a fine-grained analysis of its meaning structure. Only montón turned out to be that frequent.179

|| 177 Etymologically, the compound nombre de pila refers to a person’s name as given by baptism. Note also that pila (de bautismo (or pila bautismal) refers to the baptismal font. 178 If a sample contained more than 250 examples but less than 500, it would have been reduced to 250 instances. Yet no QN presented that frequency pattern. 179 I thank Prof. Dr. D. Speelman (Quantitative Lexicology and Variational Linguistics, KU Leuven) for the in-depth discussion and his helpful suggestions on the procedure to arrive at the most representative sample possible.

306 | QN-related schematization and N2-profiling

(3)

El primer hijo lo tuvieron hace dos años y dos semanas y se llama Ignacio, como el segundo nombre de pila de su padre. (2001, press) ‘They had their first child two years and two weeks ago and it is named Ignacio, after the second first name of his father.’

(4)

Un aficionado a las setas sabe un montón de botánica y un aficionado a la astronomía, de astronomía. (2003, press) ̴ Sabe de botánica un montón. ‘A mushroom lover knows a lot of botanic and a devotee of astronomy of astronomy.’ ‘ ̴He is acquainted with botanic.’

Tab. 1.: Proportions of raw data and manually filtered binominal constructions in CREA ‘QN+de’ in CREA Alud Aludes Aluvión Aluviones Barbaridad Barbaridades Hatajo Hatajos Letanía Letanías Mogollón Mogollones Montón Montones Pila Pilas Racimo Racimos

96 8 162 3 25 9 18 1 83 15 53 1 1400 258 162 102 54 49

Binominal constructions Representative sample 86 6 150 2 9 2 16 1 26 3 46 1 977 244 80 42 48 42

86 6 150 2 9 2 16 1 26 3 46 1 500 244 80 42 48 42

The criteria for the synchronic data analysis replicate the parameters used in the diachronic one. In other words, the examples are coded for an array of semantic-pragmatic and morphosyntactic features. The semantic-pragmatic criteria regard the function of N1, the degree of CIP, the polarity of the semantic prosody or affective load surrounding the binominal. Morphosyntactic criteria concern the co-selection pattern of the QN with regard to the determiner (to N1 and in rare cases to N2), the modification of N1, the verb and phoric agreement, and the type of N2.

Quantification construal in BQs | 307

In an attempt to grasp the individual meaning profiled by each QN, I turn to three sources of information: (1) the intuition of native speakers who are asked to comment on how they conceive of the N2-entities in authentic contexts and how (or whether) this conceptualization would change by replacing the QNs by other QNs or montón de;180 (2) the definitions proposed by reference dictionaries (DRAE, DUE, CLAVE)181 and (3) the (recurrent) contextual clues in the actual attestations. Per QN, a list of semantic facets is established, the presence of which is carefully checked and counted in the corpus. For instance, for alud de, I verify for each attestation whether the facets ‘all at once’, ‘all of a sudden’, ‘uncontrollable, unstoppable’, ‘overwhelming’, ‘reaction’, ‘different sources’, ‘unexpected’ etc. are profiled, and if so, how they are encoded.

7.2 Quantification construal in BQs The basic assumption of the semantic analysis is that, within the BQ-construction, the meaning of the QNs can be described in terms of an image-schematic structure that N1 imposes on N2.182 This working hypothesis builds on the iconicity principle (García 2009, 291; Van Langendonck 2007, 411–412; Wierzbicka 1985, 334): as discourse goes along, the QN – which is mentioned first in the flow of speech and processed before the N2 is uttered – evokes a schematic configuration of the mass or the group of entities, while N2 simply adjusts itself to this image activated by N1.183 The schematic configuration of the mass is an

|| 180 Lydia Fernández Pereda, Marían Gomez Castejón, Pedro Gras Manzano, Nuria Herrera Coronas, Daniel Michaud Maturana, María Sol Sansiñena Pascual and Ingeborg Verelst-Pérez were so kind as to repeatedly and benevolently comment on my interpretation of the Spanish examples. The conceptual facets listed per QN in this section were checked primarily by the intuition of these native speakers. 181 Full titles: Diccionario de la lengua española (RAE 2001), Diccionario de Uso del Español (Moliner 2008), Clave-Diccionario de USO Del Español (Maldonado González 1999). 182 The distinction made in lexical semantics between ‘image schemas’ (a.o. Lakoff/Johnson 1980), ‘mental spaces’ (a.o. Fauconnier 1985; Fauconnier/Turner 1998) and ‘frames’ (a.o. Fillmore 1985) is mainly linked with theorizing on conceptual metaphor and lexicography. I therefore do not consider it necessary to go into detail. 183 See also Langacker (2014, 3) on focusing and anchoring: in language processing, structures are assessed in a certain sequence. The first element accessed is the anchor and has a measure of inherent salience: “already active when other structures are accessed, it unavoidably has some influence on subsequent processing” to the extent of having a framing function. The second element is the focus and is interpreted in the conceptual domain activated by the anchor. In BQ-structure, N1 is the anchor and N2 the focus.

308 | QN-related schematization and N2-profiling

abstraction of the QN’s source semantics or frame and can highlight one or more particular subparts of the original frame depending on the surrounding contextual elements. The starting assumption is in line with Langacker’s account of the construal processes in BQs (2011, see Section 2.1) who reserves a particular role to each component of the construction – i.e. the determiner, the QN, the preposition and the N2 – in the construction of meaning. At the composite structure level, a mapping is posited between the image schematic structure imposed by the QN and the N2-entities via the preposition de which essentially expresses coextensiveness. The theoretical discussion in Chapter 6 has indicated that the very same coextensive relationship allows BQs to overcome the distinction between type specification and quantification (Langacker 1991, 53), viz. to combine quantity and quality assessment. While the quantification in BQs initially thrives on pragmatic inferences, conceptual image persistence of N1 warrants additional type specification. The semantic analysis in the present section outlines the image schema of the individual QNs and specifies the conceptual facets these image-schematic structures contain. The analysis draws on Paradis’ (2011) distinction between the ‘ontological structure’ and ‘construal operations’. The cognitive semantics literature presents various models differentiating the ‘basic’ lexical meaning of an item and different construal processes (a.o. Langacker 1987; Talmy 2000a, 2000b; Croft/Cruse 2004; see also Section 1.1.1.2), yet Paradis explicitly leaves room for the discourse context and demonstrates the diachronic value of her semantic model. In her view, “[m]eanings of linguistic expressions are emergent and get their final interpretation in context. There is no principled difference between creative form-meaning pairings and conventionalized pairings from the point of view of how they are construed.” (2011, 65).184 By ontologies, Paradis refers to a set of pre-meaning structures (instead of a set of lexical meanings) of a particular linguistic element, which can be both contentful (e.g. what is traditionally referred to as encyclopaedic knowledge) and configurational (e.g. PART/WHOLE RELATION, PATH, MODALITY, etc.). For instance, the earliest use of the intransitive verb to crawl in The aged Symeon cralls to kyrk comprises at least the following meaning structures: “actual MOTION in physical SPACE, presupposing an agentive discourse PARTICIPANT, such as the aged Symeon, slow SPEED, low

|| 184   See also Delbecque’s (2012) analysis of the highly polyfunctional string en el fondo ‘deep down, actually, basically’. The underlying starting assumption is that en el fondo presents a particular image-schematic structure (composed of three basic interrelated facets) and acquires its ultimate conceptualization in interaction with the specific context the string occurs in.

Quantification construal in BQs | 309

and horizontal ORIENTATION.” (Paradis 2011, 72). Her notion of ‘ontologies’ corresponds to my understanding of image schemas. With ‘construal operations’, she refers to the cognitive processes that operate on the imageschematic structures which are responsible for the final interpretation of linguistic items in context. While Paradis (2011) sets up a battery of construal operations selected from the literature, namely Gestalt (constitution), salience (focus of attention), comparison (judgment) and perspective (situatedness), I will be only concerned here with the image schemas of the QNs. Recall that a.o. Verhagen (2007, 56) argues that attempts at exhaustive classifications of construal operations by definition present a considerable degree of arbitrariness and overlap and often rather constitute classifications of the linguistic item under analysis. It goes without saying that the quantification construal in BQs involves specific choices as to the selection and salience of conceptual facets, the perspective and degree of subjectivity. Since one sees more in contrast or comparison (Delbecque, p.c.), an appropriate method to retrieve the image schemas of individual QNs, is to start from two near-synonyms: if differences in contexts of usage can be detected for QNs whose source semantics are closely related, evidence is found for their individual conceptual structure. For instance, spot checks in CREA reveal that mar de ‘sea of’, océano de ‘ocean of’ and piélago de ‘(deep) ocean’ profile the N2entities as a horizontally oriented infinitude, as in (5)–(7).

LOCATION

(5)

(…) después de todo, incluso en los lugares y entre los hombres de los que menos cabría esperar, se revela no un pozo, sino todo un piélago de bondad, sino que se nos informa, en fin, de una maquinaria judicial en la que la afirmación, la negación y la duda son intercambiables, (…). (1985, press) ‘After all, even in those places and among those men of which one would expect less, not a pool but an entire ocean of goodness reveals itself, yet that informs us, eventually, of a judicial machine in which affirmation, negation and doubt are interchangeable, (…).’

(6)

La angustia de sus tribulaciones le enrojeció la piel, y le hizo engordar como si las velas con las que alumbraba sus noches infinitas entre un océano de papeles fueran un raro alimento que pusiera el sebo en su carne, (…). (1992, novel) ‘The fear of his tribulations turned his skin red, and made him gain weight, as if the candles with which he lit his infinite nights among an ocean of papers were a rare food that irritated him.’

310 | QN-related schematization and N2-profiling

(7)

Todo parece indicar, así pues, que nuestro universo es plano – o con una ligerísima curvatura negativa – y que su final llegará mediante una expansión infinita, en la que los soles se apagarán los planetas se consumirán y nada quedará sino un inacabable mar de microondas en perpetuo proceso de enfriamiento. (2004, books) ‘Everything thus seems to indicate that our universe is flat – or with a very slight negative curvature – and that its end will come via an infinite expansion, in which the suns will extinguish, the planets will burn down and nothing will be left but a neverending sea of microwaves in perpetual process of cooling.’

They differ however as to which conceptual facets of the source semantics – a vast amount of water that covers the greatest part of the earth’s surface – they preferably highlight. The elaborated premodification pattern of océano de, for instance, almost exclusively involves adjectives stressing the endlessness of the mass of N2-entities such as insondable ‘bottomless’, vastísimo ‘very wide’, gran ‘big’ or inmenso ‘immense’, while mar de also combines with adjectives highlighting its dangerous (peligroso) or tempestuous (proceloso) nature. When comparing the contexts where mar de and océano de quantify the same N2 dudas ‘doubts’ (respectively 26 and 2 instances), the contexts of océano de tend to present the protagonist as completely (and irreversibly) overwhelmed by the N2s, as if already drown (e.g. (8)), while the contexts of mar de seem to picture the protagonist as an antagonist that offers resistance, as if caught in the middle of the drowning process (e.g. (9)–(10)): mar de is repeatedly preceded by locational verbs such as estar en ‘to be in (the middle of) or encontrarse en ‘to be’ in the present tense.185 Further, la mar de appears to be the only variant that comes to intensify adjectives as well (as in la mar de bonita ‘incredibly beautiful’), which demonstrates again that pure analogical thinking cannot account for the development and synchronic use of the BQ-construction. (8)

Parece incluso que ambos films podrían inaugurar un nuevo género, aquel que afirma una incógnita del significado, donde las razones se diluyen en un océano de duda e incertidumbre que lo domina todo. (2002, press)

|| 185 The considerable differences in frequency – the query for mar de returns 1005 instances, for océano de 97, and for piélago de only 13 – plausibly explain why the binominal uses of océano and piélago contain relatively more highly metaphorical uses than mar de, such as nadar en ‘to swim in’, navegar en ‘to sail in’, sumergirse en ‘to submerge in’, hundirse en ‘to sink in’, etc. Note also that the source semantics of mar de is more general and conceptually more frequent, viz. mar is the typical QN within the maritime domain.

Quantification construal in BQs | 311

‘It even seems that both films could open a new genre, that which asserts a mystery of the meaning, where the ratios are dissolved in an ocean of doubt and uncertainty that controls everything.’ (9)

Es decir, vamos a precisar un poco algo, ¿no?, para no dejar todo siempre en un mar de dudas y confusiones. (1991, oral) ‘I mean, we are going to specify something a little, aren’t we? In order not to leave always everything in a sea of doubts and confusions.’

(10) – Exijo una explicación. – Y te la vamos a dar, Floro – concedió Pacho a regañadientes –, aunque de poco sirven las explicaciones. Estamos en un mar de dudas. (1986, novel) ‘– I demand an explanation. – And we will give it to you, Floro – agreed Pacho unwillingly –, although explanations are of little use. We are in a sea of doubts.’ Additional evidence that adds up to the individual conceptualization of the QNs is indeed to be sought in the contextual clues. It bears pointing out that the following sections will not include a quantitative analysis of the different conceptual facets per QN, since such quantification would inevitably force me to decide whether only indicating explicitly coded conceptual facets or whether counting conceptual facets profiled by the larger discourse context as well. However, I assume in line with García (2009) that “[w]hat amount or part of context is relevant to the interpretation of a particular syntagm is itself contextdependent, and becomes clear only in the process of formulating and interpreting an utterance” (2009, 14). I maintain the differentiation made by Delbecque/ Verveckken (2010) between two types of contextual levels that intervene in meaning construction, i.e. the elaboration by the immediate linguistic context and host class, and actualization by the “mise en discours”. In the following paragraphs, I will first zoom in on the minimal pair alud de – aluvión de (Section 7.2.1), which is the only pair of QNs (of our ad-hoc) selection considered by the native informants to be near-synonymous and freely interchangeable. Given that literal montón de and pila de are near-synonymous and montón de and mogollón de (in informal speech) are considered to be the most frequent and general QNs by native informants, they form the second minimal trio (Section 7.2.2). Section 7.2.3 will briefly overview the image schemas of the remaining QNs analyzed. The observed tendencies will enable me to refine the notion of ‘lexical persistence’ as a gradual and to some extent unpredictable process in Chapter 9.

312 | QN-related schematization and N2-profiling

7.2.1 Construal difference in the minimal pair un alud / aluvión de N2 Strictly speaking, aluvión refers to a flood (possibly caused by heavy rain) and alud depicts a snow slide. Both pertain to the domain of natural phenomena (disasters in this case) which are beyond human control. When used quantitatively, native speakers see only little difference in receiving un aluvión de or un alud de críticas ‘criticisms’. On closer scrutiny, however, un aluvión de críticas appears to be less face threatening than un alud de críticas.186 The diverging combinatorial preferences with verbs and with N2s reveal that both QNs induce their own conceptualization of the mass. While both alud de and aluvión de combine with llamadas ‘phone calls’ and protestas ‘protests’, for instance, the following section will adduce evidence for the unmarkedness of alud de protestas and aluvión de llamadas.187

7.2.1.1 Lexicographical accounts Generalizing over the three dictionaries consulted, alud de is primarily defined as a large mass of snow loosened from a mountain side and descending swiftly and violently from it, as in (11). (11) El 27 de agosto, Xavier Lamas, médico de 31 años, se quedó para siempre allá después de que un alud de nieve lo arrastrara 900 metros desde el pico del Changzeng, situado a 6.800 metros. (1995, press) ‘The 27th of august, Xavier Lamas, a 31-year old doctor, remained there forever, after a snowslide had dragged him 900 meters from the Changzeng peak, situated at 6.800 meters.’ In addition, while DRAE merely mentions that alud can also be used figuratively, CLAVE adds a new separate definition for alud de’s quantifying interpretation as a large quantity of something that appears swiftly, as in Esta semana nos ha llegado un alud de trabajo ‘This week an avalanche of work has reached us’.

|| 186 See also Verveckken/Delbecque (2014). 187 The notion of ‘marked(ness)’ relates to the ‘commonness’ of a linguistic item. A linguistic item is ‘marked’ when it is the unexpected variant for some reason (e.g. register variation, frequency of use…). As to the combinations [aluvión/alud] de llamadas and [aluvión/alud] de protestas, I will explain in the remainder of this section that aluvión de llamadas and alud de protestas are the unmarked options: aluvión de easily combines with neutral or positive N2 entities, whereas nouns such as protestas imply a certain degree of antagonism and opposite views and therefore almost call for alud de.

Quantification construal in BQs | 313

By contrast, the quantifying interpretation of aluvión de is amply recognized in the dictionaries. The first definition yielded for aluvión is the strong or violent flood of water, as in (12). While DUE adds that the flood is typically caused by heavy rainfall, CLAVE highlights its sudden character. In addition, aluvión can refer to the alluvium or deposit of earth or other matter transported by rain or water flowing, as in (13). (12) No obstante, la nieve en las zonas esquiables situadas por encima de la cota 2.500 – Borreguiles, Veleta y Laguna – ha soportado bien el aluvión de agua, aunque únicamente las pistas de Borreguiles están abiertas al público desde el sábado. (1995, press) ‘Nevertheless, the snow in the zones where skiing is allowed situated above the height 2500 – Borreguiles, Veleta and Laguna – has beared well the flood of water, although only the ski runs of Borreguilles were open to the public since last Saturday.’ (13) Los fuertes temporales del invierno han provocado un aluvión de arenas en la desembocadura del río Miño que durante los últimos meses impide faenar con seguridad a los pescadores de A Guarda y la villa portuguesa de Caminha. (2001, press) ‘The heavy winterstorms have provoked a deposit of sand in the Miño estuary that during the last months unables the fishermen of A Guarda and the Portuguese town Caminha to fish safely.’ All three dictionaries add a separate definition line for the quantifying reading of aluvión de (which suggests that the quantifying use of aluvión de is more entrenched than the one of alud de). Both DRAE and CLAVE suggests that the quantifying use typically refers to a large influx of persons or things, by using the examples un aluvión de insultos ‘a flood of insults’ and de felicitaciones ‘of felicitations’ respectively. While DUE emphasizes the ‘extraordinary’ character of the flood of things (such as un aluvión de noticias ‘a flood of news’), CLAVE adds the specification that the entities gather suddenly.

7.2.1.2 The conceptual image of alud and aluvión de Both aluvión de and alud de construe an overwhelming number of N2-entities which suddenly arise all at once, out of nothing, mostly triggered by an external cause (mainly with aluvión de) and generally resulting in drastic consequences

314 | QN-related schematization and N2-profiling

(with alud de). The N2-entities are conceptualized as unstoppable, dynamic188 and antagonistic forces directed towards one single victim or affected person,189 as if they were inundating, respectively snowing under, this person literally. Consequently, N2 is often of unpleasant nature (in particular in combination with alud de). In (14), the new companies suddenly arrive (súbitamente, de repente) and seem to inundate (inundar) the country. From the fact that they fill the newspapers’ cover pages (relegando a los politicos a las páginas interiores de los períodicos) and from the contrast with the previous situation of unemployment, the number of new companies can be inferred to be impressively high. The agricultural and industrial products in (15) are equally conceived as suddenly (repentino) arising, inundating the speaker (inundados) and overwhelming the farmers and businessmen (preparados para competir). This sudden dramatic increase in products is brought about by an external cause (fruto de) and has drastic consequences. (14) Un país tradicionalmente parado, se ponía súbitamente en movimiento y experimentaba en primavera una inesperada floración de tipos y de ideas nuevas, algunas de las cuales madurarán y darán fruto en el otoño e invierno siguientes. Un aluvión de nuevos negocios, y nuevos empresarios, desconocidos meses atrás, parece inundar de repente la escena nacional, relegando a los políticos a las páginas interiores de los periódicos. (1988, books) ‘A country which was traditionally at a standstill got suddenly on the move and experienced in springtime an unexpected flowering of types and new ideas, some of which will come to fruition and bear fruits in the following fall or winter. A flood of new companies, and new businessmen, which only some months before were unknown, seems to suddenly inundate the national scene, relegating the politicians to the inside pages of the newspapers.’ (15) ¿No nos hemos visto "inundados" por un repentino alud de productos agrícolas e industriales, fruto del radical desarme arancelario, ante el que ni nuestros agricultores ni nuestros empresarios estaban || 188 The dynamic nature of N2 is often explicitly mentioned in (1) the type of N2, e.g. visitantes ‘visitors’, inmigrantes ‘immigrants’; (2) postmodifying adjectives, e.g. nuevos ‘new’; (3) the main verb, eg. se le cayeron encima ‘they fell on him’, le vino encima ‘it came upon him’, etc. 189 The notion of ‘victim’ is not the most accurate one, since it evokes a negative halo, whereas aluvión de is also frequently used in positive or neutral contexts. It refers to the endpoint or target of the dynamic forces (cf. infra). However, ‘endpoint’ or ‘direction’ are too vague; they suggest a finished/bounded movement or a controlling subject respectively.

Quantification construal in BQs | 315

preparados para competir? (1995, books) ‘Did we not see ourselves “inundated” by a sudden avalanche of agricultural and industrial products, a result of the radical dismantling of custom barriers, in front of which nor our farmers nor our businessmen were prepared to compete?’ The contexts of (14) and (15) look similar at first sight: both evoke an unexpected and spectacular economic development. Still, replacing the respective QNs by their near-synonym sounds awkward. The fatal consequences of the sudden appearance of the new products in (14) turns alud in the only appropriate QN, since alud de is more decisive, suggesting irreversible consequences. While with aluvión de the N2-entities are also considered to overwhelm the socalled victim, the latter is an antagonist of equal value, able to fight back. Such a positive entailment is incongruent with the specification in (15) that the farmers were not prepared to compete (cf. Section 7.2.1.4). In addition to their dynamic character, the N2-entities quantified by aluvión/alud de are generally considered as uniform items, as forming one homogeneous mass moving into the same direction. Certainly, in (16)–(17), the complaints, respectively news items, are advanced by different traders and businessmen, respectively journalists, but they are all well founded and similar with regard to content. The ‘uniformity’ of the N2s rendered in (18)–(19) is more obvious as their nature or common origin are specified. The homogeneity of the group of N2 becomes particularly clear when replacing aluvión/alud by montón in these examples: the complaints referred to by el montón de reclamaciones would turn into a chaotic heap of claims, the news items would lose in union and strength, the refugees would come from different places and the reactions would be less harmonious. In other words, in these contexts, replacing aluvión/alud by montón has a significant bearing on the imagic conceptualization. (16) La Oficina de Recaudación Municipal atiende estos días el aluvión de reclamaciones de comerciantes y empresarios de Zamora capital a los que se les ha girado más de un recibo de la tasa por recogida de basura en el mismo domicilio fiscal. (2000, press) ‘The office of municipal tax collections attends these days the flood of complaints from shopkeepers and businessmen of Zamora capital who were sent more than one receipt of the tax for garbage collection in the same fiscal domicile. ’ (17) Frente a un “alud de noticias tendenciosas que no encierran siquiera una verdad”, observa, toda tentativa de rectificar resulta inútil. El

316 | QN-related schematization and N2-profiling

propósito de los programadores de la información consiste, como es obvio, en crear una imagen-espantajo de la víctima designada, imagen que debe pegarle a la piel y acompañarle, si es posible a la sepultura. (1977, press) ‘In front of the “flood of tendentious news items that do not even contain any truth”, he observes, any intent to rectify is useless. The aim of the information programmers consists, as is obvious, in creating a phantom-image of the victim, an image that will stick to his skin and accompany him, if possible, to his grave.’ (18) La preocupación de los dominicanos por la inestable situación en Haití hace que estos mensajes tengan un gran efecto. Temen los dominicanos un aluvión de refugiados procedentes del país más pobre del hemisferio. (1994, press) ‘The concern of the Dominicans with respect to the unstable situation in Haiti make those messages have a great effect. The Dominicans fear a flood of refugees coming from the poorest country of the hemisphere.’ (19) Ni el comunicado conjunto entre el director del Instituto Musulmán de la Mezquita de París y el director general de Chanel, firmado el pasado sábado, ha podido frenar el alud de reacciones de la comunidad musulmana internacional, (...). (1994, press) ‘Not even the shared communiqué between the headmaster of the Muslim Institute of the Mosque in Paris and the manager of Chanel, signed last Saturday, could hold back the flood of reactions of the international Muslim community, (…).’

7.2.1.3 The conceptual images broken down into conceptual facets The corpus analysis reveals not only that different (sub)parts of the original frame of alud/aluvión de can become highlighted, but also that the persistence of the original frame is not always equally manifest. In examples (17) and (19), for instance, the drastic consequences (toda tentativa de rectificar resulta inútil and no ha podido frenar) are profiled, while the sudden appearance of the N2entities is less prominent. Similarly, examples (20)–(22) display a different degree of metaphorization of alud de as a huge, sudden number of N2 entities that overwhelm the ‘victim’ as if snowing him under literally. In (20) the modifying adjective verdadero enhances alud’s metaphorical interpretation. In (21) the multiple arrivals of the family members are still felt as an unavoidable and unpleasant reaction triggered by the false assumptions. However, since the family

Quantification construal in BQs | 317

visits are not necessarily spatiotemporally contiguous, the strictly metaphorical interpretation of alud is less straightforward. In (22), the N2 ‘prohibitions’ is completely devoid of proper dynamicity and thus a priori precludes alud’s metaphorical reading. In other words, a wide array of conceptual facets that may persist and come to be highlighted to different degrees suggests that CIP is not a matter of all or nothing. (20) Como anécdota marginal hay que constatar que en los días siguientes a sus apariciones en televisión, llegaba a la oficina del Defensor un verdadero alud de cartas y de reclamaciones, unas justificadas y otras peregrinas. (1986, books) ‘As a marginal anecdote it has to be observed that the days following his television appearances, a real avalanche of letters and complaints, some justified and others outlandish, arrived at the defense counsel's office.’ (21) Sabiendo que esos detalles mínimos, una vez en poder de sus hijos, iban a transformarse en volcánicas y descabelladas hipótesis que escandalizarían al resto de la familia provocando un alud de tíos y primas viniendo a decirle tú estás loca, a quién se le ocurre consentir semejante aventura, la madre debió hablar seriamente con la abuela Africa en su dormitorio. (1987, novel) ‘Since she knew that those tiny details, once in the hands of her children, would transform into volcanic and crazy hypotheses that would shock the rest of the family and provoke an avalanche of uncles and cousins coming to tell her ‘you are mad, to whom does it occur to consent to such an adventure’, the mother had to talk seriously with grandmother Africa in her bedroom.’ (22) Es decir, el Vaticano sabe que el terreno en el que se ha atrevido a opinar es todavía muy movedizo. (...) Quedan, además, abiertas algunas puertas al diálogo. En medio del alud de prohibiciones se dejan ver diferentes grados de rechazo entre lo mismo que se señala como negativo; (...). (1987, press) ‘In fact, the Vatican knows that the field in which it dared express an opinion is still very unstable. (...) Besides, some doors remain open for dialogue. In the midst of an avalanche of prohibitions there is room for varying degrees of rejection even for what is identified as negative; (...).’ In order to compare alud de and aluvión de and to determine which particular conceptual source elements shade through in the quantifying uses, the concep-

318 | QN-related schematization and N2-profiling

tual image evoked can be broken up into several semantic facets. Sometimes all elements are profiled, evidencing ‘strong CIP’ (see infra), but more frequently, particular conceptual facets/dimensions are picked out and stressed (by the choice of N2 or by the additional contextual clues). The context of (23), for instance, shows that the speaker may choose to further adjust (in the following context) the initial image (only vaguely) profiled by the sequence alud de/aluvión de N2, depending on his/her communicative goals or needs: without subsequent clarifying context, it is unclear which element gets profiled in the sequence el aluvión de canales temáticos. Only the specification of the possibility to restrict the children’s access allows the hearer to know for sure that aluvión was meant to profile the collection of thematic channels as an unstoppable mass. (23) La nueva oferta digital sirve a domicilio, entre el aluvión de canales temáticos, deporte de pago y libros de instrucciones, numerosos productos no recomendables para el público infantil, en la mayoría de los casos principales consumidores de televisión en el núcleo familiar. Sin embargo, la innovadora tecnología ofrece a los padres la posibilidad de restringir el acceso a determinados programas mediante la introducción de una clave, un procedimiento similar al que emplean las tarjetas de crédito y que ya está disponible en Vía Digital y en Canal Satélite. Para utilizar este útil dispositivo, denominado "Control paterno", basta con una sencilla operación a través del mando a distancia. (1997, press) ‘The new digital offer delivers home, among the flood of thematic channels, pay sport and manuals, lots of products which are not recommendable for children, which are mostly the principle television consumers within the family unit. However, the innovative technology offers the parents the possibility to restrict the access to certain programs via the introduction of a key, a procedure similar to the one that credit cards use and which is already available in Vía Digital and in Canal Satélite. To use this useful device, called “parent’s control”, only a simple action on the remote control is needed. ’ In the following paragraphs, the observed semantic facets for alud de and aluvión de are listed. It bears pointing out that the semantic facets metaphorically related to alud/aluvión’s source semantics can be profiled at different levels of schematicity, depending on to what extent the speaker abstracts away from the

Quantification construal in BQs | 319

image a literal avalanche/flood activates. This observation fits the gradual character of persistence illustrated in examples (20)–(22).190 In metaphoric quantifying uses of alud/aluvión de N2, the ALL AT ONCE (facet (a)) component may become stressed, reflecting the idea that all N2-entities emerge or arise at the same time. Floods and avalanches can only overflow or submerge country-sides or beings when the vast amount of water/snow arrives at the same moment. The context of (24) makes clear that the utility of the call center relies in the ability to handle various calls simultaneously. In (25), the context pictures a nearly life transmission of the process, broadcasted simultaneously in the whole world. When the speaker abstracts further away from the original frame, the immediacy (‘all at once’) may shade through in the idea that the N2-entities come from DIFFERENT SIDES/SOURCES (facet (h)) at the same time, as in (26). However, the latter metonymic extension is almost exclusively evoked by aluvión de and reminds of amounts of water in the literal frame which wash over the victim(s) from different sides at the same time (in contrast to avalanches which prototypically have a clear starting point). (24) En tal sentido, ha colaborado con Cruz Roja Española en dos proyectos muy útiles: el equipamiento con la tecnología más avanzada de una unidad móvil para casos de desastre y en la instalación de un centro telefónico de última generación que permite gestionar a la vez un aluvión de llamadas. (2002, press) ‘In the same field, it [the company] has collaborated with the Spanish Red Cross in two very useful projects: the equipment with more advanced technology of a mobile unit for cases of disaster and the installation of a call center of the latest generation that allows to handle simultaneously a flood of calls.’ (25) Hoy se cumple medio siglo: a las 10,03 del 20 de noviembre de 1945, en una sala decorada con temas bíblicos, se iniciaba en el palacio de Justicia de Nuremberg el proceso que millones de personas seguirían durante 218 días a través del alud de palabras transmitidas vía teletipo por 450 corresponsales de todo el mundo, (…). (1995, press) || 190 Only few frequency tables of the conceptual facets will be presented in this section, since the exhaustive quantification of the conceptual facets is not instructive nor surveyable. In the following sections, it will become clear that both the context and the choice of N2 co-determine which (set of) facet(s) are profiled in particular occurrences. Technically speaking, the selection of facets can be different for each instance. The tendency to primarily highlight facets metaphorically or only metonymically related to the QN’s source semantics can be derived from the frequency Table 1 in Chapter 9 representing the degree of CIP per QN.

320 | QN-related schematization and N2-profiling

‘Today half a century has gone by: at 10.03 o’clock, on the 20th of November of 1945, in a courtroom decorated with biblical themes, the process began in the courthouse of Nuremberg which millions of person would follow during 218 days via a flood of words transmitted via telex by 450 correspondents over the entire world, (…).’ (26) Como si fuera señal convenida, con la apertura del curso universitario se levantó la veda de los actos culturales. Entre lo que se inauguró y lo que a punto está dentro de unos días, el aluvión de actividades nos llega por distintos frentes, con ímpetu renovado después de la tregua veraniega: exposiciones a lo grande, conciertos, encuentros de teatro, ciclo de conferencias, Congresos internacionales, simposio sobre Murillo... (1982, press) ‘As if it were a signal agreed upon, with the opening of the academic year the season of the cultural events got started. Between what was inaugurated and what was about to be within a few days, the flood of activities reaches us from various fronts, with renewed violence after the summer truce: exhibitions in style, concerts, theater meetings, conference cycles, international conferences, a symposium on Murillo...’ Literal floods and avalanches typically occur suddenly. The abruptness with which the N2-entities appear is another facet metaphorically related to the literal frame that can be activated in quantifying uses. In (27)–(28), de repente and de inmediato are self-evident clues of the fact that N2-entities show up ALL OF A SUDDEN (facet (b)). In (29), the starting point of the flood of congratulations is specified in the subordinate clause initiated by desde que. Often, the triggering factor is profiled rather than the start of the flood itself, or its abruptness, as in (30). Since floods and avalanches are suddenly evoked, they usually take the stricken area by surprise. The semantic facet UNEXPECTEDNESS (facet (g)) can therefore by metonymically related to the QN’s original frame, as illustrated in (31)–(32) A further abstraction of this facet is that the N2-entities are NEW (facet (i)), as illustrated by the premodifier nuevos in (14) or nunca oídas hasta aquellos momentos in (27). (27) ‘Ya no era sólo el padre Antonio el que hablaba descaradamente de “libertad”, de amor, de caridad y de pureza de intenciones. De repente, un aluvión de curas nuevos planteaban soluciones nunca oídas hasta aquellos momentos. (1975, novel) ‘It was no longer only Father Antonio who spoke frankly of “freedom”,

Quantification construal in BQs | 321

of love, of charity and the pureness of intentions. Suddenly, a flood of new priests set out solutions that had never been heard of before.’ (28) Es claro que este tipo de ideas, lanzadas hoy al aire, reciben de inmediato un alud de críticas descalificadoras. (2001, books) ‘It is obvious that this type of ideas, launched in the air today immediately receive a flood of disqualifying criticisms.’ (29) La Familia Real ha recibido un aluvión de felicitaciones desde que se anunció el compromiso de Doña Elena. (1994, press) ‘The Royal Family has received a flood of felicitations since the announcement of the engagement of Doña Elena.’ (30) Los recientes sobresaltos del sistema financiero argentino, con la quiebra de un banco y la intervención estatal de otros tres bancos más, ha provocado un aluvión de visitas a Europa de los responsables económicos. (1980, press) ‘The recent shocks in Argentina's financial system, with the breakdown of a bank and the state intervention for three more banks, has provoked a flood of visits to Europe by the responsible economists.’ (31) A las 9 horas, tres médicos del cuerpo de sanidad militar comenzaron a atender las seis líneas habilitadas del teléfono 91-395-54-85. Sin embargo, el aluvión de llamadas desbordó todas las previsiones, por lo que Defensa reforzó el servicio con otros siete facultativos más, (…). (2001, press) ‘At 9 o’clock, three doctors from the military health force started to operate the six telephone lines supplied for 91-395-54-85. Nevertheless, the flood of phone calls exceeded all expectations. That’s why the Department of Defense reinforced the service with another seven physicians, (…).’ (32) Seat ha decidido aumentar en un 20 % la producción del Marbella, que a partir de hoy pasará de 150 a 180 unidades diarias. La subida de producción de este modelo, que se fabrica en la Zona Franca, se debe al fuerte crecimiento de las ventas, que han superado las previsiones tanto en España como en algunos países europeos, como Francia y Holanda. La fabricación del Marbella debía haber finalizado en 1994, pero se prolongó ante el alud de pedidos recibidos. (1995, press) ‘Seat has decided to increase with 20% the production of the Marbella, that starting from today will go up from 150 to 180 daily units. The rise in production of this model, that is built in the duty free Zone, is

322 | QN-related schematization and N2-profiling

caused by the strong growth in the sales, that have exceeded all previsions both in Spain and in some European countries, such as France and Holland. The production of the Marbella should have stopped in 1994, but has been extended in view of the flood of orders received.’ Depending on whether the aluvión/alud of N2s is experienced from the victim’s or from the trigger’s perspective, the flood/avalanche may be felt as OVERWHELMING (facet (c1)) or UNCONTROLLABLE (facet (c2)). In (33), the uncertainty cannot block or prevent the spectacular emergence of new models, while in (34) measures are taken in an attempt to ‘control’ the continuous flow of visitors. In (31) and (32), the flood of calls and the avalanche of orders exceeds the previsions. In (35) and (36), the attribute impactante and the predicate amenaza give proof of the overwhelming nature of the information and the strangers respectively. Since it is only a small leap from an OVERWHELMING process to the resultant OVERWHELMED state, the conceptual facet OVERWHELMING can be further schematized in the idea that (different) force(s) are DIRECTED TOWARDS ONE VICTIM (facet (e)), as foregrounded in (37). Likewise, the metaphorical interpretation of the N2s as an uncontrollable mass may lead to their characterization as OBTRUSIVE, INSISTENT entities (facet (f)), as illustrated in (38). The most schematized realization of the facet OVERWHELMING is the inference of EXCESSIVENESS, OF TOO MANY/MUCH N2S (facet (j)), as in (39). Usually, this ‘oversizedness’ or ‘extravagance’ is only displayed at discourse level: given that the N2-entities are generally felt as unwelcome, it is just a small leap from unpleasant entities to too many. In fact, any single ‘undesired’ N2 is automatically felt as too much to deal with. (33) La incertidumbre no ha impedido que en Ginebra saliera a la luz un auténtico aluvión de modelos nuevos: las marcas siguen empeñadas en ocupar todos los huecos del mercado para ampliar las ventas y no dar ventajas a la competencia. (2003, press) ‘The uncertainty has not prevented an authentic flood of new models to come out in Geneva: the brands keep determined to occupy all the slots of the market in order to increase the sales and not to give advantages to the competition.’ (34) Estas medidas están ideadas para un mejor disfrute de la colección y también para contener el previsible aluvión de visitas. (1997, press) ‘Those measures were devised for a better enjoyment of the collection and also to restrain the foreseeable flood of visits.’ (35) Nos detenemos un instante, ya que el aluvión de información durante estas siete horas es, cuando menos, impactante. (2002, press)

Quantification construal in BQs | 323

‘We will stop for a moment, since the flood of information during those seven hours is, to say the least, impressive.’ (36) La AFE demanda resultados inmediatos porque considera que el aluvión de extranjeros es una seria amenaza al fútbol autóctono. (1997, press) ‘The AFE requires immediate results as she considers the flood of strangers to be a serious treath for autochthonous football.’ (37) Los primeros días de enero, fue tal el aluvión de llamadas que sufrió dicha compañía, que llegaron a puntas de 100.000 llamadas en un día, con una media de mas de 40.000. (2002, press) ‘The first days of January, the flood of calls that the abovementioned company suffered, was that high that they reached peaks of 100.000 calls in one day, with an average of more than 40.000.’ (38) El aplauso al final de la primera parte, (…), fue el premio al reencuentro de los fans con las viejas canciones del intérprete, que en la segunda mitad del concierto comenzó con su ya clásico tema del primer elepé "In the air tonight". A continuación, un verdadero aluvión de temas potentes con mucha más garra que en la primera parte, lo que, al parecer, hizo despertar hasta a los propios músicos que acompañaban a Phil Collins en el escenario. (1994, press) ‘The applause at the end of the first part, (…), was the award of the reunion of the fans with the old songs of the performer, who in the second half of the concert started with the by now classical theme of the first LP “In the air tonight”. From that moment on, a real flood of powerful themes with much more spirit than in the first part, which, so it seems, woke up including his own musicians who were accompanying Phil Colins on stage.’ (39) Es entonces cuando los locales de la costa reciben un alud de clientes imposibles de atender, (…). (1996, press) ‘At that particular moment, the pubs at the seaside received an avalanche of clients impossible to serve, (…).’ The last quality which can be metaphorically related to alud/aluvión is the effect or RE-ACTION the avalanche/flood constitutes towards an external cause (facet (d2)) or the effect or consequences the avalanche/flood results in (facet (d1)). Depending on the perspective, the focus may fall on the identifiable cause, motivation or trigger of the large number of N2-entities (see (40)–(41)), on the causing role of aluvión itself (see (42)–(43)) or simply on both (as (14), (15) and (44)).

324 | QN-related schematization and N2-profiling

(40) Este cambio provocó en Madrid un aluvión de órdenes de compra y cancelaciones de ventas y llevó al departamento de supervisión de la Sociedad de Bolsas a detener el mercado poco antes de las cinco de la tarde para aliviar la congestión. (1997, press) ‘This change caused in Madrid a flood of purchase orders and sales cancellations and brought the Stock Society supervision department to stop the market shortly before five in the afternoon in order to alleviate the congestion.’ (41) La eclosión en Madrid de tantos grupos provocó un alud de dinero adolescente de cuyo atesoramiento se encargaron diversas cajas recaudadoras con forma de locales nocturnos. (2000, books) ‘The eclosion in Madrid of so many groups caused a flood of adolescent money the hoarding of which was taken care by various saving banks, under the form of nightclubs.’ (42) El anterior, “El cuento hispanoamericano ante la crítica”, publicado en 1973 por la misma editorial, había quedado ya rezagado ante el aluvión de cambios que ha sufrido la crítica del género y el peso específico que ha seguido manteniendo el cuento. (1996, press) ‘The previous one, “The hispanoamerican short story in front of the critics”, published in 1973 by the same publishing house, had fallen back before the flood of changes that the criticism of the genre underwent and before the specific weight that the short story kept on maintaining.’ (43) Pero lo que de verdad hubiera afectado a la estabilidad del país, habría sido el imparable alud de reproches que habría recaído sobre González. (1995, books) ‘But what really would have affected the stability of the country, would have been the unstoppable flood of reproaches that would have fallen upon González.’ (44) El Gobierno reaccionó ayer ante el aluvión de críticas por su comportamiento ante el caos creado por el temporal de nieve y por las acusaciones de los responsables de Interior a los conductores afectados, a los que llegaron a tachar de irresponsables. (2004, press) ‘The Government reacted yesterday to the flood of criticism for its behavior as to the chaos that was created by the snow season and the accusation of the responsibles of Home Affairs to the drivers that were affected, whom they even came to call irresponsible people.’

Quantification construal in BQs | 325

In other words, un alud/aluvión de N2 can give rise to three relational components: a) the external trigger generates the avalanche/flood of N2-entities Xcause >>> Aalud/aluvión b)

the avalanche/flood itself triggers a particular reaction Aalud/aluvión >>> Yeffect

c)

the external trigger X causes an avalanche/flood of N2s which in their turn produce a particular effect Y Xcause >>> Aalud/aluvión >>> Yeffect

Interestingly, the strength of the effect (in both directions) is gradual and often nicely mirrored in the choice of the preposition: in the examples (45)–(49)the effect ranges from an inescapable flood to a flood which barely has an impact. Whereas the preposition por in (45) clearly entails a ‘passive’ reading (the insolvencies and doubtful credits neutralized the achievements), ante is ambivalent with regard to causality as it implies an antagonist. In (46), for instance, Helmut Kohl is described as someone who tries to cope with the complaints and to defend himself. If the sentence would have been permanecer impertérrito por, Helmut Kohl would be interpreted defenseless (which would of course be incongruent with the broader context no pudo permanecer impertérrito). The same holds for frente a in (47). While gracias a depicts a welcome and positive effect of the flood of sunshine in (48), a pesar de questions the vigorousness or influence of the reactions in (49). (45) Sin embargo, estos logros quedaron neutralizados por el aluvión de insolvencias y créditos dudosos, que obligaron al banco a destinar la friolera de 10.000 millones a las dotaciones pertinentes. (1994, press) ‘However, those achievements were neutralized by the flood of insolvencies and doubtful credits, which obliged the bank to allocate no less than 10.000 million to appropriate donations.’ (46) El canciller alemán, Helmut Kohl, no pudo permanecer impertérrito ante el aluvión de quejas y expresó su desacuerdo con gestos como el que muestra la fotografía. (1994, press) ‘The German chancellor, Helmut Kohl, could not remain undisturbed facing the flood of complaints and expressed his disagreement with gestures such as the one that is shown in the photograph.’ (47) Ante casi un centenar de empresarios, españoles y extranjeros, Carlos Solchaga descartó la adopción de medidas defensivas frente al

326 | QN-related schematization and N2-profiling

aluvión de productos importados de la CE en España, “porque eso sería una trampa, en el ambiente de libre competencia que debe regir en un mercado integrado como la CE”. (1987, press) ‘In front of almost a hundred Spanish and foreign businessmen, Carlos Solchaga discarded adopting defensive measures against the flood of products that are imported from the EC to Spain, “because that would be a trap, in the spirit of free competition that should reign in an integrated market such as the EC.”’ (48) Ibiza está que arde, pero no por lo anterior, sino gracias al aluvión de rayos solares que se abate sobre cubierta. (1988, press) ‘Ibiza is on fire, but not because of what's previously mentioned, but thanks to the flood of sunbeams that falls down on the surface.’ (49) A pesar del aluvión de reacciones que han provocado sus declaraciones, el obispo de San Sebastián insistió en que la única manera inicial de buscar la pacificación del País Vasco es sentarse a dialogar con los dirigentes de la banda terrorista, (…). (1997, press) ‘In spite of the flood of reactions that his statements have provoked, the bishop of San Sebastián insisted that the only initial way of searching a pacification of the Bask Country is to sit down to dialogue with the leaders of the terrorist band, (…).’ Further, plenty of examples illustrate that stressing one facet does not necessarily entail that the other ones cease to be profiled. In (50), it is clear that the criticism arose all of a sudden as a reaction to the murder of Yasín, that various countries uttered their disapproval simultaneously towards one person (or ‘victim’ of the criticism) who responded to their opinion by threatening even more persons. (50) El jefe del Ejército de Israel, Moshe Yaalon, aseguró ayer que el presidente palestino, Yaser Arafat, y el líder de Hizbulá, Hasan Nasrala, "han entendido que son los siguientes de la lista", tras el asesinato el pasado lunes del dirigente y fundador de Hamas, el jeque Ahmed Yasín. Lejos de replegar su ofensiva, después de un aluvión de críticas internacionales por el asesinato de Yasín, Israel acrecentó sus advertencias incluyendo, además, a todos los dirigentes del grupo Hamas en su "lista negra" de objetivos, entre ellos el futuro jefe de la organización: Abdel Aziz Rantisi. (2004, press) ‘The head of the Army of Israle, Moshe Yaalon, affirmed yesterday that that the Palestine president, Yasser Arafet, the leader of Hezbola, Hasan Nasrala, “have understood that they are the next ones on the list”, after the murder last Monday of the leader and founder of Ha-

Quantification construal in BQs | 327

mas, the sheik Ahmed Yasín. Far from drawing back their offensive, after a flood of international criticism because of the murder of Yasín, Israel increased its warnings including, in addition, all leaders of the Hamas group in their “black list” of objectives, among them the future head of the organization: Abdel Aziz Rantisi.’ Before turning to the more configurational and generalizable facets that belong to the image schemas of alud de and aluvión de, Figure 1 provides an overview of the more contentful facets and distinguishes between three levels of schematicity. The summary generalizes over the different preferences of alud de and aluvión de. The overview for alud de is almost identical, but does not contain facet (h). (a) all at once (b) all of a sudden (c1) uncontrollable (c2) overwhelming

(h) DIFFERENT SOURCES (g) UNEXPECTED (f) OBTRUSIVE, INSISTENT N2 (e) DIRECTED TOWARDS A SINGLE VICTIM

(i) NEWNESS OF N2 (j) TOO MANY/MUCH

(d1) CAUSER: aluvión is trigger of (re)action/situation (d2) CAUSEE: identifiable cause triggers/motivates aluvión Fig. 1: Conceptual facets in the conceptual image of aluvión de

In the search for possible differences in the conceptualization of alud and aluvión, more configurational features such as ‘aspect’, ‘gradualness’ and ‘perspective’ have been taken into account as well. With regard to ‘aspect’, the underlying hypothesis is that the arrival of N2s introduced by aluvión can occupy a certain time span and may even be conceptualized as a gradual phenomenon (in contrast to alud). In line with the event phases and aspectual modalities as defined in Delbecque/Verveckken (2010), I observed instances of ‘globalization’ (the entire process is profiled, abstracting away from the different event phases) as in (51); of ‘ingressive aspect’ (the threshold is reached, the process is not completed but has definitely taken off) as in (52); and ‘realization’ (achievement of some duration) as in (53), in addition to instances where no particular aspect is profiled. When profiled, the aspect is generally indicated by the tense of the main verb, the auxiliary used (as in pasado ya el aluvión de ‘gone by already the flood of’, empiezan a llegar ‘start to reach’, ya se inició el alud ‘the alud already started’, etc.) or by the lexeme of the main verb (e.g. continúa el aluvión ‘the flood goes on’, se desató en un aluvión de críticas ‘a flood of criticism had broken out’ ), and by prepositional phrases or adverbs (e.g. mientras tanto ‘in the meantime’).

328 | QN-related schematization and N2-profiling

(51) Mientras tanto, continúa el aluvión de mamíferos clonados o de experimentos con clones. (1998, press) ‘In the meantime, the flood of cloned mammals or of experiments with clones goes on.’ (52) Cuando Mossén Ballarín (Barcelona, 1920) sale de los estudios de televisión donde ha sido entrevistado, un aluvión de personas se le acercan. (1996, press) ‘When Mossén Ballerín (Barcelona, 1920) leaves the television studios where he has been interviewed, a flood of persons come to him.’ (53) “Ahora, a trabajar”, volvió a ordenar personalmente, al menos a un par de consejeros, entre el aluvión de plácemes y abrazos. (2001, press) ‘“And now, let’s work”, he ordered again in person, at least to some ministers, in the middle of the flood of congratulations and embraces.’ Frequency Table 2 reveals that aluvión de indeed more frequently exploits its aspectual potential, whereas alud de seems to stick to a globalizing view, and less frequently also evokes realization. This finding is congruent with the observation that aluvión de is more easily conceived as a gradual phenomenon. The facet ‘gradual’ refers to whether the avalanche/flood is conceptualized as a single point on a time scale (as in (54)–(55)) or occupies a certain time slot (as in (56)–(57)), regardless of the aspectual value realized. While the unexpected appearance of the N2-entities quantified by aluvión de is profiled as a gradual process in 85% of the occurrences (in 112 of 132 grammaticalized occ.), only 37% (24 of 65 occ.) of the quantifying/specifying uses of alud de refer to gradual processes. (54) En el número de hoy del citado Journal se publica un aluvión de cartas en las que se reconoce la importancia del estudio y se enfatiza la necesidad de nuevas investigaciones que (…). (1997, press) ‘In today’s issue of the cited Journal a flood of letters is published in which the importance of the analysis is recognized and the necessity of new investigations that (…) is stressed.’ (55) Pero ha triunfado la fuerza mercantil – (…) – sobre toda lógica, y los indefensos escolares se ven cada año sepultados por un alud de volúmenes que les obliga a llevar gigantescas mochilas de excursionistas alpinos. (1997, press) ‘But the mercantile force – (…) – has triumphed over logic, and the defenseless schoolchildren see themselves buried every year by an ava-

Quantification construal in BQs | 329

lanche of volumes that obliges them to bring along giant backpacks of Alpine hikers.’ (56) Con el evidente propósito de neutralizar el aluvión de voces críticas que desde hace semanas le acusan de traicionar su programa electoral, dar marcha atrás y desdecirse continuamente, el presidente afirmó tajante: “…”. (1996, press) ‘With the obvious aim of neutralizing the flood of critical voices that since weeks are accusing him from betraying his electoral program, from constantly going backwards and coming back on his words, the president sharply affirmed: “…”.’ (57) A partir de ahí, un alud de papeles, que en los dos primeros meses de este año se ha incrementado en un 40% con respecto a 1983, y unos personajes que parecen sacados de una película de Berlanga. (1984, press) ‘From there on, a flood of paper, that in the first two months of this year has grown with 40% with respect to 1893, and some characters seem to be taken from a movie of Berlanga.’ Tab. 2: Distribution of the aspectual values in the functional uses of alud de and aluvión de Alud de Globalization Ingressive Realization Not applicable Total

# 43 6 15 1 65

% 0.66 0.09 0.23 0.02

Aluvión de # 48 30 43 11 132

% 0.36 0.23 0.33 0.08

The final conceptual facet quantified concerns the perspective or the viewpoint from which the alud/aluvión is approached: the viewpoint may either be the victim’s perspective, the perspective of (the source of) N2, of both or of neither of them. Table 3 provides a detailed picture of the perspective preferences per QN. Given the general focus on the drastic consequences of the alud/aluvión, it does not come as a surprise that the victim’s perspective is most frequently adopted. However, since alud de is claimed to be more decisive (see infra)) than aluvión de, I had expected a relatively more outspoken preference for the victim’s viewpoint with alud de (in contrast with aluvión de). (58) Un aluvión de enfermos, la mayoría personas mayores con infecciones respiratorias de carácter vírico, han sobrecargado los

330 | QN-related schematization and N2-profiling

servicios de Urgencias y han llenado los hospitales de Gipuzkoa. (2004, press) ‘A flood of ill people, the majority elderly people with viral respiratory infections, have overloaded the Emergency services and have filled the hospitals of Gipuzkoa.’ (59) Eso trata de probar esta muestra, que la fotografía de hoy se ha dotado de movimiento alejándose, en parte, de objetivos conceptuales, pero ofreciendo un aluvión de sugerencias. (2003, press) ‘That is what this sample tries to demonstrate, that today’s photography has endowed itself with a movement further away, partially, from the conceptual objectives, but offering a flood of suggestions.’ (60) Así, el fin del “apartheid” –(…)– ha empujado al país a una apertura económica que tiene su reflejo también en el mercado de valores, sometido a una liberalización parecida al “big bang” londinense que atrae un aluvión de nuevos inversores extranjeros. (1994, press) ‘That’s how the end of the “apartheid” –(…)– has pushed the country towards an economic opening that is also reflected in the stock market, subject to a liberalization similar to the London “big bang” that attracts a flood of new foreign investors.’ (61) A partir del 92, ha habido un aluvión de elecciones. Dos años en los que casi todos los países del Africa negra han adoptado la solución que Occidente les imponía. (1995, press) ‘From 92 onwards, there has been a flood of elections. Two years in which almost all countries of black Africa have adopted the solutions that the West imposed on them.’ Tab. 3: Distribution of the viewpoint in the grammaticalized uses of alud de and aluvión de Alud de Victim (Source of) N2 Both Not applicable Total

# 42 8 11 4 65

% 0.65 0.12 0.17 0.06

Aluvión de # 84 19 9 20 132

% 0.64 0.14 0.07 0.15

In sum, the conceptual image as outlined for aluvión de in Figure 1 and the conceptual image of alud de resemble each other. In order to fit the image of alud de, two minor modifications are in order, viz.: (1) facet (h) – profiling the N2-

Quantification construal in BQs | 331

entities as arriving from different sources at the same time – does not appear; (2) facet (f) – which refers to the insistent or decisive nature of the N2-entities – is more prominent in the case of alud de. Recall that the examples illustrating variation in preposition (which corresponds to weakening the impact of the victim, see (47)–(51)) are all instances of aluvión de. The following section zooms in on the major differences in conceptualization between both nearsynonymous QNs.

7.2.1.4 Differences in the image schemas of alud de and aluvión de This section highlights the main differences between the conceptual images evoked by aluvión de and alud de. More precisely, the image-schematic structures of both QNs diverge in (1) the orientation and decisiveness of the forces or dynamic N2-entities and in (2) graduality. First, the orientation of the forces or dynamic entities crucially differs. In line with the source semantics of aluvión, the N2-entities are conceptualized as coming from different sides or starting points at the same time (“overflowing” very much like water does, viz., from all parts). The quantity assessment by alud de, in contrast, evokes a common starting point for the N2s. Examples (62)–(63) both illustrate a frequent usage context of these near-synonymous QNs: a politician receives a lot of criticism for having made a polemical statement or taken a questionable measure, and afterwards feels obliged to react to the criticism in an attempt to save his face. The subtle differences brought about by the choice of the QN reside in the conceptualization of all parties settling in the opposition as one single block in (63), whereas in (62) various government spokesmen are conceived of as emitting their proper criticism, possibly at different moments in time. This tendency does not imply that the opposition in (63) merely consists of one party nor that all parties expressed their criticism at the same time. The use of alud de simply conceptualizes them as one criticizing entity, which makes the criticizing reaction appear as even more decisive. Generalizing over the conceptual facets that can live on in alud de and aluvión de, Figure 2 visualizes the starting point(s) of the N2-entities and their orientation towards the ‘victim’. (62) Diversos portavoces del Ejecutivo subrayaron el compromiso de Bonn de cumplir los acuerdos con los socios europeos y garantizaron la solidez de la moneda única. Scharping volvió a la carga ayer con una entrevista concedida al diario “Bild”. Ante el aluvión de críticas recibidas, sostuvo que “el SPD debe alertar sobre los posibles riesgos de la introducción de la moneda europea. Ese es nuestro deber”. (1995, press)

332 | QN-related schematization and N2-profiling

‘Various spokespersons of the Government stressed Bonn’s commitment to keep the agreements with the European partners and guaranteed the solidity of the common currency. Scharping returned to the attack yesterday with an interview conceded to the newspaper “Bild”. In view of the flood of criticisms received, he sustained the idea that “the SPD has to alert about the possible risks of introducing the European currency. That’s our duty”.’ (63) Su discurso, adornado con alabanzas a Irlanda por rechazar el Tratado de Niza y el anuncio de su intención de eliminar los delitos de injurias a la bandera y ataque a la unidad del Estado, desató ayer un alud de críticas en la oposición. (2001, press) ‘His speech, embellished with words of praise on Ireland for rejecting the Treaty of Nice and his announcement of the intention to eliminate the crimes of insult to the flag and of attack of the State’s unity unleashed yesterday an avalanche of criticism in the opposition.’

un alud de N2

un aluvión de N2

Fig. 2: Different construal of the source of N2s and direction of the forces

Another major difference concerns the gradualness of N2’s arrival. Alud de presents the sudden appearance of N2 as occupying only a limited time slot. The quantity assessment by aluvión de, in contrast, is typically associated with a sense of continuity, which endows the QN with a wide aspectual potential. In (64), the judge's decision to cancel the press conference is tantamount to admitting that the criticisms have attained their goal and that the battle is over, so to say.191 In (65), issuing an ‘avalanche’ of denials is presented as a governmental strategy (possibly repeated on other occasions): appearing as bounded, with a clear starting and end point, it forces the interlocutor to process the denials as a whole. In contrast, producing un aluvión is not necessarily delineated, as is || 191 The temporal specification (tras 'after') further confirms that the criticisms were decisive for the judge to revise his plans.

Quantification construal in BQs | 333

corroborated by contextual indices in (66)–(67). In (66) ((51), repeated for convenience), both mientras tanto and the predicate continúa suggest an ongoing process. In (67) ((37), repeated for convenience), the aluvión of phone calls extends over various days (los primeros días de enero), with an average number of calls per day. By way of comparison, using alud de in (67) would have suggested that the calls were over immediately after the first days of January. (64) Se da la circunstancia de que la entrevista publicada ayer fue concedida a “ABC” el sábado, el mismo día que el magistrado decidió no llevar a cabo una rueda de prensa que él mismo había convocado en la Audiencia Nacional, tras el alud de críticas producido contra él la semana pasada. (1995, press) ‘It so happened that the interview published yesterday was conceded to “ABC” on Saturday, the same day that the judge decided not to hold a press conference that he had himself convoked at the National Court, after the avalanche of critiques produced against him last week.’ (65) El proceder seguido por el Gobierno es el de producir un alud de supuestos desmentidos y acusaciones de falsedad, formulados en un tono de aparente firmeza para crear confusión. En unos casos, refuta detalles nimios y en otros, da muestras de clara mala fe. (1994, press) ‘The Government’s way of acting is to produce an avalanche of false denials and accusations of falseness, formulated in a tone of apparent firmness to create confusion. On some occasions, they refute trivial details, on other occasions, they give proof of bad faith.’ (66) Mientras tanto, continúa el aluvión de mamíferos clonados o de experimentos con clones. (1998, press) ‘In the meantime, the flood of cloned mammals or of experiments with clones goes on.’ (67) Los primeros días de enero, fue tal el aluvión de llamadas que sufrió dicha compañía, que llegaron a puntas de 100.000 llamadas en un día, con una media de más de 40.000. (2002, press) ‘The first days of January, the flood of calls that the abovementioned company suffered, was that high that they reached peaks of 100.000 calls in one day, with an average of more than 40.000.’

7.2.1.5 Context and combinatorial pattern of alud de and aluvión de The commonalities as well as the differences in the conceptualization conveyed by the near-synonyms alud de and aluvión de are reflected in their contexts of

334 | QN-related schematization and N2-profiling

usage, more specifically in the verbs and right collocates they frequently combine with. First, there is a large overlap in verb combinations. Against their common background of profiling the sudden, usually provoked, arrival of dynamic (antagonist) N2-entities which take the stricken area or ‘victim’ by surprise with drastic consequences, it is no surprise that alud de and aluvión de share a number of semantically related verb clusters: as shown in Table 4, they both combine with verbs of reception, verbs of triggering, verbs of re-action, verbs specifying the drastic consequences, and verbs reflecting the source semantics of the QN, though in different proportions.192 The most frequent combinations for both QNs are recibir ‘to receive’ un alud/aluvión de N2 (yielding 5 and 10 occ. respectively, cf. (62)) and provocar ‘to provoke’ un aluvión/alud de N2 (with 4 and 8 occ., cf. (68)–(69)). Tab. 4: Verb clusters associated with alud de and aluvión de Alud de Vs of reception

Aluvión de

(5 occ.) recibir ‘to receive’

(12 occ.) recibir ‘to receive’, conocer ‘to know, to experience’, ofrecer ‘to offer’ Vs of triggering (12 occ.) provocar ‘to provoke’, (10 occ.) provocar ‘to provoke’, producir ‘to produce’, ocasionar ‘to atraer ‘to attract’, levantar ‘to raise’ bring about’, disparar ‘to shoot’, desatar ‘to unleash’, descargar ‘to fire’, lanzar to launch’ Vs of re-action (5 occ.) frenar ‘to brake’, compensar (19 occ.) reaccionar to react’, volver a ‘to compensate’, ceder el paso a ‘to la carga ‘to return to the attack’, step aside for’, soportar ‘to stand’, enfrentarse ‘to face’, atender ‘to attend’, parar ‘to stop’ aguantar ‘to take, to endure’ Vs pointing to (17 occ.) obligar ‘to force’, quedar + (8 occ.) hacer ‘to make, to cause’, drastic ADJ ‘to remain, become ADJ’, hacer quedar +ADJ ‘to turn, become ADJ’, consequences que +INF ‘to cause’, afectar ‘to affect’ impedir ‘to prevent’, venir determinado por ‘to be determined by’ Vs reflecting the (12 occ.) colapsarse ‘to collapse’, (25 occ.) sobrecargar ‘to overload’, QN’s source hundir ‘to sink, to destroy’, caer ‘to llenar ‘to fill’, vertirse sobre ‘to flow semantics fall’, volcar sobre ‘to tip, dump onto’, over’, caer sobre ‘to fall over’, venir invadir ‘to invade’ encima a ‘to overtake’, sobrevenir ‘to strike’, invadir ‘to invade’, inundar ‘to inundate’, perturbar ‘to disturb’

|| 192 Aluvión de N2 also co-occurs with verbs of ‘expectation’, thereby highlighting the victim's vantage point vis-à-vis the arrival of an oversized number of N2 entities (e.g. esperarse ‘to be expected’, temer ‘to fear’, sorprenderse de ‘to be surprised of’).

Quantification construal in BQs | 335

(68) Los recientes sobresaltos del sistema financiero argentino, con la quiebra de un banco y la intervención estatal de otros tres bancos más, ha provocado un aluvión de visitas a Europa de los responsables económicos. (1980, press) ‘The recent shocks in Argentina’s financial system, with the breakdown of a bank and the state intervention for three more banks, has provoked a flood of visits to Europe from the responsible economists. (69) Es su obstinación por acabar la legislatura, careciendo del apoyo social necesario para ello, lo que puede provocar que esto termine como el rosario de la aurora y un alud de nuevos escándalos y revelaciones le obligue a agachar infamantemente la cabeza. (1995, books) ‘It is his stubbornness to finish the term, without having the social support required to do so, which could well make that this ends in disaster and that an avalanche of new scandals and revelations forces him to defamatorily bend his head.’ The major, though subtle, differences surface in the lexical richness of the verbs of triggering and in the type of re-action verbs.193 First, the cluster of verbs meaning ‘to trigger’ displays more lexical variety in the case of alud de, which is in line with the importance alud’s original frame attributes to the three participant roles ‘cause – avalanche – consequence’. Second, the verbs expressing some kind of reaction by the ‘victim’ (or the participant affected by the aluvión/alud de N2’s) profile a different type of ‘victim’: the verbs combining with aluvión de depict an antagonist of equal value, who faces the aluvión de N2s; the verbs patterning with alud de, in contrast, evoke a beatable victim, who undergoes the alud de N2s. One final observation does not concern the verbal predicate in the first place, but rather the participles (sometimes preceded by e.g. quedarse ‘to remain, to become’ or another (pseudo)copular verb) modifying alud’s victim: in over a dozen cases they highlight its weakness (e.g. sepultado ‘burried’, inundado ‘inundated’, preocupado ‘worried’, paralizado ‘paralyzed’, etc.). The focus on the victim’s state is in line with alud’s tendency to produce fatal consequences.

|| 193 It is not worthwhile dwelling on the exact number of combinations, since the total varies not only in function of the communicative needs of the speakers in the particular context but also in view of the total number of occurrences of the QNs (the grammaticalized uses of aluvión de are more or less two times as frequent as the grammaticalized uses of alud de (132 and 65 occ. respectively). For the same reason, Tables 4 and 5 do not exhaustively represent the combinations attested: per cluster, the items listed in Tables 4 and 5 appear according to decreasing token-frequency; for the items yielding only one attestation in the corpus, a representative selection has been made.

336 | QN-related schematization and N2-profiling

The collocational pattern to the right similarly reflects aluvión’s and alud’s original frame. In line with the enhanced antagonism profiled by alud, alud de leaves less room for positive or neutral N2s. For both QNs, the most important N2-cluster denotes verbal reactions such as llamadas ‘phone calls’, felicitaciones ‘felicitations’, palabras ‘words’, etc., which are conceptualized as a large number of N2s addressed by a certain addresser to an addressee (the victim), generally as a reaction to impactful events. An important subcluster which groups ‘negative verbal reactions’ such as protestas ‘protests’, acusaciones ‘accusations’, quejas ‘complaints’, etc., constitutes for alud de the almost exclusive subcluster of verbal reactions. Interestingly, the most frequent N2 observed for both QNs is críticas ‘criticisms’ (with 13 and 4 occ. respectively). The second most frequent cluster for aluvión consists of people invading the victim’s domain or interrupting him, such as turistas ‘tourists’, colonizadores ‘colonists’, inmigrantes ‘immigrants’, etc., while alud de only combines with intruder N2s on four occasions. The second most frequent N2-cluster for alud contains political or economic actions or products, such as productos ‘products’, proyectos ‘projects’, ofertas ‘offers’, elecciones ‘elections’, negocios ‘business deals’, nombramientos ‘appointments’, escándalos ‘scandals’, pactos ‘pacts’, etc., and is also frequently associated with aluvión de. Tab. 5: N2-clusters associated with alud de and aluvión de Alud de

Aluvión de

(49 occ.) llamadas ‘phone calls’, Verbal reactions (19 occ.) llamadas ‘phone calls’, cartas ‘letters’, palabras ‘words’ carcajadas ‘loud laughter’ negative: (15 of 19 occ.) hipocresías negative: (24 of 49 occ.) críticas ‘criticism’, quejas ‘complaints’, ‘hipocrisies’, insultos ‘insults’, críticas ‘criticism’ protestas, ‘protests’ People (4 occ.) clientes ‘clients’, tíos y tías (21 occ.) enfermos ‘sick people’, ‘uncles and aunts’, corresponsales colonizadores ‘colonizers’, turistas ‘correspondents’ ‘tourists’, extranjeros ‘strangers’ (sources of) (10 occ.) noticias ‘news’, (7 occ.) datos ‘data’, documentos information propaganda ‘propaganda’, ‘documents’, páginas web ‘web informaciones ‘informations’ pages’ Politic/economic (14 occ.) pactos ‘pacts’, (29 occ.) productos ‘products’, actions or products prohibiciones ‘bannings’, demandas offertas ‘offers’, proyectos ‘projects’, ‘demands’, normas ‘norms’ elecciones ‘elections’, cambios money: (1 of 14 occ.) dinero ‘money’ ‘changes’ money: (4 of 29 occ.) dinero ‘money’, anticipos ‘advances’, fondos ‘funds’ Abstract N2s (6 occ.) ideas ‘ideas’, miradas (5 occ.) pensamiento ‘thoughts’, (usually ‘looks’, voces ‘voices’, imágenes hermosura ‘beauty’ experiences) ‘images’, sentimientos ‘feelings’ other

(12 occ.)

(21 occ.)

Quantification construal in BQs | 337

It goes without saying that the internal focus on antagonism or cause-reactionconsequence correlations and the potential to construe oversized masses of N2 appearing out of nothing make aluvión de and alud de particularly appropriate quantifiers in political and economic contexts, where participants are usually diametrically opposed and remarkable phenomena concern spectacular growth or inflation. While aluvión de rather stresses the idea that the huge number of N2-entities is triggered or brought about, alud de rather focuses on the huge consequences the N2-entities bring along. The final shared important N2-cluster contains the N2s referring to some source of information, such as informaciones ‘information’, noticias ‘news’, páginas web ‘web pages’, canales ‘channels’, etc. The remaining N2s, which at first sight seem to fall outside the above mentioned clusters, appear in contexts that also make them being conceived of as generally unpleasant, dynamic entities which overtake the victim by surprise. The treatment avalancha ‘avalanche’ receives in REDES dictionary (Bosque 2004) corroborates our analysis. REDES sets out to outline the combinatorial pattern of its entries. Of all QNs analyzed in this monograph, only the combinatorial preferences of alud and aluvión are recognized as such. For both lexemes, the dictionary refers to the entry of avalancha. As to “avalanche (de)”, REDES (2004, 361) stipulates that it combines both with plural count nouns and with singular mass nouns. In its “physical” reading, it combines with nieve ‘snow’, lluvia ‘rain’, hielo ‘ice’, lodo ‘mud’, etc. and other material which can move, loosen and come down. In addition, avalancha combines with human entities, mainly nouns designating human beings who are moving (e.g. turista ‘tourist’, viajero ‘traveller’, inmigrante ‘immigrant’, etc.) or who are going to a specific place (e.g. votante ‘voter’, curioso ‘curious person’, paciente ‘patient’, cliente ‘client’, periodista ‘journalist’, etc.). Further, avalancha frequently combines with nouns designating “verbal manifestations” (e.g. rumores ‘rumors’, palabras ‘words’, etc.), other “objects of information” (e.g. libros ‘books’, películas ‘movies’, periódicos ‘newspapers’, etc.) and with “sensations or vivid feelings” (e.g. alegría ‘happiness’, entusiasmo ‘enthusiasm’, interés ‘interest’, etc.). The dictionary concludes by listing three particularly frequent “lexical families”: (a) nouns which designate manifestations of disagreement as well as some official or administrative actions that usually result from those manifestations (e.g. crítica ‘criticism’, denuncia ‘report, formal complaint’, queja ‘complaint’, protesta ‘protest’, etc.); (b) nouns which designate requests or applications, offers and varied ways of notification: (e.g. petición ‘request’, pedido ‘order’, llamada ‘call’, etc.); (c) nouns which designate something considered as new or presented as such (e.g. novedad ‘novelty’, noticia ‘news item’, revelación ‘revelation’, etc.).

338 | QN-related schematization and N2-profiling

7.2.1.6 Concluding remarks In sum, the conceptual images alud de and aluvión de impose on the N2-entities is thus far more complex and contentful than simply ‘a large number of N2s that arrives in a violent and turbulent way’, as suggested by the reference dictionaries. Although alud de and aluvión de evoke similar conceptual images, and consequently show up in similar contexts, they diverge as to preferred usage contexts and to the set of conceptual facets preferably highlighted. Returning to the most frequent shared right collocate críticas, the combination with aluvión de pictures the addressee as capable of confronting them and, by doing so, of saving his face, whereas in un alud de críticas, the criticism definitively ruins his reputation. The major differences in conceptualization can be summarized as follows: a) Both QNs focus on the abruptness of N2’s rise in number. Whereas aluvión de generally implies that there is an identifiable source, alud de stresses the drastic consequences. Alud de further adds the idea that these consequences are irreversible, viz. that the clock cannot be turned back. b) Aluvión de may also be used to quantify ‘pleasant’ entities. c) The source of N2-entities (or at least the direction they come from) is usually conceptualized as multiple for aluvión de (N2s come from different sides at the same time) and as single for alud de.194 d) The sudden appearance of N2s quantified by aluvión de may occupy a certain time span, which endows aluvión de with an aspectual potential. One final issue concerns two quantifying uses of aluvión de which are equally reminiscent of its reading as a ‘flood’ or as a ‘alluvium, deposit’. In (70), metaphorical flood refers to the interference between the sentimental uneasiness and the literary tradition. The ambiguity is triggered in (70) by the predicates petrificación and sedimentos sucesivos and in (71) by conllevar and acumulación respectively. Note that in (71), the N2s are clearly not conceptualized as dynamic. (70) La desazón sentimental ha interferido de tal modo en la tarea literaria, que son obvios el aluvión de situaciones forzadas y la petrificación de las figuraciones en sedimentos sucesivos. (1987, novel) || 194 More precisely, the total number of instances where the feature ‘different sources’ is conceptualized for aluvión de amounts to 92/132 (69,7%), in contrast to the total number of instances where the feature ‘single starting point’ is conceptualized for alud de, viz. 45/65 (69,2%).

Quantification construal in BQs | 339

‘The sentimental unease has interfered in the literary task in such a way that the flood of forced situations and the petrifaction of the figurations in successive sediments are obvious.’ (71) Esto es innegable, pero el español crece en forma más afirmada y profunda. Los mexicanos, argentinos, españoles y colombianos que lo hablan comparten mucho más que un mero sistema de señales fonéticas. Conlleva un gran aluvión de materia cultural, una heteróclita acumulación de valores, restos y renuevos de metafísica y religiosidad. (1989, press) ‘That is undeniable, but Spanish grows in a more steady and deep form. The Mexicans, Argentinians, Spaniards and Colombians who speak it share much more than the mere system of phonetic signs. It brings a big flood of cultural material, a heteroclite accumulation of values, the remains and renewals of metaphysics and religiosity.’ The examples raise an important theoretical question: do (70) and (71) instantiate different uses (or even constructions) of quantifying aluvión? Since similar instances do not occur in the diachronic analysis and those instances have emerged later than the quantifying instances of aluvión (as a flood), the quantifying instances of aluvión (as alluvium) could be considered a further extension of aluviónflood, after the latter became conventionalized. Instances of aluviónalluvium would then constitute a further exploitation of the entrenched quantifier aluviónflood in an attempt to satisfy the expressive needs of the speaker. However, it goes without saying that the quantifying potential of aluviónalluvium is less obvious than the scalar implicatures of aluviónflood. It can be also argued that the quantifying interpretation in the examples is triggered by the contextual clues de tal modo que son obvios, mucho más and gran. This reasoning immediately nuances the grammaticalized character of quantifying aluviónalluvium. Furthermore, the fact that only two (ambiguous) instances are encountered in our corpus and that, conceptually speaking, it is only a small leap from a flood to the metonymically related alluvium (which is always caused by a flood), makes me argue for considering them peripheral examples of quantifying aluvión (instead of as a separate use).

7.2.2 Construal difference in the minimal pairs una pila/un montón de N2 and un montón/mogollón de N2 The present section zooms in on the construal differences between the prototypical montón de and two near-synonyms, one of which is literal and one of which

340 | QN-related schematization and N2-profiling

is functional. More precisely, literal pila de ‘pile of’ is synonymous to the configurational sense of montón de as a literal heap of things put one above another. Mogollón de is semantically near to the quantity assessment expressed by montón de, and usually considered informal or colloquial.195

7.2.2.1 Lexicographical accounts All three reference dictionaries distinguish between two major senses for pila, depending on whether it refers to a kind of sink or to a stack of things. Although pila as a sink refers to a container and by definition has a quantifying potential, the quantifying use build on its configurational reading as a pile or solid cone. Literally, pila refers to a manmade collection or heap of things put in (serial (DUE)) order one above another, as in pila de lana ‘wool’, ladrillos ‘bricks’, platos ‘dishes’. It is usually of considerable height and obviously vertically oriented. Interestingly, both DUE and DRAE list the quantifying reading of pila in the cluster of readings related with the configuration sense, which is at the same time the first cluster described in both dictionaries. While DUE defines pila as ‘an accumulation of things’ (as in Tengo una pila de cosas que hacer ‘I have a lot of things to do’), DRAE considers quantifying pila as synonymous to montón which means ‘a considerable number’ (as in pila de años, de gente ‘a lot of years, people’) and the last definition listed by CLAVE defines una pila as ‘large quantity’ (as in Este coche tiene una pila de años ‘That car is already very (lit. a lot of years) old’).196 In other words, all three reference dictionaries suggest that the quantity assessment in pila de has already conventionalized. In addition, all three indicate that quantifying pila de belongs to informal or colloquial language use. The entries of montón are organized in a similar way. All three dictionaries first describe its literal reading as a collection of things put one above another without order, as in the example provided by CLAVE Tenía en la mesa un montón de papeles ‘He had on his desk a heap of paper’. DUE further specifies that montón is roughly conical in form and that the constituting elements may be of any form or size. Interestingly, DUE also considers montón as a more general term for a collection of entities in the form of a pile or column. || 195 The QN mogollón has been added to my database after several native speakers pointed out that in informal language use, mogollón de is as general and as frequent as montón de. 196 Interestingly, both in DRAE and in DUE, the third definition in the ‘configuration’-cluster refers to the collection of wool of one shear-season per owner, which also corresponds to the first (binominal) occurrence of pila de observed in CORDE (cf. Section 5.2.1).

Quantification construal in BQs | 341

Again, all three dictionaries relegate the quantifying reading to colloquial or informal language use. While DRAE describes montón as ‘a considerable number’ as in Tengo que decirte un montón de cosas. ‘I have to tell you a lot of things’, CLAVE defines the quantity assessment as ‘large quantity or abundance’, as in Se gastó un montón de dinero en regalos ‘He wasted a lot of his money on gifts’. DUE lists ‘a large number or quantity of things’ (e.g. Tengo un montón de cosas que decirte ‘I have to tell you a lot of things’), synonymous to mucho ‘many, much’, especially when referring to human entities. With respect to mogollón, all three dictionaries proceed in a similar way. The first definition for mogollón as a noun is ‘large quantity of something’. While CLAVE adds the idea that the collection is in disorder, DUE mentions a preference for persons and things (as in Al concierto fue un mogollón de gente ‘A heap of people went to the concert’). In addition, the reference dictionaries mention the reading of noise, bustle or mess caused by the agglomeration of a many people. Finally, mogollón can also be used adverbially, as in Me gustó mogollón ‘It pleased me a lot’, and means ‘a lot, abundantly’.197 Crucially, all senses of mogollón are attributed to colloquial speech. In sum, the fact that all three dictionaries list the quantifying readings of pila, montón and mogollón in a prominent way in addition to providing binominal examples, makes me argue that the quantity assessment of pila de, montón de and mogollón de has conventionalized, viz. spread over the entire speech community. According to the dictionaries, the QNs differ in two respects, viz. ‘verticality’ and ‘order’ (cf. Figure 3). However, the following sections will provide a more fine-grained analysis of the individual conceptual images and indicate that pila de, montón de and mogollón de not only conceptualize the N2entities in a proper and individual way, but also have their own preferences as to context of appearance.

|| 197 With respect to the precise origin of mogollón, there seems to be some confusion. While both DRAE and DUE refer to Italian moccobello ‘tip’, Catalan mogobells ‘tip’ and Arabic muqabil ‘compensation’, Corominas/Pascual (1991, 106–108) argue that the previously proposed formal and semantic similarities are accidental. The etymological dictionary suggests that mogollón derives from miga ‘crumb, portion of bread’ to which the augmentative -ón was added. Comer de mogollón soon shifted from ‘to eat fast’ to ‘to eat hungrily’ and finally to ‘(to be a) parasite’. When used as an adjective, mogollón, -ona indeed refers to a lazy person. The explanation by Corominas/Pascual, viz. the addition of the augmentative -ón to miga, brings to the fore a morphological commonality between montón de and mogollón de.

342 | QN-related schematization and N2-profiling

pila de

montón de

mogollón de

+ vertically oriented

± vertically oriented (conical configuration)

-vertically oriented

order

lack of order

inherent disorder

Fig. 3: Differences in the quantity assessment of pila de, montón de, mogollón de according to the reference dictionaries

7.2.2.2 Conceptual image of (una) pila de N2 The grammaticalized uses of pila de obviously root in pila’s source semantics.198 In the latter original frame, the concepts of ‘orderly (stacked)’, ‘vertical orientation’ are of crucial importance, which is illustrated in (72) by the contextual clues con perfecto equilibrio and se alzaba respectively. In quantifying uses, the very same features lead to ambiguity (e.g. (73), where the focus can be either on the literal pile or on the number of files that are to be dealt with) or indeterminacy (e.g. (74), where the pile of murders evokes both a pile of corpses piled up (cadáver tras cadaver) and a huge number of dead people (cf. the exaggerated llegaba al cielo and the contrast with millones de infortunados, etc.)). (72) No pude por menos que decírselo, deslumbrado al observar la pila de platos y pocillos que se alzaba con perfecto equilibrio sobre el fregadero. (1992, novel) ‘I could not help telling him, blinded by observing the pile of dishes and cups that rose in perfect balance on the kitchen sink.’ (73) Mira luego desganado la pila de expedientes que le esperan encima de la mesa. ‘Afterwards he looks reluctantly to the pile of files that are awaiting him on the table.’ (1991, books) (74) Y en esa eternidad de unos meses el mundo seguía empequeñeciendo, nadie estaba de acuerdo con nadie, todos escribían canciones de protesta, todos las cantaban, todos publicaban libros cargados de razón, pero todos seguían siendo insolidarios, cadáver tras cadáver la

|| 198   Recall that in only 16 of 80 occ., pila de is unambiguously used as a quantifier, while its specifier use is totally marginal (with 1 occ.). In other words, the literal reading of binominal pila de is still the unmarked one.

Quantification construal in BQs | 343

pila de asesinatos llegaba al cielo, desaparecido tras desaparecido llegaba a desaparecer el mismo Dios, y mientras tanto nacían millones de infortunados, el mar y el campo se morían de vergüenza y la conciencia humana entraba en bancarrota. (1981, novel) ‘And in that eternity of some months the world kept on getting smaller, nobody agreed with nobody, everybody was writing songs of protests, everybody was singing them, everybody published books loaded with reason, but everybody kept unsupportive, corpse after corpse the pile of murders reached the sky, missing person after missing person God himself came to disappear, and in the meantime millions of unfortunate people were born, the sea and the country were dying of embarrassment and the human conscience was going bankrupt.’ In quantifying uses, pila de typically profiles the N2s as a bounded group of N2s which is conceptualized as huge, but which in absolute numbers is not necessarily extraordinarily high. In (75), the speaker of course knows the precise time interval of years he lived in Rome. In spite of not necessarily being many years in absolute terms, they are conceptualized as such, since the speaker obviously got tired of living there (harto de). What matters is that the time interval was finite, clearly delineated and uninterrupted: he lived in Rome for a specific number of successive years. (75) – Soy siciliano – dijo –, siciliano de Catania. Estudié arquitectura, ejercí esa espantosa profesión en Roma durante una pila de años, me casé, me descasé y un buen día, harto de esclavitud, de cemento y del mal gusto de los ricos, junté los ahorros, compré un billete de avión (sólo de ida) para Ciudad de México y (…). (1990, novel) ‘I am Sicilian, he said, Sicilian from Catania. I studied architecture, I practiced that horrific profession in Rome for a large number of years, I got married, I got divorced, and on a specific day, tired of slavery, of cement and of the bad taste of the rich people, I collected my savings, bought a plane ticket (one way only) to the city of Mexico and (…).’ The conceptual image of pila de can be broken down into four conceptual facets which may come to be profiled in actual usage. First, the vertical orientation of the original accumulation lives on in the facet SUCCESSION (facet (a)). More precisely, the N2s are pictured as following each other immediately. In (76), the years devoted to the construction form an uninterrupted chain. The protagonist intentionally dedicates year after year to the meticulous construction of the church (piedra por piedra). In (77), pila de pictures the guys as walking in rows

344 | QN-related schematization and N2-profiling

behind the other actresses. In other words, the verticality of the original frame got lost in the grammaticalized uses, but some sense of continuity shades through. (76) (…), está en Libro de los récords, Guiness, por ser la construcción más grande que hay de tipo medieval, dedicado a esto y porque tiene la iglesia más pequeña. La iglesia más peque bueno. Cuando yo bueno, este hombre lleva la pila de años dedicado a hacer piedra por piedra, a base de de estilo muy barroco todo, o sea, lleno de colmenas, (…). (1991, oral) ‘(…), it figures in the Book of records, Guiness, for being the biggest construction there is of medieval kind, devoted to that and because it has a smaller church. The church is smaller. Well when I, anyway, that man has dedicate a lot/series of years to build, stone after stone, on the basis of a very baroque style, I mean, full of beehives (…).’ (77) ¡Dios mío! Mira, que soy una estrella que tiene luz propia, José Luis. Las otras van con linterna con una pila de tíos detrás enchufando. Bueno, ya conocen a Lola. (1986, oral) ‘My God! Look, I really am a star that has its own light, José Luis. The others [KV: actresses] are walking with a lantern and a lot of guys behind them turning them on. Well, they already know Lola.’ The literal concept of order can be mapped onto the domain of SOCIAL and ESTAB(facet (b)). In (78), the number of children corresponds to some social norm regarding the number of children each woman should have. The number is of course limited for each single woman. Yet since every child implies certain sacrifices of his mother, pila de is used to profile a lot of burden (instead of a lot of children in absolute terms).

LISHED VALUES

(78) (…) o que para que las mujeres tengamos nuestro día tenemos que ser muy esforzadas y muy meritorias. Nada de ir por ahí con cara de disfrutar de la vida: que se note que la mujer es mujer cuando es sufrida; cuando la tía se lo merece; cuando tiene una pila de hijos por los que se deja los higadillos y un marido al que lleva muy bien planchado y un trabajo en el que da la talla como si fuera una máquina y una casa como una patena. A lo largo de los siglos, se nos ha vendido la idea de que para ser mujer no basta con nacer siéndolo: hay que hacerse mujer, hay que demostrar que una es mujer. (2001, press) ‘(…) or that in order for (us) women to have our day we have to be very hard-working and very meritorious. Nothing like walking around here

Quantification construal in BQs | 345

with a look of enjoying life: it ought to appear that a woman is a woman when she is uncomplaining; when the woman deserves it; when she has a lot of children to whom she leaves the pieces of liver and a husband who she dolls up really well and a job which she fulfills as if she were a machine and a house (as clean) as a new pin. Over the years, the idea was sold to us that to be a woman it is not enough to be born one: one has to become a woman, one has to show that one is woman.’ In the same way literal piles are by definition manmade and intentionally stacked (in contrast to literal heaps, see infra), quantifying pila de equally evokes an INTENTIONAL grouping of N2s (facet (c)). In (79), the succession of bombs are intentionally dropped by Reagan on Gadhafi’s house. The choice for pila de also profiles the vertical movement in the process of dropping bombs. In (80), the heap of excrements is not accidentally left behind on the speaker’s bed. Whether or not the excrements have a conical configuration and certain height, pila de highlights the fact that they have been intentionally dropped there, which perfectly fits the context of intruders turning a residence into a mess. (79) Como el propio Gadafi que desde que Reagan le soltó una pila de bombas a domicilio se ha tomado con mucha más parsimonia la práctica del santo terrorismo yihádico. (2001, press) ‘Like Gadhafi himself who, since Reagan dropped a pile/lot/series of bombs on his residence, interpreted with much more calmness the practice of the whole jihad terrorism.’ (80) Al llegar a casa me encontré un mar de escombros. Los hombres de Fumero habían estado allí. Entre muebles caídos, cajones y estanterías derribadas, encontré mi ropa hecha jirones y los libros de Miquel destrozados. Sobre mi cama encontré una pila de heces y sobre la pared, escrito con excrementos, se leía “Puta”. (2001, novel) ‘When arriving at home I discovered a sea of rubble. Fumero’s men had been there. Among fallen furniture, boxes and shelves pulled down, I found my clothes in shreds and Miquel’s books destroyed. On my bed I found a pile of excrements, and on the wall, written with excrements, could be read “whore”.’ Minimally, quantifying pila de profiles the number of N2 as FINITE or DELINEATED (facet (d)). This tendency is particularly obvious in examples such as (76) where the QN indicates the extension of a time period. Similarly, in (81), the precise

346 | QN-related schematization and N2-profiling

starting point (the year 1936) and end point (now) of the time interval are known. (81) Según eso, ya haría una pila de años que empezó a flaquear tu fidelidad a la ideología que te convocó en el 36. (1978, novel) ‘According to that, it would already be a lot of years ago that you started to lose your loyalty to the ideology that appealed to you in 1936.’ Figure 4 sketches the conceptual facets that can be highlighted in quantifying uses of pila de and the correspondences to the source frame. Literal frame

Grammaticalized uses

verticality

(a) SUCCESSION

order

(b) SOCIAL NORMS, ESTABLISHED VALUES

manmade

(c) INTENTION (d) FINITE, DELINEATED

Fig. 4: Conceptual facets in the conceptual image of pila de

The conceptual facet SUCCESSION makes abstract notions of time an appropriate candidate for quantity assessment by pila de, since time by definition flows continuously. The corpus analysis reveals that in 8 of 16 quantifying instances, pila de combines with años. The second most important cluster concerns human N2s (in 6 of 16 occ.). In combination with años, the impersonal presentative hay is repeatedly used (in 5 of 8 occ.). Note that in contrast to other presentational verbs, hay is considered to function as existential quantifier (Delbecque 2002a, 107; Suñer 1982). The specific semantic contribution of the QN becomes particularly obvious in comparison to possible variants. First, montón is semantically near to pila in the literal interpretation of a vertical configuration of things heaped up. However, the vertical orientation is less prominent for montón, which refers to a conical configuration, while pila refers to a real stack. If (79) would have read as le soltó un montón de bombas, the bomb droppings would not necessarily have been conceived of as successive or temporarily contiguous. It could then be envisaged that after dropping some bombs on Gadhafi’s house, Reagan dropped some bombs on somebody else’s house before returning to Gadhafi’s place. Crucially, the bombs quantified by montón de bombas would largely exceed pila

Quantification construal in BQs | 347

de bombas in number. In the case of montón de bombas, the speaker definitely wouldn’t know exactly how many bombs fell on the house. In (80), the substitution by montón de heces would also picture a bigger mass of excrements and the bad intention of the intruder would not be highlighted by the QN itself. If pila would be substituted by montón in (81), the immediate succession of the years would not be foregrounded by the QN and the primary focus would be on the excessive number of years. Finally, in (82), the QNs montón and pila are used within the same sentence: while montón de perfectly fits the inherent indeterminateness of the N2 cosas, pila is congruent with the orderly way in which the names are appointed in the diary. (82) Aproveché para mirar mi agenda, cosa que rara vez hago y, como siempre que la examino, me di cuenta de que había un montón de cosas que debía haber hecho esa semana que no hice y una pila de gente a la que debía haber llamado y no llamé, con lo que he quedado fatal para siempre. (1990, novel) ‘I took the opportunity to look at my diary, something I rarely do and, as always when I check it, I realized that there were a lot of things that I should have done this week (but) which I didn’t, and a lot (lit. pile) of people which I should have called (but) which I didn’t call, which means that I screwed it up (lit. I remained terrible) forever.’

7.2.2.3 Conceptual image of (un) montón de N2 According to the lexicographical treatment of montón, its quantifying use is restricted to indicating a considerable number or excessive quantity. On closer scrutiny, however, the occurrences observed in the corpus reveal several slightly different configurations of the quantity assessment. More precisely, a set of conceptual facets can be distinguished for montón de as well, which originate from its source semantics and are exploited in different proportions. Generally, montón de is used when the exact number of N2s or the exact identity of the N2s is (temporarily) less crucial or relevant. In (83), for instance, the N2 cosas is depicted as heterogeneous: un montón de cosas refers to a lot of different things (N2), yet which particular things (N2) are meant does not matter. The idea is that, although one thinks of a lot of things, one can never foresee what actually happens in the end. The same holds for example (84): montón de adds to the alternative mucha gente the idea that the N2 gente concerns a lot of individuals, but the precise identity of the individuals is not relevant. (83) – Uno piensa un montón de cosas, pero lo que realmente le sucede es lo que no ha pensado. (1987, theatre)

348 | QN-related schematization and N2-profiling

‘– One thinks of a lot of things, but what really happens to one is what one hasn’t thought of.’ (84) A mí me daba una grima espantosa cuando, el Viernes Santo, tenía que besar los pies de un crucifijo que antes había besado un montón de gente y que tenían que estar llenos de microbios, por mucho que los monaguillos pasaran un trapito después de cada beso. (1991, novel) ‘It gave me a terrified disgust when, on Good Friday, I had to kiss the feet of a crucifix that a lot of people had kissed before and that had to be full of microbes, no matter how many times the acolytes passed a small cloth after each kiss.’ In the literal frame of montón de, the N2-entities are usually heaped up by human endeavor (e.g. the particularly frequent combination un montón de libros ‘a heap of books’). In its grammaticalized uses as well, the facet of ACCUMULATION is omni-present. If no other facet of montón de’s image-schematic structure is highlighted, the N2s can usually be conceptualized as somehow accumulated or gathered on a specific ground. In (85), the speaker describes his father as ‘possessing’ a lot of diplomas, which a particular person gathers or accumulates one by one by definition. The facet of ACCUMULATION is often the only one that persists in sentences where montón de quantifies abstract notions of time, as in (86): the addressee has already kept silent for a long (and uninterrupted) time. (85) Aparte de ser mi padre, es un tío muy listo. ¿Opináis lo mismo los dos? ¿Vuestros padres son tíos muy listos? Sí, el mío es una lumbrera . ¿verdad? Piensa que tiene un montón de títulos, ya es académico de todo. (1991, oral) ‘In addition to being my father, he is a very smart guy. Do you both think the same? Are your fathers intelligent guys? Yes, mine is a genius. Really? He thinks he has a lot of titles, he is already academician in everything.’ (86) Llevas un montón de tiempo sin dar ni clavo, podría haber dicho también en el mismo tono reprobatorio. (1988, novel) ‘For a lot of time you didn’t do anything (lit. you spend a lot of time without doing anything), he could have said as well on the same disapproving tone.’ As a faint vestige of this act of accumulation, the N2s in the quantifying uses can be construed as SPATIOTEMPORALLY CONTIGUOUS ((facet (a); recall the chain of schematizations visualized in Figure 1 of Chapter 3). In (87), montón de quanti-

Quantification construal in BQs | 349

fies a group of people which are gathered on the same spot in the zoo (where one can watch the chimpanzees) at the same time (un domingo por la tarde). They occupy a specific physical space and the speaker is located right behind them. In (88), the stars are all intentionally accumulated on the hat of the women. (87) Era un domingo por la tarde y decidí invertir un poco de mi tiempo en una rápida visita al parque. Recuerdo que había un montón de gente viendo los chimpancés, yo me coloqué detrás. (2001, books) ‘It was a Sunday afternoon and I decided to invest some of my time in a quick visit to the park. I remember that there was a heap of people looking at the chimpanzees, I put myself behind them.’ (88) Elsa convenció a sus ciento cincuenta invitados de que llevasen cada uno un tocado en la cabeza que les identificase. (…) Una dama llevaba en la cabeza un montón de estrellas plateadas y se presentaba como la reina de los marcianos. ‘Elsa convinced her hundred and fifty guests to each wear a headdress on the head that would identify them. (…) A lady wore on the head a heap of silvered stars and presented herself as the queen of the Martians.’ (2001, books) Abstracting away from strict spatiotemporal contiguity, the N2-entities can simply be depicted as MEMBERS OF ONE AND THE SAME CATEGORY (facet (e)). In (89), the relative clause specifies the category the N2s belong to and motivates why they are taken together: there were a lot of different things which the protagonist told and which all receive the label que ninguna de nosotras entendió demasiado bien. In (90), montón de more or less simply functions as a unit of measurement grouping a lot of different authors which all qualify for narrando la situación muy verídicamente y sacándose unas pelas. In view of its potential to point at the common denominator or category the N2s belong to, montón de frequently introduces or concludes enumerations (of stars in (91), vitamins in (92), elements in (93)). (89) Y volvió a hablar del crimen (ya no recuerdo en qué país, en qué año), de bandas internacionales, de un montón de cosas que ninguna de nosotras entendió demasiado bien. Y de nuevo de su sufrimiento, de las amenazas, de su terrible periplo. (1994, books) ‘And he started to talk again about the crime (I don't remember anymore in which country, in which year) of international gangs, of a whole bunch of (lit. heap of) things that none of us understood really

350 | QN-related schematization and N2-profiling

well. And again about his suffering, the threats, his terrible wanderings.’ (90) Psicólogos, sociólogos, científicos, más la sabiduría popular y las tertulias de la radio abarrotadas de listillos y famosos dando su versión del puzzle, dando su opinión sobre la incompatibilidad de los sexos. Más un montón de escritores narrando la situación muy verídicamente y sacándose unas pelas. (2001, novel) ‘Psychologists, sociologist, scientists, and in addition popular wisdom and radio talk packed with smart asses and famous people giving their version of the puzzle giving their opinion on the incompatibility of the sexes. And in addition, a bunch of (lit. heap of) writers narrating the situation in a very truthful way and lining their pockets.’ (91) Se ven muchas cosas desde aquí. Hay un montón de estrellas; la osa mayor, las siete hermanas... ¡La luna! (1990, theatre) ‘One sees a lot of things from over here. There are a lot of stars; the big bear, the seven sisters… The moon!’ (92) Grupo B. Al grupo B pertenecen un montón de vitaminas; las más importantes son la B1 (tiamina), la B2 (riboflavina), la B3 (nicotinamida, niacina o vitamina PP), la B6 (piridoxina), la B12 (cianocobalamina), la PGA o ácido fólico, el ácido pantoténico y la lecitina. (2001, books) ‘Group B. Group B includes a lot of vitamins. The most important ones are B1 (thiamine), B2 (riboflavin), B3 (nicotinamide, niacin or vitamin PP), B6 ( piridoxin), B12 (cyanocobalamin), PGA or folic acid, pantothenic acid and lecithin.’ (93) Es decir, el teatro moderno ya no solo es el texto, sino también es la forma de poner las luces, también es el decorado, también es el vestuario, también son los actores, también es la mímica, también es la danza. Es decir, participan un montón de elementos en el teatro que no son estrictamente literarios,(…). (unknown, oral) ‘I mean, the modern theatre isn’t but text anymore, yet it is also the way of arranging the lights, it is also the set, it is also the costumes, it are also the actors, it also is the mime, it is also the dance. In other words, a lot of elements that are not literary strictly speaking engage in theatre, (…).’ A second conceptual facet which can be directly related to the original ‘accumulation’-sense concerns the LACK OF INDIVIDUALITY of the N2-entities or the homog-

Quantification construal in BQs | 351

enization of the N2s (facet (b)), which results from bringing different things together and throwing them on a single heap without any further distinction. In examples (94) and (95), several contextual cues indicate that the individuality and identity of the N2-entities get totally absorbed by the whole. The idea is that by using montón de, the quantifier generalizes over all different N2s and the exact identity does not interest nor matter to the speaker. In (96), there are no contextual clues strictly speaking, but the use of the gerund (están llamando) indicates that what matters is that these listeners (oyentes) are all calling at the same time. (94) Cuando intento pensar en lo sucedido antes de aquel día, todo se me junta en la memoria como un confuso montón de gestos y rostros y lugares, sin acciones, ni formas, ni perspectivas, en que es difícil identificar y distinguir los matices. (1985, novel) ‘When I try to think about what happened before that day, everything gets together in my memory as a confused heap of gestures and faces and places, without actions, without forms, without perspectives, in which it is difficult to identify and distinguish the nuances.’ (95) Por el suelo había desperdigados restos más o menos identificables de mobiliario y en una zona oscura, entre un montón de objetos sin identidad, rescaté algo que podía parecerse al árbol de un perchero y también una desarmada corona de la que pendía un único cuerno. (1992, novel) ‘On the floor there were scattered more or less identifiable rests of the furniture and in an obscure zone, between a lot of objects without identity, I saved something that could resemble the pole of a coatrack and also a dismantled crown from which a single horn was hanging.’ (96) Buenos días, Iñaki. Muchas gracias por estar aquí. Antes que nada, el capítulo de dudas, que nos están llamando un montón de oyentes. (1999, oral) ‘Good morning, Iñaki. Thanks a lot for being here. First of all, the chapter of doubts, since a lot of listeners are calling us (right now).’ Another major facet directly related to the literal interpretation of montón is the Recall that the reference dictionaries (see Section 7.2.2.1) indicate that the use of montón – in contrast to pila – suggests the absence of structure or of any kind of other ordering principle. While the vertical orientation is inherent in pila, the conical configuration of montón is less prominent and results from continuously throwing things, whether or not intentionally, on the same spot. This facet is also prominent in montón de’s LACK OF ORDERING PRINCIPLE OR STRUCTURE.

352 | QN-related schematization and N2-profiling

quantifying uses. In the examples (97)–(100), plenty of contextual clues refer to the chaotic or unorganized nature of the (accumulation of) N2s. (97) No, debía hacer un poco de retiro mental y dedicarme a atar cabos, recapacitar sobre el aluvión de datos no concluyentes, buscar claves que explicaran el montón de sospechas sin fundamento claro que se abatían sobre el caso sin orden ni concierto. (2002, novel) ‘No, I had to do some mental retreat and dedicate myself to joining loose ends, to reconsider the flood of inconclusive data, searching keys that would explain the heap of suspicions without clear foundation, that spilled over the case without order nor harmony.’ (98) Nuestro libro iba ganando páginas, pero había que estructurarlo de alguna manera. Hasta aquel momento teníamos un montón de citas dispersas, además de la fábula cruel de la sábana y la medusa, la sospecha razonable de que fue un pez quien inventó las orejas, y el relato costumbrista de unos floristas enamorados que follaban con gran escándalo. (1994, novel) ‘Our book was getting more pages, but it was necessary to structure it some way. Till that moment, we had a lot of dispersed quotes, in addition to the cruel fable of the sheet and the jellyfish; the reasonable suspicion that it was a fish who invented the ears, and the story based on local customs of some florists in love, who were fooling around with a big scandal.’ (99) Pero de repente su inmovilidad te contraría. Tú con un montón de trabajo, con cantidad de sensaciones que no logras ordenar, y ella inmóvil, ensimismada junto a la puerta. (1994, books) ‘But all of a sudden, her immobility upsets you. You with a heap of work, with a quantity of sensations that you don't manage to classify; and she immobile, lost in thought next to the door.’ (100) David se aturulla, un montón de preguntas se atropellan en su boca. (2000, novel) ‘David gets confused, a heap of questions run over each other in his mouth.’ It is only a small leap from the absence of any structuring principle or chaotic character to VAGUENESS (facet (c)). Typically, the exact identity of the N2s quantified by montón de is profiled as irrelevant. In (101), the speaker may or may not know which order the doctor gave, but the identity of the prescription actually doesn’t matter, since it was the mere presence of the doctor that tranquilized his

Quantification construal in BQs | 353

grandmother. Similarly, in (102), montón de camareros does not refer to a group of particular waiters the speaker has in mind. The idea is that a lot of waiters appear from everywhere and that is hard to keep track of them. The facet vagueness inherent in montón de explains why the highly general noun cosas ‘things’ is the most frequent N2 observed in combination with montón de (see infra). (101) Cuando llegó el médico de la familia de Olvido, la abuela ya estaba sentada en su cama, incorporada con ayuda de unos almohadones porque decía que no podía respirar. El médico tranquilizó a mi madre y recetó un montón de cosas. La presencia del médico animó a la abuela. (1994, novel) ‘When the doctor of the Olvido family arrived, the grandmother was already sitting in her bed, up straight with the help of some big pillow because she said she couldn't breath. The doctor calmed my mother and prescribed a lot of things. The presence of the doctor gave grandmother courage.’ (102) En la comida, ni primero ni segundo, pero sus fritos y sus guisos calientes, fiambres, de todo, las cosas más buenas que tú te puedas calculá y un montón de camareros p'arriba y p'abajo, luego te cuento la que se formó con los camareros. (1979, novel) ‘For food, nor a first nor a second plate, but its fried food and hot stew casseroles, cold meats, and everything, the best things you could imagine, and a lot of waiters upstairs and downstairs, then I will tell you the quarrel that arose with by the waiters.’ In addition to vagueness as to the exact identity of N2, montón de can profile VAGUENESS AS TO THE EXACT QUANTITY or number of N2. In (103), montón de certainly profiles a lot of years, but how many exactly depends on the personality of each single poet or painter apprentice. Likewise, in (104), the exact number of albums the protagonist could fill if he would collect caricatures does not matter since the album-collection is hypothetical from the start. The idea is that the campaign is omnipresent and that Franco’s caricatures are showing up in large numbers to the extent that they could fill a lot of albums, but the precise extent remains unspecified. The potential of montón de to profile a number or mass of N2 which is not clearly delineated enables the QN to simply profile an INDETERMINATE MASS (facet (f)), more or less equivalent to the unspecific quantifier some in English. In (105), montón de evokes indeterminacy: the speaker argues that his dancing act may turn out wrong for a variety of reasons, but that his bad performance is never caused by the fact of wearing yellow clothes. The number

354 | QN-related schematization and N2-profiling

of possible causes of bad performance is indeterminate: there may be more reasons for a bad performance than the ones enumerated, but which and how many does not matter. The principle idea is that he does not believe in the superstitious reason of yellow tissues. (103) Un aprendiz de poeta, o de pintor, podía acabar con facilidad en manos de la brigada político-social y ser detenido y amedrentado y golpeado sin mucho motivo, porque motivos no hacían falta, ninguno, y quedarse asustado durante un montón de años. (1995, novel) ‘A poet apprentice or a painter apprentice, could easily end up in the hands of the political-social brigade and be imprisoned and terrified and beaten without much reason, because there was no need for reasons, none, and remain terrified for a lot of years.’ (104) La campaña internacional contra el Régimen español aumenta por días. Si coleccionara caricaturas de Franco, con su barriguita y su fajín de general, podría llenar un montón de álbumes. Es un asunto que está muy liado y no sé lo que va a pasar. (1986, novel) ‘The international campaign against the Spanish government is gaining force every day. If I were to collect caricatures of Franco, with his belly and his general sash, I could fill a heap of albums. It is a very complicate matter and I don’t know what is going to happen.’ (105) Es raro, ¿verdad?, porque en el mundo del flamenco hay mucha superstición. Mira, cuando bailo sólo pienso en que todo salga bien. Me pueden salir las cosas mal porque me duelan los pies, el estómago o por un montón de cosas más, pero no porque lleve algo amarillo. (1990, press) ‘It's weird, isn't it?, because there is a lot of superstition in the world of flamenco. Look, when I dance, I only care about all turning out well. Things can go wrong for me because my feet hurt or my stomach or because of a lot of other things, but not because I'd be wearing something yellow.’ A final conceptual facet that can be linked to montón de’s source semantic is the factor COINCIDENCE or accidental nature of the accumulation of the N2s (facet (d)). Recall that in contrast to literal piles, a literal montón does not necessarily result from human intention. In (106), the speaker explicitly mentions that he doesn’t know why exactly so many people were present. The use of the reflexive clitic se in the relative clause se me han occurrido in (107) also indicates that the N2s happened beyond the speaker’s will and thus unintentionally. This facet is however less frequently exploited.

Quantification construal in BQs | 355

(106) La entrevista con Shultz tuvo lugar en la suite del hotel y no sé muy bien por qué estuvimos presentes un montón de gente: el presidente, el ministro, Juan Antonio Yáñez y yo, Chencho Arias, Santi Salas y Javier Solana. (1993, books) ‘The interview with Shultz took place in the suite of the hotel and I don't know very well why we were a lot of people being present: the president, the secretary, Juan Antonio Yáñez and I, Chencho Arias, Santi Salas and Javier Solana.’ (107) Tengo previsto sacar un libro, Un millán de cosas, que, como su propio nombre indica, se trata de un montón de tonterías que se me han ocurrido. (1996, press) ‘I plan to publish a book, Un millán de cosas, which, as its own name indicates, is about a lot of nonsense that occurred to me.’ Figure 5 sketches the conceptual facets that can be highlighted in the quantifying uses of montón de and how they correspond to the source frame. Literal frame

Grammaticalized uses (a) SPATIOTEMPORAL CONTIGUITY

accumulation

(b) LACK OF INDIVIDUALITY,

(e) MEMBERSHIP OF ONE AND THE SAME CATEGORY

HOMOGENIZATION

(c) VAGUENESS AS TO IDENTITY / lack of structure or order

EXACT QUANTITY

(f) INDETERMINATE QUANTITY

(d) COINCIDENCE, LACK OF INTENTIONALITY

Fig. 5: Conceptual facets in the conceptual image of montón de

Interestingly, the grammaticalized uses diverge as to the degree of schematicity typically associated to montón de. Whereas the quantifying uses of montón de have developed a wide array of conceptual facets which may come to be highlighted, the specifying uses typically combine with a high degree of conceptual persistence (see also Chapter 9 on the gradual character of persistence). In (108)–(109), N2 is depicted as an insignificant heap of residual or left-over material. Yet this heap is purely metaphorical and does not correspond to any actual heap of waste or of rotten meat. In contrast, the N2s in (110)–(111) constitute spatiotemporally contiguous accumulations, yet the lesser value of these objects is of primordial importance. Interestingly, all specifier uses of montón de realize the conceptual facet of homogeneization: grouping several entities under a single concept or category is of course in harmony with both the semantics

356 | QN-related schematization and N2-profiling

of the specifier reading and with the primary function of BQs, viz. that of singling out a set of N2s. (108) – Hay gente simpática – dijo. Pati se volvió para seguir la dirección de su mirada. – Vaya – gruñó escéptica –. A mí me parecen todos un montón de basura. (2002, novel) ‘– There are nice people – he said. Pati turned around to follow the direction of his gaze. – Well, she grunted skeptically –. To me, they all seem a heap of trash.’ (109) Si os despojaran de vuestro dinero, los cien supermagníficos juntos no valdríais ni dos centavos. Seríais un montón de carne enferma y de mentes inservibles. (1990, theatre) ‘If they strip you of your money, you hundred supermagnificent guys together wouldn't be worth two centavos. You would be a heap of ill meat and useless minds.’ (110) El Juez, en el silencio expectante de las miradas del público, que confluían sobre aquel montón de trapos que a duras penas conservaba cierta verticalidad humana, le pidió al acusado que, si había lugar, completara los testimonios aportados con alguna circunstancia que (…). (1995, novel) ‘The Judge, in the expectant silence of the gazes of the public, which came together on that heap of cloths that, with a hardly maintained some human verticality, asked to the accused to complete, if it was suitable, the testimonies that had been given with some circumstance that (….).’ (111) El teclado es básicamente un montón de botones que se resisten a ser manipulados coordinadamente. El ratón es básicamente un puntero. Maquinitas sorprendentes pero, esencialmente, son mecánica pura. Son como un cuerpo perfecto sin cerebro. (1993, books) ‘The keyboard is basically a heap of buttons that resist to being manipulated in a coordinated manner. The mouse is basically a pointer. Little surprising machines, but, essentially, they are pure mechanics. They are like a perfect body without brain.’ In view of the highly general nature of the conceptual image of montón de, it does not come as a surprise that the preferred N2-collocate in the quantifying uses is the the dummy noun cosas ‘things’ (in 40 (or 13%) of 311 quantifying

Quantification construal in BQs | 357

occ.). In addition, in contrast to the majority of the QNs, montón de can combine with any type of N2 (concrete as well as abstract; human, animate and inanimate; mass nouns as well as count nouns). Among the most frequent right collocates figure gente ‘people’ (22 occ.), años ‘years’ (21 occ.), dinero ‘money’ and veces ‘times’ (both 10 occ.). As to particular predicates quantifying montón de repeatedly combines with, the 36 occurrences of the quantifying presentative verb hay ‘there is/are’ stand out.199 In addition to existential verbs, the quantifying uses combine with a wide variety of verbs referring to ‘possession’: either the possession in itself (tener ‘to have’ (20 occ.), llevar ‘to carry’, pertenecer a ‘to belong to’, etc.) or to the process of acquirement (coger ‘to take’, ganar ‘to win, to earn’, obtener ‘to get’, ocupar ‘to occupy’, sacar ‘to take out, to get’, etc.). Montón de can also be introduced by the preposition con ‘with’, typically related with possession, as in (99). A third cluster concerns predicates referring to the process of gathering or accumulation (acumularse ‘to accumulate’, juntarse ‘to assemble, to come together’, recoger ‘to collect’, reunir ‘to bring together’, etc.). With regard to the specifying uses of montón de, they are almost always introduced by a combination of the copular verbs ser ‘to be’ or aparecer ‘to seem’ in combination with the adverb sólo ‘only’ or the structure no … más que ‘nothing more than’ and ya no … sino … ‘not anymore … but …’. In addition, specifying montón de combines three times with convertir ‘to turn into’ and haber reducido a ‘to have reduced to’. A final argument for the conceptual image of montón de outlined in Figure 5 can be found in comparing the specific semantic contribution of a QN and its near-synonyms. The corpus yields several examples which – either because of the N2 or contextual cues – remind of the image schematic structure of other QNs. First, Section 7.2.2.2 revealed that the N2 años is the most frequent right collocate of pila de. In (115), montón de can be replaced by pila de. The latter QN suggests however that the speaker perfectly knows how many years ago Margarita emigrated. Yet the precise number of years, let alone the precise circumstances (cf. she married a Korean, but who exactly does not matter), seems of

|| 199 By way of comparison, the other QNs combine far less frequently with the existential quantifier hay. While the combination with the near-synonymous pila de (5 occ.) and mogollón de (8 occ.) sounds natural, the QNs with an essentially contentful conceptual image only rarely combine with hay. Whereas the combination is attested twice for aluvión de and only once for racimo de, hay does not occur in combination with hatajo de, barbaridad de, alud de or letanía de. Interestingly, these QNs less easily give way to ‘pure’ quantity assessment and usually add a qualifying component to the quantification construal, which makes them less susceptible of occurring with hay.

358 | QN-related schematization and N2-profiling

little importance in this example. The vagueness of the description is however perfectly congruent with montón de’s image schema. (112) Cuando los norteamericanos se vayan de Torrejón, Margarita Chen, la propietaria del restaurante chino Happy House, no tendrá que abrir sus puertas para la cena a las seis de la tarde, pero perderá una parte de su clientela habitual. Margarita llegó de Taiwan hace un montón de años y ha decido echar raíces en Torrejón. Aquí se casó y se divorció de un coreano. (1988, press) ‘When the North-Americans leave Torrejón, Margarita Chen, the owner of the Chinese restaurant Happy House, will not have to open her doors for dinner at six in the evening, but will lose part of her usual clients. Margarita arrived a lot of years ago from Taiwan and has decided to put down roots in Torrejón. Here, she married and divorced a Corean.’ As will become clear in Section 7.2.2.4, the other ‘near-synonym’ (or partial functional equivalent) mogollón de combines with N2s referring to negatively evaluated human nouns. On these grounds, mogollón de could easily replace montón de in (113). However, with mogollón de locos, the argument would be less convincing. While with montón de locos, the interlocutors are more or less depicted as a large number of similar fools, mogollón de locos does not profile a large homogeneous mass of foolish family members yet primarily focuses on their individuality. In other words, with mogollón de, a list of the names of the so-called fools in the following context would be perfectly appropriate. (113) Existe un tipo de familia que se esmera durante la consulta en afirmaciones contradictorias y discusiones interminables entre ellos. Son muy frecuentes: “No, la que estuvo internada en un manicomio es tu tía Ignacia, hermana de tu abuela.” “Ni hablar, no era hermana de mi abuela, si lo sabré yo, era la mujer de un hermano de mi abuela, así que no soy descendiente; en la familia en la que hay un montón de locos es en la tuya”. (1987, books) ‘There is a type of family which, during the consultation, puts a lot of effort in contradictory statements and endless discussions between themselves. Very frequent are statements such as: “No, the one who was put in an asylum was your aunt Ignacia, your grandmother’s sister.” “No way! She wasn't one of my grandmother’s sisters, I perfectly know that she was the wife of one of my grandmother’s brothers, which means that I do not descend from her; it’s your family that has a lot of lunatics.”

Quantification construal in BQs | 359

Finally, the context of (114), where montón de peros is preceded by several sentences introduced by the adversative marker pero ‘but’, reminds of the QN letanía de which typically combines with N2 referring to small parts of discourse (see also Section 7.2.3.3). The substitution of montón by letanía in (114) would profile a long-winded sequence of the adversative marker causing boredom, while with montón de peros, the sequence of adversative markers rather invokes disbelief: since a lot of peros have been said out loud, the discursive and convincing power of every single one weakens. In addition, with letanía de peros it would feel more natural if the adversative markers followed one another immediately. On the other hand, the fact that large parts of discourse are inserted in between the sequence of peros does not inconvenience the choice for montón de. (114) – Pero, ¿de verdad crees que te sería útil en Exteriores? – Por Dios, Ramón, no seas modesto. Tu viaje a Berlín fue la mejor prueba. Desde hace más de un mes estás actuando de hecho como ministro de Asuntos Exteriores. – Pero,… – ya llevas dichos un montón de peros, hombre. ¿Tanto te molesta el cargo? (1981, novel) ‘– But, do you really think I would be useful to you in Foreign Affairs? – For God's sake, Ramón, don't be modest. Your trip to Berlin was the best proof. Since more than a month, you are effectively acting as Foreign Affairs secretary. – But,... – You've already said a lot of “buts”, man. Does the job bother you that much?’

7.2.2.4 Conceptual image of (un) mogollón de N2 Given the etymological uncertainty with respect to mogollón, it cannot be empirically verified whether the conceptual image of its quantifying uses originates in its source semantics. This is not to deny that mogollón de also has a proper conceptual image and a corresponding combinatorial pattern. The majority of the conceptual facets highlighted in the corpus data relate to the immanent disorder of the collection of N2s (mentioned by CLAVE only). Typically, mogollón de profiles the N2-entities as a collection in complete chaos or DISORDER (facet (a)). In (115), the fragment pictures a vacillating speaker (¿no?, yo qué sé, yo creo, más o menos) who complains about the dissolution of social stratification which makes it more difficult to draw straightforward con-

360 | QN-related schematization and N2-profiling

clusions. The focus is on the heterogeneity of the N2 gente (revuelta): the people come from different places, they speak different languages, etc. In other words, large towns not only accommodate a lot of people, but a lot of different types of people. By way of comparison, the substitution by un montón de gente would primarily affect the facet of heterogeneity and simply profile a lot of people (which happen to lack any kind of organization). (115) Estas cosas yo creo que antes las tenían más fácil, ¿no? Antes hace, yo qué sé, veinte años, pues te ibas y sabías que que en una zona rural, en cualquier pueblecillo, bueno pues podías hacer un estudio, más o menos, pero es que... una ciudad grande, pero es que ahora que en cualquier sitio hay un mogollón de gente revuelta, gente de todos los sitios, gentes cada uno de... con sus lenguas, con una... que ya no... que no es tanto que la gente que tiene dinero tiene estudios o la gente que tiene menos pelas es de una determinada *clases*. Bueno, pero de todas formas... Ya está todo más revuelto, ¿no?, ya es más difícil sacar conclusiones así. (unknown, oral) ‘Those things, I believe that formerly people could deal more easily with them, isn’t it? Before, about, what do I know, twenty years ago, well you left and knew that that in a rural zone, in any little village, well, you could make a study, more or less, but the thing is, a big city, but the thing is that now that in every place there is a lot of people in a mess, people of all places, people everybody of… with their own languages, with a… that no more … that it is not the case anymore that the people who have money have studies and the people who have less money belong to a specific *class*. Well, in any case … Everything is now more confused, isn’t it?, and it is now more difficult to draw conclusions that way.’ The immanent lack of organization can be further schematized into the facet LACK OF CLARITY or VAGUENESS (facet (b)). In (116), the speaker contrasts his proper search for a definite answer about extraterrestrial life with the few people who do know the results (and amount) of research carried out. In 19 of 44 quantifying uses of mogollón de, the discourse context pictures a speaker in search of confirmation (cf. in 14 of 44 uses, the question tag ¿no? ‘isn’t it?’ shows up in the (usually immediate) context of mogollón de), e.g. (116)–(117). In (118)–(119), truth is simply hidden. (116) Lo saben lo saben unos cuantos sólo, pero... Ya lo sé. ... los más importantes. Y que siempre que sale esta conversación yo siempre me acuerdo de de de bueno... del tema de los ovnis y tal, ¿no? Joder, sobre

Quantification construal in BQs | 361

eso , si... por si... sobre si existe vida en algún otro planeta y tal, ¿¿¿hombre??? estoy convencido convencido de que de que no a lo mejor que se sepa dónde, y tal, pero se tienen mogollón de pruebas, mogollón de cosas y mogollón de estudios, que actualmente nosotros nunca vamos a saber, ¿no? (unknown, oral) ‘Only a few know it know it, but… I already know. … the most important ones. And every time that conversation comes up I always remember the well… the issue of the UFOs and so, don’t you? For heaven’s sake, about that, if … if only… about if life on another planet exists and so on, well, I am convinced convinced that that may be one doesn’t know, and so on, but we have a lot of tests, a lot of things and a lot of analyses, that in fact we will never know for sure, do we?’ (117) Pues en el momento que sea solo, a ver quién le hace la sombra, o sea quién además, eso es curioso, ¿no? porque eso La caída una guerra después de mogollón de tiempo, una guerra donde han al fin y al cabo, han han participado pero mogollón de países, ¿no?, y con unas armas impresionantes. Y el y el número de muertos pues ha sido No, no ha sido una cosa mínimo, ¿no? (unknown, oral) ‘Because at the moment it would be only, let’s see who would overshadow him, that is, who else, it’s strange, isn’t it? Because that The collapse a war after a lot of time, a war in which in the end a lot of countries have have participated, isn’t it? And with some impressive armed forces. And the and the number of dead people has been It hasn’t hasn’t been a tiny thing, has it?’ (118) Viendo la película de Lara Polop se comprende que no puede uno fiarse de nadie, porque tras las apariencias se esconde mogollón de cosas, es decir, más o menos lo que las bienintencionadas películas sociales que la derecha filmaba en los años cincuenta trataban de explicarle al público español, a falta de poder hablarle directamente, de su propio desencanto. (1984, press) ‘By seeing the movie of Lara Polop one comes to understand that one cannot trust anybody, because behind the appearances a lot of things are hidden, that is, more or less what the well-intentioned social film which the Right Wing filmed in the fifties tried to explain to the Spanish public, failing to be able to speak to them directly, of their own disenchantment.’ (119) – Qué va – exclamó –. Me ha contado un mogollón de historias. Cada palabra que dice el tío, mentira. (1984, novel)

362 | QN-related schematization and N2-profiling

‘– Well no, she shouted –. He has told me a lot of histories. Every word this guy says, is a lie (lit. he lies). ’ Abstracting further away from the concept of VAGUENESS, mogollón de can simply profile an INDETERMINATE QUANTITY (facet (e)). In (120), the exact number of acids does not matter, yet their remarkable high quality is highlighted. In (121), the exact number of activities organized by the professor is not even known since the quantity appreciation is built on far memories only. Likewise, in the 3 occurrences where mogollón de intensifies an adjective, the exact degree of ‘importance’ (in (122)) and of ‘attractiveness’ (in (123)) does not matter. What is more, the degree cannot be objectively measured. Instead of saying that LT (in (123)) is ‘highly’ or ‘very’ attractive, the speaker says that he is ‘quite’ attractive. (120) Miraba a su alrededor para asegurarse de que no podría ser oído por nadie. Se acercó a Antonio y le susurró: – Tengo algo que te puede interesar. Ayer me pasaron un mogollón de ácidos. – ¿Son buenos? – preguntó Antonio. – Quitan la cabeza. Lo que yo te diga. (1984, novel) ‘He looked around him to make sure that he could not be heard by nobody. He came closer to Antonio and whispered to him: – I have something that might interest you. Yersterday they gave me a couple of acids. – Are they good? – Antonio asked. – They are adorable. I can tell you.’ (121) ¿Profesores que me hayan marcado? Pues, en la e-ge-be no tengo mucho es eso, que se te va olvidando, no tengo mucho recuerdo, pero, fundamentalmente, un profesor mío de historia y mi profesora de inglés, porque, desde quinto de e-ge-be, fue mi sexto, séptimo y octavo me dio inglés. Entonces, son cuatro años, que es casi que con ella hacía mogollón de actividades, pues yo me quedaba a comer, y todo eso, tenía mogollón de actividades de extraescolares, estaba muy metida en eso. (unknown, oral) ‘Teachers who have marked me? Well, in the EGB [primary education] I don’t have a lot here’s the thing, that you start to forget, I don’t have a lot of memories, but, fundamentally, a teacher of history of mine and my English teacher, because, since the fifth grade, that was my sixth, seventh and eighth year she taught me English. That are thus four years, the thing is that she did a lot of activities, so I stayed over

Quantification construal in BQs | 363

there to eat, and all that, she had a lot of extracurricular activities, she was very into that.’ (122) Pero un buen día, te visten de novia, y todo el mundo te felicita, y te da besos, y te hacen fotos, y te hablan de una serie de historias que suenan mogollón de importantes, y entonces se te empiezan a caer los palos del sombrajo, y (…). (1991, theatre) ‘But then one day, they dress you like a bride, and everybody congratulates you, and kisses you, and they make pictures of you, and they talk to you about a series of histories that sound quite important, and then the poles of the sunshade start to fall down, and (…)’ (123) El otro día oí por la calle: “Luis Tosar no es guapo, pero es mogollón de atractivo”. ¿Estás de acuerdo? (2002, press) ‘The other day I heard on the street: “Luis Tosar is not handsome, but he is quite attractive”. Do you agree?’ In addition, the inherent disorder or lack of structure profiled by mogollón de can also give way to a focus on the HETEROGENEITY of N2 (facet (c)). In (127), the fact that the N2s are different individuals, not spatiotemporally contiguous and that the protagonist maintains irregular and unstable relations with them, is of crucial importance for a good understanding of the sentence. (124) Normalmente es eso, cuando un tío va con muchas tías, es todo un macho, es en plan es que yo me lo imagino con todas las cabezas de las tías, disecadas en casa. Sí. Rollo: me he ido de cacería y me he liado con tantas. Y las tías una tía que se lía con mogollón de tíos en plan rollo una semana o tanto, es una puta, o sea Yo no pienso eso. Yo me estoy metiendo ya en el tema Tú no, pero Machismo-feminismo, pero, o sea, es verdad, o sea, eso es un tema que es tabú todavía. (unknown, oral) ‘Normally that is what it is, when a guy goes out with a lot of women, he is all of a man, it is like the thing is that I can already imagine him with the heads of all the women dissected at home. Yes. Affair: I went hunting and I got involved with so many. And the women a woman who has a romance with a lot of guys seeing it as an affair for like a week or so, she is a whore, I mean That is not what I think. I am getting into an issue You not, but machismo-feminism, but, I mean, it is true, that is, this is an issue which is still taboo.’ Finally, the conceptual facet of disorder in mogollón de can also be mapped onto the domain of social values, where mogollón de profiles the (number of) N2

364 | QN-related schematization and N2-profiling

as AGAINST THE RULES or established code (facet (d)). In (128), the work the husband brings along is not appreciated by the speaker. Likewise, in (129), mogollón de quantifies an uncommon punishment: N2 refers to sculptures of saints whose hands have been cut off. (125) Dijo que iba a echar conmigo unos diítas de descanso, pero vaya con el descanso: tenía que ver a un señor (el teléfono), y luego otra cosa de los negocios (el teléfono), y conferencia con Bilbao y comer con otros dos señores (el teléfono): fíjate tú que descanso el suyo, en vez de darle un martillazo al teléfono y quedarse un poquito tranquilo, ¿no?; él ya con un malhumó grande y tomando pastillas pa los nervios cuando él mismo se estaba echando encima aquel mogollón de trabajo y se había ido a Málaga a descansar conmigo. Y eso dicen que es bueno, andar como los locos, que ya parece tonto el que no anda como un loco y eso es el pogrezo, o sea, el adelanto. (1979, novel) ‘He said that he would take some days of rest with me, but what a great rest: he had to see some gentleman (the telephone), and then two other business things (the telephone) and a conference with Bilbao and to have lunch with two other men (the telephone): imagine the rest he had, instead of hitting the telephone with a hammer and staying a little calm, isn’t it?; he already was in a very bad mood and taking pills against the nerves when he himself was accepting that mass of work and left for Málaga to take a break with me. And they say it is good, to behave like fools, that the one who doesn’t behave like a fool already seems stupid and that is progress, that is, the advance.’ (126) En una iglesia había mogollón de santos cristianos con las manos cortadas. Al parecer fue un castigo de los feligreses por haber permitido que en cierta ocasión se incendiara la iglesia. (2003, press) ‘In a church there were a lot of Christian Saints whose hands had been cut off. Apparently it was a punishment of the parishioners for having allowed on some occasion that the church burned down.’ Figure 6 lists the conceptual facets which shape the image of mogollón de. Given the etymological uncertainty about mogollón’s origins, the correspondences to mogollón’s source frame cannot be visualized. However, the conceptual image is observed to present two degrees of abstraction from the basic facet DISORDER.

Quantification construal in BQs | 365 

 (b) LACK OF CLARITY, VAGUENESS

(a) DISORDER

(e) INDETERMINACY

(c) HETEROGENEITY OF N2 (d) BREAKING RULES, NOT DONE

Fig. 6: Conceptual facets in the conceptual image of mogollón de

Strictly collocationally speaking, the combinatorial pattern of quantifying mogollón de is less delineated. With regard to verbal predicates, it strikes the eye that quantifying mogollón de combines in 12 (of 44) occurrences with existential verbs (and with the impersonal quantifying presentative hay (8 occ.) in particular) and in 8 occurrences with verbs referring to possession or acquirement (e.g. tener ‘to have’ (7 occ.), chorizar ‘to steal’, sacarse ‘to take out’). Similarly, with regard to the N2-combinations, two frequent clusters stand out. Quantifying mogollón de combines 13 times with human nouns (e.g. gente (5 occ.) ‘people’, tíos (3 occ.) ‘guys’, etc.) and 6 times with N2s referring to ‘money’ (e.g. dinero (2 occ.) ‘money’, pasta (2 occ.) ‘money’, oro ‘gold’, etc.). Yet more importantly, instances of quantifying mogollón de commonly (in 26 of 44 occ.) show up in contexts with a negative semantic prosody: N2 does not have a negative affect by definition, but the context profiles it as being unwelcome or unpleasant. The etymological uncertainty, the very elaborated context expansion of mogollón de200 and the fact that the QN is always used as a quantifier or a specifier to N2 (see Chapter 3) make me argue that the degree of GR reached by mogollón is quite high and comparable to the status of montón de. Consequently, the persistence of conceptual elements can sometimes be rather low and the difference with absolute quantifiers such as mucho is sometimes negligible. In (130), mogollón de is reformulated by mucho ‘much, many’. Except for the fact that mogollón de fits the informal register of the utterance and adds a hyperbolic flavour, no difference in meaning can be derived from the context. (127) ¡Es que a las tías os da mogollón de corte! ¿Entiendes? No es que le ¡A las tías os da mucho corte! (unknown, oral) ‘The thing is that women are interminably embarrassed (lit. gives you

|| 200   See also Chapter 3. In addition to binominal quantifier uses, mogollón de can be used to intensify adjectives and as degree modifier (i). (i) ¿Qué es de tu vida?, que nunca, hace mogollón que te llamo y no te localizo. (1991, oral) ‘What about your life, because I never, I have been calling you for quite some while without localizing you’.

366 | QN-related schematization and N2-profiling

helluva of embarrassment)! Do you understand? It is not that. Woman are very embarrassed (lit. it gives you much embarrassment)!’. The specific facets constituting the conceptual image of mogollón de can be illustrated by comparing mogollón to other QNs. In (128), the contextual cues de repente ‘all of a sudden’ and aparece ‘appears’ form part of aluvión/alud de’s conceptual image and therefore usually invite to use those QNs. Replacing mogollón de gente by aluvión de gente would picture the demonstrants as people which keep on coming from all sides at the same time. Both aluvión de gente and alud de gente stress the impact of the manifestation on the speaker, while in (128) the speaker is not the principle addressee and the precise impact is not profiled nor relevant. What matters in (128) is the fact that the manifestation is not organized too well: the speaker cannot immediately retrieve the aim or demand of N2. This lack of clarity makes mogollón de the best fit. (128) Hoy no es mi día. De pronto aparece por la calle un mogollón de gente con pancartas dando gritos. Sí, señores, una manifestación. Me quedo allí, en medio del follón, intentando descifrar lo que dicen las pancartas. Los tíos las llevan tan mal, que van todas arrugadas y no hay forma de enterarse de nada. Al fin me aclaro; se trata de una manifestación de parados. (1981, novel) ‘Today is not my day. All of a sudden a heap of people appears on the street with placards and shouting. Yes, gentlemen, a demonstration. I stayed there, in the middle of commotion, and tried to work out what the placards said. The guys were holding them so badly, that they were all crumpled and there was no way to get to know anything. In the end I got it straight; it concerned a demonstration of unemployed persons.’ Likewise, in (129), the near-synonymous mogollón de and montón de are used in the same sentence. Although the difference in meaning is minimal, the quantity assessment diverges with regard to the importance attributed to the heterogeneity of N2. With mogollón de gente, the speaker refers to a lot of different individuals. The wine bottles, in contrast, may be identical. This does not imply that the N2-entities quantified by montón de cannot be heterogeneous. Yet with montón de the focus is primarily on the large quantity, while the use of mogollón de emphasizes that, in addition to being highly abundant, the N2s are of different kind. If the sentence would have been ves a un montón de gente, an overall picture of the people met would be given, as if the exact (and different) identity of the persons spoken to did not matter. Further, while chaos immanently rules in mogollón de N2, the messy character in un montón de N2 is rather an unfortu-

Quantification construal in BQs | 367

nate coincidence. Similarly, the context in (130) reminds the scenery described in example (124). Yet mogollón de tíos and montón de tías focus on different aspects: in (130), montón is more appropriate since the context globalizes over all different girls the interlocutor (tú) is sleeping with and does not focus on the possibly unstable or irregular nature of the relations. The tías are of course heterogeneous, i.e. different individuals, but that is not of primordial importance. What matters is that, generalizing over all girls, there has been more than one, the exact number as well as identity of the tías being simply irrelevant in (130). (129) Beber está socialmente aceptado, o sea, tú sales un sábado y ves a mogollón de gente con montón de botellas y que están ahí, y está bien, o sea, está bien, y ya está y entonces, pues, con el problema de la droga o sea, pienso que (…). (unknown, oral) ‘Drinking is socially accepted, that is, you go out on Saturday and meet a lot of people with a lot of bottles who are there, and that’s alright, that is, it’s ok and that’s it and then, well, with the drug problem, I mean, I think (…)’ (130) Lo que pasa es que la gente no está realmente concienciada en que una relación heterosexual, que se supone que es segura porque se sigue pensando que es un tema de gays o de logatas, como suele decir la gente, no suele usar tanto el condón como se debería usar. ¡Hombre!, es segura si tu si tú tienes relaciones habitualmente con esa pareja y esa con y esa pareja contigo, simplemente. Está claro que si tú te vas acostando por ahí con un montón de tías y luego vienes a mí y te acuestas conmigo, pues que tengo más posibilidades de cogerlo. (unknown, oral) ‘What happens is that people are not really aware of what a heterosexual relation is, that one supposes that it is safe because one keeps thinking that it is an issue of gays and [logatas], as people use to say, the condom is usually not used as much as it should be used. Well hey, it is safe if you if you have daily intimacies with that girl and that with and that girl with you, put simply. It is obvious that if your go around sleeping with a lot of girls and afterwards you come to me and you sleep with me, well that I have more chances to get it.’ Finally, the three adjective intensifier uses of mogollón de remind of the typical use of la mar de. Again the difference in meaning is subtle and relates to the concepts of indeterminacy or vagueness profiled by mogollón de. In (122), it is as if the exact degree of ‘importance’ does not matter and is possibly not even known to the speaker, but the histories are ‘quite’ important. If the sentence

368 | QN-related schematization and N2-profiling

would be la mar de importantes, the speaker is rather sure and presents the histories as extremely or infinitely important. In addition, the degree of importance would seem to be common knowledge. In (126), la mar de atractivo does not fit in with the context, because with la mar de atractivo, it is as if this infinite attractiveness could not be denied. However, estás de acuerdo precisely questions the fact that the public would unanimously agree with this judgment. This subtle difference between mogollón de ADJ and la mar de ADJ originates in the conceptual image of both QNs – which at least for la mar de relates to its source semantics – and is morphosyntactically mirrored in the determiner variation. While mogollón does not combine with determiners when intensifying adjectives, mar exclusively combines with the definite article la. While the absence of determiners yields ungrounded or unidentified instances, definite determiners typically introduce ‘known’ entities, i.e. the interlocutor already can establish the identity of the entity referred to.201 In sum, although several native speakers pointed out to me that they simply use mogollón de as a colloquial variant of montón de, the data discussed in this section indicate that the alternation between mogollón and montón confirms yet also exceeds the variation in register.

7.2.3 Outline of the conceptual images of the remaining QNs The present section provides an overview of the conceptual image of four other QNs. More precisely, I will outline the image schemas of racimo de ‘bunch, cluster of’, hatajo de ‘(little) herd of’, letanía de ‘litany of’ and barbaridad de ‘barbarity of’. Although it would be an interesting line of research to contrast racimo with the binominal uses of ramo de ‘bouquet of’, ramillete de ‘posy, bunch’ or clúster de ‘cluster of’, to compare hatajo to the quantifying uses of manada || 201 In line with the cognitive-constructional premise that variation in form corresponds to variation in meaning, a different conceptualization depending on whether mogollón is preceded by the indefinite determiner (un mogollón de N2) or lacks a determiner (Ø mogollón de N2) is plausible. However, except for a slight preference of the concept INDETERMINACY to combine with Ø mogollón de (in 6 of 9 occ. where the INDETERMINACY is stressed) and the slight tendency of Ø mogollón de to stress the messy or heterogeneous character of N2 (in 16 of 25 occ.)., I have not found obvious differences in meaning. Since the variant without determiner is more frequent (25 of 45 occ., in addition to 20 combinations with the indefinite determiner and 1 occurrence of aquel mogollón de N2), it can be argued that the variant Ø mogollón de N2 is in the middle of a specialization process and will end up as the unmarked variant at the end of the GR process. I have too little data, however, to verify this hypothesis: according to the corpora used, the use of mogollón de emerged only recently.

Quantification construal in BQs | 369

‘herd’, hato ‘bundle’ and atajo ‘short cut’; etc., the present section does not describe them in terms of specific minimal pairs for reasons of space.

7.2.3.1 Conceptual image of (un) racimo de N2 7.2.3.1.1 Lexicographical accounts The first definition given by all three reference dictionaries DUE, DRAE and CLAVE concerns the collection of grapes which are growing closely together, fastened to the same stalk.202 In varying order of appearance, three more definitions are provided. Only DUE and DRAE indicate that racimo can also apply to other kinds of fruit growing on a branch, even if they do not present the typical ‘bunch of grapes’-like disposition, e.g. racimo de ciruelas ‘plums’, de guindas ‘morello cherry’, de cerezas ‘of cherries’. In addition, racimo can refer to a collection of any kind of small things fastened together which present the same tangled configuration of a bunch of grapes, as in Adornó la puerta con un racimo de bolas de Navidad ‘She decorated the door with a bunch of baubles’. Finally, within the botanic domain, racimo refers to a specific type of inflorescence, where a group of flowers or pieces of fruit of similar size are growing together along a common axis.

7.2.3.1.2 The conceptual image of (un) racimo de N2 As was the case for the QNs of the minimal pairs described in the previous sections, the image-schematic structure of racimo de consists of facets which abstract away from specific facets of the literal source frame in which racimo refers to a bunch of grapes growing together closely on the same branch. This typical ‘bunch of grapes’-like configuration obviously persists in the quantifying use in (131) where small and similar objects (casitas) are grouped along a single core (e.g. the city center of Jerusalem) and present a clustered structure (rotas en mil direcciones) and cramped configuration (apretadas). In contrast to the intensifying, more precisely amplifying, effect of QNs evoking an immense and exaggerated number of N2 (e.g. alud, aluvión, montón and mogollón), racimo de usually profiles the N2s as a clearly delineated and bounded group.

|| 202 Etymologically, this definition corresponds to the Latin origin racimus (Corominas/ Pascual 1991, vol. 4, 746–747).

370 | QN-related schematization and N2-profiling

(131) El sol naciente había apagado las antorchas de Jerusalén, ofreciendo a nuestros atónitos ojos un inmenso racimo de casitas blancas y ocres, apretadas las unas contra las otras y rotas en mil direcciones por quebradas callejuelas. (1984, novel) ‘The rising sun has extinguished the torches of Jerusalem, offering to our astonished eyes an immense bunch of little white and ocher houses, squeezed together and broken down into a thousand of directions by zigzagging alleys.’ In addition to such examples which perfectly mirror the original, clustered disposition, several examples are observed where the source frame persists only partially. The conceptual image of racimo de can indeed by broken down into several conceptual facets which can become singled out in particular contexts. In line with the similarity as to appearance and size of the grapes or pieces of fruit in general, or of the flowers growing together, the grammaticalized uses can highlight the UNIFORMITY OR EQUIVALENCE (facet (a)) among the N2s as to quality (in (132)) or function (in (133)). (132) Directores de calidad. A pesar de que su calidad no haya ido pareja a la de las orquestas europeas, la abundancia de formaciones ha permitido que sí se pueda encontrar un buen racimo de directores iberoamericanos. Dos de nuestras orquestas, las Sinfónicas de Córdoba y Asturias, cuentan con titulares provenientes de allí: Leo Brouwer de Cuba y Maximiano Valdés de Chile. (1996, press) ‘Quality conductors. Although their quality has not been in keeping with that of European orchestras, the abundance of groups has made it possible to find a good bunch of Latin American conductors. Two of our orchestras, the Symphonic Orchestra of Córdoba and that of Asturias host conductors from there: Leo Brouwer from Cuba and Maximiano Valdés from Chile.’ (133) En 1901, el grupo Sota controlaba sesenta y siete de los ciento cincuenta y dos vapores vizcaínos, seguido por el de Martínez Rodas y un racimo de pequeñas compañías especializadas en el transporte de mercancías. (1994, books) ‘En 1901, the Sota group controlled sixty seven of the hundred fifty two Biscayan ocean steamers, followed by the Martínez Rodas group and a bunch of little companies specialized in goods transport.’ Further, the single stalk or axis within the source frame of racimo lives on in the accumulation or bundling inherent in the quantifying uses of racimo de. More precisely, when racimo de is interpreted dynamically (as a process), the N2s are

Quantification construal in BQs | 371

conceptualized as an INTENTIONAL GATHERING OR ACCUMULATION (facet (b1)). In (134), for instance, the coach Zambrano invites (pidiendo) his players to join their ambition, to actively bundle their forces. Yet in the stative interpretation of racimo de as a result of a previous bundling operation, the N2s are depicted as consciously accumulated for a particular reason or by an identifiable source, i.e. as constituting a WELL-DESIGNED AGGREGATE (facet (b2)). In (135), the bunch of values are carefully handled by the publicity. In (136), the artist has combined several philosophical issues in a single painting for a particular reason. (134) Zambrano se jugó las cuerdas vocales pidiendo un racimo de ambición, pero lo único que consiguió fue comprobar la premura de un equipo que llega a la promoción con una profunda herida moral. (1996, press) ‘Zambrano put at stake his vocal cords while asking for a bunch of ambition, but the only thing that he obtained was to confirm the haste of a team that makes it to the play-off with a deep moral wound.’ (135) Un racimo de valores son, en consecuencia, manejados por la publicidad, pero su captación no es fácil, pues cada mensaje publicitario suele ser monotónico, reflejando sólo un cierto aspecto del alma juvenil, requiriéndose como siempre un examen del conjunto de anuncios si se quiere advertir el trabajo verdaderamente sinfónico ejecutado por la publicidad. (2001, books) ‘A bunch of values are consequently handled by the publicity, but capturing them is not easy, as every commercial tends to be monotonous, reflecting only a particular aspect of youthfulness, thus requiring, as usually, an analysis of the whole set of commercials if one wants to notice the really symphonic work carried out by the advertising agencies.' (136) En cada cuadro hay un racimo de problemas filosóficos. “Me preocupa lo que al hombre en general: lo transitorio de la vida, el tiempo que nos ordena en pasado, presente y futuro, el yo enfrentado al mundo, lo masculino y lo femenino y el concepto mismo del arte, el yo creador y contemplador, el gusto y la memoria, personal y colectiva”. (1995, press) ‘In every painting there is a bunch of philosophical problems. “I am concerned with the human being in general: the temporariness of life, the time that arranges us in history, present and future, the ‘I’ confronted with the world, the masculine and the feminine and the concept itself of art, the ‘I’ who creates and who contemplates, taste and memory, personal and collective.’

372 | QN-related schematization and N2-profiling

A final conceptual facet that can persist concerns the original clustered structure of a bunch. As to the mutual relation among the individual N2s, the clustered structure shades through in the TANGLED OR TIGHT DISPOSITION of the N2s among each other (facet (c1)). In (137), the adjective apretados ‘cramped’, which indicates the cramped nature of the gardens, suggests that the houses as well are built closely together. Globalizing over the interrelation among different N2s, quantifying racimo de can also picture the grouping of N2s as HIGHLY CONCENTRATED OR DENSE (facet (c2)). In example (138), for instance, racimo de introduces a group of problems which particularly – if not exclusively – affect the wine sector. In (139), the context suggest that the N2s (exiliados latinoamericanos) crowd together and exclusively concentrate in a particular neighborhood of Barcelone. (137) Caminaba despacio, entusiasmado por el encanto de la noche en aquel racimo de casas blancas y de apretados jardines colgados de la ladera de la sierra. (1990, novel) ‘I was walking slowly, excited by the charm of the night in that bunch of white houses and of squeezed gardens hanging down on the slope of the mountain range.’ (138) En el carrusel de botellas encontré soberbios tintos del 2000, cuya estructura herreriana presagiaba el mejor futuro. Junto a frágiles y recientes acuarelas de trazo mágico, latentes levaduras e insinuado carbónico, para bebérselas de inmediato. Tan lúdica y maratoniana experiencia tuvo un paréntesis (que agradeció mi hígado) para que sabias voces departieran sobre el racimo de problemas que afectan al sector. (2003, press) ‘In the carousel of bottles I found superb 2000 red wines, whose Herrera-like structure foreboded the best future. In addition to fragile and recent aquarelles of a magic touch, latent leavenings and insinuated carbonic, to be drunk immediately. Such a playful and marathonian experience had a parenthesis (that my liver appreciated) in order for wise voices to discuss about the bunch/cluster of problems that affect the sector.’ (139) Aquellos barrios se habían convertido en la antesala del cementerio para las viejas generaciones condenadas a morir entre sus humedades, (…). Junto a los viejos supervivientes de la posguerra, los maduros con sensación de fracaso por no haber salido a tiempo de la trama estrecha y satánica de la ciudad vencida. Y luego gentes de paso, recientes inmigrantes del país marroquí, algún que otro racimo de

Quantification construal in BQs | 373

exiliados latinoamericanos forzados al alquiler barato. (1977, novel) ‘Those neighborhoods had become the anteroom of the cemetery for the old generations condemned to die among its damps, (…). Next to old survivors of the postwar period, the mature ones with a feeling of failure for not having left at time the narrow and satanic plot of the defeated city. And then people who just pass through, recent immigrants of the Moroccan country, one or another bunch of Latin-American exiles forced to low rent.’ The different conceptual facets of the image schema of racimo de and their relations to the literal frame are outlined in Figure 7. Literal frame

Grammaticalized uses

small and similar objects

(a) EQUIVALENCE (b1) INTENTIONAL AND ACTIVE GATHERING

common axis (b2) MOTIVATED, WELL-DESIGNED AGGREGATE (c1) TANGLED, CRAMPED, TIGHT CONFIGURATION clustered structure (c2) CONCENTRATION

Fig. 7: Conceptual facets in the conceptual image of racimo de

Interestingly, except for the right collocate casas ‘houses’ and problemas ‘problems’, which combine twice with quantifying racimo de, no preference for a particular type of N2 can be observed. Since the conceptual facets within racimo de’s conceptual image are rather of configurational than contentful nature (see supra), it does not come as a surprise that racimo de combines with almost any kind of N2 (abstract as well as concrete, animate as well as inanimate, mass as well as count nouns) provided that they can be made to fit racimo de’s conceptual image. Further, no particular recurrent predicate has been observed in combination with racimo de. In keeping with the hypothesis that one sees more in comparison, one final argument for the conceptual image of racimo de can be found in contrasting racimo de with other suitable QNs. If in (138), the construction would have been sobre el montón de problemas que afectan al sector, the N2s would not have been profiled as interrelated. The substitution by montón would also bring about an increase in number of the problems and they would be of different nature. If the sentence would be sobre el alud de problemas que afectan al sector, the N2s would not be conceptualized as a clearly delineated group of enti-

374 | QN-related schematization and N2-profiling

ties but as an insurmountable number of problems that will irreversibly tear down the wine sector. The substitution by montón de and alud de thus shows that the close relation among the N2s and the concentrated and limited nature of the N2s – which are not evoked in the alternative constructions – are conceptual facets inherent in the conceptual image of racimo de.

7.2.3.2 Conceptual image of (un) hatajo de N2 7.2.3.2.1 Lexicographical accounts The definitions for hatajo provided by DUE, DRAE and CLAVE are very concise. Both DRAE and DUE first mention its literal reading as a little group of cattle. All three mention the more desemanticized or quantifying reading of a group of persons or things (e.g. un hatajo de pillos ‘a bunch of rascal’, de disparates ‘of silly things’, de pícaros ‘of rogues, villains’, de bobadas ‘of nonsense’) and underline its pejorative nature. Interestingly, while CLAVE points out that hatajo is partially synonymous to atajo, DUE defines hatajo as hato pequeño de ganado ‘little bundle/flock of cattle’. In addition, the dictionary entries of hato and atajo usually consider the graphical variant hatajo as a synonym. The former primarily refers to a bundle of cloths, while atajo is etymologically linked to the verb atajar (‘to take short cut, to intercept, to cut something/somebody short’). Although CLAVE points out that hatajo is not synonymous to atajo in the sense of ‘short cut’, the graphical and phonological similarity between both variants might be motivated. The Diccionario de Español Actual (Seco et al. 1999), for instance, explicitly indicates that hatajo refers to a little group of cattle, especially a group which has been separated from the (encompassing) flock (e.g. El mayoral divide el rebaño en hatajos (…) ‘The farm manager divides the flock in little groups’).

7.2.3.2.2 The conceptual image of (un) hatajo de N2 In the grammaticalized uses, hatajo de typically refers to a rather small group of despicable or negatively evaluated human beings. In (140), the group of interlocutors is clearly delineated since the speaker knows, of course, how many people he is addressing. They are depicted as a bunch of swines. The quantifying use in (141) profiles a relatively small group of bandits, which are by definition unpleasant persons. The focus on the unexpected nature of the hold-up (antes de que pudieran reaccionar) suggests that the crooks proceeded via surprise-attack since the soldiers largely exceeded them in number.

Quantification construal in BQs | 375

(140) – Dos cosas, por lo menos, debíais aprender de este hecho, hatajo de cabritos. (1986, novel) ‘– Two things, at least, you ought to learn from this event, bunch of swines (lit. little goats).’ (141) Allí le previnieron que la chusma del Arrabal era a la sazón ciertamente peligrosa, pese a lo cual continuó su camino al atardecer. Ya noche cerrada bordeó el inmenso puerto de Maó, escoltado por una patrulla de ingleses. No habrían recorrido media milla hacia San Felipe cuando toparon en un recodo con un hatajo de bandidos que despacharon a los soldados antes de que pudieran reaccionar. (1986, novel) ‘There they warned him that the rabble of Arrabal was absolutely dangerous at that time. In spite of that he continued his journey at dusk. When it was already completely dark, he came close to the immense harbor of Maó, escorted by a patrol of Englishmen. They had not yet travelled half a mile towards San Felipe when, at a bend in the road, they ran into bunch of (lit. herd of) bandits that got rid of the soldiers before they could even react.’ Minimally, the N2s quantified by hatajo de are profiled as a group of SPATIOTEMPORALLY CONTIGUOUS entities (facet (a)). Two major facets within the conceptual image of hatajo de precisely build on the spatiotemporal contiguity and are primarily related –but not exclusively– to the quantifying and the specifying use respectively. First, in line with the specific composition of a herd by a farmer, hatajo de can highlight the idea that the N2s are DELIBERATELY ACCUMULATED (facet (f)) or gathered on purpose. In (142), for instance, the possessive su enhances the idea that the woman meticulously selected the saints to take care of her salvation. From the moment other mercies are needed, other saint(s) will be preferred. Likewise, if a set of N2s is intentionally composed by an identifiable source, the accumulation of N2s results in a CLEARLY DELINEATED (facet (g)) set (as in (140)–(142)). (142) Porque esa hambre de iglesia, de misas y novenas, de bodas y bautizos pendientes, la tenían lo mismo los vencedores que los vencidos. Yo mismo vi a aquella mujer que se llevó en un asno su hatajo de santos para el fogón ponerle ahora una vela a la Virgen, promesa de algún beneficio o demanda de algún favor perentorio. Decían que, durante la guerra, los ateos hacían blasfemar a los cristianos antes de matarlos para que así se fueran derechos al infierno. (1982, novel)

376 | QN-related schematization and N2-profiling

‘Because that hunger for Church, for masses and novenas, for weddings and baptisms expected for the near future, the victors and vanquished had it equally. I saw with my own eyes that woman who took away on a donkey her bunch of saints to (burn them in) the stove, put now a candle for the Blessed Virgin, promise of some benefit or supply for a certain urgent favor. It seems that, during the war, the atheists made the Christians blaspheme before killing them so they would go straight to hell.’ Second, as a faint vestige of the herd-instinct guiding the movements of a group of cattle, the grammaticalized uses of hatajo de can foreground the idea that all N2-entities are moving into the SAME DIRECTION (facet (c)) or are performing the same action (e.g. (141) and (143)). Similarly, the broader context of (144) pictures a collective crime (tropelía) whereby one particular person goaded the rest of the group to join him in crime (to eat a particular sacred chicken). In (145), the context stresses the STUPIDITY (su estupidez) of the N2s (facet (d)): without exception, the soldiers are qualified as stupid and incompetent, they seem to be incapable of making their own decisions and unquestioningly swallow their superior’s orders. The final facet which originates in the literal herd-instinct concerns the LACK OF INDIVIDUALITY OR EQUIVALENCE among the N2s (facet (e)) which can be highlighted in the grammaticalized uses of hatajo de. Example (146) illustrates the homogeneization evoked by hatajo de: in the view of that woman, all authorities present are complete careerists and traitors, regardless of their rank. In her eyes, they were all the same. It goes without saying that the facet of EQUIVALENCE AMONG THE N2-ENTITIES lends itself perfectly to specifying uses. (143) Las tabernas habían cerrado a toque de oración y los vecinos tuvieron luces prendidas en las ventanas hasta el amanecer. Aun así la turba y un hatajo de penados fugados entraron a saco y cometieron desmanes. (1986, novel) ‘The inns were closed at the prayer call and the neighbors held their lights turned on at the windows until dawn. Still the mob and a bunch of (lit. herd of) escaped convicts manage to enter plundering and to commit outrages.’ (144) Bernardo Olleros estaba con el resto de la recua en el Viña Honorio, que fue el restaurante donde se cometió la tropelía. Y allí comulgué, Fermín – volvió a sollozar –, allí me atiborré de la carne culpable... Éramos una tropa de caníbales, un hatajo de borrachos devorando el muslo mutilado... (1992, novel) ‘Bernard Olleros was with the rest of the group (lit. train) in the Viña

Quantification construal in BQs | 377

Honorio, which was the restaurant where the outrage happened. And there I got fooled, Fermín – he started to sob again –, there I stuffed myself full of the culpable meat… We were a troop of cannibals, a bunch of (lit. herd of) drunkards devouring the mutilated leg.’ (145) Me salí a los diecinueve, porque no encontré lo que buscaba: el espíritu puro y guerrero de la Roma mejor, de la gran Roma. Lo que allí había era un hatajo de tarados y de incompetentes. A la vez que una falta de seguridad en sí mismos y de profesionalidad en los superiores, que siempre actuaban procurando disimular su estupidez con el ordeno y mando. (2002, books) ‘I left at the age of nineteen, because I did not found what I was looking for: the pure and warlike spirit of the better Rome, of the great Rome. What was there was a bunch of (lit. herd of) stupid and incompetent people. In addition to a lack of self-confidence and of professionalism of the superiors, who were always acting while trying to hide their stupidity with regard to assuming leadership and command.’ (146) – Aquella humilde mujer, que se había quedado sin marido y cuyo valor más preciado era el hijo que había llegado a sargento, me miró con un desprecio infinito, casi con asco. Para ella, todas las autoridades presentes en el acto, con uniforme y sin él, no éramos más que un hatajo de arribistas y traidores. (2002, novel) ‘That humble woman, who lost her husband and whose most precious merit was her son who made it to sergeant, looked at me with infinite disdain, almost with disgust. To her, all the authorities assisting at the funeral, with and without uniform, we were nothing more than a bunch of (lit. herd of) careerists and traitors.’ Another facet typical of the grammaticalized uses of hatajo de is some sense of ANTAGONISM (facet (b)). Typically, the interests or viewpoints of two entities are diametrically opposed (or at least compared), usually between a single individual and a group of (similar) people. In example (146), the perspective of the woman (aquella humilde mujer) is contrasted with the intentions of the authorities. In example (147), the antagonism holds between the speaker and his addressees. In four examples, hatajo de is used in the context of a war or battle, as in (148). (147) – No os va a ser tan fácil, hatajo de galopines. Todavía no nació quien me pueda tomar el pelo. (1986, novel)

378 | QN-related schematization and N2-profiling

‘I will not be easy for you, bunch of (lit. herd of) rogues. The one who can tease me has to be born still.’ (148) Lo demás es zarandaja, ¿o usted cree que Milwaukee sacrificaría a uno sólo de sus hijos por un hatajo de kurdos que murieron hace veinte años? (2003, press) ‘Everything else is tittle-tattle, or do you really think Milwaukee would sacrifice a single one of its sons for a bunch of (lit. herd of) Kurds who died twenty years ago?’ The link between the facet ANTAGONISM and the source semantics of a literal hatajo is not straightforward. Whether it relates to the tension between the small group (hatajo) and the larger flock (rebaño) from which it has been separated, to the opposition between a leading entity (e.g. herdsman) and the faceless crowd, to the fact that the N2-entities are animate and strictly speaking could have a will of their own, or whether the facet is exclusive of the grammaticalized uses is an interesting question, but which cannot be resolved on the basis of 16 occurrences only (which contain only 1 literal use). In addition, the etymological uncertainty about hatajo makes multiple influences in the development of hatajo de possible. The context of the following example (149), for instance recalls both the source frame of the verb atajar (restos de imágenes) as the frame of a literal herd in the sense that the pictures were gathered and selected carefully. (149) (…) y yo me volví a la catedral, ahora vacía, sembrada de partituras musicales que habían volado del coro, de libros y misales con las hojas arrancadas sacados de los anaqueles de la sacristía. Una mujer, indiferente, arramblaba tranquila con un hatajo de restos de imágenes (troncos, brazos, piernas, cabezas mutiladas que te miraban con ojos sonrientes). Tomé del suelo un misal de cantos dorados y (…) y eché a correr camino de mi casa grande felicísimo con mis trofeos. (1982, novel) ‘(…) and I went back to the cathedral, which was not empty, covered with music scores that had blown off from the choir, with books and missals with the sheets toured out, taken from the shelves of the sacristy. A woman, indifferent, made off calmly with a bundle of (lit. herd of) remnants of images (trunks, arms, legs, mutilated heads that stared at you with smiling eyes). I picked up from the ground a missal with golden edged chants and (…) and I started running in the direction of my big home very pleased with my trophies.

Quantification construal in BQs | 379

Figure 8 visualizes the conceptual facets which form part of the imageschematic structure of hatajo de. In view of the exploitation of the literal ‘herdinstinct’, it does not come as a surprise that hatajo de primarily combines with despicable N2s, usually nouns referring to stupid, low-rank or criminal people. Literal frame

Grammaticalized uses (a) SPATIOTEMPORAL CONTIGUITY (b) ANTAGONISM (c) SAME DIRECTION, SAME ACTION

herd instinct

(d) STUPIDITY (e) LACK OF INDIVIDUALITY (f) ACCUMULATION

composition herd (g) DELINEATED, CLEAR BOUNDARIES

Fig. 8: Conceptual facets in the conceptual image of hatajo de

Interestingly, I hardly observed differences in the grammaticalized binominal uses of hatajo de, hato de and atajo de. By way of comparison, the total number of occurrences per uses are listed in Table 6. Tab. 6: Distribution per reading for hatajo de, hato de and atajo de

Hatajo de Hato de Atajo de

H

Q

S

Ambig.

Indet.

TOTAL

1 10 2

6 4 2

7 3 8

0 0 2

2 0 0

16 17 14

It strikes the eye that hato de preferably combines with literal uses, as in (150)– (151), while atajo de mainly yields specifier uses, as in (152). As to the combinatorial pattern, the grammaticalized uses of hato de only combine with nouns referring to negatively evaluated human beings (e.g. demonios ‘devils’, putas ‘whores’, gañanes ‘boors’, etc.) while both hatajo de and atajo de also combine with inanimate nouns. In some examples, the source frame obviously persists in the grammaticalized uses. For instance, in the specifying use of atajo de in (153), the bundling or compiling of the N2-entities is particularly obvious: the interlocutor in (153) accuses the speaker of considering himself as a bundle of perfec-

380 | QN-related schematization and N2-profiling

tions only, and the snoring does not square with his perfect character. However, the substitution of atajo in (152) and hato in (154) by hatajo does not bring about a substantial difference in conceptualization.203 (150) ¿Habrá algo más ingenuo y bucólico que un pastorcillo soplando la flauta en un ribazo, mientras apacienta su hato de ganado? (1987, press) ‘Will there be anything more ingenuous and bucolic than a little shepherd playing his flute in a steep bank, while his herd of cattle grazes?’ (151) La MADRE ya entregó a ANTONIO un hato de ropa y libros, le peina, arregla la ropa, le da un beso muy fuerte y ANTONIO avanza hacia su TÍA. (1979, theatre) ‘The mother already returned to Antonio a bundle of clothes and books, she combs his hair, she dresses him well, gives him a big kiss and Antonio steps forward to his aunt.’ (152) – Nuestro lugar está con el pueblo – dijeron. – Esto no es el pueblo – repliqué –, es la chusma, y no sabéis de lo que es capaz este atajo de bestias. (1975, novel) ‘– Our place is here with the people – they said. – These are not the people – I replicated, they are the plebs, and you don’t know what this bunch of beasts is capable of.’ (153) – ¿Que yo ronco? ¿Estaba roncando? – Sí, roncabas. Tú, atajo de perfecciones, roncabas y además he de decirte que con toda la orquesta: flauta, trombón, oboe y hasta violonchelo. – Estás loca, estás de atar. O sea, que como yo ronco – que es mentira – tú te pones a cantar a voz en grito. (1990, novel) ‘– That I snore? I was snoring? – Yes, you snored. You, bundle of perfections, you snored and in addition I have to tell you that with the entire orchestra: the flute, the trombone, the oboe and even the violoncello. – You are insane, you are to be tied. I mean, that as I snore – which is a lie – you start to sing out loud.’ || 203 Interestingly, my Spanish informants who were so kind as to comment on the differences in meaning between the QNs in (152) and (154) as well as on the soundness of substituting the respective QNs by hatajo unanimously observed that (1) the QNs were perfectly mutually substitutable in these contexts and (2) that the constructions were highly marked and infrequent, especially hato de which is associated with more cultivated speech. Interestingly, atajo is intuitively associated with its meaning as a short cut.

Quantification construal in BQs | 381

(154) Hay hombres también y se charla animadamente, estallando de vez en cuando una risita aguda. Una de ellas ha dejado de tocar el piano, (…), como las demás, mira al viejo que, junto con Andrea y la directora del Club, permanecen en el umbral. A su vez, el viejo las mira: “¿Mujeres? ¡Un hato de viejas!... Onduladas, maquilladas, emperifolladas..., ¡pero todas viejas!” (1985, novel) ‘There are also men and there is vivid talking going on, what sometimes bursts into a wicked little laugh. One of them [the women] has stopped playing the piano (…), like the other ones, she looks at the old man who, next to Andrea and the directress of the Club, remains at the threshold. At his turn, the old mean looked at them: “Women? A bunch of old wives!... With wavy hair, with makeup, dolled up…, but all old women!”’

7.2.3.3 Conceptual image of (una) letanía de N2 7.2.3.3.1 Lexicographical accounts In all three reference dictionaries consulted, the definition of letanía concerns the Christian prayer which consists of a series of petitions to the Trinity and/or the blessed Virgin and which is sung after the rosary. Within the same frame, the second definition in DRAE concerns the public procession which is held on a regular basis as a rogation by singing the litanies. The procession-sense only occupies the final place in DUE. In addition, letanía is defined as a series or enumeration of words, names or expressions with an insistent effect, as in No vengas con esa letanía ‘Do not come again with that litany’. Crucially, DRAE, DUE and CLAVE all specify that this use belongs to colloquial or informal speech. No mention is made of the quantifying or binominal use of letanía de, not even indirectly in the form of illustrations. The only binominal example is proposed by CLAVE, yet does not correspond to the coextensive quantifier construction. In Me sé de memoria la letanía de los reyes godos ‘I know by heart the litany of the Gothic kings’, the PNP (de los reyes godos) specifies which litany exactly is meant without exploiting the QN’s quantifying potential.

7.2.3.3.2 The conceptual image of (una) letanía de N2 Within the frame of the Christian Church, a litany is a prayer which is traditionally led by a priest and responded by the people in the form of, usually short, response formulae. It consists of a long list of supplications to the Trinity (or

382 | QN-related schematization and N2-profiling

other Saints) to implore their help in precarious situations. Several facets of this original scenery live on in the grammaticalized uses of letanía de, which is nicely illustrated by the quantifying use in (155). The context does not only evoke the frame of the catholic church (cánticos, blasfemias, pasión sagrada, adorable redentor), but also a ‘leading’ person (señor Francisco) and (an) ‘interlocutor(s)’ (su mujer, en público) who is expected to agree. Typically, letanía de profiles the N2-entities as a long and dreadful sequence – either an enumeration or a repetition – of expressions, uttered with a specific purpose or in self-defense. Crucially, the (re)action by the ‘speaker’ who pronounces the letanía de N2 is usually felt as exaggerated and shocking (clímax, apogeo, catástrofe familiar), while the speaker himself rather expects affirmative reactions from his interlocutor(s). In other words, the grammaticalized uses can copy the following elements of the source frame: the speech participants, the internal focus on antagonism (actionreaction, opening lines-response formula) and the sequential and repetitive character. Apart from being less congruent with the context, the substitution of letanía by montón in (155), the facets ‘successive’, ‘adressed to a specific public in an attempt to elicit a confirmative reaction’ and ‘remarkable or exaggerated’ would get lost. Nor would the antagonism between Francisco and his wife be profiled in the quantity assessment. Un antológico montón de blasfemias profiles the blasphemies as a chaotic grouping of expressions which are not even necessarily pronounced within a single time span. (155) El señor Francisco, recién desvelado por los cánticos, corría furioso, en camiseta y calzoncillos, de un lado a otro, vomitando a voz en grito una antológica letanía de blasfemias, mientras era sujetado por su mujer para impedirle que llevara a cabo su insensato propósito de salir a la calle a cantarlas en público. Si, entre la luz de las antorchas, divisaba algún vecino “rico y facha”, cantando aquello de “... por vuestra pasión sagrada, adorable redentor.”, el clímax llegaba a su apogeo, y la señora Ángela tenía que esgrimir sus mejores dotes de persuasión para evitar que aquello degenerara en catástrofe familiar. (2001, theatre) ‘Mr. Francisco, recently woken by the canticles, was running furiously, in his undershirt and underpants, from one side to another, bringing up with a shouting voice an anthological litany of blasphemies, while he was held back by his wife in an attempt to prevent him from bringing to an end his foolish intention to go out on the street and sing them in public. If, between the light of the torches, he saw any “rich and fascist” neighbor, singing that of “… by your holy Passion, adorable Redeemer”, the climax would come to its apogee, and Mrs. Ángela

Quantification construal in BQs | 383

had to use her best powers of persuasion to avoid that this would degenerate into a family disaster.’ The image schema of letanía de can be broken down into at least four conceptual facets which can be related to facets of the source frame and present two levels of schematization. First, in line with the prayer-frame where the clergy leads and the people respond, the N2-entities quantified by letanía de can usually be attributed to a particular SPEAKER (by extension also writer) and are addressed to a particular PUBLIC (facet (a)). In (156), the swearwords are pronounced by David and meant for his mother. Since 14 of 17 grammaticalized instances of letanía de quantify a kind of discourse, the speaker-hearer interaction is at least implicitly present. In (157), it can be inferred that the N2-entities were uttered by politicians and addressed to the electorate audience. When abstracting away from the discourse-space, the N2-entities continue to have an identifiable SOURCE (and/or AFFECTEE) (facet (e)). In (158), the litany of arrests, kidnapping, etc. is initiated, whether or not intentionally, by the Moroccan government. (156) Enseguida saldrá con el cesto a recoger la ropa seca, y David reanuda su letanía de tacos en voz baja. – ¿Todavía con esta monserga? – dice mamá, fingiendo un malhumor –. Tuve que hacerlo, hijo. Tú nunca habrías consentido que se lo [KV: perro] llevaran. – ¡Claro que no! ¿Cómo te dejaste convencer? (2000, novel) ‘She will come out in a minute with a basket to fold the dry laundry, and David resumes silently his litany of swearwords. – Still that drivel? says mum, faking a bad mood – I had to do it, son. You would never have allowed that they take him [KV: dog] away. – Of course not! How could you let you convince?’ (157) Me pone más cachondo esa publicidad disuasiva de los agentes macroeconómicos basada en que la única razón es la tensión, que toda esa letanía de razones microeconómicas de los convocantes para justificar y desdramatizar el acontecimiento social del siglo, o del medio siglo, como lo bautizó Gutiérrez. (1988, press) ‘I get more hot-blooded by that misleading publicity by the macroeconomic agencies based on the idea that the only motive is the tension, than by this whole litany of micro-economic reasons of the demonstrators to justify and mitigate the social event of the century, or the half century, as Gutiérrez baptizes it.’

384 | QN-related schematization and N2-profiling

(158) Está, en primer lugar, el tema pesquero, que desde hace años envenena unas relaciones que podrían ser mejores. Los marroquíes decidieron un día, unilateralmente, aumentar a doce millas el mar territorial y a setenta la zona pesquera, con un "dahir" (decreto real) que hizo historia. Así comenzó una extensa letanía de detenciones, secuestros, protestas diplomáticas y multas. (1976, press) ‘In the first place, there is the fishing-related issue which since several years poisons a relationship that could be better. One day, the Moroccans decided unilaterally to raise to twelve miles the territorial sea and to seventy the fishing zone, with a “dahir” (a royal decree) that has written history. That’s how an extensive litany of arrests, kidnappings, diplomatic protests and fines has started.’ As a faint vestige of the precise composition of the prayer, letanía minimally profiles the N2-entities as the elements of an ENUMERATION (facet (b2)) or REPETITION (facet (b1), as in (159)). The structure of the enumeration(/repetition) can be twofold: either N2 is the umbrella-term for the constituting items of the enumeration, as in (160), (157) and (158), or the umbrella-N2 is immediately followed by the actual enumeration, as in (161). It goes without saying that long enumerations or repetitions can be felt as LONG-WINDED or tedious (facet (f)), when abstracting away from the original frame, as in (162). (159) El primer valor del libro consiste en el hallazgo de su estructura formal: la de un diario hablado. Un bombardeo informativo va imprimiendo a diario en nuestro subconsciente una letanía de frases, que, por efecto de su acumulación y reiteración, acaban por perder la dureza de su referencia primera. (1996, press) ‘The first value of the book consists in the finding of his formal structure: the one of a spoken diary. An informative bombardment is stamping every day a litany of sentences in our subconscious, which, as an effect of their accumulation and reiteration, end up losing the firmness of their first reference.’ (160) Biblia en mano, María comenzó su letanía de acusaciones, contra el presidente y los "curuleros" (congresistas) vinculados al Proceso 8.000. Después fue encarcelada. (1996, press) ‘Holding the Bible in her hand, María started her litany of charges, against the president and the “curuleros” (congress members) linked to the 8.000 Process. Afterwards she was imprisoned.’ (161) No ha sido fácil vencer su resistencia, para que se aviniese a esta entrevista. Le he desgranado la letanía de colegas que le han precedido:

Quantification construal in BQs | 385

Eduardo Móner, Roberto García Calvo, José Augusto de Vega, Javier Delgado Barrio, Clemente Auger, Carlos Jiménez Villarejo... (1996, press) ‘It was not easy to beat his resistance, so that he would agree on that interview. I have rattled off the litany of colleagues that preceded him: Eduardo Móner, Roberto García Calvo, José Augusto de Vega, Javier Delgado Barrio, Clemento Auger, Carlos Jiménez Villarejo…’ (162) Sólo un vago populismo, impreciso y contradictorio, con un fuerte olor a naftalina que repite sin rubor una cansina letanía de viejos dogmas marchitos. (1987, press) ‘Just a vague populism, imprecise and contradictory, with a strong smell like mothballs that in a shameless way repeats a weary litany of old and wilted dogmas.’ Similar to the specific purpose (the help of the Trinity) and occasion (e.g. during an epidemic or war) of literal litanies, the grammaticalized uses typically present the litany of N2s as a REACTION to a particular situation or event (facet (c1)) or as pronounced with a specific PURPOSE in mind (facet (c2)). In (156), for instance, the insults David addresses to his mum are meant as a reaction to her giving up on his dog. In (161), the speaker previously mentions the aim of his enumerating the colleagues (para que se aviniese a esta entrevista). In (157) and (163) the STRONG IMPACT of the litany on the affectee (facet (g)) is highlighted (me pone más cachondo, tiene a los canarios en un estado de exasperación), which is a further abstraction of the literal frame. (163) Desde Madrid se ha perpetrado una larga letanía de agravios y dejadeces que tiene a los canarios en un estado de exasperación. (1977, press) ‘Madrid has perpetrated a long litany of insults and neglects which holds the habitants of the Canary Islands in a state of exasperation.’ The three preceding conceptual facets can be directly related to a subpart of the prayer-frame. The final facet (h) observed in the instances of quantifying/ specifying letanía de, however, is diametrically opposed to the corresponding facet in the literal frame. While in a literal litany, the people respond in a docile way to the introductory sentences by the priest, the grammaticalized uses evoke a (public of) interlocutor(s) that disagree or resist (cf. vencer su resistencia en (161)). The interlocutors or affectees are usually shocked or surprised by the letanía de N2. In (155), the wife is doing everything she can to prevent the speaker from continuing his shocking list of insults on the street. When no discourse-frame is evoked, the context of letanía de always involves some element of EXAGGERATION, SURPRISE or IRONY, as in (164).

386 | QN-related schematization and N2-profiling

(164) Expertos canónicos corrigen al episcopado y creen que los malos tratos anulan el matrimonio. JUAN G. BEDOYA – Madrid La letanía de tragedias que causan cada día los malos tratos domésticos ha metido a la Iglesia católica en un nuevo jardín de contradicciones y críticas. Externas e internas. (2002, press) ‘Canonic experts correct the episcopate and believe that the illtreatments annul marriage. JUAN G. BEDOA – MADRID The litany of tragedies that the domestic maltreatments cause every day has put the Catholic Church in a new garden of contradictions and criticsm. External and internal.’ Figure 9 visualizes the conceptual facets constituting the conceptual image which letanía imposes to the N2. The left-most column lists the subparts of the source frame the conceptual facets correspond to. The two following columns reflect how the facets of the prayer-frame can persist in the grammaticalized uses, distinguishing between two levels of abstraction. Literal frame

Grammaticalized uses

priest - public

(a)

opening sentence –

(b1)

REPETITION

response formula

(b2)

ENUMERATION

(c1)

REACTION TO A SITUATION/EVENT

(c2)

PARTICULAR PURPOSE IN MIND

supply for help

agreement

(d)

SPEAKER < > INTERLOCUTOR

DISAGREEMENT, RESISTANCE

(e) SOURCE < > AFFECTEE (f) LONG-WINDED

(g) STRONG IMPACT ON THE AFFECTEE

(h) EXAGGERATION / IRONY / SURPRISE

Fig. 9: Conceptual facets in the conceptual image of letanía de

In view of the image-schematic structure of letanía de and the importance of the discourse space including all its participants (speaker-message-interlocutor), it should not come as a surprise that una letanía de N2 repeatedly combines with verbs of saying such as pronunciar ‘to pronounce’, gritar ‘to shout’, cantar ‘to sing’, escribir ‘to write’ and desgranar ‘to reel off’ in particular. The immanent succession of N2s quantified by letanía endows the QN with a certain aspectual potential: it combines with verbs or adverbial adjuncts focusing on the end of the enumeration (acabar con ‘to stop’), on its beginning (comenzar ‘to start’), its repetitiveness (todas las tardes ‘all the afternoons’) or its duration (durante siete años ‘during seven years’).

Quantification construal in BQs | 387

The combinatorial pattern concerning the N2 equally reflects the initial frame. The binominal instances of letanía de almost exclusively – in 23 of 26 occ. – combine with N2s which either literally refer to speech acts or expressions (e.g. interrogantes ‘questions’, acusaciones ‘charges’, consejos ‘pieces of advice’, blasfemias ‘blasphemies’, tacos ‘swearwords’, etc.) or are contextually pictured as pronounced out loud (as in (165)). (165) Desde la altura de Canada Dry escucho, en una babélica confusión de lenguas, a los guías que, en inglés, alemán, francés o italiano, convierten esta primera maravilla del mundo que es Keops, en una letanía de cifras: “En sus orígenes, la pirámide tenía 146 metros de altura, es decir, dos veces la altura de las torres de Nôtre-Dame de París. Su base cubre más de cinco hectáreas (…). Son 2.250.000 bloques de piedra, (…).” (1995, books) ‘From the height of Canada Dry I listen, in a Babel-like confusion of languages, to guides who, in English, German, French and Italian turn this first Wonder of the World that Keops is, in a litany of numbers: “Originally, the pyramid was 146 meters high, that is, two times the height of the towers of the Nôtre-Dame in Paris. Its base covers more than 5 hectares (…). It counts 2.250.000 stone blocks, (…).’

7.2.3.4 Conceptual image of (una) barbaridad de N2 7.2.3.4.1 Lexicographical accounts The first definition formulated both by DRAE and DUE links barbaridad to the quality bárbaro. Historically, the ancient Greek and Romans referred to foreigners as ‘Barbarians’. Since ‘foreign’ is spontaneously considered the opposite of civilization, as the absence of culture, it is only a small leap from ‘Barbarian’ to either ‘brute, cruel’ or ‘idiot’. ‘Ignorance’ and ‘rudeness or brute cruelty’ correspond indeed to the two following definitions listed by DRAE and DUE and the first two definitions by CLAVE. In addition, both DRAE and CLAVE insist on the excessive character of barbaridad. According to the former, barbaridad can refer to the excess in itself or to an exaggerated action. Further, all three reference dictionaries conclude with una barbaridad ‘a barbarity’, which in informal or colloquial speech means “mucho” ‘many, a lot’ (DUE), “cantidad grande o excesiva” ‘big or excessive quantity’ (DRAE) or “gran cantidad” ‘big quantity’ (CLAVE). Only DUE and CLAVE add a binominal example (Tengo una barbaridad de trabajo ‘I have a barbarity of work’ (DUE); Hizo una barbaridad de fotografías ‘He took a barbarity of pictures’ (CLAVE).

388 | QN-related schematization and N2-profiling

7.2.3.4.2 The conceptual image of (una) barbaridad de N2 Again, the motivation of the image-schematic structure of the QN is to be sought in the source frame. Barbaridad de typically profiles an unusual and excessive quantity of N2s which somehow transgresses ethical or established norms, whether or not intended by the protagonist. In (166), the agent commits the indecency of leaving an excessive tip, even though he read in the travel guide that such gratuities are “not done” in London. The excess is usually committed by an outsider, who is unaware of the established norm (similar to the ‘foreigner’ which are considered ‘uncivilized’ in ancient Roman empires). The substitution of barbaridad by montón would profile the amount as more acceptable and less excessive, which consequently would conflict with the contextual cues (underlined in (166)) that picture the gratuity as completely outrageous. (166) Pagó a regañadientes la abusiva nota del té completo dejando incluso una propina excesiva para aliviar de algún modo la afrenta de aquella situación. Pansy le regañó al salir. En la guía Fodor's había leído que las propinas en Londres no debían ser superiores en ningún caso al 15 por ciento suponiendo que no estuviera ya incluida en la factura. Y él había dejado una barbaridad de propina que podría haberse destinado a la compra de otra jarrita del té (…). (1995, novel) ‘He paid unwillingly the unfair bill of the tea, while including an excessive tip to relieve somehow the embarrassment of that situation. Pansy quarreled with him when leaving the place. She had read in the Fodor’s guide that in London, gratuities should in no way exceed 15 per cent, provided that it is not yet included in the bill. And he had left a barbarity of a tip that would have sufficed to by another jug of tea.’ The conceptual image of barbaridad de can be broken down in several conceptual facets which either belong to the background in the literal frame or are abstractions of originally profiled conceptual elements. As a faint vestige of the participant structure in the literal interpretation of barbaridad as a crime, comprising the agent, the patient or victim and the crime itself, the focus can be on the antagonism in quantifying readings as well. In (167), the ANTAGONISM (facet (a)) is twofold: the suggestion of the government not only conflicts with the vision of the medical staff (enfrenta X con Y), it is also diametrically opposed to the speaker’s point of view (yo, pero, absolutamente inaceptable). The importance of the participant roles in the original frame may also lead to the idea that the IMPACT of the barbaridad de N2s on the victim (facet (f)) is huge. In (168), the context indicates that the effects of bad government and the crisis have reached the coachmen of Sevilla. The impact of new taxes on their miserable pittance leads them to organize a manifestation.

Quantification construal in BQs | 389

(167) Lo más terrible es que ha surgido un problema nuevo, un problema que enfrenta, concretamente, a la organización médica colegial con la administración. Yo le entiendo perfectamente la necesidad que tiene el insalud de reducir el enorme consumo de medicamentos que se produce en España, pero la vía que ha elegido para conseguir esta reducción es absolutamente inaceptable, es la vía de limitar la independencia y la libertad del médico para recetar. Es que se gasta una barbaridad de dinero en medicinas inútiles, porque la asistencia primaria es de muy baja calidad y, entonces, el médico se defiende dando infinidad de medicinas que no hacen ninguna falta, que sería un ahorro tremendo para el hospital, (…). (1983, oral) ‘What is most terrible is that a new problem arose, a problem that, to be precise, puts the medical staff face to face with the government. I perfectly understand the necessity of the National Health Institute to reduce the enormous consumption of medicaments that is current in Spain, but the manner they have chosen to achieve this reduction is completely unacceptable, it is the option of limiting the independence and the freedom of the doctor to prescribe. The thing is that a barbarity of money is spent in unnecessary medicines, because first assistance is of very low quality, and, consequently, the doctor defends himself giving an infinity of medicaments which are not needed, and which would be a tremendous saving for the hospital.’ (168) Hasta a los coches de caballos les ha llegado la crisis y los efectos del mal gobierno. A los cocheros de los peseteros de Sevilla les quieren cobrar no sé cuánto al mes, una barbaridad, de cuotas de la Seguridad Social, (…) Dicen que el trabajo está tan achuchado que hay días que no sacan ni para quitar las moscas al caballo. Y se han manifestado. (1994, press) ‘The crisis and the effects of bad government have reached even the horse carriages. They even want to charge the coachmen of the skinflints of Sevilla I don’t know how much a month, a barbarity of fees of Social Security, (…). They say that the work is so though that there are days that they do not even earn enough to keep the flies away from the horse. And they have taken part in a demonstration.’ In line with the crime or stupidity of a literal barbarity, the large quantity of N2 usually implies the TRANSGRESSION of an ETHICAL OR ESTABLISHED MORAL BOUNDARY (facet (b)). In (166), the protagonist intentionally ignores the general practices mentioned in the guide. In (167), the reduction proposed by the government is inacceptable in view of the immense amount of unnecessary expenses made. In

390 | QN-related schematization and N2-profiling

(168), the introduction of new taxes on miserable salaries is pictured as unethical. Interestingly, the context typically plays with the idea of whether or not the barbaridad de N2 occurs OUT OF IGNORANCE (facet (c)), which is in line with the original frame of barbaridad de in which strangers are considered unfamiliar with general rules and standards. Abstracting away from concrete social norms, barbaridad de minimally profiles the number of N2s as EXCESSIVE (facet (g)). In (169), the barbarity of things one could do does not imply any unethical practice. The context pictures a dream in which the fictional and huge sum of money would suffice to solve almost any problem, and still the protagonist would have something left to spare. (169) Oye, soñemos un poco, ¿qué harías tú o qué plantearías tú que se podría hacer si de repente los cien millones de dólares de que se va a dar a la contra, pues, no sé . lo los tuvieras tú o o se dieran definitivamente al pueblo de Nicaragua, qué se podría hacer con cien millones de dólares? , una barbaridad de cosas. Se podrían hacer cien millones de proyectos aquí, para levantar esto. Se podría hacer casi todo, yo creo. (…) Pues, no sé, el problema del agua, el problema de las viviendas, que todavía no tienen apenas casas, el problema de la de la comida todo pues. Aquí se podría hacer todo y sobraría plata para hacerlo en otros doscientos asentamientos, yo creo. (1986, oral) ‘Hey, let’s dream a little bit. What would you do or what would you propose that could be done if all of a sudden the hundred million dollars that will be given in return, well, I don’t know. If you would have them of if they would definitely be given to the people of Nicaragua, what could be done with hundred million dollars? A barbarity of things. Hundred million projects could be done here, to build it up. Almost everything could be done, I think. (…). Well, I don’t know, the water problem, the housing problem, the fact that they still hardly have houses, the food problem, so everything. Here everything could be done and there would still be money to spare to do it in two hundred more settlements, I believe so.’ In addition, the original focus on ethical norms or social standards can persist in the quantifying reading in the form of an identifiable EVALUATOR or JUDGE (facet (e)). Example (167), for instance, nicely illustrates that the quantity assessment by barbaridad de reflects the speaker’s subjective viewpoint (Yo entiendo perfectamente … pero…). Similarly, example (170) presents an interview where the woman’s opinion is asked on the employment of her husband. Although doing overtime is not a wrong practice strictly speaking, it is felt as UNPLEASANT by his wife (facet (i)).

Quantification construal in BQs | 391

(170) Mi marido es químico. (…) Bueno ahora ya, claro, al ser la dirección ha dejado un poco lo de los cacharritos, la investigación, (…), pero, claro, ya sabe usted lo que pasa en estas cosas, ahora ya es más desde el despacho y lo que es una dirección, claro. ¿Le ocupa mucho tiempo? No, le ocupa todo, le ocupa una barbaridad de tiempo. Y además lo que pasa en estas cosas que es que cuando se termina la hora de trabajo, suena la campana, para muchos es el momento de dejarlo, pero él, a lo mejor en ese momento, tiene que entrar en un despacho, (…). (unknown, oral) ‘My husband is a chemist. (…). Well now, of course, since he is the management he has left a bit aside the pots and tubes, the investigation, (…), but, of course, you know which things happen then, now he is more (working) from the office and from what management is about, of course. Does it occupy him a lot of time? Now, it fully occupies him (lit. all [his time]), it occupies him a barbarity of time. And what is more, what happens then is that when the working hour comes to an end, the bell rings, for a lot of people that is the moment to leave it [work], but he might have to go into an office at that moment.’ Minimally, barbaridad de profiles an UNUSUAL number of N2s (facet (d)), which reflects the original transgression of moral boundaries or norms. In (170), the husband is explicitly compared to the usual husband (para muchos es el momento de dejarlo) who comes home at a reasonable hour. In (169), the number of projects the protagonist could achieve is fictive right from the start (soñemos un poco). In two occurrences only, barbaridad de does not involve breaking an ethical or established rule. Whereas in (170), the unpleasant consequences of the time spent at work are profiled, example (171) induces some sense of IRONY or SURPRISE (facet (h)): pieces of candy are not expected to hold vitamins, certainly not in large proportions, and cannot be considered a valid alternative to cigarettes. (171) ¿No tendrá usted un pitillo, por casualidad? – Lo siento, pero puedo ofrecerle un caramelo Sugus, que está demostrado que lleva la misma nicotina que un Montecristo y además una barbaridad de vitaminas. (2001, novel) ‘You don’t have a cigarette, by coincidence? I am sorry, but I can offer you a Sugus candy, of which it is demon-

392 | QN-related schematization and N2-profiling

strated that it contains the same nicotine as a Montecristo and also a barbarity of vitamins.’ Figure 10 represents the conceptual image of una barbaridad de N2. The first column lists the conceptual facets profiled in the literal uses. The second and third column indicate the corresponding conceptual elements possibly profiled in the quantifying uses while distinguishing between two levels of abstraction. Literal frame

Grammaticalized uses

agent (-patient)

(a) ANTAGONISM

(f) IMPACT ON PATIENT

crime/

(b) CROSSING ETHICAL VALUE

(g) EXCESS

stupidity

(c) UNINTENTIONALLY / OUT OF IGNORANCE

transgression

(d) UNUSUAL CHARACTER

(h) IRONY

moral boundary

(e) EVALUATION OR JUDGMENT

(i) UNPLEASANT

Fig. 10: Conceptual facets in the conceptual image of barbaridad de

In spite of the rich conceptual image of quantifying barbaridad de, no specific combinatorial pattern is observed. Recall that only 9 occurrences of binominal barbaridad de are found. Both tiempo ‘time’ and dinero ‘money’ combine two times with barbaridad de. Two more nouns are related to ‘money’, viz. cuotas ‘fees’ and propina ‘tip’. More importantly, in 5 (of 9) occurrences, the broader context somehow makes reference to criminal, uncivil or violent practices: e.g. example (168) immediately follows a news flash reporting a murder (asesinato) and refers to bad government, example (167) opposes the antagonists in the discussion and uses bombastic language (terrible, tremendo). I will come back to this observation in Chapter 9.

7.2.4 The contribution of the plural morpheme In line with the cognitive grammarian assumption that a difference in form corresponds to a difference in meaning, the plural variant of a specific QN can be expected to entail a different conceptualization of the N2. In view of the limited frequency of occurrence of the majority of the plural variants (see Table 7), it is difficult to ascertain empirically whether the plural variants of the QNs evoke the same conceptual image as the corresponding singular variants while

Quantification construal in BQs | 393

simply adding the specific contribution of the plural morpheme. Yet, generalizing over all individual plural QNs, it can be shown that the plural morpheme -s brings about the same effect systematically.

Aluviones de

Barbaridades de

Hatajos de

Letanías de

Mogollones de

Montones de

Pilas de

Racimos de

Head Quantifier Specifier Ambiguous Indeterminate TOTAL

Aludes de

Tab. 7: Layering of uses of the plural QNs

5 1 0 0 0 6

1 1 0 0 0 2

0 2 0 0 0 0

1 0 0 0 0 1

0 3 0 0 0 3

0 0 0 0 0 0

122 95 1 17 9 244

35 1 0 3 3 42

34 8 0 0 0 42

It can be derived from Table 7 that the majority of the plural QNs are interpreted literally, as in (172). This observation does not hold for barbaridades de nor for letanías de however, but I have argued earlier that literal binominal uses of singular barbaridad and letanía are highly marked as well. Only montón de appears to have a productive plural variant, which adds up to the evidence of its prototype status. (172) Sobre él [KV: escritorio] yacía la máquina Underwood que había adquirido con el anticipo de Cabestany y dos pilas de cuartillas, una en blanco y la otra escrita por ambas caras. (2001, novel) ‘On it [the desk] was lying the Underwood machine that he had acquired with the advance from Cabestany and two piles of sheets, one (in) blank and the other one written on both sides.’ Crucially, the effect of the plurality marker is identical for all QNs (except for montones de, see infra), regardless whether the interpretation is a literal or a quantifying one. While the QN in itself primarily serves to single out a set of N2s (i.e. a unitizing function), the plural morpheme evokes a plurality of sets (of N2s). The plurality marker is vague as to the exact number and simply supposes ‘more than one’. It is exactly the plurality of the QNs (or of the sets of N2s) which is evoked by the plural variant. In (173), for instance, the binominal syntagm aludes de investigación refers to various avalanches of investigation which have been triggered by the phenomenon ‘soap’. The singular variant alud de investi-

394 | QN-related schematization and N2-profiling

gación would be highly marked in this context, since the sudden appearance of the investigation clashes with the large period of time described (desde la creación … hasta nuestros días). Further, polarization can only occur if several sets of investigation are presupposed. The same holds for example (174), where the bunches of faces are necessarily spatially separated, since they occupy different balconies. In (175), the subordinate clause donde quiera que voy indicates that the speaker has gone to several places, which means that the gatherings of people have taken place at different places and at different moments. In other words, abstracting away over all plural QNs, it can be argued that the plural variant evokes at least two and probably many more not spatiotemporally contiguous sets of N2s. It goes without saying that this mere plurality enhances the quantifying potential of the BQ: since a single set of N2s already implies a (large) quantity of N2s, more than one / several sets definitely imply a lot more N2-entities. (173) Desde la creación de la primera narrativa radiofónica ya calificable como radionovela [Painted Dreams (1930)] hasta nuestros días, el soap ha sido –inconsecuentemente – objeto de aludes de investigación. Pero curiosamente, el análisis se ha ido polarizando a lo largo del tiempo en estudios de marketing por un lado, y por otro en especulaciones de tipo sociológico-empirista que procuran detectar el protagonismo del soap, como motivador de destructivos modelos de comportamiento. (1987, books) ‘Since the creation of the first radio play which could already be called radio novel [Painted Dreams (1930)] until nowadays, the soap has been – inconsistently – the subject of avalanches of research. But interestingly, the analysis has been polarizing throughout the years in marketing studies on the one side, and on the other in speculations of sociological-empiricist kind which try to detect the prominence of the soap, as motivating destructive models of behavior.’ (174) Y sin embargo, en la ciudad, qué júbilo, qué sonar de campanas, qué gentío esperando en las calles, espiando en las ventanas, qué racimos de rostros encaramados a balcones como en tiempo de bodas o bautizos reales. (1978, novel) ‘And nevertheless, in the city, what a jubilation, what a bell ringing, what a crowd waiting in the streets, spying from the windows, what bunches of faces climbed up the balconies as in times of royal weddings or baptisms.’

Quantification construal in BQs | 395

(175) (…) que su mayor satisfacción es su larguísima carrera de docente en la universidad española y el montón de libros de texto que han circulado por ahí, entre las generaciones de estudiantes: “Donde quiera que voy, se me acercan montones de personas para decirme que han aprendido en aquellos libritos azules, en los años 50, 60, 70, 80... ”. (2003, press) ‘(…) that his major satisfaction is his very long career of professor in the Spanish university and the heap of textbooks that have circulated there, among the generations of students: “Wherever I go, heaps of persons come to me to tell me that they have learned from those little blue books, in the fifties, sixties, seventies, eighties…”.’ Interestingly, for montones de only, several instances can be observed where abstraction is made of the specific semantic contribution of the plural morpheme. The context of (176), for instance, does not make explicit reference to different groupings of singers to the extent that one might wonder whether there is still a difference between hay un montón de cantantes and hay montones de cantantes. Similarly, in (177), it cannot be derived from the context to which groupings of appointments (N2s) the plural marker refers to. Yet I still argue that a subtle difference in conceptualization distinguishes between the two variants. If in (176) the sentence would have been hay un montón de cantantes, the speaker would somehow have generalized over a set of singers he has in mind and which he contrasts with the small number of presenters. In absolute terms, the quantity assessment by montones de definitely evokes more singers than un montón de cantantes would. Yet the exact number of singers remains vague (recall that the plural morpheme merely implies more than one) and the quantifier uses of montones usually evoke an infinite number (of sets of N2s). For instance, while the amount of appointments in (177) would still appear somehow feasible in the case of un montón de citas, the plural variant evokes an impossible mission. Summarizing, the plural morpheme continues to neutralize to a certain extent the unitizing function of BQs which normally results in a homogeneous set of N2s: montones de evokes not spatiotemporally (sets of) possibly not contiguous N2s. (176) – ¿Se gana más como cantante o de presentadora? – Como cantante se puede vivir muy bien. Si se fija, hay montones de cantantes millonarios y muy pocos presentadores de televisión que se forran. (1990, press) ‘– Does one earn more as a singer or as a presenter? – One can live fairly well as a singer. If you look at it closely, there are

396 | QN-related schematization and N2-profiling

heaps of millionaire singers and very little television hostesses which make a lot of money.’ (177) Melchor, Gaspar y Baltasar apenas si dan abasto. El día 5 tienen montones de citas y, cómo sólo unos reyes realmente "magos" podrían hacer, acudirán a todas ellas. (1995, press) ‘Melchior, Caspar and Balthazar can hardly cope with it. On the 5th they have heaps of appointments, and, as only three really “wise” kings [or magicians] can do, they will keep them all.’ It has to be noticed that only a restricted number of uses of montones de seems to abstract away from the specific conceptual contribution of the plural morpheme, which is plausibly to be related with montón de’s prototype role. This observation reminds of Brems’ claim that the plural morpheme of English QNs is conducive to GR since “the plurality marker enhances the hyperbolic quantifier value of these SN-expressions [KV: Size Noun]” (2007b, 193). More precisely, the plural variants of heaps of and loads of “have grammaticalized the most” (2007b, 185), in contrast to bunches of which does not present a single quantifier use. Interestingly, although the weakening of the contribution of the plural morpheme -s only occurs with the most grammaticalized QN montón de, the plural variants of the Spanish QNs do not necessarily entail a quantifying reading and are generally infrequent. Finally, it bears pointing out that the plural morpheme in itself indeed somehow intensifies the QN. In the following examples, the binominal syntagm obviously refers to a large number of N2, yet the quantifying inferences primarily originate in contextual cues and the QN is used literally. In (178), the scenery contains at least two walls which are filled with piles of books. Although pilas y pilas are interpreted literally, the repetition of the QN, the plural morpheme and the contextual setting evoke a lot of (piles of) books. The same holds for (179), where the protagonists are depicted as surrounded (rodeados) by various piles of boxes. Although the piles maintain their vertical configuration, the fact that they surrounded someone implies that they come in large number. (178) Contra las paredes hay pilas y pilas de novelas baratas del Oeste y la colección entera de "Claro de luna", las lecturas favoritas de mi tío. (1992, novel) ‘Against the walls there are piles and piles of cheap novels from the West and the entire collection of “Claro de luna [Moonlight]”, the favorite lectures of my uncle.’

Conclusion | 397

(179) – No te preocupes por lo del viejo – le dijo su hermano aquella noche, cuando se desplomaron, agotados, sobre sus camas nuevas, rodeadas de pilas de cajas sin abrir –. (2002, novel) ‘– Don’t you worry about the old man – his brother said to him that night, when they collapsed, exhausted, on the new beds, surrounded by piles of unopened boxes.’

7.3 Conclusion This chapter has illustrated the rather concise character of the dictionary descriptions as to the quantifying reading of QNs. If the quantifying reading is recognized at all, the definition is usually limited to hyperbolic quantification or to the indication of its pejorative nature. In this chapter, however, I have argued that every QN evokes its proper schematic conceptualization of the mass, which is to be concretized by the meaning of N2. By contrasting near-synonymous QNs and by meticulously inspecting the context for additional specifications, I have shown that the conceptual images of the quantifying uses are reminiscent of the QN’s source semantics. As a consequence, QNs are only rarely mutually interchangeable. Mutual substitutions usually bring about – at least slightly – different conceptualizations of N2. The fact that native speakers do not observe any considerable difference between the specifying uses of hato de, hatajo de and atajo de does not necessarily contradict this hypothesis, since their etymologies are considered uncertain and interrelated to a certain degree. Further, even among the members of nearsynonymous minimal pairings such as alud de–aluvión de and pila de–montón de–mogollón de, the subtle difference in the conceptual image seems to lead to individual preferences as to collocational patterns. Chapter 8 will elaborate on the fluctuations in the morphosyntactic co-selection pattern depending on the QN chosen and the reading activated (head, quantifier or specifier). The conceptual images of the QNs have systematically been broken up into different conceptual facets which can be more contentful or more configurational. Four recurring facets can be observed, which are all in line with the primary unitizing function of BQs: QN primarily single out a set of N2s (cf. Chapters 2 and 3). First, the conceptual image of several QNs – as well as the particular contribution of the plural morpheme – addresses the spatiotemporal contiguity of the N2s. Second, the similarity among the N2s was presupposed in the cases of pila de, racimo de, letanía, alud de and aluvión de, yet not specifically required with montón de. Mogollón de, on the other hand, preferably quantifies heterogeneous N2s. The third recurrent facet which revolves around the primary

398 | QN-related schematization and N2-profiling

function of the binominal construction as such is the implication of some (preceding) accumulation or gathering event. In addition, several QNs are apt at profiling the N2s as members of a single category, such as montón de, racimo de and hatajo de, which recalls the premises of the specifer reading. Finally, the facet ‘antagonism’, without being directly related to the constructional semantics of the BQ-construction, also plays a prominent role in the conceptual image of aluvión de, alud de, hatajo de, letanía de and barbaridad de. Crucially, the sets of conceptual facets constituting the conceptual image of each QN are not to be seen as necessary or conditional facets. Depending on the specific contextual setting, the construal operations of the mise-en-discours and the N2 quantified, one or more facets can be highlighted. Even when the focus relies on one particular facet, this does not mean that the other ones cease to be profiled. These observations lead me to operationalize conceptual persistence as a gradual, unpredictable and two-dimensional discourse process in Chapter 9. The distinction of QN-related conceptual images throws up an interesting theoretical question, viz. to what extent are the conceptual images or imageschematic structures of QNs conventionalized? This question relates to the semantics-pragmatics distinction. Traditionally, the pragmatic element is considered to disappear after it has conventionalized, that is after it has become an encoded semantic meaning. However, if one assumes that meaning is emergent (cf. a.o. Paradis 2011) and that the meaning structures of a specific linguistic item and of the elements that constitute its immediate context mutually accommodate (cf. a.o. Pottier’s (1998) concept of isosemy, see infra), one cannot but maintain that the boundaries between semantic meaning and pragmatic inference are fuzzy, i.e. that there is a continuum between conventionalized and inferred meaning. Likewise, Brisard (2009) claims that the semanticspragmatics distinction is simply not recognized in cognitive semantics, since the meaning of a specific item cannot be separated from its usage context. Instead of adding conceptual content at the level of semantic representations, construal operations are a matter of “adopting a certain marked stance towards such representations” (Brisard 2009, 66). As to the conceptual images of QNs, I propose to view them as the QN’s meaning potential which gets its final interpretation in specific utterances. In other words, the QN’s meaning is somehow instantiated in specific usage contexts. Yet the QN-related conceptual image is not to be seen as an undetermined semantic representation: even without actualizing context, una pila de gente, un tropel de gente and un aluvión de gente evoke different conceptualizations of the N2 gente (by adding respectively some sense of succession, of hierarchy or dynamicity). Further, I do not equate the chains of schematizations with chains of ‘pragmatic inferences’ from a core set of facets metaphorically related to the QN’s source semantics. Instead, I view

Conclusion | 399

the abstractions as chains of metonymical extensions. The notion of ‘pragmatic inferences’ is thus to be reserved to the inference of quantity assessment that is susceptible of triggering the change into quantifier. However, more interesting is the answer provided by the present case-study to the initial functional-pragmatic question of this volume, i.e. in what way do BQ-constructions differ from canonical quantity assessment by quantifiers? In other words, what is the pragmatic use of BQs given that Spanish already disposes of a large set of standard quantifiers? The short answer to that question is that BQs not only single out a set of N2s, but also systematically add a qualifying component to quantity assessment. For singling out a set of N2s, BQs are considered to act like indefinite quantifiers (Rigau 1991, 338). Within the semantic map of indefiniteness, Van der Auwera/ Van Alsenoy (2011) invite to add the category ‘singular-plural count’ next to ‘singular count’ and ‘plural count’. With the notion of ‘singular-plural count’, they refer to a singular indefinite determiner that scopes over a plural entity, as in Any four men will be able to lift this (in contrast to the indefinite plural count in *Any men will be able to lift this). In my view, the quantity assessment by BQs fits in with the idea of ‘singular-plural count indefiniteness’. Further, BQs add a qualifying component to quantification. In general, quantifiers serve to denote a number or a certain extension along a quantity scale (Langacker 2002, 340). In addition to subsuming instantiation and grounding – like regular quantifiers do (see Taylor 2002, 355) –, BQs also assume a type-specifying function: the QNrelated conceptual images co-designate the type of the N2-entities.

8 The motivated nature of the QN-related coselection patterns Chapter 6 was primarily concerned with the impact of conceptual persistence on the semantics of each BQ. In the present chapter, the claim will be made that conceptual persistence, the constructional semantics – viz. the ability to construe literal, quantifying and specifying uses – and the mise-en-discours also constrain or direct the morphosyntactic behavior of the construction. In other words, the QN-related co-selection patterns are not accidental, but conceptually and discourse-pragmatically motivated while also QN-related to a certain extent.204 The diachronic case-studies have shown that the GR of BQ-constructions did not result in the complete decategorialization or total loss of noun-features of N1. The formal persistence is also attested synchronically, in the form of variation in the determiner to N1, the premodifier to N1, the preposition, the right collocate N2 and verb agreement.205 In context, the examples (1)–(5) sound fine. (1)

Recibió un alud de críticas. // El alud de críticas que recibió. ‘He received a flood of criticism. // The flood of criticism he received.

(2)

Su discurso provocó un imparable alud de críticas. ‘His discourse evoked an unstoppable flood of criticism.’

(3)

Un alud de cartas y [Ø/de] llamadas invadió su despacho. ‘A flood of letters and (of) calls invaded his office.’

(4)

Un alud de nieve / cartas / protestas / innovación / tristeza / etc. ‘A flood of snow / letters / protests / innovation / sadness / etc.’

(5)

Un alud de recuerdos se le [cayósg/cayeronpl] encima. ‘A flood of memories fell upon him.’

|| 204 The reasoning might appear circular. While recurring collocates and contextual cues are taken to be additional evidence for CIP in Chapter 7, it is rather the semantic-pragmatic characterization of the context and CIP which are considered to motivate the combinatorial patterns of the construction in the present chapter. As I have argued previously (in Section 6.1.2), the iconicity principle explains the primary role of CIP. 205 See also De Smet (2014) who considers pseudopartitive constructions hybrid cases of reanalysis: “Hybrids manifest conflicting behavioural features, providing evidence that some surface forms instantiate different underlying structures at the same time.” (2014, 30).

Morphosyntactic description of the BQ in the literature | 401

The present chapter will argue that the noun-ness of the QN in quantifying and specifying uses is only partial – or in RAE’s terms, QNs present a defective syntax (2009, 826). Although the different constituents of the binominal construction continue to have a functional contribution to the BQ as a whole (cf. Section 2.1.2), the morphosyntactic variation is limited and relies heavily on semanticpragmatic motivations. The structure will be as follows. While Section 8.1 provides an overview of the few existing syntactic descriptions of the BQ-construction in the literature, Section 8.2 examines the paradigmatic variation as attested in the corpus study. The constructional slots of the BQ-construction are discussed from left to right. In a nutshell, the paradigmatic variation of these slots seems subject to the interaction of N1’s function, discourse-pragmatic motivations and CIP. Depending on the paradigm analyzed, some particular factors can play a more prominent role.

8.1 Morphosyntactic description of the BQ in the literature Since QNs are rarely considered a grammatical category in itself in the literature, detailed syntactic descriptions are scarce. The only exception – to my knowledge – is the dissertation of Gutiérrez Rodríguez (2008) on the grammatical aspects of weak quantifiers (cuantificadores débiles) in Spanish. The last chapter of her PhD is exclusively devoted to ‘complex weak quantifiers’, i.e. pseudopartitive constructions (2008, 315–387), in which she carefully analyses the syntactic behavior of QNs such as montón, mogollón, aluvión, ejército, etc. In reference grammars of Spanish, quantifiers are generally classified according to the notions they quantify (individuals (e.g. cualquier ciudadano ‘any citizen’), substance (e.g. demasiado trabajo ‘too much work’) or degree (e.g. bastante despacio ‘quite slow’)); to the grammatical class they belong to (adjectival (e.g. libro alguno ‘some book’), pronominal (e.g. cualquiera lo entendería ‘anyone would understand it’), adverbial (e.g. duerme demasiado ‘he sleeps to much’)) or to their semantic nature (cf. RAE 2009, 1381). Two semantic types of quantifiers are usually distinguished: ‘strong’ – also called ‘definite’ – such as ambas ‘both’, cada ‘every’, todo ‘all’, etc. as opposed to ‘weak’ quantifiers – also referred to as ‘indefinite’ – such as algo ‘some’, nada ‘nothing’, ninguno ‘no’, etc. A common criterion to separate strong from weak quantifiers is the former’s inappropriateness as complement of impersonal haber ((6a), as is also the case for definite, demonstrative and possessive determiners (6b)). Standard weak quantifiers, by contrast, easily combine with the existential predicate hay (6c).

402 | The motivated nature of the QN-related co-selection patterns

The same holds for pseudopartitive constructions, which make the QNs analyzed in this volume fit the category of ‘weak’ quantifiers. (6)

a. *Había {todos los/ambos/cada} niño(s) en el jardín. ‘There were {all the / both / every} child(ren) in the garden. (Gutiérrez Rodríguez 2008, 9) b. *Había {los/estos/sus} niño(s) en el jardín. ‘There were {the / those / his} children in the garden. c. Había {muchos/tres/algunos/más/Ø} niños en el jardín. ‘There were {many/three/some/more/Ø} in the garden. (Gutiérrez Rodríguez 2008, 9) d. Había un montón de niños en el jardín. ‘There were a heap of children in the garden.’ Hubo un aluvión de críticas de la oposición. ‘There has been a flood of criticism from the opposition.’ Hay una barbaridad de dinero en su cuenta. There is an awful lot (lit. barbarity) of money on his bank account.’

Gutiérrez Rodríguez (2008, 316) describes pseudopartitive constructions as a complex type of weak quantifiers. The first noun within the binominal syntagm is systematically considered a ‘quantifying noun’ (nombre cuantificativo).206 Building on Bosque’s (1999) classification (cf. Section 2.2.1), she also distinguishes bounding nouns (acotadores, e.g. una botella de agua ‘a bottle of water’), measure nouns (de medida, e.g. un litro de agua ‘a liter of water’) and group nouns (de grupo, e.g. un grupo de niños ‘a group of kids’). In addition, she adds a fourth type of lexicalized quantifying nouns (nombres cuantificativos lexicalizados) which she defines as nouns which either have completely lost their etymological meaning and only express quantity (subtype 1), or exclusively present a quantifying reading (subtype 2).207 The first subtype of lexicalized quantifying nouns thus encompasses nouns such as un montón ‘a heap’, la mar ‘the sea’, la tira ‘(lit.) the strip’ or un sinfín ‘lit. an endlessness’, whereas quanti-

|| 206 Recall that Bosque (1999) uses the notion of quantifying noun in a broad sense, thereby including bounding nouns, measure nouns, group nouns, quantity notions, whereas my understanding of QN is limited to nouns with scalar implicatures in addition to their lexical meaning (as a single noun). 207 Note that ‘lexicalized’ is to be interpreted as ‘part of the lexicon’ and not as the end result of a process opposed to GR. Instead, ‘lexicalized’ means that the quantifying interpretation of the noun in question is conventionalized and thus mentioned in a separate definition line in reference dictionaries. Further, Gutiérrez Rodríguez briefly mentions (2008, 385) that ‘lexicalized quantifying nouns’ are involved in a process of GR which has yet not completed.

Morphosyntactic description of the BQ in the literature | 403

ty expressions such as (gran) cantidad ‘(large) quantity’, gran número ‘large number’ or infinitude ‘infinity’, which have a quantity-related interpretation by definition, belong to the second subtype the author distinguishes (2008, 320– 321). Interestingly, in the description of the morphosyntactic characteristics of the quantifying nouns, she repeatedly observes that the behavior of group nouns – especially the metaphorically used group nouns – and lexicalized quantifying nouns diverges from the other classes of quantifying nouns. METAPHORIC GROUP nouns (such as ejército ‘army’) differ from genuine group nouns (such as grupo ‘group’) in that they typically present a quantifying reading when they are used metaphorically and, when doing so, extend the range of entities they can quantify (as in un rebaño de turistas ‘a herd of tourists’, un rosario de quejas ‘a rosary of complaints). While some can express quantity both when used literally and metaphorically (e.g. un rebaño de ovejas ‘a herd of sheep’ vs. un rebaño de turistas), many only quantify N2 when used figuratively (e.g. un alud de insultos ‘an avalanche of insults’). Further, true group nouns can be ambiguous between a ‘referential’ and a ‘quantifying’ reading, as in (7) where both readings of grupo are equally plausible: i) the referential reading interprets grupo as a collective noun and suggests that there has only been one speech; ii) the quantifying or distributive reading implies that there have been as many speeches as deputies. The metaphorically used ejército in un ejército de periodistas curiosos ‘an army of curious journalists’, however, cannot refer literally to a group of soldiers anymore and automatically yields a quantifying reading, i.e. a lot of journalists. LEXICALIZED QUANTIFYING NOUNS then slightly differ from metaphorically used group nouns in that (i) they often behave like genuine quantifiers, while group nouns remain more restricted by their nominal source construction (2008, 384–385) and (ii) in that they can combine with any N2 (whereas other group nouns usually impose lexical restrictions).208 Still, in spite of several common syntactic features, the author refrains from considering ‘lexicalized quantitative nouns’ as true weak quantifiers, since they cannot be used in partitive constructions – viz. with another determiner preceding N2 – as in (8) and, as a result of their uncompleted GR, still partially behave like nouns (2008, 354, 385). || 208 As will become clear in Section 8.2, I believe that the discrepancies in syntactic behavior Gutiérrez Rodríguez points at, can be explained (i) when considering both metaphorical group nouns and lexicalized quantifying nouns as a single grammatical category, which corresponds to my category of quantifying nouns – note that they are both involved in GR and they construe the quantity of N2 in the same way – and (ii) when accepting that pseudopartitive constructions express coextensiveness between N1 and N2.

404 | The motivated nature of the QN-related co-selection patterns

(7)

Solo un grupo de diputados de IU logró pronunciar su discurso. ‘Only one group of deputies of IU has been able to give their speech.’

(8)

*gran número de los niños (*big number of the children) *infinidad de las casas (*infinity of the houses) *la mar de los adultos (*the sea of adults) (Guitérrez Rodríguez 2008, 354)

In the next sections (8.1.1–8.1.10), I will briefly touch upon the recurrent topics in the few existing syntactic descriptions of BQs.

8.1.1 Collective or distributive reading Many studies differentiate between a collective (also called referential in the case of QNs) and distributive interpretation of collective nouns, group nouns or pseudopartitive constructions (among many others Bosque 1999; Brucart 1997; Gutiérrez Rodríguez 2008; Mihatsch 2000; RAE 2009; Vos 1999). As shown for (7), the collective interpretation implies that there is only one event (in spite of the plural nature of N1). In the distributive reading, the number of events corresponds to the quantifier (Gutiérrez Rodríguez 2008, 326). With respect to QNs, Gutiérrez Rodríguez states that group nouns are often ambiguous (2008, 325– 326), whereas lexicalized quantifying nouns always evoke quantity, with montón de as only exception (since montón can also be complemented by entities which actually compose the literal heap, as in El montón de arena del jardín ‘The heap of sand of the garden’). The distinction between the collective and the distributive reading of a quantity expression – or more carefully, the quantifier’s potential to evoke a collective, respectively a distributive interpretation – can be revealed by the predicates it is apt to combine with (Bosque 1999, 45; Gutiérrez Rodríguez 2008, 128; RAE 2009, 807). Verbal predicates which refer to collective arguments such as amontonarse ‘to heap up’, acumularse ‘to accumulate’ and reunir ‘to bring together’, adjectives such as numeroso ‘numeroso’ or the adverb juntos ‘together’ trigger a collective or referential reading of the quantifying noun, while verbal predicates which necessarily apply to individual entities (whether or not they constitute a larger group) invite for a distributive reading, such as hablar francés ‘to speak French’, ser inteligente ‘to be intelligent’, parecerse ‘to be alike’, adjectives such as alto ‘high’, the preposition entre ‘between’ and the adverbs separadamente ‘separately’ or uno por uno ‘one by one’. To ‘test’ whether the referential, respectively distributive, reading is activated, Gutiérrez

Morphosyntactic description of the BQ in the literature | 405

Rodríguez (2008) systematically verifies whether verb agreement with N1, respectively N2, is possible. As explained in Section 8.1, Gutiérrez Rodríguez claims that group nouns are generally ambiguous between a referential and a quantifying interpretation; that the subgroup of metaphoric group nouns is in most cases incompatible with a referential reading and that lexicalized quantifying nouns by definition evoke a quantifying reading. However, examples of metaphoric group nouns (9a) and lexicalized quantifying nouns (9b) in which N1 has full referential value can easily be found – note that these instances are systematically referred to as ‘literal’ or ‘head’ uses in the present volume. Note also that definite determiners such as the definite article, demonstratives or possessives tend to impose the referential reading (see also Brucart 1997): while in (10a), there might be as many bicycles as children, there is only one bicycle in (10b). (9)

a. un alud de nieve, un hatajo de ovejas ‘an avalanche of snow’, ‘a herd of sheep’ b. un montón de libros (en la mesa) ‘a heap of books (on the table)’

(10) a. Un grupo de niños trajo su bicicleta. (Gutiérrez Rodríguez 2008, 347) ‘A group of children brought their bike.’ b. El grupo de niños trajo su bicicleta. ‘The group of children brought their bike’. If BQs indeed primarily serve to single out a set of N2s while also adding a qualifying component – as hypothesized in Chapters 2 and 7 – they seem particularly apt to evoke a collective reading. Further, since a particular conceptual image is metaphorically projected on N2, it does not come as a surprise that the QNs analyzed in our corpus can combine with the predicates commonly said to evoke a collective reading only (11a–b). However, the examples in (12) show that BQs can also combine with a distributive reading. Note that in (12c), the combination with uno por uno or todos is excluded, not because aluvión de would be incompatible with the distributive reading, yet because it is technically impossible to individually overload an Emergency service. The fact that un racimo de problemas can be construed both with a collective (13a) and a distributive (13b) reading makes me argue that BQs are in fact flexible: since the binominal pattern explicitly mentions the set-level – in N1 – as well as the plurality – in N2 –, I argue that BQ-constructions can foreground both readings depending on the precise contextual setting or ‘mise-en-discours’. Gutiérrez Rodríguez also claims that, in (14a), montón can only have a quantifying interpretation in view of the N2 niños, which does not prevent this ‘lexicalized quantifying noun’ from evok-

406 | The motivated nature of the QN-related co-selection patterns

ing a collective reading however (2008, 347). She also states that montón is the only lexicalized quantifying noun apt to display a collective reading: with infinidad, the combination with juntos is excluded (14b). (11) a. Al final, se acumula(/n) un montón de problemas. ‘Eventually, a heap of problems is(/are) accumulating.’ b. Este múltiple racimo de actuaciones está(/n) llamado a compensar el probable declive. ‘That multiple bunch of proceedings is(/are) said to compensate the probable decline.’ (12) a. Entre el alud de críticas recibidas, hubo una particularmente ruinosa. ‘Between/Within the avalanche of critical remarks received, there was one particularly ruinous.’ b. He desgranado uno por uno la letanía de compañeros que le han precedido. ‘I have rattled off one by one the litany of colleagues that have preceded him.’ c. *Un aluvión de enfermos han sobrecargado uno por uno los servicios de Urgencias. ‘*A flood of ill people have overloaded one by one the Emergency services.’ (13) a. Un racimo de problemas afectan todos juntos al sector. ‘A bunch of problems all together affect the sector.’ b. Un racimo de problemas afectan uno por uno al sector. ‘A bunch of problems affect the sector one by one.’ (14) a. Un montón de niños escribieron juntos un poema a la profesora. ‘A heap of children wrote the professor together a poem.’ b. Una infinidad de niños trajeron un regalo a la profesora. ‘An infinity of children brought the professor a gift.’ (Gutiérrez Rodríguez 2008, 347)

8.1.2 Pluralization of N1 Bounding nouns and measure nouns are usually preceded by a quantifier and have a regular plural, as in tres tazas de café ‘two cups of coffee’, respectively dos kilos de patatas ‘two kilos of potatoes’ (Bosque 1999; Gutiérrez Rodríguez 2008). Group nouns, however, can also have a plural form, but are said to au-

Morphosyntactic description of the BQ in the literature | 407

tomatically activate their referential or collective reading when N1 is plural: in (15b), there are two different groups and two different speeches. Further, metaphoric group nouns (16) and the majority of the lexicalized quantifying nouns (17) can only be used in singular. Some can occur in plural while quantifying N2, yet in those cases, the plural morpheme does not convert N1 in a plural entity but simply reinforces the quantifying interpretation ((18), Vos 1999, 267; Gutiérrez Rodríguez 2008, 330).209 (15) a. Un grupo de diputados leyeron su discurso. ‘A group of deputies read their speech.’ b. Dos grupos de diputados leyeron su discurso. ‘Two groups of deputies read their speeches.’ (16) a. Un ejército de amigos {se acercó/acercaron} a saludarle. ‘An army of friends came closer to greet him.’ b. *Dos ejércitos de amigos se acercaron a saludarle. ‘*Two armies (…).’ (17) a. Había {mogollón/*mogollones} de niños. ‘There were a lot of children.’ b. Tiene {infinidad/*infinidades} de amigos. ‘He has an infinity of friends.’ (18) a. Había {un montón/montones} de gente. ‘There were a lot/lots of people.’ b. Comí {gran cantidad/grandes cantidades} de pasteles. ‘I have eaten a large quantity of cakes. At first sight, the corpus data seem to corroborate Vos’ and Gutiérrez Rodríguez’ intuitions on the possibility of a plural N1 and on the possible contribution of the plural morpheme. However, the data do not confirm the strict differentiation between metaphoric group nouns and lexicalized quantifying nouns. Section 7.2.4 has shown that plural forms of alud, aluvión, barbaridades, letanías, mares, pilas and racimos can also combine with a quantifying reading of N1 and that the contribution of the plural morpheme precisely resides in evoking ‘a plurality of sets of N2s’. In other words, the pluralization of N1 yields an intensification of the quantifying reading.

|| 209 Examples (15)–(18) are copied from Gutiérrez Rodríguez (2008, 329–330).

408 | The motivated nature of the QN-related co-selection patterns

8.1.3 Suffixation of N1 Bounding nouns as well as measure nouns – though to a lesser extent – can combine with diminutive suffixes which in those cases have an affective value and mean muy poco ‘very little’ (Gutiérrez Rodríguez 2008, 330–331). Vasito in (19a) can be used to reinforce the excuse and suggests that the speaker drank less than un vaso de whisky ‘a glass of whisky’; kilito in (19b) refers to barely a kilo. Group nouns also accept both the diminutive (20a) and augmentative suffix (20b), but only if they are not metaphoric. In general, lexicalized quantifying nouns are also incompatible with diminutive and/or augmentative suffixation, except for montón which can combine with the augmentative without giving up its quantifying interpretation (21a). The diminutive, however, automatically evokes a small heap, i.e. the literal reading (21b), which is of course incompatible with the N2 niños (Gutiérrez Rodríguez 2008, 331).210 (19) a. Me bebí un vasito de whisky. ‘I drank (only) a little glass of whisky.’ b. ?Ponme un kilito de patatas. ‘Give me barely a kilo of potatoes.’ (20) a. Había un grupito de alumnos en el patio. ‘There was a small group of students in the schoolyard.’ b. #Recibió un aluvionazo de disculpas. ‘*He received a huge flood of excuses.’ (21) a. Había un montonazo de niños. ‘There was a big heap of children.’ b. *Había un montoncito de niños. ‘*There was a small heap of children.’ In line with the import of the plural morpheme, the augmentative suffix in (21a) can be said to reinforce the quantifying reading. The fact that only the augmentative can be used in combination with montón dovetails of course with its intensifying potential, which is of course absent in the case of diminutives.

8.1.4 Determiner pattern of N1 and N2 The N1-determiner always agrees with N1. According to Gutiérrez Rodríguez (2008, 336–337, 352–354; see also Brucart 1997; Vos 1999), group nouns and lexicalized quantifying nouns are restricted to the indefinite determiner, since

|| 210 Examples (19)–(21) are taken from (Gutiérrez Rodríguez 2008, 331).

Morphosyntactic description of the BQ in the literature | 409

definite determiners or additional quantifiers automatically provide N1 with referential value and thus block the quantifying interpretation. While un grupo de niños in (22a) allows both for a collective and a distributive reading, there has only been one letter in (22b) yet three in (22c), independently of the total number of students. (22) a. Un grupo de alumnos {escribió/escribieron} una carta al director. ‘A group of students wrote a letter to the director.’ b. El grupo de alumnos escribió una carta al director. ‘The group (…).’ c. Tres grupos de alumnos escribieron una carta al director. ‘Three groups (…).’ Metaphoric group nouns and lexicalized quantifying nouns exclusively combine with the indefinite determiner. While (23a) refers to a large group of agonistic journalists, (23b) necessarily refers to a real army composed of journalists (which explains the marked character of the sentence; example taken from Gutiérrez Rodríguez 2008, 337). Gutiérrez Rodríguez also argues that lexicalized quantifying nouns such as mogollón and gran número do not accept any determiner variation and can only combine with un or no determiner at all, whereas pasada only occurs in combination with una and mar only if preceded by the definite la. 211 (23) a. Un ejército de periodistas llenó la sala. ‘An army of journalist filledsg the room.’ b. #Tres ejércitos de periodistas llenaron la sala. ‘Three armies of journalists filledpl the room.’ However, some specific morphosyntactic contexts seem to allow for determiner variation with group nouns or lexicalized quantifying nouns which express a quantifying reading. More precisely, the combination with the definite determiner is possible when BQs are followed by a specifying relative clause (Bosque 2012, p.c.; Rodríguez Espiñeira 2012, p.c.; RAE 2009, 890), as in (24b). Similarly, when the referent of the BQ is topicalized, a definite determiner has to be used, as in (25). Further, the functional uses of BQs also seem compatible with exclamative ¡qué!, which endorses the quantity and quality assessment by N1, as in (26).

|| 211 Note that my corpus data do not corroborate Gutiérrez Rodríguez’ (2008) intuition on mar. Although only la mar de, as a fixed, lexicalized expression, has come to develop adjective and adverb intensifier uses, both la mar de ‘thefeminine sea of’ and un/el/… mar de ‘a/the/…masculine sea of’ are used to assess the quantity of N2 in CREA.

410 | The motivated nature of the QN-related co-selection patterns

(24) a. Hay una/*la pila de gente en la puerta. ‘There is a/*the lot (lit. pile) of people at the door.’ b. Una/la pila de gente que hay en la puerta. ‘A lot (lit. pile) of people/The many people who are at the door.’ (25) Los servicios de emergencia reciben diariamente una cantidad aplastante de llamadas. El/*un aluvión de llamadas recibidas ayer desbordó las previsiones. ‘The Emergency services receive an overwhelming number of calls every day. The/*A flood of calls received yesterday exceeded all previsions.’ (26) Al ver tanta comida decía: ¡Qué montón de comida! ‘When seeing so much food he said: “What a heap of food!”.’ As to the N2-determiner, the absolute absence of a(n) (additional) determiner is the only grammatically correct option in quantifying uses of the BQ. When any determiner or numeral is added to N2, the binominal turns into a partitive construction, in which both N1 and N2 have referential value (Gutiérrez Rodríguez 2008, 337, 353). That is why, according to the author, the quantifying reading of N1, and hence plural verb agreement, are impossible in (27).212 The underlying motivation is of course that in BQs, the entire syntagm [det. + QN + de] as a whole grounds N2. Similarly, while quantifiers to N2 are possible in the literal uses – they occur only marginally in the data however (see also Mihatsch 2000, 52, on the homogeneous/heterogeneous character of N1) – as in (28a), the quantifying reading blocks any numeral to N2 (28b). (27) Un grupo de {diez/varios/algunos} diputados {leyósg/*leyeronpl} su discurso. ‘A group of {ten/various/some} deputies read their discourse. (Gutiérrez Rodríguez 2008, 338) (28) a. un tropel de cuatro cómicos ‘a mob of four comedians’ un racimo de 30 motores ‘a cluster of 30 motors’ una pila de 17 libros ‘a pile of 17 books’

|| 212 Note that the analysis proposed by Gutiérrez Rodríguez (2008) implies that the possessive su is coreferential with the subject. Without disambiguating contextual cues, it is however possible to imagine a context in which every deputy reads a fragment of a single, larger discourse, in other words, that all the fragments read by the different deputies together constitute one large discourse on a specific topic.

Morphosyntactic description of the BQ in the literature | 411

b. *un tropel de 27 recuerdos ‘*a mob of 27 memories’ *un montón de 120 niños ‘*a heap of 120 children’ *un hatajo de 23 egoístas ‘* bunch of 23 egoists’

8.1.5 Adjectival modification to N1 and N2 N2s are free to combine with any adjective, provided the adjective occurs within the N2-slot. The modification pattern within the syntagm [det + QN + de] is highly restricted however, at least in the quantifying uses of BQs. Gutiérrez Rodríguez (2008, 339–345) argues that the N1-slot is restricted to adjectives which either modify the quantity assessment by N1 (for instance configurational adjectives indicating the height such as grande ‘big’ or evaluative adjectives such as buen ‘good’) or which modify N2 (and thus turn N1 in some kind of ‘transparent’ noun). Further, the adjective necessarily precedes the QN, as in (29a), since the postposition of the adjective actualizes the referential or collective reading of grupo in (29b).213 (29) a. Un gran grupo de niños {ganósg/ganaronpl} un premio. ‘A big group of children won a price.’ b. Un grupo grande de niños {ganósg/*ganaronpl} un premio. ‘A big group of children won a price.’ If the adjective cannot be related to quantification, as is the case for interesante in (30), the contrast between preposition and postposition is even more pronounced. According to Gutiérrez Rodríguez (2008, 344), the postmodifying interesante in (30b) modifies the N1 grupo and thus automatically provides the N1 with referential value (thereby blocking the quantifying interpretation and plural verb). However, in (30c), interesante precedes N1 yet actually modifies N2: it are the deputies who are qualified as interesting. For an adjective which morphologically agrees with N1 to scope over N2, the adjective has to precede N1 within the binominal syntagm. (30) a. Un grupo de diputados {leyósg/leyeronpl} su discurso. ‘A group of deputies read their speech.’ b. Un grupo interesante de diputados {leyó/?leyeron} su discurso. ‘An interesting group of deputies read their speech.’ c. Un interesante grupo de diputados {leyó/leyeron} su discurso. || 213 Examples (29)–(30) are taken from Gutiérrez Rodríguez (2008, 343), yet the acceptability judgement has been slightly adapted.

412 | The motivated nature of the QN-related co-selection patterns

Finally, Gutiérrez Rodríguez notes that, since lexicalized quantifying nouns are unambiguous, i.e. they do no admit a referential reading within the binominal construction, there are even less combinatorial possibilities as to the N1adjective. She also relegates the adjectival pattern of cantidades (ingentes) ‘immense quantities’, (gran) cantidad ‘(huge) amount’ and (gran) número ‘(huge) number’ to lexicalized preferences and states that la mar and infinidad are incompatible with any adjectival modification.

8.1.6 (Lexical) restriction on N2 Quantifying nouns generally impose lexical restrictions on the (type of) N2 they combine with in function of their own semantics (Bosque 1999; RAE 2009; Gutiérrez Rodríguez 2008). Minimally, N2 has to refer to some plurality, which can be expressed either by a plural (e.g. personas ‘persons’) or mass noun (e.g. dinero ‘money’). As to the category of group nouns, RAE (2009, 832) suggests to distinguish between nouns which do not impose lexical restrictions on N2 (such as cantidad ‘quantity’, conjunto ‘group’, grupo ‘group’, etc.) and nouns which do require a specific type of N2 (such as enjambre ‘swarm’ referring to a group of bees). Gutiérrez Rodríguez (2008, 326–328) hypothesizes distinct restrictions per type of quantifying noun: (1) the prototypical grupo is said to be the less restricted group noun and to freely combine with any N2 referring to persons or animals; (2) metaphoric group nouns extend the combinatorial pattern of their referential use, but continue to impose restrictions in their quantifying use (rebaño ‘herd’ extends its range of N2 from sheep to human beings as in un rebaño de turistas ‘a herd of tourists’); (3) lexicalized quantifying nouns can practically combine with any N2 since they are intrinsically quantifying notions, to the extent that their function is similar to the absolute quantifier mucho/a(s) ‘many’, which establishes a purely functional relation between the quantifier and the quantified entities. Finally, group nouns are claimed to be relational (Bosque 1999; Gutiérrez Rodríguez 2008) in quantifying uses, which means that the prepositional complement is necessary. Without prepositional phrase indicating the constituting entities, QNs such as ejército in (31) automatically evoke a referential reading. Unless the N2s have been previously mentioned in the context, el ejército cannot refer to a lot of curious journalists for instance, yet necessarily evokes a group of soldiers. In other words, ‘bare’ or uncomplemented QNs can sound fine in anaphorical contexts as in (32a) (cf. Bosque 2012, p.c.) but are only marginally used as such: (32b) shows that the semantically more general montón is general-

Morphosyntactic description of the BQ in the literature | 413

ly appropriate, while QNs with a rich source semantics such as aluvión require a contextual setting evoking their source frame. (31) El ejército se acercaba por el camino. ‘The army approached via the road’. (32) a. ¿Cuántas cervezas bebiste? – Un montón. How many beers did you drink? – A lot.’ b. No consiguieron contestar a la cantidad aplastante de llamadas. Recibieron {?un aluvión/un montón/montones/*una pila}. ‘They couldn’t answer the overwhelming amount of calls. They received {?a flood/a lot/lots /*a pile}.

8.1.7 Verb agreement As has already been indicated in Chapter 2, the chapters on verb agreement are often the only sections in Spanish grammars where reference is made to the BQconstruction. In contrast to collective nouns, (pseudo-)partitive constructions are considered to trigger plural verb agreement by definition. The addition of the prepositional phrase makes the ‘plurality’ more visible. However, the examples given are almost always restricted to the literal uses of QNs. Further, grammars of Spanish largely focus on factors which attract plural verb agreement of singular collective nouns, viz. the well-known concordantia ad sensum (Section 8.1.7.1), while the criteria determining verb agreement of pseudopartitive constructions are seldom taken into account. Section 8.1.7.2 describes the tendencies detectedd by four extensive studies on agreement, which are – to my knowledge – the only ones which consider verb agreement of binominal constructions worth analyzing.

8.1.7.1 Traditional account of the verb agreement of collective nouns Basically, two types of criteria are considered to attract plural agreement with N2. From a morphosyntactic point of view, the distance between the collective and the predicate might impact the number of the verb, in addition to the position of the verb with respect to the collective and the specification of the constituting entities in a prepositional phrase or parenthetic clause. From a semantic point of view, the indeterminacy of the collective noun and its conceptualization as a plurality (in contrast to a single, homogeneous unit) can influence verb agreement. In the Spanish literature, a shift in focus can be observed from for-

414 | The motivated nature of the QN-related co-selection patterns

mal criteria to semantic factors and, eventually, to the interaction between both types of criteria. First, the underlying idea of the parameter of distance is that the more words are inserted in between the verb and the subject, the more chance the verb stands to agree in plural (Gili Gaya 1993). Martínez (1999, 2767) takes this reasoning one step further and argues that the separation of the subject and the verb (in that order!) by parenthetic clauses is conditional for the verb to be plural. According to Bello (1988) and Salvá (1988), the fact that the verb does not immediately follow the collective is one of the conditions sine qua non for a plural verb (cf. (33)). Bosque (1999, 37) states more carefully that plural agreement with N2 should not be considered erroneous when the verb and the subject figure in different clauses (as in (34)). (33) *La gente huyeron. [La gentesg huyeronpl] ‘The people flew.’ (34) La pareja de enamorados montará su hogar en Roma, ciudad en la que viven. [La parejasg de enamoradospl montarásg … vivenpl] ‘The couple of lovers will set up home in Rome, city in which they live.’ Second, the post-position of the verb with respect to the collective noun also invites to use a plural verb, since if “the verb follows the subject, the plurality is so visible and near that it is very unusual that the verb is singular” (Gili Gaya 1993, 55, translation KV; see also Bosque/Gutiérrez-Rexach 2009, 541). Likewise, Martínez (1999, 2768) considers a plural verb which precedes the subject ungrammatical (as in (35)). (35) Medio país, unos asustados y otros excitados por el terremoto, se echaron a la calle.214 *Se echaron a la calle, unos asustados y otros excitados por el terremoto, medio país. [medio paíssg … se echaronpl] ‘Half the country, some frightened and others excited by the earthquake, got out on the street.’ || 214 Note that several of the examples listed in the literature are rather infelicitous for not clearly separating the influence of different criteria. The plural of echaron can also be triggered by the the content of the parenthetic clause. The fact that the constituting entities of the collective noun país are explicitly referred to, might as well enhance their visibility – and thus the plurality of the subject.

Morphosyntactic description of the BQ in the literature | 415

The third and final morphosyntactic parameter relates to the specification of the constituting entities, which is commonly considered sufficient to attract a plural verb (Alarcos LLorach 1999; 1995; Bello 1988; Butt/Benjamin 1993; Alcina Franch/Blecua 2001; RAE 1973), as in (36). More importantly, the plural verb appears to be the unmarked variant “when the collective noun is specified by a plural complement that indicates the persons or things the whole consists of” (RAE 1973, 388, translation KV). (36) Considerable número de los indios murieron. (Bello 1988, 819 c) *Considerable número murieron. (Bello 1988, 819 c) [númerosg de los indiospl… murieronpl] ‘A considerable number of the Indians died.’ As to the semantic criteria, the parameter of indeterminacy (Bello 1988; Salvá 1988) concerns those collective nouns whose constituting entities are unknown. For instance, the entities forming un grupo ‘a group’ can be of any type (of soldiers, of ladies, of dogs…), while the nature of the entities of determinate collectives is known, such as un enjambre ‘a swarm’ (consisting of bees), un gobierno ‘a government’ (consisting of ministers). According to Bello (1988), only indeterminate collectives can combine with plural verbs (as in (37)). (37) Amotinóse la gente, pero a la primera descarga de la tropa huyeron despavoridos. [amotinósesg la gentesg… huyeronpl] ‘The people rose up, but at the first shot of the troop they flew terrified.’ Finally, the conceptualization of the collective can trigger a plural verb (or concordantia ad sensum) when (the plurality of) the individual constituting entities are highlighted instead of (the singularity of) the set as a whole (Gili Gaya 1993; RAE 1973). 215 For instance, while (38a) profiles several independent arrivals, (38b) profiles a group of demonstrants who all arrive en bloc.216

|| 215 Note that many authors (a.o. Gutiérrez Rodríguez 2008; RAE 2009), interpret la concordancia ad sensum ‘the agreement ad sensum’ in a less strict way as the subject’s possibility to combine with a plural verb. 216 Note that the reading in which all demonstrants arrive together at the same time is the most probable one, but nothing more than a global conceptualization which possibly generalizes over exceptions. In other words, it is not necessary that the demonstrants all arrive at the same time strictly speaking, but they are conceptualized in that way.

416 | The motivated nature of the QN-related co-selection patterns

(38) a. La mayoría de los manifestantes vinieron de Madrid. [la mayoríasg de los manifestantespl vinieronpl b. La mayoría de los manifestantes vino de Madrid. (Gili Gaya 1993) [la mayoríasg de los manifestantespl vinosg] ‘The majority of the demonstrants came from Madrid.’ The new reference grammar by RAE (2009, 818) adds that the context can contain adjectives (numeroso ‘numerous’, nutrido ‘large, abundant’, etc.), verbs (agruparse ‘to group’, reunirse ‘to gather’, etc.), prepositions (entre ‘among, between’) and adverbial complements (por unanimidad ‘unanimously’, de común acuerdo ‘by mutual agreement’, colectivamente ‘collectively’, etc. ) which enhance the plurality inherent in collective nouns.

8.1.7.2 Studies on verb agreement of (pseudo-)partitive constructions Four studies provide a detailed discussion on verb agreement of (pseudo-) partitive constructions: Fält (1972) concentrates on verb agreement in Spanish; Brucart (1997) investigates the concordantia ad sensum with partitive and pseudopartitive constructions; Chapters 33.8 and 33.9 of the most recent reference grammar, viz. Nueva Gramática de la lengua española (RAE 2009), focus on the difference between both types of binominal constructions; Gutiérrez Rodríguez (2008) analyses verb agreement of quantifying nouns. Fält (1972) is in general rather loyal to the Spanish tradition. In his view, six factors impact verb agreement with singular collectives: (1) the degree of organization of the collective, (2) contextual cues which enhance the unity or the plurality of the subject, (3) the conceptualization of the collective (as a plurality or as a unit), (4) the distance between the verb and the subject, (5) the position of the verb with respect to the subject and (6) the determiner to the collective. The underlying idea of the last criterion is that the definite article combines “preferably with singular agreement, and the indefinite article, or zero, with plural agreement” (Fält 1972, 100, translation KV), e.g. (39)–(40)). (39) Desdichadamente, en la edad escolar la cantidad de proteínas suele ser insuficiente. [la cantidadsg de proteínaspl suelesg] ‘Unfortunately, at school age, the amount of proteins is usually not sufficient.’ (40) La enfermedad se ha extendido por las regiones más densamente pobladas del país a las que han sido enviadas una gran cantidad de dosis de vacuna.

Morphosyntactic description of the BQ in the literature | 417

[hanpl una gran cantidadsg de dosispl] ‘The disease had spread to more densely populated regions of the country to which a large quantity of doses of vaccines have been sent.’ As to the (pseudo)partitive construction, Fält’s data analysis reveals the following tendencies. Basically, verb agreement with (pseudo)partitive constructions is not subject to the factors affecting agreement with collective nouns. He distinguishes between organized and unorganized collectives, thereby referring to the internal structure of the collective.217 According to Fält, organized collectives which are specified by a prepositional phrase, tend to attract plural agreement. However, in the description of (41), he takes the criterion of the collective’s conceptualization into account as well. While the group of soldiers in (41a) is conceptualized as a single block or a whole, the women in (41b) are endowed with proper individuality to a certain degree. The plural agreement in (41c) is attributed to the quantifying reading of the N1. (41) a. Un grupo de soldados norcoreanos cruzó anoche la zona desmilitarizada y tendió una emboscada a una patrulla del ejército sudcoreano. [un gruposg de soldadospl cruzósg] ‘A group of North Korean soldiers crossed the demilitarized zone last night and laid an ambush to a patrol of the South Korean army.’ b. Un grupo de señoras rodeaban a su amiga, ayudándola a secarse. [un gruposg de señoraspl rodeabanpl] ‘A group of women surrounded their friend, helping her to dry.’ c. Durante estos últimos seis meses han salido, en España, Francia, Italia y Alemania, un montón de libros dedicados a alabarla o descuartizarla. [han salidopl un montónsg de librospl] ‘During these last six months a heap of books devoted to praise and to ruin her have been published in Spain, France, Italy and Germany.’ Among the various types of unorganized collectives Fält distinguishes, only two types preferably combine with singular verb agreement in binominal constructions: the collectives denoting ‘succession’ (when preceded by the indefinite

|| 217 Fält’s parameter of ‘organization’ should not be equated with Bello’s criterion of ‘determinacy’. While the former refers to the internal constitution of the mass, the latter regards the known or unknown character of the components.

418 | The motivated nature of the QN-related co-selection patterns

article, e.g. (42)) and the collectives such as avalancha ‘avalanche’ and the like, which indicate nature phenomena (Fält 1973, 149), as in (43). (42) Se junta a mi alrededor una serie de tipos lamentables, verdaderos enemigos de la sociedad. [se juntasg una seriesg de tipospl] ‘A series of pitiful guys gathered around me, real enemies of society.’ (43) ¡Qué torrente de recuerdos, atropellándose unos a otros, le mantenía en desasosegada vigilia! [qué torrentesg de ideaspl manteníasg] ‘What a stream of memories, pushing and shoving, held him in restless vigil!’ Brucart (1997) approaches verb agreement of binominal syntagms from a generative perspective and relegates the factors determining verb agreement to the semantic component of grammar. According to him, two requirements need to be fulfilled in order for the verb to be plural: (1) a (pseudo-)partitive construction where N2 modifies N1 by specifying the domain of entities N1 is a subpart of, and (2), the distributive reading of the verb, which means that the number of events equals the number of subject entities. For instance, since in (43), the number of victories achieved corresponds to the number of subject entities, plural verb agreement is correct as well.218 (44) La mayoría consiguiósg/consiguieronpl el triunfo. ‘The majority obtained the victory.’ In the section on verb-subject agreement, the new reference grammar by RAE (2009) zooms in on the difference between partitive and pseudopartitive constructions. Both are binominal constructions, but they differ as to interpretation and as to the degree of determinacy of the second noun. In partitive constructions, a subpart of a larger whole is singled out or selected (RAE 2009, 1442) and the second noun is preceded by a definite determiner (e.g. un grupo de los profesores ‘a group of the professors’). Pseudopartitives, in contrast, “provide the number corresponding to a calculation or the size corresponding to magnitude” ((RAE 2009, 1448, translation KV); e.g. un grupo de profesores ‘a group of professors’). Generally, while partitive constructions can always equally combine with singular and plural agreement – i.e. display doble concordancia ‘double agreement’–, the agreement pattern of pseudopartitive constructions is || 218 See also Bosque/Gutiérrez-Rexach (2009, 537–548) for a finegrained description of ‘distributive’ and ‘collective’ readings.

Morphosyntactic description of the BQ in the literature | 419

determined by the difference between collective and quantifying nouns, i.e. by the grammatical category of N1. RAE (2009, 1446) mentions some tendencies related to the type of collective. Although singular agreement always remains possible, double concordance is observed with fraction nouns (e.g. mitad ‘half’, tercio ‘third’) and “other nouns with an intrinsically quantifying interpretation” (RAE 2009, 2578, translation KV), such as cantidad ‘quantity’, mayoría ‘majority’, número ‘number’, parte ‘part’, resto ‘rest’, etc. However, nouns that denote groups or sets such as grupo de ‘group of’ tend to combine with a singular verb, which agrees with N1 representing the subpart that is extracted from a larger group (RAE 2009, 2577). With respect to pseudopartitive constructions, double agreement can be explained in the light of the quantitative interpretation and as “a syntactic consequence of the classification of the nouns” (RAE 2009, 2574, translation KV). More precisely, if N1 is interpreted as a collective, the verb is singular. Recall that collectives refer to a homogeneous group of individuals, which are usually left implicit (e.g. bosque ‘forest’, jurado ‘jury’, familia ‘family’, etc.). However, if N1 is interpreted as a quantifying noun (“cuantificador de grupo” ‘(lit.) group quantifier”), the verb is plural. The notion of quantifying nouns refers to a plurality of entities which can be of any kind (and which, for discourse-pragmatic reasons, are usually specified). RAE then adds that some nouns show a “strong resistance to be interpreted as collective nouns” (2009, 2575, translation KV), and by extension, to combine with singular verbs, such as infinidad ‘infinity’ and multitud ‘multitude’. Where the new Spanish reference grammar of the RAE (2009) remains rather vague as to which quantifying nouns “have an intrinsically quantifying interpretation” (2009, 2578) or apparently resist being “interpreted as collective nouns” (2009, 2575), Gutiérrez Rodríguez (2008, 344–348) starts from the established classification of quantifying nouns (Bosque 1999), to which she adds the category of lexicalized quantifying nouns (cf. infra, Section 8.1). The first two classes of quantifying nouns, viz. bounding and measure nouns, exclusively combine with singular verb agreement (45). With group nouns (46a) and metaphoric group nouns (46b), plural verb agreement is usually possible and in the case of lexicalized quantifying nouns, plural agreement might be the only grammatically correct option (46c).219 (45) a. *Una cestasg de manzanaspl se cayeronpl. ‘*A basketsg of applespl fellpl down.’

|| 219 Examples (45)–(47) are taken from Gutiérrez Rodríguez (2008, 344–348).

420 | The motivated nature of the QN-related co-selection patterns

b. *Un kilosg de peraspl se desparramaronpl por el suelo. ‘*A kilosg of pearspl spreadpl out over the floor.’ (46) a. Un grupo de niños {trajosg/trajeronpl} su bicicleta. ‘A group of children broughtsg/pl their bycicle.’ b. Un ejército de periodistas {invadiósg/invadieronpl} la sala. ‘An army of journalists invadedsg/pl the room.’ c. Un montón de niños {??trajosg/trajeronpl} su bicicleta. ‘A heap of children {??broughtsg/broughtpl} their bike.’ In line with the generative framework the author subscribes to, she disagrees with the association established in the literature between the possibility of ‘double agreement’ and the ambiguity between the referential (collective) and quantifying (distributive) reading of N1. The traditional explanation of the agreement ad sensum implies that in the quantifying uses, verb agreement is made with N2, which necessarily induces incorrect predictions according to Gutiérrez Rodríguez (2008, 346) for several reasons. First, the traditional account fails to explain why only group nouns can combine with a plural verb, instead of all nouns interpreted as a quantifier. Second, metaphoric group nouns such as alud ‘avalanche’ also give way to double agreement yet only present a quantifying interpretation. Third, many examples can be found where a collective reading combines with a plural verb (47a), respectively a quantitative reading with a singular verb (47b).220 That is why the author claims that agreement is always made with N1; even in cases of plural verb agreement, agreement is not made with N2 but with the intrinsic plural nature of N1. The presence of plural N2 helps of course triggering plural verb agreement (2008, 382–383) – which is not the case for *La gentesg vinieronpl ‘*The peoplesg camepl’. (47) a. Un montón de niños escribieron juntos un poema a la profesora. ‘A heap of children wrote together a poem to the professor.’ b. Un grupo de niños trajo su bicicleta. ‘A group of children brought their bicycle.’ However, Gutiérrez Rodríguez (2008, 348) concludes her analysis on verb agreement of quantifying nouns by arguing that in case of a collective reading, verb agreement is made with N1, and in case of quantifying reading, plural verb agreement is possible yet not necessary. In other words, she leaves the reader in the dark as to which factor(s) then determine(s) verb agreement in quantifying uses of N1.

|| 220 In (47b), both a collective and a distributive reading are actually possible.

Morphosyntactic description of the BQ in the literature | 421

8.1.8 Topicalization and extraposition of the prepositional phrase The topicalization and extraposition tests carefully carried out by Gutiérrez Rodríguez (2008, 356–363) show that the prepositional phrase is less autonomous in pseudopartitve than in partitive constructions, which somehow confirms my intuition of considering the syntagm [det. + N1 + de] as a single chunk. First, extraposition means that a constituent is moved from its canonical position to the right periphery of the clause. With partitive constructions, the PNP can only be moved rightwards if the binominal construction is indefinite and occurs after the verbal predicate (48a). Yet even in those circumstances, the extraposition of the PNP of pseudopartitive constructions is ungrammatical (48b).221 (48) a. ?De repente, apareció un grupo en la sala de los niños desaparecidos. ‘All of a sudden, a group appeared in the room of the disappeared children.’ b. *De repente, apareció un grupo en la sala de niños desaparecidos. ‘(…) of disappeared children.’ Second, topicalization or left dislocation means that the topic of the clause is situated at the left periphery of the clause. While the PNP of partitive constructions can easily function as a topic (49a) (see also Brucart 1997), pseudopartitive constructions do not allow topicalization of the PNP (49b–c), except with bounding, measure and (metaphoric) group nouns (preferably in object position). Lexicalized quantifying nouns, however, block any attempt of focalization, independently of the subject or object position of the BQ in the clause (50). (49) a. De los manifestantes, un grupo se dirigieron al gobierno civil. ‘Of the demonstrates, a group addressed the civil government.’ b. *De manifestantes, un centenar se dirigieron al gobierno civil. ‘Of demonstrates, a hundred addressed the civil government.’ c. ?De turistas, en el museo había un ejército. ‘?Of tourists, in the museum there was an army.’ (50) a. *De amigos, María tiene un montón. ‘Of friends, María has a heap.’ b. *De libros, una barbaridad se estropearon. ‘Of books, a barbarity got ruined.’

|| 221 Examples (47)–(49) are taken from Gutiérrez Rodríguez (2008, 256–363).

422 | The motivated nature of the QN-related co-selection patterns

An interesting question which remains unsolved until now is why some native speakers consider the uncomplemented use of QNs correct in specific anaphoric contexts (as in (32)), while they generally agree on the anomaly of contexts in which the PNP gets topicalized.

8.1.9 The scope of quantification by BQs The scope of quantification is the syntactic domain or part of the sentence which is affected by the quantity expression. In the case of BQs, the scope of the ‘quantifier’ by definition includes the immediate right context, viz. N2. In other words, the quantity assessment expressed by un N1 de always minimally holds for N2. In general, the remaining semantic information at the right side of the quantifying sequence falls within the scope of the quantifier (RAE 2009, 1495). The exact scope of a quantifier can be less straightforward in combination with another expression of variable scope – such as another quantifying expression, a marker of negation or interrogation, or modal or intentional verbs – since both scopes can come to interact. For instance, in (51a), two interpretations are possible (Bosque 2012, p.c.): i) the negation can be interpreted to scope over the quantifier, which means that, generalizing over the distinct individuals, the group of people who showed up was not numerous; ii) the quantifier mucha can be interpreted to scope over the negative element no, which implies that a lot of specific persons did not come. However, for some reason, un montón de gente in (51b) definitely takes scope over the negation and the only reading possible is that a lot of specific individuals did not show up. Interestingly, BQs do not per se evoke specific N2s: in (50c), without context, un montón de gente is ambiguous as to the specificity of N2. (51) a. No vino mucha gente. ‘There were not many people which came./ Many people did not come.’ b. No vino un montón de gente. ‘A lot of people didn’t come.’ c. Vino un montón de gente. ‘A lot of people came.’ Interestingly, RAE (2009, 3659) mentions that, just like the indefinite algo ‘{some/any}thing’, alguno ‘some, any’ and alguien ‘someone’, nominal syntagms preceded by the indefinite article un(a) can be interpreted both within and outside the scope of negation, although the former is more frequent. In (52a), the common interpretation is that the speaker didn’t tell the interlocutor anything at all, but it is also possible that the speaker had one particular topic in mind which he didn’t talk the interlocutor about the day before. However, if the indefinite una is replaced by the BQ un montón de as in (52b), the indefinite

Morphosyntactic description of the BQ in the literature | 423

article loses its potential ambiguity as suggested by RAE and the only possible interpretation is the one in which the quantifier takes scope over the negation, i.e. there are a lot of specific topics the speaker didn’t talk the interlocutor about. (52) a. Ayer no te dije una cosa. ‘Yesterday, I didn’t tell you {anything/ something}.’ b. Ayer no te dije un montón de cosas. ‘Yesterday, I didn’t tell you a lot of things.’ The matter becomes even more complicated when switching the BQ and the negation around within the sentence. Iconically, in view of the linear order, the interpretation in which the left variable scopes over the variable to the right is the most natural one. However, in (51)–(52), the negation is always mentioned before the quantifying expression. Still, with absolute quantifiers the interaction produces ambiguity, with QNs, on the other hand, the interaction results in the inverted order: the QN takes scope over the negation. In (53a), on the contrary, the quantifier un montón de precedes the negation and, against the iconicity principle, the outcome is ambiguity again between an interpretation in which the group of (unspecified) students that don’t answer is numerous and the interpretation in which a lot of specific individuals do not answer. The same analysis holds for the absolute quantifier muchos in (53b). (53) a. Un montón de alumnos no contestan. ‘A heap of students don’t answer.’ b. Muchos alumnos no contestan. ‘Many students don’t answer.’ The situation of QNs which present a rich source semantics – and by consequence, are less prone to desemantization – is equally intriguing. In (54a), the negation is mentioned before the quantifier un aluvión de and the negation necessarily scopes over the BQ: the sentence means that in Madrid (in contrast to other cities for instance), the change did not lead to an overwhelming number of purchase orders. In (54b), the BQ not only precedes but also takes scope over the negative adverb no: even a lot of calls (all together and unspecified since hypothetical) will not exceed the previsions. In other words, in both cases, in spite of possible interaction with the negation, the focus is on the collectionlevel of N1. The speaker does not think of specific purchase orders nor calls – they are merely hypothetical anyway – but emphasizes the vast amount of orders, respectively calls.

424 | The motivated nature of the QN-related co-selection patterns

(54) a. En Madrid, ese cambio no provocó un aluvión de órdenes de compra. ‘In Madrid, that change did not provoke a flood of purchase orders.’ b. Incluso un alud de llamadas no desbordará las previsiones esta vez. ‘Even an avalanche of calls will not exceed the previsions this time.’ The diverging behavior of QNs with a rich source semantics on the one hand and montón on the other might be related to the diverging tendencies as to primarily assessing quantity or quality. Chapter 7 has observed that QNs with a high metaphorical load are more prone to conceptual persistence, whereas in many cases, montón de is close to the absolute quantifier mucho/a(s) and only gives way to highly schematic persistence. In other words, quantification expressed by alud de or aluvión de is situated closer to the classification pole, viz. involves a higher degree of quality assessment, than BQs based on montón de. Bosque (2012, p.c.) suggests that the diverging behavior between BQs (51b) and typical quantifiers (51a) might be related to the fact that QNs tend to express extreme values on a scale, which might prevent them from falling within the scope of negation. In other words, extreme values leave less room for interaction. Another interesting domain of interaction is the combination of the BQ with another quantifier. In (55), prototypical montón de is combined with another quantifier in distinct contextual settings. In (55a), the BQ is mentioned first and the most common interpretation is that every single child of the numerous group received two novels, viz. that montón de scopes over dos. The reading in which there were only two novels for the entire group of children is not impossible yet highly marked. In (55b), montón de is again mentioned first and combines with a verbal predicate that imposes a distributive reading. As a consequence, the only possible reading is the one where every single child of the numerous group is bilingual. The only variable altered in (55b) is the verbal predicate which now imposes a collective reading. The example shows that the collective reading of montón de can be imposed by the context and that a quantifier following montón de can take scope over the latter. In (55c), dos is mentioned before montón de, and both readings are possible: either there is a specific group of children that has been divided over two schools, or two schools both receive a numerous group of students. The examples in (56) do not contain a QN referring to an extreme degree as is the case for montón de but rather a configurationally oriented QN. The example in (56a) is ambiguous, yet the most common interpretation is the iconically motivated one in which cada takes scope over racimo de, viz. in which every painting contains a distinct cluster of philo-

Morphosyntactic description of the BQ in the literature | 425

sophical problems. The reading in which the same group of philosophical issues is repeated in every painting is highly marked. If the order is inverted, as in (56b), the example is still ambiguous, but the most common interpretation is again the one in which there is a separate and individual cluster of problems for every painting. Finally, the examples in (57) present QNs with a rich source semantics. In line with the analysis of (54), in spite of possible interaction with another quantifier and the importance of linear order, the highly metaphoric QNs do not favour the distributive reading. In both sentences, the more common interpretation is the one in which all the criticisms are considered as a whole, and the BQ thus scopes over dos. (55) a. Un montón de niños recibieron dos novelas. ‘A lot of children received two books.’ b. Un montón de niños hablan dos lenguas. ‘A lot of children speak two languages.’ c. Dos escuelas acogen a un montón de niños. ‘Two schools receive a lot of children.’ (56) a. En cada cuadro hay un racimo de problemas filosóficos. ‘In every painting, there is a cluster of philosophical problems.’ b. Hay un racimo de problemas filosóficos en cada cuadro. ‘There is a cluster of philosophical problems in every painting.’ (57) a. Dos políticos reciben un aluvión de críticas. ‘Two politicians receive a flood of criticism.’ b. Un aluvión de críticas está dirigido a dos políticos. ‘A flood of criticism is directed towards two politicians.’ In conclusion, the scope of quantification by BQs is a complicated matter in which distinct factors seem to interact, a.o. the presence of other scope markers (e.g. no, a quantifier, etc.), the source semantics of the QN, the linear order and the precise discourse context. Oversimplifying, QNs which express an extreme degree such as montón de tend to scope over negation or additional quantification. QNs which tend to combine with a high conceptual image persistence usually block a distributive reading in contexts of negation or additional quantification.

8.1.10 Conclusion The previous sections have shown that the morphosyntactic behavior of BQs is described in the literature as not fully predictable. The variation is however not

426 | The motivated nature of the QN-related co-selection patterns

unlimited and often governed by lexical restrictions. A common way-out to explain the constructional intricacies is the distinction between a referential – or literal – and quantifying reading. A major disadvantage of focusing on the morphosyntactic possibilities of the construction is that, from a certain moment, the linguist has to turn to artificial, decontextualized examples. As necessary as such analyses may be for the understanding of the construction, this strategy forces us to neglect a crucial piece of information, viz. the precise mise-en-discours, which I consider to be the most crucial factor in the construction’s formal make-up. The focus of the next section is therefore diametrically opposed: starting from actually attested variation in the corpus data, I will claim that many of the fluctuations described as impossible or exceptional in the literature are perfectly motivated when taking the specific discourse context into account.

8.2 Morphosyntactic variation of BQs in the corpus data In the present section, the attested morphosyntactic characteristics of the construction are systematically analyzed in function of the use of N1 – respectively head, quantifier and specifier – and the precise discourse context of the occurrences. The paradigmatic variation within the constructional slots of the BQ will be claimed to be conceptually and discourse-pragmatically motivated.

8.2.1 Fluctuation in the N1-determiner pattern In line with the GR hypothesis, N1 serves as the head of the binominal construction in its literal uses, while N2 fulfills this function in the grammaticalized readings. Recall that in the functional uses, the determiner to N1, the QN and the preposition – as a single chunk – are considered to quantify, respectively modify, the N2. In view of the head-status of the second noun, determiner variation in the grammaticalized uses is unexpected. The more since the nominal elements are necessarily co-extensive, which means that N1 cannot have a proper discoursereferential value. Yet precisely the co-extensiveness between both nominal elements allows the N1-determiner to scope over the entire BQ-construction. Although the determiner agrees morphologically with N1, it semantically scopes over N2. By way of illustration, the three quantifying uses of letanía de in (57)–(58) combine with the indefinite una, the definite la and the possessive su. Interest-

Morphosyntactic variation of BQs in the corpus data | 427

ingly, the choice for the N1-determiner is discourse-pragmatically motivated, independent of the QN used. If the BQ were omitted, the sentence would stick to the same determiners: the plural indefinite zero or unas (note that unas combines several determiner interpretations) in (57) because of the rhematic value of frases, the plural definite las in (58) because of the thematic value (not some of the convinced colleagues, but all of them) and the possessive in (59) which identifies Arrabal as the person pronouncing the litany. (58) Un bombardeo informativo va imprimiendo a diario en nuestro subconsciente una letanía de frases, que, (…). (1996, press) ‘An informative bombardment is stamping every day a litany of sentences in our subconsciousness, which, (…)’. …va imprimiendo en nuestro subconsciente [Ø/ unas] frases que… (59) Le he desgranado la letanía de colegas que le han precedido: Eduardo Móner, Roberto García Calvo, (…). (1996, press) ‘I have rattled off the litany of colleagues that preceded him: Eduardo Móner, Roberto García Calvo, (…). ’ ... he desgranado las colegos que le han precedido: … (60) ¿Se trata de un hombre y una muñeca? ¿De un pelele y una mujer? – escribe Arrabal en su cadenciosa letanía de interrogantes – ... ‘Is it about a man or a doll? About a wimp or a woman? – Arrabal writes in his rhythmic litany of questions – …’ (1966, press) …escribe Arrabal en sus interrogantes – … With regard to the determiner variation in functional uses, four major tendencies can be observed (cf. Table 1, see also Delbecque/Verveckken 2014). First, generalizing over the different readings, the indefinite determiner is by far the preferred combination and is only marginally liable to alternate either with definite determiners, viz. the definite article (as in (61)), the demonstrative (as in (62)) and the possessive (as in (63)) or no determiner at all. A fine-grained qualitative analysis has shown that the topicalized nature of the BQ is preconditional to the use of definite determiners. In (61), the announcement of the release of cultural events allows the speaker to zoom in on el aluvión de actividades and to go on listing them. In (62), este alud de ofrecimientos echoes the previously introduced sucesivas ofertas de trofeos. In (63), relentlessly complaining gets subsumed as a personality trait in su letanía de agravios y rencores.

428 | The motivated nature of the QN-related co-selection patterns

El/la Definite

Aquel(la) demonstrative

Ese/esa Demonstrative

Este/esta Demonstrative

Su Possessive

¡Qué! Exclamative

Otro/otra Indefinite

No determiner

H

# %

138 0.65

59 0.28

2 0.01

4 0.02

1 0.005

1 0.005

0 0

1 0.005

6 0.03

212

Q

# %

459 0.75

102 0.17

4 0.01

2 0.003

7 0.01

4 0.01

1 0.001

2 0.003

29 0.05

610

S

# %

37 0.7 690

3 0.06 181

4 0.08 13

3 0.06 14

2 0.04 12

0 0 7

0 0 1

0 0 3

4 0.08 44

53

TOTAL

TOTAL

Un(a) Indefinte

Tab. 1: Distribution of determiner variation per reading

875

(61) Como si fuera señal convenida, con la apertura del curso universitario se levantó la veda de los actos culturales. Entre lo que se inauguró y lo que a punto está dentro de unos días, el aluvión de actividades nos llega por distintos frentes, con ímpetu renovado después de la tregua veraniega: (…). (1982, press) ‘As if it were an agreed sign, with the opening of the academic year the season of the cultural events got started. Between what was inaugurated and what was about to be within a few days, the flood of activities reaches us from various fronts, with renewed violence after the summer truce: (…)’ (62) No contenta con el éxito inicial, Editorial Office volvió a la carga y siguió bombardeando infructuosamente a Quintela con sucesivas ofertas de trofeos los años 1985, 1986 y 1987. (…). Abrumado por este alud de ofrecimientos, Quintela pasó al contraataque en clave de humor. (1995, press) ‘Not satisfied with the initial success, Editorial Office came back and continued in vain to bombard Quintela with successive trophy offers in 1985, 1986 and 1987. (...) Overwhelmed by this avalanche of offers, Quintela struck back humorously.’ (63) Ella lo recriminaba todavía, a la menor oportunidad, por lo que consideraba su cobardía para haberse lanzado a esa vorágine, su incapacidad para haber resuelto debidamente la situación. Esta tarde ha tenido que oírle de nuevo todas esas quejas, soportar su letanía de agravios y rencores, (…). (1977, press) ‘She still incriminated him, at the slightest opportunity, for what she considered his cowardice to

Morphosyntactic variation of BQs in the corpus data | 429

have flung himself in this whirl, for his inability to solve the situation properly. This evening he had to hear all those complaints again, to stand her litany of insults and hard feelings, (…).’ As mentioned previously for (58)–(60), the same determiner can continue to be used when leaving out the BQ. The preference for the indefinite determine un(a) is also discourse-pragmatically motivated, since it goes hand in hand with the typical rhematic status of the BQ-construction (cf. Mihatsch Fc., 15), to the extent that the indefinite determiner is the default option (in the grammaticalized uses) unless the BQ is topicalized. The second and third tendency concern the amount of variation. Table 1 illustrates that only three co-selections are productive: the indefinite determiner, the definite determiner and no determiner. Note that the productivity of ‘no determiner’ is mainly QN-related or limited to the quantifying uses of mogollón de. Further, Table 1 shows that that the head uses do not present more structural variation than the quantifying uses, which even display one combination that does not occur in the literal uses, viz. the exclamative ¡qué! ‘what a!’; this form marks the presence of some illocutionary force (surprise, enthusiasm, confusion, perplexity, indignation etc.) and maximizes both the quantity and the quality of the (type of) properties of the nominal entity it modifies. In this context, the QN may seem pleonastic, since the construction without QN is also registered as such (cf. Alonso-Cortés 1999, 3998). However, whereas e.g. (64) makes a huge number of impressive memories appear as a disparate mass, the QN in (65) additionally conveys the idea that to the speaker they form an aggregate, comparable to an unordered stack, albeit in a vague and abstract way, given that montón ‘heap, pile’ is the most grammaticalized of the QNs. Yet, the untidiness and immobility which characterize its source semantics nicely fit in with the remote and untouched image of the N2 entities (‘memories’ from ‘thirty, thirty-five years’ ago). (64) – Veros vestidas así… ¡Qué montón de recuerdos! ¿Cuánto tiempo hace? ¿Treinta? ¿Treinta y cinco años? (1993, theatre) ‘– To see you dressed like this… What a heap of memories! How long ago is it? Thirty? Thirty five years?’ (65) ¡Qué de recuerdos! ‘How many (lit. what of) memories!’ Finally, the three demonstratives aquel, ese and este, appear to be equally frequent and are mainly used in specifying readings, which is in line with discourse-pragmatic function the N1-determiner continues to fulfill. Recall that specifying uses primarily specify N2 and are therefore often rooted in the subjective vantage point of the speaker. Delbecque (2011) illustrates that, in addition to situational deictic marking, the choice of the demonstrative can be revealing

430 | The motivated nature of the QN-related co-selection patterns

for the stance of the speaker. Consequently, in (66), e.g. ese is not simply a situational deictic marker suggesting that the speaker is pointing to a bunch of crooks, it primarily signals the speaker's endorsement of the categorization (cf. Delbecque 2011). In other words, even when N1 denotes concrete entities, there is more at stake than the spatial configuration of the QN. With est- ‘this’, speaker involvement is rather situated at the experiential level. This can be readily seen when N2 is abstract, i.e. not spatially anchored; e.g. in (67), the demonstrative indicates that the speaker is personally affected and that the nonconventional qualification of a large series of misfortunes as a litany is rooted in her own experiential domain. The demonstrative underlines that the equation she comes up with is telling for her mindset. (66) No pierdas tiempo en maldiciones. Enfílame a ese racimo de bandidos y pon tu pincho al servicio del rey. (1976, theatre) ‘Don’t waste time in curses. Get me that bunch of crooks and put your gun at the king’s service.’ (67) ¿Por qué no me matas, Fernando, y acabamos de una vez con esta letanía de angustias y de lágrimas? (1990, theatre) ‘Why don’t you kill me, Fernando, and then we end once and for all this litany of fears and tears?’ When zooming in on the determiner distribution per QN (see Table 2), slight QNrelated preferences can be observed. First, three QNs appear to combine particularly easily with the definite determiner: aluvión in 23 (or 38%), alud in 63 (or 51%) and letanía in 6 (or 35%) of its quantifying uses. Intuitively, I hypothesize that the productive topicalized uses of these QNs is rooted in their rich source semantics or lexical content which is harder to desemanticize. Second, while the determiner patterns of hatajo, montón, pila and racimo are in harmony with the general tendencies described per reading, the exclusive combination with the indefinite determiner of barbaridad and the strong preference for no determiner to mogollón strike the eye. The preference for una barbaridad de can be related to the limited token frequency and the exclusive exploitation of quantifying uses in the synchronic corpus. Further, I ascribe the preference for no determiner of mogollón to its high degree of GR in terms of context expansion and to the uncertainty regarding its source semantics. In line with the subtle difference between la mar de and un mar de (cf. Chapter 5, where la is considered to be a marker of quantity), I hypothesize that the systematic absence of determiner in the case of mogollón signals the quantifying use and may become more pronounced in the future. Why some QNs prefer the definite determiner (e.g. la mar, la tira) and others preferably do not combine with a determiner to N1 is an interesting question, but one which falls outside the scope of this analysis.

Morphosyntactic variation of BQs in the corpus data | 431

TOTAL

No determiner

Racimo

Otro/otra Indefinite

Pila

¡Qué! Exclamative

Montón

Su Possessive

Mogollón

Este/esta Demonstrative

Letanía

Ese/esa Demonstrative

Barbaridad Hatajo

Aquel(la) demonstrative

Aluvión

H Q S V H Q S A V Q H Q S V H Q S A V Q AI DM S V H Q S A V H Q S A V H Q S

El/la Definite

Alud

Un(a) Indefinte

Tab. 2: Distribution of determiner variation per QN per reading222

15 35 1 4 3 56 3 0 1 9 0 5 4 2 1 9 1 0 1 18 0 1 1 0 74 302 25 27 11 32 14 1 7 2 13 11 1 690

1 23 1 1 2 63 1 1 4 0 0 0 0 0 2 6 0 0 0 0 0 0 0 0 35 5 1 3 2 14 2 0 2 4 5 3 0 181

0 1 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 1 2 2 0 0 0 0 0 0 1 1 0 13

0 0 0 0 0 1 1 0 3 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 3 0 1 1 0 1 0 0 0 0 0 1 1 14

0 3 0 0 0 2 1 0 2 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 0 12

0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 2 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 7

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 1 0 3

0 0 0 0 0 2 1 0 1 0 1 0 3 0 0 0 0 0 0 25 3 0 0 1 0 2 0 0 0 0 0 0 0 0 5 0 0 44

|| 222 The abbreviations AI and DM refer to the adjective intensifier and degree modifier uses of mogollón de.

432 | The motivated nature of the QN-related co-selection patterns

In sum, although determiner variation is a typical nominal feature, the fluctuation in this pattern is semantically and discourse-pragmatically motivated. In other words, the formal persistence is not considered to counteract the grammaticalization process. On the contrary, given that the determiner pattern is useful and rooted in the discourse integration of the BQ, the alternation is not likely to be given up by native speakers in the near future.

8.2.2 Fluctuation in the premodification pattern The presence of adjectives modifying N1, which is a typical nominal feature, also seems to counteract the GR process and testifies to a certain degree of formal persistence. Again, the ‘noun-ness’ is only partial, since the paradigm of premodifying adjectives is certainly not an open-ended nor a productive one, yet constrained by structural and conceptual selection restrictions. In (68), for instance, it is perfectly fine to add the adjectives verdadero and imparable, which intensify the quantifying reading. N1 does not admit, however, relational adjectives (e.g. local ‘local’) nor descriptive adjectives (e.g. acuoso ‘watery’). (68) Recibí un [verdadero/imparable] alud de llamadas. ‘ I received a [real/unstoppable] avalanche of calls.’ *Recibí un local alud de llamadas. *Recibí un acuoso alud de llamadas. In addition, the intensifying adjectives are restricted to the prenominal slot: while it is perfectly fine to say un verdadero alud de llamadas, postmodifying adjectives as in *un alud verdadero de llamadas are not observed in CREA.223 This tendency is functionally and conceptually motivated: Delbecque (1990, 374–376) shows that by preposing the adjective, the property attributed to the noun is presented as unquestionable and presuppositional knowledge, whereas the postmodifying adjective invites to compare the modified noun to other instances of its type for the property expressed by the adjective.224 Consequently,

|| 223 The only postmodifying adjectives observed are past participle forms with complements, which necessarily occupy the postnominal slot (Delbecque 1990, 384). E.g. Para entonces el campamento ya sólo era un montón de escombros, rodeado de enemigos por todas partes. ‘By that time, the camp was already nothing more than a heap of ruins, surrounded by enemies everywhere’. 224 This interpretation is in line with the iconicity principle, since “the processing of the given property is completed when “arriving” at the noun”, and as such, “the idea of any external comparison is overridden.” (Delbecque 1990, 272).

Morphosyntactic variation of BQs in the corpus data | 433

the reverse order in ?un alud imparable de llamadas is highly marked and would to some extent imply an external comparison with other avalanches (e.g. (69)). In line with Delbecque’s analysis, I argue that the preposed adjective is the unmarked and only option in grammaticalized uses, since, strictly speaking, N1 has no proper referential value and external comparison is excluded. Further, since only intensifiers or adjectives profiling a facet of the QN’s conceptual image are observed, the few premodifying adjectives observed in the corpus merely intensify the QN’s image schematic structure by presenting the adjective as the essence of N1. (69) Recibí un inmenso alud de llamadas. (‘immense’) ?Recibí un alud inmenso de llamadas. >< Un alud inmenso (de nieve) aplastó los alpinistas. ‘An immense snow slide crushed the mountain climbers.’ As to the conceptual restrictions on the adjective co-selection pattern, they are partially QN-related. Table 3, which presents the adjective variation per QN per reading, shows that alud, letanía and especially aluvión, have an extensive adjective paradigm in their quantifying use. The productive adjective co-selection pattern in quantifying uses thus corresponds to the QNs with a particularly rich source semantics. Although montón also presents 6 adjective-combinations (and 11 occ.), the proportion of premodifying adjectives in quantifying uses is rather small since montón de is a high-frequency item. Further, in the case of montón and mogollón, the limited number of adjectives can even be considered an additional signal of an advanced stage of GR. Tab. 3: Variation in the premodification pattern per QN per reading Adjective-combinations in CREA

QN Alud

Aluvión

Barbaridad Hatajo

H Q

S H Q

S Q H Q S

catastrófico ‘disastrous’ (1), provocado ‘provoked, elicited’ (1) verdadero ‘real’ (2), encauzado ‘channelled’ (1), imparable ‘unstoppable’ (1), incontenible ‘uncontainable’ (1), producido ‘produced’ (1), riquísimo ‘very rich’ (1) ----verdadero ‘real’ (3), gran ‘big’ (2), auténtico ‘authentic’ (1), importante ‘important’ (1), inmenso ‘immense’ (1), nuevo ‘new’ (1), pleno ‘full’ (1), previsible ‘foreseeable’ (1), previsto ‘foreseen’ (1), pasado ‘past’ (1) semejante ‘similar’ (1) --medio ‘half’ -----

434 | The motivated nature of the QN-related co-selection patterns

Adjective-combinations in CREA

QN Letanía

Mogollón

Montón

Pila

Racimo

H Q

--antológica ‘anthological’ (1), cadenciosa ‘rythmical’ (1), dulce ‘sweet’ (1), extenso ‘lengthy, wide’ (1), innumerable ‘innumerable’ (1), larga ‘long’ (1), cansina ‘weary’ (1) S --Q buen ‘good’ (1) AI --DM --S --H enorme ‘enormous’ (2), determinado ‘certain’ (1), gran ‘big’ (1), hermoso ‘beautiful, lovely’ (1), pequeño ‘little’ (1), solo ‘only, single’ (1) Q buen ‘good’ (7), ingente ‘huge’ (1), semejante ‘similar’ (1), súbito e efímero ‘sudden and ephemeral’ (1), confuso ‘confused’ (1) S impreciso ‘imprecise’ (1), inmenso ‘immense’ (1), simple ‘simple’ (1) H enorme ‘enormous’ (3), abultada ‘huge’ (1), compuesta de ‘composed of’ (1), construida ‘construed’ (1), gran ‘big’ (1), grueso ‘big, thick’ (1) Q --S depositada ‘placed, put, deposited’ (1) H --Q buen ‘good’ (1), inmenso ‘immense’ (1), múltiple ‘multiple’ (1) S ---

Yet the constraints on the adjectival pattern are mainly linked to the reading of the QNs and the kind of possible qualifications is highly predictable. Strictly speaking, the literal or head uses are free to combine with any type of adjective (cf. Di Tullio/Kornfeld 2008, 2). However, except for montón and pila, the adjective paradigm in literal uses appears to be largely unproductive (cf. Table 3). Likewise, only few instances are observed where specifier uses have a premodifying adjective to N1. Quantifying uses seem to combine slightly more easily with adjectives, yet only with a restricted set, i.e. with intensifying adjectives, which either focus on the volume (e.g. gran ‘big’, inmenso ‘immense’, enorme ‘enormous’) or the authenticity (e.g. verdadero ‘real’, buen ‘good’) or profile a specific facet of the QN’s conceptual image which gets particularly emphasized (e.g. imparable ‘unstoppable’ in the case of alud). The latter adjective-combinations are of course QN-related: it is highly doubtful that other QNs than alud or aluvión combine with imparable, incontenible, previsible, etc., or other QNs than letanía with antológica, cadenciosa or cansina, etc. Only two exceptions to the above-mentioned tendencies are observed in my corpus. Both in (70) and in (71), a descriptive adjective is added to a quantifying N1. In (70), letanía is ironically used and dulce can be considered antonymous to the facets typically attributed to letanía’s conceptual image (e.g. weary, dreadful, etc.). The context shows that the dictionary described is indeed par-

Morphosyntactic variation of BQs in the corpus data | 435

ticularly extensive, yet the unusual length does not bother the speaker. In (71), the adjectives súbito y efímero actually scope over the entire binominal construction: the idea is that the disciples came simultaneously as quickly as they went, as if forming a block of disciples. Rather than further specifying the type of montón, the coordinated adjectives endow the QN with an aspectual reading. (70) Definido como "libro herramienta" por Camilo José Cela – al final de un largo y divertido prólogo en el que regala al lector tres breves cuentos con la tradicional factura de "la casa" –, el volumen, de casi 500 páginas, es algo más que un último recurso de consulta: es un exquisito pasatiempo, un entretenido matarratos [sic], un compañero del tendido, la grada o la andanada, sabio y viejo aficionado que nos desgranara la dulce letanía de lo olvidado o lo nunca bien aprendido. (1996, press) ‘Defined as “tool book” by C.J.C. – at the end of a large and funny prologue in which he offers the reader three short stories with the traditional “house” style –, the volume, of almost 500 pages, is a bit more than an ultimate consulting resource: it is an exquisite pastime, an entertaining raticide, a partner of the arena, the amphitheater or the covered stand, wise and old maniac that reels off the sweet litany of what has been forgotten or what has never been properly learned.’ (71) Por lo demás, parece ocioso indicar que, dada la agitada vida y, aún más, el conflictivo temperamento de Caravaggio, fue impensable que éste tuviera discípulos. Tuvo, no obstante, un súbito y efímero montón de seguidores, que, (…), cambiaron relativamente pronto de orientación. (1997, books) ‘Apart from that, it seems pointless to indicate that, given the hectic life and, what’s more, the controversial temperament of Caravaggio, it was unthinkable that the latter would have disciples. He has had, however, a sudden and ephemeral lot of followers, who (…) relatively quickly changed orientation.’

8.2.3 Fluctuation in the N2-pattern As the discussion at the end of Chapter 5 points out, the issue of the nature of N2-restrictions is a delicate one which cannot be reduced to the mere presence or absence of lexical restrictions as suggested in literature (Cf. Section 8.1.6). I will illustrate in the present section that the variation in the N2-pattern is conceptually constrained. Basically, the N2-collocates can vary along five dimen-

436 | The motivated nature of the QN-related co-selection patterns

sions: the morphological number of N2, the type of N2, N2’s concreteness or animacy, the lexical items in the N2-slot and, to a limited extent, the grammatical category of the N2-slot. As to the variation in morphological number, the grammaticalized uses can equally combine with singular or plural nouns, as in (72) and (73) respectively, provided that N2 refers to a multiplex referent, i.e. made up of different parts. Since multiplex referents can be expressed by singular and plural lexical items, this parameter is not considered relevant unless different types of multiplex referents are distinguished. (72) Sabemos que hay un montón de personas en Libia que piensan que su país estaría mejor sin Gadafi. (1986, press) ‘We know that there are a lot of persons in Libya who think that their country would be better off without Gaddafi.’ (73) Hija, hay un montón de gente en la puerta, (…). (1995, oral) ‘Dear child, there is a lot of people at the door, (…).’ The discussion on the typology of masses or multiplex referents generally evokes the two-dimensional continuum of boundedness/unboundedness (cf. the count vs mass distinction) and homogeneity/heterogeneity (cf. Bach 1986; Langacker 1991). I based myself on the distinction made by Langacker (1991, 78; 2004, 81) between singular count nouns, plural mass nouns and non-plural mass nouns. Singular count nouns, such as apple, designate one discrete (and bounded) entity. Mass nouns, however, profile a homogeneous mass, which can be spatially continuous (non-plural mass noun, e.g. water) or discontinuous and replicate, i.e. when the mass consists of many discrete “particles” that can be of any size and that are often discernible (plural mass noun, e.g. sand). Figure 1 presents Langacker’s visualization of the mass-count noun distinction. In addition, plural count nouns profile entities being bounded (e.g. apples). The difference between plural count nouns and plural mass nouns resides in the fact that, despite their particulate nature, the latter construe masses as being internally homogeneous.

Morphosyntactic variation of BQs in the corpus data | 437

Singular Count Noun

Plural Mass Noun

Non-Plural Mass Noun

Discrete Entity

Replicate Mass

Non-replicate Mass

Fig. 1: Langacker’s interpretation of the count-mass noun distinction (1991, 78)

For the sake of systematicity, I replaced the notion of non-plural mass noun by singular mass noun. Crucially, the Langackerian typology of count and mass nouns is conceptually based and does not necessarily correspond to the morphological number of the noun. For instance, gente ‘people’ is usually classified as a plural mass noun in my data. The difference between singular and plural mass nouns is a gradient one, however: while arena ‘sand’ is generally considered non-replicate, it can be considered replicate to a certain degree in the context of children sifting it. Size and social habits (in how approaching the mass) usually determine whether the replicate or non-replicate conceptualization is profiled (Wierzbicka 1985, 314). In addition, Wierzbicka (1985) illustrates that the apparently arbitrary grammatical behavior (as to number) of primarily countable or primarily uncountable nouns is sensitive to changes in conceptualizations, to the extent that “grammatical forms are semantically motivated, because, while they are not necessarily correlated with any ‘real world attributes’, they are correlated with different conceptualizations” (1985, 327). Since I expect BQs to quantify both plural count nouns and singular or plural mass nouns, the dimension of morphological number of N2 is irrelevant when determining collocation preferences of N1. By way of illustration, primarily countable nouns usually shift to mass noun interpretations in the BQ-construction (e.g. (74)). (74) Recalaron asimismo en el resguardado puerto de Fornells, (…). Asaron montones de sardineta para la tripulación, y guisaron además una gigantesca tortuga con alcachofas, habas, guisantes y patatas tiernas, (….). (1986, novel) ‘They also put in at the protect port of Fornells, (…). They roasted heaps of sardine for the crew, and they also cooked a giant turtle with artichokes, beans, peas and tender potatoes (…).’

438 | The motivated nature of the QN-related co-selection patterns

When zooming in on the variation in the type of N2 attested in my corpus, the corpus data corroborate the initial hypothesis according to which the BQ-construction combines (almost) exclusively with multiplex referents, viz. singular mass nouns, plural mass nouns or plural count nouns. Table 4 indicates that only 5 instances are found where N2 is a singular count noun: the only premodifier use followed by a singular count noun is given in (75). The four head-uses are identical and concern the string un racimo de uva in recipes.225 Further, specifier uses appear to combine slightly more easily with singular mass nouns than quantifying uses. Interestingly, in 29 of 47 quantifying instances, the singular mass noun is an abstract notion such as tiempo ‘time’, pensamiento ‘thought(s)’, ambición ‘ambition’, amargura ‘bitterness’, propaganda ‘propaganda’, etc. (in contrast to 2 of 12 specifying uses). Yet overall, the reading of QN does not seem to correspond to a specific type of N2 and all readings combine with count and mass nouns (see also Bosque 2007, 193; Di Tullio/ Kornfeld 2008, 2). These findings more strongly suggest that the combinatorial pattern of BQs is not restricted to multiplex referents but rather the other way round, that the BQ-construction imposes a multiplex reading on N2. (75) Es un mogollón de tren; está lleno de moros, huele mal... (1985, theatre) ‘It is a mess of a train; it’s full of Muslims, it smells bad… ’ Tab. 4: Variation in the N2-pattern according to the type of N2 (per reading)226

Sg c N Sg m N Pl m N Pl c N

H

Q

S

A

V

Total

4 33 27 148

0 47 47 516

1 12 5 35

0 1 2 42

0 3 1 37

5 96 82 778

Table 5 shows that the variation as to the type of N2 is not sensitive to the QN neither. In line with the overall QN-independent preference for plural count nouns in the right collocate position, the majority of the QNs preferably combine with plural count nouns. However, mogollón and barbaridad combine with

|| 225 Note that the classification of uva as singular count noun is questionable: if uva is meant to indicate the type specification of racimo, it is a singular count noun. Yet when racimo de uva is rather interpreted as a measure expression, uva is a singular mass noun. Anyway, in view of the small size of grapes and the human habit to eat more than one of them, measure expressions rather combine with the plural count noun uvas. 226 For the sake of simplicity, only the instances of singular QNs are taken into account.

Morphosyntactic variation of BQs in the corpus data | 439

singular mass nouns in more or less 30% of their uses. Yet I still refrain from considering the variation in the N2-paradigm as to the type of N2 as sensitive to the QN chosen.

Alud

Aluvión

Barbaridad

Hatajo

Letanía

Mogollóng

Montón

Pila

Racimo

Tab. 5: Variation in the N2-pattern according to the type of N2 (of the functional uses per QN)

Sg c N

Q S

0 0

0 0

0 0

0 0

0 0

0 1

0 0

0 0

0 0

Sg m N

Q S

3 1

3 3

3 0

0 0

1 0

12 0

22 8

0 0

3 0

Pl m N

Q S

0 0

2 1

3 0

1 0

0 0

7 0

31 4

2 0

1 0

Pl c N

Q S

59 2

119 4

3 0

5 7

16 2

25 0

258 17

14 1

17 2

The third dimension of variation attested in the N2-pattern relates to the degree of concreteness and animacy of the N2-entities. Three major QN-related tendencies can be drawn from Table 6, which can be conceptually motivated. First, generalizing over the different readings, the BQ-construction is primarily used to quantify concrete and inanimate entities. This tendency is most obvious with the head readings, which almost exclusively combine with concrete and inanimate entities, and is in harmony with the function the QN fulfills in literal uses, viz. specifying the configuration of a set of nearly-identical N2-entities. The few abstract entities which are observed are literal uses of letanía, where N2 concerns a type of discourse (e.g. cuestiones ‘questions’, predicciones ‘predictions’). The human N2-entities observed concern heaps or piles of dead corpses which for that reason are maybe more properly categorized as concrete inanimate entities. The second tendency concerns the preference of quantifying uses for abstract entities. Reversely, abstract N2s are almost only observed with quantifying readings and mainly concern types of feelings, thoughts, discourse and time indications. Third, the QNs selected in my database are rarely used to quantify animals. Note however that, originally, hatajo refers to a set of farm animals.

440 | The motivated nature of the QN-related co-selection patterns

Tab. 6: Variation in the N2-pattern according to the concreteness and animacy of N2 (per reading)

Concrete, [-animate] Concrete, [+animate] Concrete, [+ human] Abstract

H

Q

S

A

V

Total

202 1 6 3

198 5 115 292

34 1 9 9

43 0 2 0

17 1 8 15

494 8 140 319

Table 7 presents the N2-variation according to the concreteness and animacy of N2 per QN. It shows that hatajo is particularly useful for quantifying and specifying a set of human entities and letanía de for quantifying discourse entities. Further, the QN-related tendencies simply reflect the QN-independent preference for concrete inanimate and for abstract entities.

Alud

Aluvión

Barbaridad

Hatajo

Letanía

Mogollóng

Montón

Pila

Racimo

Tab. 7: Variation in the N2-pattern according to the concreteness and animacy of N2 (of the grammaticalized uses per QN)

Concrete [-animate]

Q S

28 3

36 4

4 0

3 0

6 0

17 1

91 24

2 1

11 1

Concrete, [+animate]

Q S

0 0

1 0

0 0

0 1

0 0

2 0

2 0

0 0

0 0

Concrete, [+ human]

Q S

5 0

19 0

0 0

3 5

1 0

12 0

65 3

6 0

4 1

Abstract

Q S

29 0

68 4

5 0

0 1

10 2

13 0

153 2

8 0

6 0

The fourth type of variation observed in the N2-pattern concerns the lexical variation in the N2-slot. No less than 531 different lexical items are observed in the N2-slot (of a total number of 961 occ.). The most frequent recurrent N2collocates are cosas ‘things’ (44 occ.), gente ‘people’ (31 occ.), años ‘years’ (29 occ.), críticas ‘criticism’ (18 occ.), libros ‘books’ (17 occ.) and dinero ‘money’ (17 occ.).227 It bears pointing out, however, that the choice for a particular lexical item is always discourse-pragmatically motivated: N2 functions as the head of || 227 Note that cosas, gente and años are the most frequent N2-combinations of high-frequency montón, yielding 39, 22 and 21 instances respectively. Likewise, críticas is primarily observed with aluvión (12 occ.) and alud (4 occ.).

Morphosyntactic variation of BQs in the corpus data | 441

the construction, and it is the kind of profile the speaker wants to convey of the set of N2-entities that guides the choice for a particular QN. In other words, although QN-related preferences – rather than restrictions (see infra) – can be observed, the choice for N2 is rooted in the discourse context. As I illustrated in the diachronic case-studies, per QN, the N2-combinations can be organized in clusters of semantically related nouns. The clusters themselves echo the QN’s source semantics. Further, N2s which fall outside the scope of the recurrent clusters are usually profiled by the context as an element of a specific cluster. For instance, letanía combines with parts or genres of speech (e.g. interrogantes ‘questions’, frases ‘sentences’, etc. ) or with kinds of discourse (e.g. dogmas ‘dogmas’, acusaciones ‘accusations’, consejos ‘advices’, etc.). In addition, letanía is observed with nouns which are contextually depicted as something somebody is saying (or writing), as in (76) and (77). (76) (…) con una poderosa fuerza lírica, sobre la que ya se pronunciaron los críticos de su época, esgrimiendo una letanía de influencias y antecedentes en la que figuraban los grandes nombres de la poesía inglesa, (…). (1977, press) ‘(…) with a strong lyric power, on which the critics of his time gave their opinion, enumerating a litany of influences and antecedents in which the great names of the English poetry appeared, (…).’ (77) (…) y la [KV: la mercancía] vende a velocidad vertiginosa en apenas media hora, en una letanía de nombres hermosos, precios y cantidades adjudicados exactamente a cada uno de los presentes, (…). (1997, press) ‘(…) and he sells it [KV: the merchandise] at vertiginous speed in hardly half an hour, in a litany of beautiful names, prices and quantities exactly allocated to each one of the persons present, (…).’ Since the synchronic N2-pattern continues the clusters observed per QN in the diachronic analysis (cf. Chapter 5) and the main combinatorial preferences have been indicated in the analysis of the conceptual image per QN (cf. Chapter 7), I will not comment on them extensively in the present section. Very briefly, alud and aluvión mainly quantify verbal reactions, human beings conceived of as immigrants, (sources of) information, politic/economic actions or products and experiences. All five clusters profile a specific subpart of the QN’s conceptual image. Pila exclusively combines with stackable objects in the literal uses and usually quantifies notions of time or human beings. Hatajo combines with animals only in its head uses, and almost exclusively quantifies or specifies (negatively evaluated) human beings. The clusters observed with barbaridad concern

442 | The motivated nature of the QN-related co-selection patterns

nouns referring to money or time. In its head use, racimo evokes bunches of fruit (particularly grapes) and flowers. Since its conceptual image is primarily configurational (and only to a limited extent contentful), the clustered structure of the N2s is less pronounced, except for two N2-clusters referring to houses and human beings of equal value. Yet all N2-combinations are compatible with racimo’s conceptual image. For mogollón, three N2-categories are observed: human beings (12 occ.), money (6 occ.) and feelings (3 occ.). As is the case for racimo, the remaining N2-combinations cannot be organized in clusters of conceptually similar nouns yet perfectly fit mogollón’s conceptual image. In addition to the clusters of human entities, objects made of paper, time indications, nouns referring to money and (sources of) information, the high-frequency item montón combines with a lot of conceptually unrelated nouns, which is in line with its advanced stage of GR.228  In sum, the variation in the N2-paradigm appears highly susceptible to conceptualization, and CIP in particular. The recurrent patterns are thus QN-related and also susceptible to the reading of N1. Since the number of N2s which do not directly relate to a specific N2-cluster but which are contextually construed as doing so increases over time, I choose the notion of QN-related ‘N2-preferences’ to ‘N2-restrictions’. Bosque (2007) comes to a similar conclusion in arguing that Algunas de las nuevas formaciones pasan a estar casi totalmente lexicalizadas, como mar de dudas o rosario de desdichas, pero otras muchas no lo están, y dan lugar a un gran número de grupos nominales, siempre que la noción abstracta que aporta el sustantivo cuantificativo pueda aplicarse. (2007, 193)229

Likewise, Di Tullio/Kornfeld (2008, 5) claim that the BQ-construction is the locus of a complex interaction between GR and lexicalization. The BQs engage in processes of GR in the sense that they come to express functional content. || 228 Di Tullio/Kornfeld (2008) distinguish between intrinsic QNs (such as cantidad ‘quantity’, grupo ‘group’ and montón ‘heap’) and QNs by metaphor/metonymy (such as tropel ‘mob’, montaña ‘mountain’, catarata ‘waterfall’). Recall that the analysis in Chapter 4 has shown that montón as well has acquired its quantifying reading via metonymic shifts in profile. Di Tullio/ Kornfeld (2008, 2) argue that the intrinsic QNs do not present combinatorial restrictions, while the latter group of QNs presents “mayores restricciones semánticas y categoriales” (‘major semantic and categorial restrictions’): e.g. un montón de amigos versus una ??montaña de amigos ‘friends’, Me divertí un montón ‘I had a lot of fun?’ versus ??Me divertí una montaña ‘I had a mountain of fun’. 229 Some of the new formations become almost completely lexicalized, as sea of doubts and rosary of misfortunes, but many others do not, and give way to a large number of nominal groups, provided that the abstract notion brought about by the quantifying noun is applicable, (Translation KV).

Morphosyntactic variation of BQs in the corpus data | 443

The BQs engage in processes of lexicalization in so far as they present strong lexical restrictions which are related to the QN. Although I do not consider the lexical combinatorial pattern as restrictions in the strict sense, I agree that the combinatorial preferences are QN-related and echo the QN’s source semantics. Finally, the N2-paradigm also varies as to the grammatical category of the element in the prepositional phrase. With la mar de and mogollón de, the N2element can be an adjective or adverb as well. Further, una barbaridad and (un) montón can occur without PNP and function as adverbs by themselves. I consider this paradigmatic variation as a sign of an advanced stage of GR, which will not be achieved easily by the other QNs analyzed which present a rich source semantics and are therefore less likely to desemanticize. In sum, the N2-variation is generally not restricted to a specific number or type of noun (in the end not even to a specific grammatical category) and it does not correspond to the distinction between head, quantifier or specifier readings. However, the observed tendencies are highly conceptually motivated and QNrelated.

8.2.4 Fluctuation as to verb agreement The factors influencing verb agreement deserve special mention for two reasons. First, verb agreement is a well-known test in determining the syntactic head of clauses and binominal syntagms in particular (cf. Aarts 1998; Keizer 2007; Mihatsch 2000; Vos 2002). Second, it is the combinatorial pattern where the impact of CIP is most obvious. Recall also that in the majority of Spanish reference grammars, the section on verb agreement is usually the only one which mentions the pseudopartitive construction (cf. Chapter 2).230 However, linguists disagree as to the essentially syntactic, semantic or lexical nature of agreement (cf. Acuña-Fariña 2009 for a detailed overview). The syntactic view is based on the assumption that agreement is “dysfunctional” and “superfluous” (Acuña-Fariña 2009, 391). When comparing the lack of additional morphological coding to the determiner and adjective in the English sentence in (78) to its morphologically rich Spanish translation in (79), one might conclude that the Spanish agreement system is indeed exotic and does not add new nor other information than the grammatical relationship between the elements of the noun phrases. However, the purely syntactic characterization of agreement cannot be maintained from a cognitive perspective, according to

|| 230 This section is adapted from Verveckken/Cornillie (2012).

444 | The motivated nature of the QN-related co-selection patterns

which grammar is inherently meaningful. Relating the preference for singular or plural agreement to the lexicon is also hardly justifiably in view of the immense cost it implies for the speaker’s memory. More recently, several linguists have come to suggest that syntactic and semantic factors impacting agreement overlap and compete (Acuña-Fariña 2009; Berg 1998; Eberhard et al. 2005). Spanish literature on collective nouns and verb agreement gradually shifts from a focus on morphosyntactic factors to a prominent role for semantics (cf. Section 8.1.7). The following case-study builds on the interaction between syntax and semantics, while slightly giving more prominence to the latter. (78) [We will keep] the little white candlesticks and the comfortable red chairs. (79) [Nos quedamos con] lo-s pequeño-s candelabro-s blanco-s y la-s cómoda-s silla-s roja-s. (Acuña-Fariña 2009, 391) Although several types of combinations exist, the case-study carried out in Verveckkken/Cornillie (2012) is primarily concerned with singular verb agreement with N1 (cf. the structure [[N1sg de N2pl] + Vsg], as in (80)) and with plural verb agreement with N2 (cf. the structure [[N1sg de N2pl] + Vpl], as in (81)).231 (80) Hay etapa en principio pero es que no ha llegado la mayoría de los equipos. (COREC, 1991, oral) [hasg llegado la mayoríasg de los equipospl] ‘There is a stage, in principle, but the thing is that the majority of the teams has not arrived.’ (81) Un aluvión de enfermos, la mayoría personas mayores con infecciones respiratorias de carácter vírico, han sobrecargado los servicios de Urgencias y han llenado los hospitales de Gipuzkoa. (2004, press) [un aluviónsg de enfermospl hanpl sobrecargado] ‘A flood of ill people, in the majority elderly people with respiratory infections of a viral character, have overloaded the Emergency services and have filled the hospitals of Gipuzkoa.’

|| 231 The case-study on verb agreement reveals once more the fuzzy boundaries between syntactic categories and interpretations. The distinction made between several types of agreement is purely formal in nature. It includes however a serious simplification: singular nouns can refer to plural entities as well (cf. the mass/count noun distinction). For not overcomplicating the matter, I will stick to the form-based typology of agreement.

Morphosyntactic variation of BQs in the corpus data | 445

These two types of agreement are particularly interesting since the verb explicitly agrees with N1 or N2, at least syntactically. This is not the case when both nominal elements agree in number. Therefore, the structures [[N1sg de N2sg] + Vsg] and [[N1pl de N2pl] + Vpl] will not retain my attention. Nor will I be concerned with the well-known concordantia ad sensum, where the verb does not agree with any of the nouns (e.g. [[N1sg de N2sg] + Vpl] and [[N1pl de N2pl] + Vsg], as in (82)). The cases of concordantia ad sensum are marginal (only 5 occ.). (82) Una base de datos, con cantidad de gente que se iban apuntando en... de acuerdo a las necesidades que tenían. (COREC, unknown, oral) [cantidadsg de gentesg se ibanpl] ‘A data base, with a lot of people that were registering in … in concordance with the needs they had.’

8.2.4.1 Methodological note The corpus used for the case-study on agreement partially differs from the one used in the other sections of this monograph. Since genre-related preferences and the difference between partitive and pseudopartitive constructions are traditionally considered to be two important factors in determining verb agreement, I added the corpus used by Van Doorslaer (2009): she analyzes verb agreement in oral language and her corpus mainly includes N1s that preferably occur in partitive constructions. Van Doorslaer (2009) extracts from the Corpus Oral de Referencia del Español Contemporáneo (COREC) all instances of, a.o., aluvión ‘flood’ (1), cantidad ‘quantity’ (13), conjunto ‘group, collection’ (8), cúmulo ‘set, series’(1), grupo ‘group’ (20), mayoría ‘majority’ (24), mitad ‘half’ (4), montón ‘heap’ (8), multitud ‘crowd’ (1), parte ‘part’ (36), pieza ‘piece’ (1), resto ‘rest’ (13), trozo ‘piece, bit, slice’ (7) where the binominal constructions functions as the subject. To this set, I added the extractions from CREA where alud ‘avalanche’ (24), aluvión ‘flood’ (32), barbaridad ‘barbarity’ (2), hatajo ‘herd, bunch’ (4), letanía ‘litany’ (7), montón ‘heap’ (7), pila ‘pile’ (15), racimo ‘racimo’ (6), tropel ‘mob’ (25) figure in the subject-position. In sum, the corpus of this case-study contains 259 examples (137 from COREC, 122 from CREA). Since verb agreement can concern both the main clause, a relative clause or both at the same time, the total number of cases of agreement amounts to 284 instances. At first sight, the corpus data seem to confirm the genre-related hypothesis of verb agreement which predicts that plural agreement with N2 is more com-

446 | The motivated nature of the QN-related co-selection patterns

mon in spoken language (cf. Bosque 1999, 38). Table 8 shows indeed that plural agreement with N2 presents 43% of the instances in the oral data, whereas the instances of plural agreement in written data only present 20%. Tab. 8: Agreement with N2 according to the genre Oral Vpl with N2 Other Total

# 67 89 156

Written % 0.43 0.57

# 25 103 128

% 0.20 0.80

The latter conclusion should be handled with care, however, since it generalizes over two limitations of the ad hoc corpus. First, while the subcorpus of Van Doorslaer consists mainly of partitive constructions, the extractions of CREA contain pseudopartitive constructions (or BQs only). Second, the subcorpus of Van Doorslaer contains oral data only, while the extractions of CREA include written data as well. The ad hoc corpus of this case-study thus not only runs the risk of comparing two different constructions, but also of invalidating the conclusion drawn from Table 8. However, three reasons justify the use of this ad hoc corpus: (1) both subcorpora are already at hand, (2) the difference between partitive and pseudopartitive constructions also figures among the traditional criteria verified in Section 8.2.4.3.4, (3) the subcorpus extracted from CREA, which technically speaking should allow to compare the genre-related proportions, only contains 3 (of 122) instances of oral language and undermines any intent of comparing both types of discourse within CREA.

8.2.4.2 Explanatroy scope of the existing accounts The structural difference between collective nouns and BQ’s, viz. the specification of the constituting entities by a prepositional phrase, is generally considered to enhance the ‘visibility’ of the plurality or the individual components, and consequently, to invite plural verb agreement with N2 (cf. the state of the art in Section 8.1.7). In contrast to plural agreement with singular collective nouns, which morphosyntactically clashes and is therefore attributed to the semantics of the construction, plural verb agreement with binominal constructions is not necessarily a purely semantic phenomenon, since BQ’s present both a single noun (N1) and a plural noun (N2). Unlike Fält (1972), who simply states that binominal constructions are not susceptible to the traditional formal criteria, I propose to determine the precise impact of the formal criteria. In view of the precise formal make-up of BQs, I hypothesize that formal criteria might

Morphosyntactic variation of BQs in the corpus data | 447

continue to play a role, though only in interaction with semantic and conceptual criteria. The following examples illustrate that each single instance of verb agreement is usually determined by a combination of several factors, which can have different weight according to the context in which they are used. The indeterminate (or unorganized) nature of montón and pila and the specification of the constituting entities in a prepositional phrase normally trigger plural verb agreement, yet the verb is singular in (83) and (84). Singular verb agreement with N1 can have been forced by the fact that the article is definite in (84) (in contrast to the indefinite in (85)) or the impossibility of a distributive verb reading (since the individual plates are not capable of rising nor of presenting a vertical configuration, the verb alzarse necessarily concerns the set of plates as a whole). (83) Las personas fotografiadas pueden parecer ridículas. Muchas lo son. Pero la foto no es ridícula como tal foto. Un montón de fotos de un fotógrafo de prensa aguanta mejor el paso del tiempo que un montón de palabras de un periodista por muy bueno que sea el periodista. (1995, novel) [un montónsg de fotospl aguantasg] ‘The people who are photographed may look ridiculous. Lots of them are. But the picture in itself is not ridiculous as a picture. A heap of pictures by a photographer stands the flowing of time better than a heap of words by a journalist no matter how good the journalist may be.’ (84) No pude por menos que decírselo, deslumbrado al observar la pila de platos y pocillos que se alzaba con perfecto equilibrio sobre el fregadero. (1992, novel) [la pilasg de platos y pocillospl se alzabasg] ‘I could not help telling him, blinded by observing the pile of dishes and cups that rose in perfect balance on the kitchen sink.’ However, the impossibility of a distributive interpretation does not prevent plural verb agreement in (85). The plural agreement is probably triggered by the fact that N2 consists of two coordinated plural nouns and that the relative clause specifies N2 and not N1. Yet, as (84) also presents these contextual characteristics, they cannot be considered dominant criteria. (85) Entra Camaleón. Lleva en sus brazos una pila de periódicos y revistas que le estorban la visión y ocultan la mitad superior de su cuerpo, (…). (1991, theatre) [una pilasg de periódicos y revistaspl estorbanpl y ocultanpl]

448 | The motivated nature of the QN-related co-selection patterns

‘Camelón enters. He holds in his arms a pile of newspapers and magazines that block his view and hide the upper half of his body, (…).’ Verb agreement is even more complicated in (86), since the binominal constructions invites both singular and plural agreement in the very same sentence. While N1 is the subject of the singular verb in the pseudo-relative clause which profiles the homogeneity of the group, the plural verb in the main clause seems to highlight the plurality of the N2-entities. Further, the predicate and the copular verb refer to the general type of the especies, and not to the particular subgroup that shows up in autumn. (86) En vera... en o... en otoño, a finales de verano, otoño el grupo de especies que... aparece son de dinoflagelados... con cromatóforos todos ellos, con cloroplastos todos ellos… (COREC, unknown, oral) [el gruposg de especiespl aparecesg … sonpl] ‘In sum < word interrupted> … in au… in autumn, at the end of summer, autumn the group of species that shows up are dinoflagellates... with chromatophores all of them, with chloroplasts all of them…’ In an attempt to determine the precise weight of the different factors listed by the Spanish tradition, the quantitative and qualitative analysis will be presented in Section 8.2.4.3. The starting assumption – and guideline of this chapter – is that semantics and construal prevail over formal constraints. From the moment QNs have adopted grammaticalized uses and indicate a large quantity of N2, it feels more natural that the verb agrees in plural with N2, the new head of the construction (e.g. (87)). However, the abundance of examples like (88), where a grammaticalized, quantifying use combines with singular verb agreement challenges this hypothesis. (87) Las palabras que siguieron luego o, mejor, el alud de frases deslabazadas [sic] que Eduardo escupió literalmente sobre mi rostro, me adentraron en una realidad sorprendente. (1982, books) [el aludsg de frasespl me adentraronpl] ‘The words that followed then or, better, the avalanche of incoherent sentences that Eduardo literally spat in my face, forced me into an astonishing reality.’ (88) (…), lo que puede provocar que esto termine como el rosario de la aurora y un alud de nuevos escándalos y revelaciones le obligue a agachar infamantemente la cabeza. (1995, books)

Morphosyntactic variation of BQs in the corpus data | 449

[el aludsg de escándalospl le obliguesg] ‘(…), which could well make that this ends in disaster and that an avalanche of new scandals and revelations forces him to defamatorily bend his head.’ In the remainder of this Section, I will argue that examples like (88) justify the importance attributed to CIP. More precisely, a high degree of CIP does not automatically result in agreement with N1, but it aptly accounts for those instances where grammaticalized uses of BQ combine with singular agreement in spite of the quantifying reading.

8.2.4.3 Towards the case-study The present case-study analyzes the scope of the traditional criteria. The structure reflects the historical shift from a focus on morphosyntactic factors to a focus on semantic criteria in the literature. I will first determine the weight of the article (Section 8.2.4.3.1), of the distance between the subject and the verb (Section 8.2.4.3.2), of the position of the subject with respect to the verb (Section 8.2.4.3.3), of the structural difference between partitive and pseudopartitive construction (Section 8.2.4.3.4) and the difference between main and relative clauses (Section 8.2.4.3.5). As to the semantic criteria, I will examine the impact of the type of verb (Section 8.2.4.3.6) and of the function of N2 (Section 8.2.4.3.7). Two general comments on the frequency tables are in order. First, as I indicated previously, I only take into account those instances where agreement with either N1 or N2 is visible. Examples with subjects such as un montónsg de gentesg or las pilaspl de librospl are filtered out. Needless to say, more types of unverifiable agreements have been observed, e.g. hesitation by the speaker or transcriber (cf. (89)), impersonal verbs (cf. (90)) or compound subjects (cf. (91)–(92)). (89) (…) y posteriormente un grupo de... exfuncionarios de correos hará(n) una demostración de manipulación de voto. (COREC, 1991, oral) [un gruposg de exfuncionariospl harásg/haránpl] ‘(…) and later a group of… ex-post office employees will held a demonstration of vote manipulation.’ (90) (…) porque hay en él iviasa montones de gente que compran el piso y ven las letras y no las pagan. (COREC, unknown, oral) ‘(…) because there are in it [iviasa] heaps of people that buy the flat and see the pay contracts and don’t pay them.’

450 | The motivated nature of the QN-related co-selection patterns

(91) Los montones de ropa mojada y los sonidos grabados son los elementos más realistas y cálidos, (…). (1994, press) [los montonespl de ropasg y los sonidospl sonpl] ‘The heaps of wet cloths and the sounds taped are the most realistic and warm elements, (…).’ (92) Aun así la turba y un hatajo de penados fugados entraron a saco y cometieron desmanes. (1986, novel) [la turbasg y un hatajosg de penadospl entraronpl] ‘Still the mob and a bunch of (lit. herd of) escaped convicts managed to enter plundering and to commit outrages.’ In addition, it should be pointed out that the instances of verb agreement in relative clauses have been systematically left out of the frequency tables. Section X will show that the predicates in relative clauses primarily combine with plural agreement (52 of 74 occ., in contrast to 40 cases of plural agreement in 128 main clauses). By excluding the relative clauses from the frequency tables, I prevent this tendency from deforming the results of the quantitative and qualitative analysis. More importantly, morphosyntactically we are dealing with a different construction. In the case of relative clauses, verb agreement is not primarily made with the antecedent, which is either N1 or N2, but with the relative pronoun which anaphorically refers to N1 or N2.

8.2.4.3.1 Definiteness of N1 According to Fält (1972, 99), (pseudo)partitive constructions preceded by the indefinite article tend to combine with plural verb agreement with N2. The underlying discourse-pragmatic reasoning is that, in contrast to the indefinite, the definite article automatically strengthens the proper referential (or extensional) value of N1, since definites primarily serve to introduce known or identifiable entities. As such, a quantifying interpretation is rather implausible in combination with a definite N1-determiner and therefore triggers verb agreement with N1, as in (93) (in contrast to the indefinite article in (94)). (93) Porque la pila de cosas por hacer me fascina tanto como me aterra. (2002, web page) [la pilasg de cosaspl me fascinasg] ‘Because the pile of things to do fascinates me as much as it terrifies me.’ (94) Cuando Mossén Ballarín (…) sale de los estudios (…), un aluvión de personas se le acercan. (1996, press)

Morphosyntactic variation of BQs in the corpus data | 451

[un aluviónsg de personaspl se le acercanpl] ‘When Mossén Ballerín (…) leaves the television studios (…), a flood of persons come to him.’ Tab. 9: Agreement according to the N1-determiner

Vsg with N1 Vpl with N2

Indefinite art.

Definite art.

36 17

52 23

However, Table 9 shows that this tendency only holds in one direction. Although the definite determiner indeed prefers singular verb agreement with N1 (in 52 of 75 occ. (or 69%)), the use of the indefinite determiner does not preferably combine with plural verb agreement with N2. On the contrary, in only 17 (or 32%) of 53 instances, the use of the indefinite article can be considered to motivate a plural verb. In other words, singular verb agreement is (proportionally speaking) equally common with indefinite and definite articles.

8.2.4.3.2 Distance between the subject and the verb Several authors argue that the probability of plural agreement with N2 increases as the distance between the subject and the verb increases. The underlying motivation is twofold: on the one hand, speakers more easily remember the meaning of an item than its form; on the other hand, speakers more easily remember the last element mentioned. In other words, when a certain time span separates the binominal syntagm from the verb (e.g. (95)), plural agreement with N2 feels more natural. (95) ...Oigan señores, esto es un juicio, no un teatro. Pero es que la mayor parte de los juicios que se ven en... en la sala, en vista oral, bueno, pues prácticamente no hacen más que reproducir lo anteriormente desarrollado y expuesto, ¿no? (COREC, 1991, oral) [la mayor partesg de los juiciospl … hacenpl] ‘ Listen gentlemen, this is a trial, not a theatre. But the majority of the trials that one sees in… in the room, life, well, the almost do nothing more than reproduce what has been previously developed and exposed, isn’t it? ’

452 | The motivated nature of the QN-related co-selection patterns

Tab. 10: Agreement according to the number of words separating the V and the SUBJ

Vsg with N1 Vpl with N2

0

1

2

3

4

5

6

32 11

22 10

7 3

10 4

0 3

4 0

13 9

For the sake of simplicity, the distance between the verb and the binominal subject of the clause is measured in terms of the number of words that separate them.232 Table 10 shows that the tendency is again only corroborated in one dimension: although the absence (or scarcity) of words separating the verb and the subject presents a strong preference for singular agreement with N1, the probability of plural agreement with N2 does not increase as more words separate the verb and the binominal subject.

8.2.4.3.3 Position of the subject with respect to the verb The third formal constraint stipulates that the ante-position of the binominal subject with regard to the verb favors plural agreement with N2, since the plurality of the N2 has already been made explicit (e.g. (96)). (96) (…) en España, eh... la mayor parte de las personas nunca incriminan a una persona cuando ... ocurre un hecho delictivo, nunca dicen... (…) (COREC, 1991, oral) [la mayor partesg de las personaspl incriminanpl] ‘(…) In Spain, eh… the majority of the persons never charge someone when … a criminal fact occurs, they never say… (…).’ Tab. 11: Agreement according to the position of the subject vis-à-vis the predicate Ante-position Post-position SUBJ SUBJ Vsg with N1 Vpl with N2

59 36

29 4

However, Table 11 shows that, again, the syntactic constraint is only verified in one direction. Post-posed subjects indeed almost exclusively combine with singular verb agreement (in 29 (or 88%) of 33 cases) and plural verb agreement

|| 232 The calculus of the number of categorial boundaries that separates the binominal subject and the verb is probably more relevant than the actual number of words (cf. Corbett’s Agreement Hierarchy). For reasons of feasibility, I leave the matter for further issue.

Morphosyntactic variation of BQs in the corpus data | 453

is more common with ante-posed subjects (in 36 (or 90 %) of 40 cases), but the tendency does not hold in the opposite direction. Ante-posed subjects do not prefer plural over singular verbs: in 59 (in 62%) of 95 instances, singular agreement with N1 is observed.

8.2.4.3.4 Partitive versus pseudopartitive constructions Although several authors clearly differentiate between partitive and pseudopartitive constructions when it comes to the function or interpretation of N1, no stringent differentiation is made between their agreement patterns (cf. the doble concordancia in RAE 2009). However, on the basis of the function of N1, different tendencies can be hypothesized. In partitive constructions, both N1 and N2 have referential value: while N1 functions as the head of the construction, the PNP indicates the larger set of entities from which N1 is extracted. In view of the syntactic and semantic head-function of N1, I expect partitive constructions to preferably combine with singular agreement with N1. Tab. 12: Agreement according to construction type

Vsg with N1 Vpl with N2 Total

Partitive

Pseudopartitive

10 22 32

78 18 96

Table 12 challenges this hypothesis however, since the proportion of plural verbs with partitive constructions (22 or 69% of 32 instances) exceeds the proportion of plural agreement with pseudopartitive constructions (18 or 19% of 96 instances). Only 10 instances corroborate the working hypothesis. When zooming in on the partitive constructions of the corpus, it immediately strikes the eye that these instances concern a limited set of N1s that recur. In other words, the choice for a partitive or pseudopartitive realization can be N1related. More precisely, generalizing over the instances of verb agreement in main clauses and relative clauses, mayoría (in 23 of 24 occ.), mitad (in 4 of 4), parte (in 35 of 36), pieza (in 1 of 1) and resto (in 4 of 7) show a clear preferences for the partitive construction. What is more, mayoría, resto and parte tend to

454 | The motivated nature of the QN-related co-selection patterns

evoke quantifying uses in spite of their partitive realization.233 The difference between the literal reading and the quantifying reading of mayoría, parte and resto is quite subtle and resides in the conceptualization of N2 as a ‘reference mass’: when the exact extension, number or size of N2 is known and profiled, the reading is strictly partitive. When the extension of the set of N2-entities is unknown, unprofiled or irrelevant, the (morphosyntactically) partitive construction yields a quantifying interpretation. While in (97) the speaker obviously knows the total number of teams, the total number of Spanish women is probably unknown by the speaker in (98). In the latter example, the quantity assessment is subjective: the conceptualizer does not know the precise extent of the group of women, nor of the group of women with a pronounced maternal instinct. (97) Hay etapa en principio pero es que no ha llegado la mayoría de los equipos. A las once y media intentaremos dar más detalles. (COREC, 1991, oral) [hasg llegado la mayoríasg de los equipospl] ‘In principle there is a stage but the thing is that the majority of the teams has not arrived. At eleven thirty we will try to provide more details.’ (98) Todas las señoras se acercan y le quieren abrazar, achuchar... yo creo que en el fondo, todas las señoras querrían que fuera hijo suyo. Sí, eh... eh... eh... el problema es que la mayoría de las mujeres pues tienen ese instinto maternal y ya se sabe, que quieren achuchar todo lo que sea. Claro y está bueno... (COREC, 1991, oral) [la mayoríasg de las mujerespl tienenpl] ‘All the women come closer and want to hug him, to cuddle him… I believe that in the end, all the women would want him to be their son. Yes, euh… euh… euh… the problem is that the majority of the women well they have this maternity instinct and as is well known, they want to cuddle anything. Of course, and that is a good thing…’ By way of illustration, Table 13 presents the distribution of uses of the partitive construction in my corpus. While the quantifying instances present a clear pref-

|| 233 In 23 examples, parte is preceded by the intensifying adjectives mayor (12 occ.), gran (7 occ.) and buena (4 occ.).

Morphosyntactic variation of BQs in the corpus data | 455

erence for plural agreement, the literal uses combine equally frequently with singular as with plural agreement. Tab. 13: Agreement according to reading in partitive constructions

Vsg with N1 Vpl with N2

N1 head

N1 quantifier

7 8

3 14

Needless to say, a corpus of 32 instances is insufficient to uncover clear tendencies. It intuitively suggests however that semantic criteria play a more prominent role in verb agreement than purely morphosyntactic criteria.

8.2.4.3.5 Verb agreement in relative clauses As I argued previously (cf. Section 8.2.4.3), the verbs in the relative clauses tend to agree with N2. Since in 52 (or 70%) of the relative clauses, plural verb agreement is observed, these instances have not been taken into account when determining the scope of the other factors traditionally considered to influence verb agreement. The reasoning is as follows: [c]uando el antecedente de la relativa es el grupo nominal que representa el conjunto más amplio del que se obtiene la parte, la característica se predica de la totalidad. De manera implícita, se atribuye también a la parte extraída, ya que los elementos que la componen son asimismo miembros del conjunto más amplio. (RAE 2009, 2584)234

The agreement with N1 in (99) and with N2 in (100) indeed suggests that the verb agrees with the antecedent of the relative clause. (99) (…), mi hijo Julián, que mañana cumple diez años, me observa sonriente e intrigado por esa pila de cuartillas que crece y crece, quizá convencido de que su padre también ha contraído esa enfermedad de los libros y las palabras. (2001, novel) [esa pilasg de cuartillaspl que crecesg] ‘(…), my son Julián, who celebrates his tenth birthday tomorrows, stares at me smiling and intrigued by that pile of sheets that grows

|| 234 When the antecedent of the relative clause is the nominal that represents the larger group of which a subpart is obtained, the characteristic is attributed to the whole. Implicitly, it is also attributed to the extracted subpart, given that the elements which compose the latter are at the same time members of the larger whole (translation KV).

456 | The motivated nature of the QN-related co-selection patterns

and grows, may be convinced of the fact that his father has also caught that disease of books and words.’ (100) Lo demás es zarandaja, ¿o usted cree que Milwaukee sacrificaría a uno sólo de sus hijos por un hatajo de kurdos que murieron hace veinte años? (2003, press) [un hatajosg de kurdospl que murieronpl] ‘Everything else is tittle-tattle, or do you really think Milwaukee would sacrifice a single one of its sons for a bunch of (lit. herd of) Kurds who died twenty years ago?’ Table 14 corroborates the correspondence between the subject referent of the relative clause and verb agreement. When N1 functions as the antecedent of the relative pronoun, agreement is made exclusively with N1. When N2 is the antecedent of the relative pronoun, the verb agrees in 50 (or 96%) of 52 with N2. Tab. 14: Agreement according to the antecedent in relative clauses

Vsg with N1 Vpl with N2

N1 = referent

N2 = referent

Ambiguous referent

19 0

2 50

1 2

Further, examples such as (101) show that verb agreement can also serve to disambiguate. In the following example, the singular verb suggests that the habitants of the Canary Islands are not put in a state of exasperation by the individual insults, but by the accumulation of insults. (101) Desde Madrid se ha perpetrado una larga letanía de agravios y dejadeces que tiene a los canarios en un estado de exasperación. (1977, press) [una larga letaníasg de agravios y dejadecespl que tienesg] ‘Madrid has perpetrated a long litany of insults and neglects which holds the habitants of the Canary Islands in a state of exasperation.’

8.2.4.3.6 Type of verb and contextual cues The present section discusses both the meaning of the verb and the contextual cues which can highlight the singular or plural character of the subject. The hypothesis is that when a verb, by its proper meaning (e.g. (102)) or by contextual cues (e.g.(103)), either evokes a singular subject or refers to N1, singular agreement with N1 is to be expected. In (102), the predicate llegar al cielo obvi-

Morphosyntactic variation of BQs in the corpus data | 457

ously relates to the pile, i.e. the set of murderers, and not to the individual murderers themselves. Similarly, in (103), the predicate poseer una vigorosa nervadura can only characterize a mass or set of entities, and not the individual sheets of paper. Among the recurrent verbs referring to a group of entities (and not to the individual entities), I observed amontonarse ‘to gather, to pile up’, desbordar ‘to exceed, to go beyond’, acumularse ‘to accumulate’, crecer ‘to grow’, consistir en ‘to consist in’, estar dividido ‘to be divided’, formar ‘to constitute, to form’, menguar ‘to diminish’, etc. (102) (…), pero todos seguían siendo insolidarios, cadáver tras cadáver la pila de asesinatos llegaba al cielo, (…). (1981, novel) [la pilasg de asesinatospl que llegabasg] ‘but everybody kept unsupportive, corpse after corpse the pile of murders reached the sky, (…).’ (103) A medida que examinaba el archivo comprendí – (…) – que aquel montón de papeles desparejos, de sus anotaciones, poseía una vigorosa nervadura interior que (…). (2004, novel) [aquel montónsg de papelespl poseíasg] ‘As I examined the archive, I understood – (…) – that that heap of irregular papers, of his notes, possessed a vigorous internal vein structure that (…).’ Reversely, when the verb, by its proper meaning or by contextual cues, only applies to N2 (e.g. (104)) or by definition evokes the plurality of its subject (e.g. (105)), the verb is plural. In (104), the context – the masculine plural of los and muchos – indicates that there have been a lot of wounded people (heridos) and not a lot of piles of injured (in that case, the context would have read las and muchas). Like ser mucho, the predicate agolparse requires a plural subject entity. Among the recurrent verbs referring to a plural subject, I observed empezar a salir ‘to start coming out’, agolparse ‘to crowd’, reunirse ‘to get together’, distribuirse ‘to divide up’, despedirse ‘to say goodbye (to each other)’, hablarse ‘to talk to each other’, etc. (104) (…), en la riña de Sierra Vista hubo nueve muertos y una pila de heridos, no los contó nadie pero fueron muchos, (…). (1988, novel) [una pilasg de heridospl fueronpl] ‘(…) in the fight of Sierra Vista there have been nine casualties and a pile of wounded, nobody counted them but there were many, (…).’

458 | The motivated nature of the QN-related co-selection patterns

(105) Un tropel de recuerdos, que creía ya olvidados, fueron agolpándose en su cabeza y sintió muy cerca la impaciencia del fuego y, (…). (1993, novel) [un tropelsg de recuerdospl fueronpl] ‘A mass of memories, which he thought to have forgotten already, were crowding together in his head and he felt the impatience of the fire nearby and (…).’ I further consider Brucart’s (1997) distinction of ‘distributive reading’ as a subtype of verbs which only relate to N2. According to Brucart, a distinction has to be made between a ‘distributive reading’ of the predicate and the ‘group reading’. The reasoning is as follows: Así, en Una parte de los turistas habían nacido en Argentina se alude a que tanto nacimientos como turistas forman el subconjunto aludido por una parte. Por lo tanto, en la interpretación distributiva el evento resulta multiplicado por el cuantificador. En cambio, en la lectura de grupo, a pesar de existir un sujeto formado por una pluralidad de individuos, el evento al que refiere el predicado es único, como sucede en La pareja obtuvo el primer premio en el concurso. (Brucart 1997, 173)235

In (106), the sentence would not make sense if the characteristic ser negros would apply to the entire set of children. The fact of being black, i.e. the origin, relates by definition to the individual children. (106) Yo viví ahí, por ejemplo, tres meses el año pasado, y... mi hijo, de 6 años iba a una escuela... eh... una escuela pública en la que la mayoría de los niños eran negros. (1991, oral) [la mayoríasg de los niñospl eran negrospl] ‘I have lived there, for instance, three months last year, and … my son, of six, went to a school… euh … a public school in which the majority of the children were black.’

|| 235 As such, A part of the tourists were born in Argentine relates to as many births as tourists which constitute the subpart alluded to by una parte. That’s why, in the distributive interpretation, the event is multiplied by the quantifier. In contrast, in the group reading, in spite of the fact that the subject is made up of a plurality of individuals, the event the predicate refers to is singular, as in The couple obtained the first price in the contest (translation KV).

Morphosyntactic variation of BQs in the corpus data | 459

V evokes plural SUBJ

V evokes singular SUBJ

Reference to N1

Reference to N2

V neutral as to N1/N2 reference

Vsg with N1 Vpl with N2

Distributive reading

Tab. 15: Agreement according to type of verb/context

3 21

1 3

37 3

19 0

3 3

25 10

Table 15 shows that the semantic criterion type of verb/contextual cues could be considered relevant to verb agreement. When the predicate requires a single subject entity (or homogeneous mass) or when the meaning of the verb obviously applies to N1, the verb is singular. This tendency is verified in 56 (or 95%) of 59 occurrences. Similarly, when the verb reading is distributive, requires a plural subject or applies to N2 only, the verb is plural. This tendency is verified in 27 (or 80%) of 34 occurrences. The stability of this global distribution suggests once more that the semantic criteria play a more prominent role than the morphosyntactic ones in determining verb agreement.

8.2.4.3.7 Function of N2 It is generally agreed upon (a.o. Aarts 1998; Keizer 2007; Traugott 2008a; Vos 2002) that verb agreement is made with the syntactic head of (binominal) noun phrases. Recall that in GR, QNs shift from head status (modified by N2) to quantifier status (quantifying N2 which has acquired head status). In other words, when interpreted literally, QNs are expected to combine with singular verb agreement with N1. For instance, when pointing at a heap of books piled up on the table, we say El montón de libros apilados está/#están en la mesa. ‘The heap of books piled up is/#are (lying) on the table’ (e.g. (107)). When functioning as a quantifier, the verb is expected to agree with N2 (cf. Di Tullio/Kornfeld 2008, 6). When noticing a number of soldiers in the street, we rather say Un montón de soldados andan/#anda en la calle. ‘A lot of soldiers are walking in the street’ (e.g. (108)). (107) En el fondo del escritorio se adivinaba una pila de cuadernos y una vasija con lápices y plumas. (2001, novel) [se adivinabasg una pilasg de cuadernospl] ‘At the back of the desk a pile of notebooks and a box of pencils and pens could be spotted.’

460 | The motivated nature of the QN-related co-selection patterns

(108) Para conseguir ese bloqueo harían falta un tropel de bombarderos de los que entonces no andábamos muy boyantes. (1981, novel) [haríanpl falta un tropelsg de bombarderospl] ‘To achieve that blockade a mob of firemen would be necessary, of the kind of which we did not have too many that time.’ Tab. 16: Agreement according to the reading of N1

Vsg with N1 Vpl with N2

N1 = head

N2 = quantifier

Ambiguous

27 12

46 28

15 0

As plausible as the semantic motivation sounds, Table 16 shows that the criterion function of N1 is not conclusive either. Although it is true that literal QNs tend to combine with singular verb agreement (in 27 (or 69%) of 39 occ.) and the cases of plural agreement generally concern quantifying readings (in 28 (or 70%) of 40 occ.), the hypothesis is challenged by the large number of quantifying uses (46 or 62% of 74 occ.) that combine with singular verb agreement.

8.2.4.3.8 Concluding remarks The previous sections have shown that the majority of the (morphosyntactic) criteria traditionally provided are not conclusive in determining verb agreement. This is not to say that they do not impact verb agreement at all, but they are often verified in one direction only. In other words, the criteria indicate tendencies and not rules strictly speaking. For instance, the ante-position of the subject with regard to the verb seems to be conditional for plural agreement, yet does not trigger plural agreement by itself. It bears pointing out that in the (counter)examples provided, it is often difficult to disentangle the scope of different parameters.236 In (109), the fact that the BQ-construction follows the verb is expected to trigger singular verb agreement. Yet in the light of the use of the indefinite article, the quantifying reading of N1, the fact that the verb and the subject immediately follow each other, etc. singular verb agreement with N1 is the unmarked option. The analysis thus suggests on the one hand that the various parameters interact, and, on the oth-

|| 236 This is exactly the reason why I decided not to carry out a statistical analysis of the criteria listed.

Morphosyntactic variation of BQs in the corpus data | 461

er, that the traditional, formal criteria are not dominant in determining verb agreement. (109) Pues mire usted, en Estados Unidos sucede, en los países que se pa que se consideran más bien más desarrollados económicamente, también suceden una cantidad de barbaridades como ésta y (…).(COREC, 1991, oral) [sucedenpl una cantidadsg de barbaridadespl] ‘Well have a look please, in the United States it happens, in the countries that that consider themselves to be rather more developed economically, a lot (lit. quantity) of barbarities like this one also happen and (…). As to the relevance of the traditional criteria in interaction, the case-study suggests that the semantic criteria scope over the formal criteria. On the one hand, the type of verb (and/or contextual cues) motivates the verb agreement in 83 (or 89%) of 93 occurrences (in the remaining 35 occ., the context and verb were neutral). On the other, the counter-examples to the formal criteria can generally be motivated in the light of the reading of N1. However, the focus on the reading of N1 still leaves 46 occurrences unexplained, where the quantifying reading combines with singular verb agreement. I will therefore argue in the following sections that conceptual persistence can shed light on the ‘unexpected’ verb agreement in those 46 occurrences.

8.2.4.4 CIP as a complementary factor in verb agreement As I have argued repeatedly in the previous chapters, I believe that the function of BQ-constructions resides in its ability to add a qualitative component to quantity assessment. This quality assessment has been argued to result from CIP. As to verb agreement, I hypothesize that the tendency towards CIP of BQconstructions can motivate cases of singular verb agreement in spite of the quantifying use of N1. In view of the metaphorical weight of N1 on the quantity assessment, it can become difficult to determine the semantic head of the construction. For instance, in (110), the image-schematic structure of letanía is so pervasive that the singular agreement with N1 feels like the most natural option. (110) En todo caso, la letanía de cuestiones era respondida con un Nunca lo harán que prohibía todo acercamiento, toda interferencia, toda aparición. (1987, novel) [la letaníasg de cuestionespl erasg] ‘In any case, the litany of questions was answered by a “They will

462 | The motivated nature of the QN-related co-selection patterns

never do that which precluded any rapprochement, any interference, any appearance”.’ Table 17 compares the tendency towards CIP of the grammaticalized uses to their agreement preferences. The frequency table shows that singular verb agreement indeed almost exclusively (in 44 or 96% of 46 cases) shows up when the conceptual image of the QN persists (at least metonymically) in the quantifying use. On the other hand, the distribution also suggests that CIP does not trigger singular verb agreement, since in 26 (or 37%) of 70 occurrences, CIP combines with plural verb agreement. Tab. 17: Agreement according to the degree of CIP High CIP

Medial CIP

Neutral CIP

30 22

14 4

2 2

Vsg with N1 Vpl with N2

In sum, the interest of the CIP-criterion resides in its complementary function. When comparing frequency tables 16 and 17, it is fair to say that the literal reading triggers singular agreement with N1. Reversely, the quantifying reading of N1 conceptually motivates plural agreement with N2, but plural agreement is not necessarily the most natural option. In the quantifying uses, conceptual persistence plays a prominent role. When the conceptual image of the QN is pervasive in the quantifying use, singular verb agreement may become more natural again. Figure 2 provides a simplified sketch of the semantic-conceptual interaction between the reading and the conceptualization of the QN. It goes without saying that the interference visualized in Figure 2 should not be interpreted as a rule governing verb agreement. Instead, the figure reflects strong tendencies only, since any instance of verb agreement is inserted in a specific (morphosyntactic) context where several factors interact. agreement with N1

N1 = head reading of N1

CIP N1 = functional no CIP

agreement with N1 (unmarked) (/with N2 possible) agreement with N2

Fig. 2: Interaction between the reading and conceptualization of N1

The relevance of CIP reminds of a particular trend in studies on agreement, especially of the English psycholinguistic tradition (Bock et al. 1999; Bock et al.

Morphosyntactic variation of BQs in the corpus data | 463

2006; Vigliocco 1996) to attribute the agreement pattern of QNs to specifications in the lexicon. Likewise, Vos (2002) relates variation in the agreement pattern to the grammatical category – which is of course fixed in the lexicon – of the N1: quantifying nouns such as montón and mayoría trigger plural agreement, while ‘partitive’ nouns such as parte and trozo combine with singular agreement and collective nouns such as familia and bosque allow both types of agreement. Instead of attributing the agreement pattern to lexical preferences of particular N1s, I believe that QNs differ as to their susceptibility of desemanticizing. The abstract, quantity-related source semantics of mayoría for instance, sharply contrasts with the rich source semantics of alud. In addition, the corpus analysis indicates that the majority of the QNs alternate singular and plural agreement.237 In sum, instead of preferences specified in the lexicon, I claim that verb agreement is determined by the interaction of a set of criteria of different kind and different weight.

8.2.4.5 Agreement hierarchy of BQ-constructions The frequency tables in the previous sections suggest that the semantic criteria, viz. the type of verb(/contextual cues), the function of N1 and the degree of CIP, play a crucial role in the verb agreement of BQ-constructions. Although marginal in number, examples such as (111) indicate that semantic criteria can still be overridden: in spite of the CIP, han is a plural verb. The facts that a parenthesis is inserted between the subject and the verb, that the N2-entities are further specified and that the verb requires a plural subject have impacted the verb agreement. (111) Un aluvión de enfermos, la mayoría personas mayores con infecciones respiratorias de carácter vírico, han sobrecargado los servicios de Urgencias y han llenado los hospitales de Gipuzkoa. (2004, press) [un aluviónsg de enfermospl hanpl sobrecargado] ‘A flood of ill people, in the majority elderly people with respiratory infections of a viral character, have overloaded the Emergency services and have filled the hospitals of Gipuzkoa.’

|| 237 In our corpus, only six N1s favor a specific agreement type: while alud, aluvión, tropel and letanía tend to combine with singular agreement with N1, mayoría and parte almost exclusively combine with plural agreement.

464 | The motivated nature of the QN-related co-selection patterns

I therefore defend the claim made by Eberhard et al. (2005) that agreement “is not only syntactic, not only semantic, and not only pragmatic, but all of these things at the same time” (2005, 531). Any attempt to establish the determinant criterion in verb agreement will fail, since several criteria appear to interact. Further, on the basis of the case-study, it is possible to establish a sort of hierarchy among the factors. Depending on the degree of stability observed in our corpus (cf. frequency tables 9–17), the relevant factors can be organized on a scale from more to less relevant. Figure 3 provides a simplified sketch. The first three factors are particularly relevant since they function in two directions (i.e. they specify the contextual configuration which triggers a specific agreement type, while the opposite context triggers the alternative agreement type). The last three factors, not by coincidence of morphosyntactic nature, define the contexts favoring singular agreement with N1. However, reversely, the opposite contexts do not necessarily trigger plural agreement with N2. a.

b.

c. d. e.

f.

Relative clause. In general, plural agreement with N2 is the unmarked option in relative clauses. In fact, the verb agrees with the syntactic antecedent of the relative pronoun: in plural when the relative clause refers to N2, in singular when the relative clause refers to N1. The function and conceptual image of N1. When N1 is interpreted literally and thus functions as the semantic head of the construction, the verb agrees in singular with N1. If N1 functions as a quantifier, the verb agrees with N2, except in cases of (a high degree of) CIP. The type of verb/contextual cue. When the verb evokes either N1 or N2 by itself or by contextual cues, it agrees with N1 or N2 respectively. The N1-determiner. When N1 is preceded by a definite article, the verb agrees with N1. The distance between the subject and the verb. The absence (or scarcity) of words inserted between the subject and the verb favors singular agreement with N1. The position of the subject with respect to the verb. The post-position of the subject combines with singular verb agreement.

Conclusion | 465

agreement with N1

N1 = referent relative clause N2 = referent

agreement with N2

literalN1

agreement with N1 CIP

main clause

agreement with N1 (unmarked)



functional N1

agreement with N2 when the V/context evoke N2 no CIP

agreement with N2 agreement with N1 when the V/context evoke N1 // agreement with N1 possible in case of



definite article distance between subject and V post-position of subject

Fig. 3: Agreement hierarchy of binominal quantifiers

In sum, verb agreement can be considered another morphosyntactic domain where form is motivated by the BQ’s constructional semantics (i.e. the reading of N1 and the degree of CIP) and partially QN-related (since the rich source semantics of some QNs block pure quantifying uses).

8.3 Conclusion The literature review shows that the few morphosyntactic descriptions of BQs, and by extension of pseudopartitive constructions, present the binominal construction as a highly constrained – and often even lexicalized and fixed – expression. The in-depth qualitative analysis of the corpus data enabled me to explain part of the syntactic peculiarities from a cognitive-functional point of view, but does not confirm the ‘defective’ nature of the BQ as presented in literature. On the contrary, the data reveal not only a lot more (morphosyntactic) variation than expected, but also the motivated character of these fluctuations. More precisely, I have argued that the construction-internal variation of BQconstructions, viz. the determiner pattern, the modification pattern and the N2pattern, is conceptually driven. More precisely, the paradigmatic variation interacts with the constructional semantics (i.e. the literal, quantifying or specifying use of N1), the QN’s conceptual image and discourse-pragmatic motivations. The impact of the conceptual factors can vary according to the paradigm analyzed (cf. Table 18). When it comes to determiner variation, the constructional

466 | The motivated nature of the QN-related co-selection patterns

slot appears to be highly susceptible to discourse-pragmatic motivations, since in the grammaticalized uses, the N1-determiner scopes over the entire BQ-construction. This functional motivation explains why determiner-variation is to a certain extent related to the reading of N1. In addition, the rich source semantics or the advanced stage of syntactic context expansion of specific QNs can correspond to a particular determiner-preference. In the adjective-paradigm, the impact of the constructional semantics and CIP appear to be equally strong. Generalizing over the different QNs, the grammaticalized uses only admit intensifying – as to quantity and to quality – adjectives. On the other hand, CIP determines which adjectives may function as intensifiers for particular QNs. The adjective-paradigm is also discourse-pragmatically motivated in the sense that the restriction to the prenominal position is functionally motivated. With regard to the N2-paradigm, it bears pointing out that in the grammaticalized uses, the N2 is the head (at least semantically) of the construction and therefore guides the speaker’s choice for an appropriate QN. Reversely, individual QNs present specific preferences in harmony with their realization as a head, quantifier or specifier. Crucially, the BQ-construction imposes a multiplex reading on N2. In addition, the reading of the QN and CIP play an important role in verb agreement as well, in interaction with morphosyntactic criteria. Tab. 18: Simplified sketch of the prominence of the conceptual factors per paradigm Reading of N1

CI(P) of N1

Discourse-pragmatics

Determiner

++

+

+++

Adjective

++

++

+

N2-collocate

++

++

+++

The strong impact of CIP and construal on the morphosyntactic characteristics of the BQ-construction strengthen my intuition on the prominent role of the QN’s image-schematic structure, both in the interpretation of specific utterances (synchronically) as in the process of GR (diachronically). Given that the pragmatic function of the BQ resides in its ability to add a qualifying component to quantity assessment and that this additional element roots in its CI, BQs are not likely to give up their tendency towards CIP. Since CIP affects, or better, guides the morphosyntactic behavior of the construction, it can explain why GR is sometimes ‘on hold’. I did not dwell on the variation in the prepositional paradigm in cases of compound right collocates, relating to whether the preposition is repeated or not. It is an interesting question, but few data are at hand in my corpus. Intuitively, I suppose that the repetition of de in cases of coordinated N2s enhances

Conclusion | 467

the individuality of both N2s. However, (112) and (113) mitigate the scope of this hypothesis: both examples are drawn from the same text and are only a few paragraphs separated. Yet their interpretation is quite similar. I leave the issue for further research. (112) De acuerdo con este principio, el público debe quedar sometido a un aluvión de informaciones y opiniones, algunas correctas, otras falsas y algunas que contienen elementos mixtos. En última instancia puede confiarse en que el público es capaz de digerir todo, para descartar lo que no esté al servicio del interés público y aceptar sólo (…). (1978, books) ‘In line with this principle, the public has to remain subject to a flood of informations and opinions, some correct, others false and some which contain mixed elements. Eventually, it can be taken for granted that the public is capable of digesting everything in order to discard what is not of public interest and to only accept (…).’ (113) El procedimiento optativo tal como lo entendían los liberalistas, fue dejar sometido el público a un aluvión de informaciones y de opiniones, unas posiblemente verdaderas, otras posiblemente falsas y otras que pueden tener parte de verdad y parte de error. En última instancia, de acuerdo con la tesis liberal, puede confiarse en que el público lo digiere todo y es capaz de descartar lo que no está al servicio del interés público y aceptar sólo (…). (1978, books) ‘The optional procedure as the liberalists interpreted it, was to leave the public submitted to a flood of informations and of opinions, some possibly true, others possibly false and others which are partially true and partially erroneous. Eventually, in line with this liberal claim, it can be taken for granted that the public digests everything and is capable of discarding what is not of public interest and only accept (…).’ I am now able to explain why I rather avoid the notion of restrictions. In my view, the specific morphosyntactic and combinatorial patterns or tendencies are driven by the construal the speaker wishes to convey of a specific entity or situation, which results in specific co-selection preferences rather than restrictions.

9 Theoretical reflection: the notion of CIP revisited The previous chapters have not only shown that each QN evokes a unique conceptualization of the N2-entities and that this conceptual image originates in the QN’s source semantics but also illustrated that the individual conceptual image of each QN is responsible for fluctuations in the morphosyntactic co-selection pattern. Against this background, the present chapter will link up with the claim made at the end of the diachronic part that persistence is not merely a side effect of GR, nor a matter of all or nothing, but will instead characterize CIP as a gradual, to some extent unpredictable and two-dimensional process. The structure of the present chapter is as follows. Section 9.1 will refine the notion of CIP in the light of the results of the synchronic analysis (see also Verveckken/Delbecque Fc.). Since the ‘mechanisms’ of analogy and CIP have often been mentioned in one breath in the diachronic analysis, Section 9.2 will dwell on the implications of the revisited notion of CIP for the role of analogical thinking and while doing so, bridge the synchronic and diachronic parts of this volume (see also Delbecque/Verveckken 2014; Verveckken Fc.). Section 9.3 will summarize the major conclusions and point to theoretical implications for GT.

9.1 The notion of CIP revisited In the present section, I argue that the phenomenon of conceptual persistence not only varies in degree (Section 9.1.1) but also in the facets of the QN’s conceptual image which get highlighted (Section 9.1.2). From a discourse perspective, the right collocate and the surrounding context play a crucial role in shaping the conceptualization the QN evokes, assuming that the speaker will try to convey a maximally coherent message (Section 9.1.3).

9.1.1 CIP as a gradual phenomenon A major reason why the term conceptual is more convenient than Hopper’s (1991) notion of lexical persistence is the gradual nature of persistence – recall that in the GR framework, “lexical” is often associated with the dichotomy lexical-grammatical. The following examples of hatajo de, for instance, illustrate different degrees of metaphorization of the QN. Example (1) provides a highly metaphorical reading of the QN: the speaker addresses a limited number of

The notion of CIP revisited | 469

people who he considers fairly ignorant (dos cosas debíais aprender) and he categorizes them all as swines without any further distinction, which mirrors the lack of individuality among the animals in a literal herd. In addition, the N2 evokes the literal ‘animal’-frame which enhances of course the conceptual persistence. By contrast, the use of hatajo de in (2) seems to have abstracted further away from the original ‘herd’-frame. Although the N2entities are reduced to a shared characteristic or action (que murieron hace viente años) and hatajo de is used pejoratively, the strictly metaphorical interpretation of hatajo de is less straightforward since the Kurds (N2) are not conceptualized as a mass of entities moving closely together. In fact, they are not even animate entities anymore. In (3), the N2s are completely devoid of animacy which rules out any metaphorical interpretation of hatajo de. (1)

– Dos cosas, por lo menos, debíais aprender de este hecho, hatajo de cabritos. (1986, novel) ‘– Two things, at least, you ought to learn from this event, bunch of swines (lit. little goats).’

(2)

Lo demás es zarandaja, ¿o usted cree que Milwaukee sacrificaría a uno sólo de sus hijos por un hatajo de kurdos que murieron hace veinte años? (2003, press) ‘Everything else is tittle-tattle, or do you really think Milwaukee would sacrifice a single one of its sons for a bunch of (lit. herd of) Kurds who died twenty years ago?’

(3)

De él dirán inmediatamente que es un moralista, un lúdico severo, teórico riguroso de la sensibilidad y sensible crítico de la duda. Visto lo anterior, conforta abandonar hatajo de originalidades tan repetidas y contemplar los modestos matices que, dentro del discurso político, viene adoptando últimamente un concepto tan poco chocante en la democracia como el liberalismo. (1995, books) ‘Of him they will immediately say that he is a moralist, a ludic yet severe man, a strict theorist of sensibility and a sensible critic of doubt. In view of that, it comforts abandoning a (lit. Ø) bunch of originalities repeated so many times and to consider the modest nuances that, within the political discourse, has recently come to adopt a concept so slightly chocking in democracy as liberalism.’

As a rule of thumb, I propose to distinguish between three degrees of CIP, i.e. high, medial and neutral CIP, depending on whether the relation of the grammaticalized QN with its source frame is a metaphorical, a metonymic or simply an implicit one. In an attempt to achieve maximal transparency in my data, I

470 | Theoretical reflection: the notion of CIP revisited

based the estimation of the degree of CIP on the degrees of abstraction established for the facets within the conceptual image of each QN. Interestingly, some facets immediately entail a metaphorical reading of the QN, whereas other facets are derived via metonymic extensions from the QN’s literal frame. The left-most column of facets in the grammaticalized uses are systematically associated with a metaphoric interpretation. HIGH CIP

MEDIAL CIP

(a) all at once

(h) DIFFERENT SOURCES

(b) all of a sudden

(g) UNEXPECTED

(c1) uncontrollable

(f) N2 is OBTRUSIVE, INSISTENT

(c2) overwhelming

(e) DIRECTED TOWARDS A SINGLE

NEUTRAL CIP

(i) NEWNESS OF N2

(j)TOO MANY/MUCH

VICTIM

(d1) causER: aluvión is trigger of (re)action/situation (d2) causEE: identifiable cause triggers/motivates aluvión

Fig. 1: Distinction between degrees of CIP on the basis of the conceptual image of aluvión de

By way of illustration, Figure 1 copies the conceptual image of aluvión de as established in Chapter 7 and indicates which sets of facets usually yield high, medial or neutral CIP. High CIP proceeds from the activation of the entire set of facets that originate in a metaphoric interpretation of aluvión de, viz. facets (a)(c) (as in (4)). Examples qualify for medial CIP either when one or more facets metonymically related to aluvión de’s original frame get highlighted (from (e) to (h), as in (5)) or when the set of metaphorically related facets is only partially profiled (cf. example (6), where the N2s are depicted as if overwhelming the Spanish road system literally, and to a lesser extent as appearing all of a sudden (comienza), yet the facet ‘all at once’ is not profiled in this particular context). The BQ displays neutral CIP when the quantity assessment is only vaguely reminiscent of aluvión de’s source semantics and profiles one or more of the facets (i) to (j). However, even the newness of N2 (facet (i)) or the idea that there are too many N2 entities (facet (j)) can only be arrived at through a chain of metonymic inferences. In other words, even in the cases categorized as displaying neutral CIP, the QNs’ conceptualization echoes its original frame.

The notion of CIP revisited | 471

(4)

Un país tradicionalmente parado, se ponía súbitamente en movimiento y experimentaba en primavera una inesperada floración de tipos y de ideas nuevas, algunas de las cuales madurarán y darán fruto en el otoño e invierno siguientes. Un aluvión de nuevos negocios, y nuevos empresarios, desconocidos meses atrás, parece inundar de repente la escena nacional, relegando a los políticos a las páginas interiores de los periódicos. (1988, books) ‘A country which was traditionally at a standstill got suddenly on the move and experienced in springtime an unexpected flowering of types and new ideas, some of which will come to fruition and bear fruits in the following fall or winter. A flood of new companies, and new businessmen, which only some months before were unknown, seems to suddenly inundate the national scene, relegating the politicians to the inside pages of the newspapers.’

(5)

Cuando Mossén Ballarín (Barcelona, 1920) sale de los estudios de televisión donde ha sido entrevistado, un aluvión de personas se le acercan. (1996, press) ‘When Mossén Ballerín (Barcelone, 1920) leaves the television studios where he has been interviewed, a flood of persons come to him.’

(6)

(…), la presencia de la bicicleta comienza a causar serios problemas en numerosos matrimonios, mayormente con hijos. En nuestro país existe del orden de medio millón de licencias de cicloturistas federados, amén del otro aluvión de esforzados de la ruta en versión no profesional que pueblan las carreteras españolas. (1994, press) ‘(…) the presence of the bicycle starts to cause serious problems in many marriages, mainly with children. In our country, there are in the order of half a million of licenses for federated touring cyclists, apart from the other flood of road maniacs in a non-professional version that populate the Spanish roads.’

It goes without saying that the conceptually rich original frame of aluvión de and alud de are harder to desemanticize than that of those QNs whose conceptual images are primarily configurational, less rich or less specific (such as pila, montón, racimo, etc). The majority of the image-schematic structures outlined in Chapter 7 present only one or two levels of schematization. Still, the distinction between high, medial and neutral CIP can be maintained. By way of illustration, Figure 2 applies this threefold distinction on the conceptual image of pila de. The grammaticalized uses of pila de qualify for high CIP when they mirror at least one of the facets related to the original frame (as in (7), where the bomb

472 | Theoretical reflection: the notion of CIP revisited

dropping reminds of the original verticality). Medial CIP proceeds from the activation of one of the facets which are metonymically related to pila de’s source frame, viz. the facets of the second column which corresponds to a first level of schematization. Neutral CIP is when none of the facets gets highlighted, as in (8). (7)

Como el propio Gadafi que desde que Reagan le soltó una pila de bombas a domicilio se ha tomado con mucha más parsimonia la práctica del santo terrorismo yihádico. (2001, press) ‘Similar to Gadhafi himself who, since Reagan dropped a lot/series of bombs on his residence, took with much more calmness the practice of the whole jihad terrorism.’

(8)

¿Qué lo haga aquí? Pues le hecho marrón parrilla, ahora mismitico Digo que lo haga aquí, ¡con la pila de gente que hay mirándome! (1996, oral) ‘That I should do it here? Well I have made roast chestnuts. Right now at the very same moment. I say that you should do it here. With the heap of (lit. pile of) people that are looking at me!’

HIGH CIP

MEDIAL CIP

verticality

(a) SUCCESSION

order

(b) SOCIAL NORMS, ESTABLISHED VALUES

manmade

(c) INTENTION (d) FINITE, DELINEATED

Fig. 2: Distinction between degrees of CIP on the basis of the conceptual image of pila de

In addition, examples present high CIP as soon as some element of the context forms part of the source frame of the QN. In (9), the context evokes the effects of N2 on the Catholic Church, which is precisely the religion to which litanies belong. In (10), the giant backpacks of Alpine hikers evoke a mountain scenery which is exactly the frame hosting snow slides as well. In (11), the quantifying inference of montón de is foregrounded by the juxtaposition with completo and todas, yet since a grouping of X-rays generally comes in a pile-like configuration, the occurrence qualifies for high CIP. (9)

La letanía de tragedias que causan cada día los malos tratos domésticos ha metido a la Iglesia católica en un nuevo jardín de

The notion of CIP revisited | 473

contradicciones y críticas. Externas e internas. (2002, press) ‘The litany of tragedies that the domestic maltreatments cause every day has put the Catholic Church in a new garden of contradictions and criticism. External and internal.’ (10) Pero ha triunfado la fuerza mercantil – que procura con harta frecuencia presentar lo superfluo como necesario – sobre toda lógica, y los indefensos escolares se ven cada año sepultados por un alud de volúmenes que les obliga a llevar gigantescas mochilas de excursionistas alpinos. (1997, press) ‘But the mercantile force – that very frequently tries to present the superfluous as necessary – has triumphed over logic, and the defenseless schoolchildren see themselves buried every year by an avalanche of volumes that obliges them to bring along giant backpacks of Alpine hikers.’ (11) En las urgencias españolas se saca mucha sangre a enfermos que no lo necesitan en la falsa creencia – por parte del paciente – de que sin un análisis no se puede hacer un buen diagnóstico, y ante la equivocada presunción – por parte del galeno – de que sin un perfil bioquímico completo del enfermo y un montón de radiografías en todas las proyecciones se suele estar indefenso ante el juez, en el caso de que el paciente realice una demanda. (1995, press) ‘In the Spanish emergencies, a lot of blood is taken from ill people who do not need it in the false belief – on behalf of the patient – that without an analysis one cannot make a good diagnosis, and on the wrong presumption – on behalf of the physician – that without a complete biochemical profile of the sick person and a heap of X-rays in all projections, one uses to be defenseless in front of the judge, in case the patient starts a lawsuit. ’ It is striking, yet not unsurprising, that there is no direct correlation between the relative degree of GR (in terms of frequency) and the degree of persistence. The traditional view on lexical persistence would expect a high degree of GR to go hand in hand with neutral CIP. Table 1 presents the distribution of the degree of CIP per QN and is compared to the relative ‘degree’ of GR per QN as presented in Table 2 (repeated here for convenience). The highest proportion of neutral CIP characterizes montón de, the least specific of the QNs as to the internal constellation of the mass. The remaining QNs are all susceptible of also adding a substantial qualitative component to the conceptual frame that is being developed, beyond their quantifying function. Mogollón de, which is highly grammatical-

474 | Theoretical reflection: the notion of CIP revisited

ized, always highlights facets of its conceptual image. The pairs aluvión de – alud de, on the one hand, and hatajo de – letanía de, on the other, present similar proportions as to the tendency towards high CIP and the proportion of grammaticalized uses. Instead of combining preferably with neutral CIP, those highly grammaticalized QNs are inclined to systematically activate their source frame. However, recall that I refuse to equate ‘relative proportion of grammaticalized uses (in contrast to literal uses)’ with ‘degree of GR’. The degree of GR can only be adequately ‘measured’ in terms of context expansion (see Chapter 3). If ‘degree of GR’ is measured in terms of ‘context expansion’, neutral CIP indeed combines with advanced context expansion in the cases of montón de and la mar de, yet not in the cases of barbaridad de and mogollón de. Likewise, in line with the usage-based assumptions on GR, a decrease in conceptual persistence over time is to be expected. However, the diachronic case-studies have shown that only the developmental paths of montón de and la mar de present this correlation. Yet while the tendency to add a qualifying component only with montón de gradually decreases, la mar de preferable combines with neutral CIP right from the start.

Alud de

Aluvión de

Barbaridad de

Hatajo de

Letanía de

Mogollón de

Montón de

Pila de

Racimo de

Tab. 1: Degree of CIP of the functional uses per QN

High CIP

# %

46 0.71

108 0.82

5 0.56

10 0.77

13 0.68

0 0

78 0.23

3 0.18

9 0.39

Medial CIP

# %

17 0.26

20 0.15

2 0.22

1 0.08

6 0.32

45 1

118 0.35

12 0.71

12 0.52

Neutral CIP

# %

2 0.03

4 0.03

2 0.22

2 0.15

0 0

0 0

144 0.42

2 0.12

2 0.09

65

132

9

13

19

45

340

17

23

Total

The notion of CIP revisited | 475

Alud de

Aluvión de

Barbaridad de

Hatajo de

Letanía de

Mogollón de

Montón de

Pila de

Racimo de

Tab. 2: Synchronic layering of uses per QN

Head

# %

16 0.19

5 0.03

0 0

1 0.06

4 0.15

0 0

114 0.23

47 0.59

25 0.52

Quantifier

# %

62 0.72

124 0.83

9 1

6 0.38

17 0.65

44 0.96

311 0.62

16 0.20

21 0.44

# %

3 0.03

8 0.05

0 0

7 0.44

2 0.08

1 0.02

29 0.06

1 0.01

2 0.04

Ambiguous

# %

0 0

1 0.01

0 0

0 0

1 0.04

0 0

33 0.07

10 0.13

0 0

Indeterminate

# %

5 0.06

12 0.08

0 0

2 0.13

2 0.08

1 0.02

13 0.03

6 0.08

0 0

86

150

9

16

26

46238

500

80

48

Total

The latter observations make me refrain from considering CIP as a useful tool for identifying GR processes, in the way it was initially conceived by Hopper (1991). Instead, it provides an additional argument for the claim made in the diachronic part of this volume that CIP is responsible for the rise in productivity of the BQconstruction, since it catalyzes the extensions towards both new QNs and new N2-combinations per QN. Although the status of ‘mechanism’ may sound exaggerated – recall that persistence is originally considered a side-effect of GR–, it certainly motivates the productivity of the construction. Further, since the BQs provide the Spanish speaker with a useful tool for hyperbolic quantification, they are not likely to end up as desemanticized quantifiers in the near future.

9.1.2 CIP as an unpredictable phenomenon The systematic splitting up of the QN’s original frame into several conceptual facets which constitute its conceptual image (see Figures 1, 4-10 in Chapter 7) brings us to a second refinement of the notion of CIP, viz. its unpredictability || 238 Recall that for mogollón de, one adverb use and three adjective intensifier uses have been observed. The exact number of adverb uses and adjective intensifier uses of montón de have not been counted for reasons of feasibility.

476 | Theoretical reflection: the notion of CIP revisited

or the impossibility to foresee which semantic property will persist in the grammaticalized uses. In setting up the conceptual images, I repeatedly illustrated that the sets of conceptual facets are not to be seen as a matter of all or nothing. Instead, the context and N2-collocate co-determine which property (or properties) of the QN’s source frame get(s) highlighted. In both examples (12) and (13), aluvión de displays high CIP for referring to a group of N2s which suddenly appear all at once and, by doing so, overwhelm a specific entity (or victim). Yet, in (12), the abruptness and newness of the N2s are particularly prominent: while de repente specifically highlights the suddenness of the appearance, the postmodifying nuevos stresses the ideas that the curas did not behave as such before and somehow just came into being. However, in (13), the predicate desbordó todas las previsiones focuses on the unexpected nature of the high number of N2s: the military health services had expected a lot of calls, but the actual number of calls exceeded all expectations. The latter facet is however not foregrounded as such in (12), and is, to some extent, not even compatible with the context. (12) ‘Ya no era sólo el padre Antonio el que hablaba descaradamente de “libertad”, de amor, de caridad y de pureza de intenciones. De repente, un aluvión de curas nuevos planteaban soluciones nunca oídas hasta aquellos momentos. ‘It was no longer only Father Antonio who spoke frankly of “freedom”, of love, of charity and the pureness of intentions. Suddenly, a flood of new priests set out solutions that had never been heard of before.’ (13) A las 9 horas, tres médicos del cuerpo de sanidad militar comenzaron a atender las seis líneas habilitadas del teléfono 91-395-54-85. Sin embargo, el aluvión de llamadas desbordó todas las previsiones, por lo que Defensa reforzó el servicio con otros siete facultativos más, (…). (2001, press) ‘At 9 o’clock, three doctors from the military health force started to operate the six telephone lines supplied for 91-395-54-85. Nevertheless, the flood of phone calls exceeded all expectations. That’s why the Department of Defense reinforced the service with another seven physicians, (…).’ As to the selection of particular facets within the conceptual image, the construal operations (Paradis 2011) play an important role. The contextual cues can belong both to the elaboration level and to the actualization level (Delbecque/ Verveckken 2010). While the former refers to the host-class of the grammaticalizing item (i.e. the N2 or adjective the QN combines with), the latter encom-

The notion of CIP revisited | 477

passes the larger construction the QN appears in (e.g. the predicate) and the “mise en discours” or its discourse integration. Since N2 (and the modifying adjective, if present) constitute the closest context of N1, its role is crucial for determining the relative prominence of the possible conceptual properties that perspire in the QN. It goes without saying that concrete nouns such as casas ‘houses’ in un racimo de casas stand more chance to highlight the entangled or ‘bunch of grapes’-like configuration than abstract notions such as ambición ‘ambition’, which can hardly combine with other facets than ‘intentional gathering or bundling’. Further, the adjectives cadenciosa ‘rhythmical’ and cansina ‘weary’ observed in combination with letanía de obviously highlight different subparts of the original frame. As to the level of actualization, examples such as (14) illustrate that contextual cues frequently cross clausal boundaries. (14) Pagó a regañadientes la abusiva nota del té completo dejando incluso una propina excesiva para aliviar de algún modo la afrenta de aquella situación. Pansy le regañó al salir. En la guía Fodor's había leído que las propinas en Londres no debían ser superiores en ningún caso al 15 por ciento suponiendo que no estuviera ya incluida en la factura. Y él había dejado una barbaridad de propina que podría haberse destinado a la compra de otra jarrita del té (…). (1995, novel) ‘He paid unwillingly the unfair bill of the tea, while including an excessive tip to relieve somehow the embarrassment of that situation. Pansy quarreled with him when leaving the place. She had read in the Fodor’s guide that in London, gratuities should in no way exceed 15 per cent, provided that it is not yet included in the bill. And he had left a barbarity of a tip that would have sufficed to by another jug of tea.’ The importance of the N2 and the context in co-determining the conceptualization of the mass of entities leads me to characterize conceptual persistence as a two-way discourse phenomenon.

9.1.3 CIP as two-way discourse phenomenon I have repeatedly observed in this volume that the notion of collocational ‘constraints’ or ‘restrictions’ in GT is not quite appropriate for characterizing the relation between the QN and its co-selection pattern. As to the ‘restrictions’ the QN imposes on the right collocate, it has been shown that the N2s can be arranged, both synchronically and diachronically, into several clusters of semantically related N2s (e.g. the cluster of ‘invaders’ of aluvión de). In addition, the observed N2-clusters generally echo the QN’s source semantics. The conceptual

478 | Theoretical reflection: the notion of CIP revisited

motivation underlying the ‘constraints’ of course weakens their restrictive character. In other words, QNs develop combinatorial preferences – commonly interpreted as ‘restrictions’ – which are in congruence with their source semantics, but the only ‘constraint’, strictly speaking, that the QN imposes on N2 is that N2 can be made to fit the conceptual image of N1. The recurrent umbrella category ‘other’ among the N2-clusters indeed contains several N2s which are not expected to occur with a specific QN on the basis of its lexical definition or observed combinatorial preferences. The latter category should not be considered a waste bin however, since the N2-combinations can only occur when the contextual setting and the construal are right. For instance, at first sight the combination of the N2 páginas web ‘web pages’, which is completely devoid of animacy, clashes with the conceptual image of aluvión de. Yet in (15), several contextual cues make for a perfect fit: the sudden emergence of new web pages is said to reflect the intense boom of alternative medicine that has crushed several zones of the country. Since those properties attributed to the web pages clearly match the source semantics of aluvión de, the latter QN is a fully coherent choice in that particular discourse context. The source semantics of N1 thus interacts with the discourse context and contributes to the overall semantic congruence. Similarly, although many QNs would be grammatically correct in (14), barbaridad ‘barbarity’ is totally appropriate in terms of contextual coherence, for two reasons at least: (1) several contextual cues, i.e. abusiva, afrenta, no debían ser, etc., refer to the transgression or violation of a norm or social standard; (2) etymologically, barbaridad stems from the ancient Greek and Roman notion of ‘barbarians’ which refers to foreigners, often considered ‘uncivilized’ in the ancient empire. The context of (14) also pictures the protagonist as a tourist in London, who, in spite of the guide, does not manage to conform to the norms. Although the enhanced coherence is particularly striking in (14), native speakers need not of course be aware of the etymologies. In other words, the similarity/contiguity relations that hold between a specific [N1 de]-pattern and the new N2-combination to which it extends can be ensured at discourse level only. If the specific discourse integration of páginas web in (15) and propina in (14) would not be taken into account, those N2-combinations would be incorrectly considered as not corresponding to the N1-frame. (15) Tras arrasar en Estados Unidos, la medicina alternativa extiende sus tentáculos en España. (…) Al otro lado del Atlántico, el boom es de tal calibre que en algunas zonas del país cerca del 50% de los pacientes recurre a estas prácticas. (…) Como era de esperar, Internet también refleja este interés social, con un verdadero aluvión de páginas web sobre muchas de las medicinas alternativas. (1997, press)

The notion of CIP revisited | 479

‘After crushing in the United States, alternative medicine extends his tentacles to Spain. (…) At the other side of the Atlantic, the boom is of such caliber that in some zones of the country, nearly 50% of the patients turn to those practices. (…) As was only to be expected, the Internet also reflects this social interest, with an authentic flood of web pages on many of the alternative medicines.’ By focusing exclusively on the ‘constraints’ N1 yields on the host-class expansion towards new N2s, the analysis of the BQ remains one-sided and does not do full justice to the role played by N2. The latter, however, is at least as crucial as that of N1, not only because inside the binominal syntagm N1 and N2 are inextricably paired up, but also because in the wider context the profile of N2 is more prominent than that of N1. Recall that N2 functions as the head of the construction and conveys the essential information. As a consequence, certain N2s may even show strong preferences for a specific QN, or at least a specific type of QN (without resulting in semi-lexicalized expressions such as un enjambre de abejas ‘a swarm of bees’, cf. Bosque 2007, 193). Nouns such as críticas ‘criticism’, llamadas ‘calls’, protestas ‘protests’, etc., which by their very nature refer to entities which (generally) arise as a negative re-action, will prefer QNs such as alud ‘avalanche’ or aluvión ‘flood’. By way of comparison, the combination with the remaining QNs analyzed (i.e. barbaridad ‘barbarity’, hatajo ‘herd’, letanía ‘litany’, mogollón ‘lot’, pila ‘pile’, racimo ‘bunch’) is not found in the CREA-corpus (except marginally for with montón ‘heap’). In addition, the corpus search for the string [de críticas] yields almost exclusively combinations with the overwhelming nature phenomena aluvión ‘flood’, alud ‘snow slide’, lluvia ‘rain’, avalanche ‘avalanche’, torrente ‘torrent’, ola ‘wave’, oleada ‘huge wave’, chaparrones ‘downpour, cloudburst’, tormenta ‘storm’ and marejada ‘heavy sea, swell’. Similarly, the query [de preguntas] (‘of questions’) returns batería de preguntas ‘a battery of questions’ as the preferred combination yielding 25 instances (next to absolute quantifiers or generally QNs such as serie ‘series’(with 63 occ.) or par ‘pair’(with 15 occ.)). It further almost exclusively combines with N1s that evoke four kinds of image: horizontal alignment (e.g. retahíla ‘string’, sarta ‘string’, madeja ‘hank’, cadena ‘chain’), violent attack (tiroteo ‘shooting’, bombardeo ‘bombing’, andanada ‘broadside’), dynamic nature phenomena beyond human control (mar ‘sea’, ola ‘wave’, chaparrón ‘downpour’, avalancha ‘avalanche’, aluvión ‘flood’), or N1s evoking a mess (montón ‘heap’, laberinto ‘labyrinth’, maraña ‘tangle’). These N1s bring to the fore the confronting, relentless, unstoppable, confused, often antagonistic and even aggressive nature immanent in the N2 preguntas, thus profiling the (face) threatening dimension of the questioning. This finding provides further evi-

480 | Theoretical reflection: the notion of CIP revisited

dence for the mutual accommodation between both nominals. It bears pointing out that not all N2s by themselves call for a specific QN. Less emotionally loaded nouns or less specific nouns such as personas ‘persons’, which do not evoke but very general frames, can of course combine with a wide range of QNs. The previous finding thus suggest that, from a discourse perspective, it is the kind of profile the speaker wants to convey of the N2 entities that guides the choice of N1, not the other way around. It is therefore warranted to approach the interplay between persistence and analogical thinking as a two-way discourse phenomenon:239 N1, on the one hand, and N2 and the context, on the other, mutually accommodate to each other.240 On the one hand, the mutual adjustment between N1 and N2(/the wider discourse context) involves the imageschematic structure of the mass N1 imposes on N2 and the resulting combinatorial constraints or preferences N1 develops. On the other hand, the N2 and the specific discourse integration of the BQ co-determine not only which conceptual facets of the QN’s conceptual image are activated – and by consequence, the degree of CIP the BQ is apt to display –, but also which QN is the most appropriate and coherent choice in a specific context. Figure 3 visualizes the two directions of influences within the binominal syntagm.

Conceptual image Combinatorial constraints

N2 (+context)

N1  

Selection N1 Actualization conceptual image

Fig. 3: CIP as a two-way discourse phenomenon

The mutual accommodation between context and QN reminds of the notions of ‘isosemy’ (Pottier’s 1998; 2008) or ‘isotopy’ (Kerbrat-Orecchioni 1979; Rastier 1987), which refer to the conceptual affinity between two or more ele-

|| 239 See also García (2009, 291) who argues that “the paradigmatic value of a symbol and its syntagmatic distribution necessarily support each other”. 240 See also Michaud Maturana (2011) on the concept of ‘conceptual accommodation’ which refers to the possible process of re-interpretation that occurs when two conceptually divergent lexical elements are combined (2011, 189, 194–196).

The notion of CIP revisited | 481

ments of the same discourse context, since speakers tend to produce maximally coherent or ‘isosemic’ messages (cf. Grice 1975).241 Within the binominal syntagm, isotopy refers to the conceptual convergence between N1, on the one hand, and N2 and/or one or more elements of the surrounding context on the other. As discourse goes along, the co-occurrence of lexical elements that share one or several meaning components with N1, enhances the relative prominence of these components in the frame evoked by the discourse fragment they are part of (cf. (14)–(15)). As they contribute to the conceptual integration of N1, they consequently enhance its degree of CIP. Put cognitively, the mental spaces or conceptual images of N1 and of N2 are blended into a single, convergent frame. In other words, the original images are only partially projected in the resulting frame: only those conceptual facets of N1 that can be made to accommodate to the frame of N2 are picked out and shade through in the grammaticalized reading, and conversely. As a consequence, the final conceptualization of a specific BQ depends on the precise context of the specific utterance – which is in line with the emergent nature of meaning assumed in cognitive semantics. Distinct QN-occurrences thus profile distinct selections of facets within the QNspace (cf. the contrast between (12) and (13)). It goes without saying that the discrepancies are more salient with QNs presenting a rich source semantics (e.g. aluvión) than with QNs such as pila (‘pile’) which primarily encompass configurational facets. Yet there is more to it. In view of the mutual adjustment between the QN and its context, the choice of N1 can be claimed to function as a coherence device. In (16), for instance, the choice for montón de, instead of aluvión de which typically combines with the N2 críticas, warrants the coherence of the sentence.242 More precisely the predicate llevarse highlights the matching properties between montón and críticas: it suggests that the criticism or reproaches have been accumulated during the day, possibly with some intervals. In this precise

|| 241 The notion of ‘isosemy’ is originally used by Pottier in the context of lexical semantics. Isosemy is claimed to guide lexical choices in that the co-occurrence of two polysemic lexical elements or constructions will result in conceptualizations that are (partially) convergent, i.e. to the “reciprocal selection of likelihood” (Pottier 2008, 94). The notion of isotopy raises the congruence implied by isosemy to the discourse level. The principle of isosemy is rather associated with the conceptual congruence between specific linguistic elements. Isotopy designates the conceptual congruence or cohesion within an entire sequence or discourse, i.e. the speaker’s urge to construe a maximally coherent discourse by means of the recurrence of a single conceptual domain in several linguistic elements. 242 In our synchronic data (limited to 9 particular QNs though), críticas combines 12 times with aluvión de, 4 times with alud de and only 2 times with montón de.

482 | Theoretical reflection: the notion of CIP revisited

context, aluvión de is inappropriate: the introduction by una vez más a priori excludes an unexpected number of criticism, tras la jornada clashes with the sudden character usually profiled by aluvión de and the enumeration of the reproaches in the following sentences is definitely too short to fit the overwhelming character of a flood. In sum, to be a fully coherent quantification device, the QN minimally shares one conceptual facet with N2 or the surrounding context. (16) Una vez más, los árbitros se llevan un montón de críticas tras la jornada. El nivel medio de los colegiados es mediocre, pero pésima es la actitud de la mayoría de los jugadores. (1996, press) ‘Once again, the referees have (gathered) a lot of criticism after the day. The average level of the referees is mediocre, but the attitude of the majority of the players is very bad.’ In sum, the schematicity of the meaning structure of (relatively) grammaticalized items does not prevent them from having a meaningful import on the conceptualization of events or impact on the overall coherence of the context, which might explain why the GR of the BQ-construction seems to be ‘on hold’: as the (partial) activation of the QN’s source semantics plays a major role in the functional uses of N1, the QNs within the BQ are not likely to desemanticize completely. Further, since the morphosyntactic behavior of the construction is shown to be conceptually and discourse-pragmatically determined, the two-way discourse pressure counteracts the QN’s formal decategorialization as well. Our broader definition of how persistence functions synchronically sheds new light on the recent trend in GT to highlight the contextual embedding of grammaticalizing constructions (Diewald 2006; Himmelmann 2004). Whereas GT is nowadays primarily concerned with the triggering function of the context as well as the resulting combinatorial restrictions, the grammaticalized item imposes on its host class, our analysis highlights two things: (1) the conceptually motivated character of combinatorial ‘restrictions’ and (2) the constraints the host class and the discourse topic, as well as the morphosyntactic and pragmatic context, in turn continue to impose on the grammaticalizing item.

9.2 Theoretical implications for the notion of analogy In the diachronic case-studies, I have argued extensively that the mechanisms of lexical persistence and analogical thinking closely interact. As a consequence, the revisited notion of persistence – as a conceptual, gradual, unpredictable and, above all, two-way discourse phenomenon – affects the tradition-

Theoretical implications for the notion of analogy | 483

al understanding of analogy or analogical thinking. In a nutshell, the present section will show that analogy can best be operationalized as a two-dimensional process: a clear distinction is made between paradigmatic analogical thinking (which motivates change) and syntagmatic analogical thinking (which constrains change and interacts with CIP). Further, analogical mechanisms are claimed to be at work both construction-internally as construction-externally.

9.2.1 The concept of analogy in a nutshell Within the framework of GR, analogy is usually considered as a precondition for change. One of the hotly debated topics is its status as a mechanism, motivation or motor in language change (Givón 1991; Fischer 2007; Wanner 2006).243 Basically, it refers to structural or semantic similarity the speaker perceives between a particular (source-)construction and a (target-)construction which invites him to parse the former as an instance of the latter. An often-cited example is morpho-syntactic hypercorrection: e.g. a speaker may use brung rather than brought “because it fits another past tense pattern: rung, stung, etc., which happens to be far more frequent than the pattern of brought.” (Fischer 2011, 34). In historical linguistics, analogy is primarily associated with structural similarity between two or more elements of a paradigm (cf. Traugott/Trousdale 2010, 35). In the GR framework, the explanatory power of analogy gained particularly in strength since Hoffmann’s (2004) analysis of low-frequency complex prepositions: he argues that low-frequency items such as by dint of grammaticalize via analogy to their more frequent structural relatives such as in view of, which are involved in standard GR. Similarly, analogical thinking to ‘conventionalized’ or entrenched modeling constructions, which are either more frequent or more schematic, forms the backbone of diachronic construction grammar approaches picturing the process of GR by distinguishing different constructional levels or degrees of schematicity (Traugott 2007; 2008a; 2008b; Trousdale 2008b). Since reanalysis has also been proved to be a relevant mechanism in language change, there is a long-running, framework-independent debate among historical linguists on which mechanism should be attributed a primary or dominant role. Oversimplifying, reanalysis or rebracketing refers to the abrupt cate-

|| 243 See also Fischer (2008, 350): “analogy should be seen as both a mechanism and a cause (cf. also Itkonen 2005). By means of analogy we may change structures and the contents of paradigmatic sets, but is also analogy that causes the learner to build up more abstract types or schemas.” See also Anttila (2003, 426) who highlights the “gap filling” potential of analogy.

484 | Theoretical reflection: the notion of CIP revisited

gory/meaning shift items undergo when they are assigned a new grammatical category/meaning in ambiguous contexts. It is thus a covert and invisible change. For instance, in un montón de libros ‘a heap of books’, it might be unclear to the hearer whether the speaker meant a specific configuration – with [un montónhead [de libros]PNP] as underlying structure – or a large quantity of books – with [[un montón de]quantifier libroshead] as underlying structure. While reanalysis operates within syntagms, analogy works across syntagms (De Smet 2009, 1728). Arguments usually cited in favor of reanalysis are the ‘looseness’ of analogy (Fischer 2008, 350) and the immanence of (re)analysis in analogy. Traugott/ Trousdale a.o. argue that “there is no change without reanalysis” (2010, 39): although a specific change can be motivated by analogy, the mechanism remains reanalysis or parsing, viz. assigning a specific category to an instance. However, those in favor of analogy as the dominant mechanism rather focus on its ‘proportional’ and ‘gap filling’ nature.244 In view of the gradual character of GR or language change in general, De Smet (2009, 1751) proposes to break down the mechanism of reanalysis into several more fundamental mechanisms which correspond to principles of synchronic grammatical organization (cf. Fischer 2011; 2012).245 While even those in favor of ‘reanalysis’ recognize that ‘analogy’ can be the motivation for change (Traugott/Trousdale 2010), Fischer (2011, 36) more strongly suggests that analogy is a prerequisite for GR: “the driving force next to (syntagmatic) context and frequency, is the availability of a (paradigmatic) category or construction pattern that shows formal and/or semantic similarities so that the new variant may fit the synchronic system of the speakerlistener.” Likewise, De Smet (2014) argues that analogy has a poor reputation as a source of structural innovation. It seems that analogy can at most account for items switching categorial allegiance, but not for the introduction of fundamentally new categories. However, the poor reputation is unjustified. Any analog-

|| 244 Cf. Fischer (2011, 34–35): “The important point about analogical extension is that it occurs proportionally. It doesn’t simply involve the ‘expansion of contexts in which a construction can occur’, ‘adding new peripheral members [e.g. new infinitives, inanimate subjects] to a category [e.g. going-to]’ (Bybee 2003, 158); it happens because, once going to is taken for the Auxcategory, it will follow the behavior of other members of this new category.” See also the quotation of De Smet (2009, 1751) at the end of Chapter 4. 245 Fischer (2011, 40) argues that ‘reanalysis’ is a mechanism that linguists can perceive in studying language changes, “but is not something that speakers actually do. Speakers do not reanalyze, they substitute one pattern holistically for another.” Wanner (2006, 150) states: “Analogy is a weak and insecurely activated bond between terms, yet it is the only force available and is present throughout”.

Theoretical implications for the notion of analogy | 485

ically-licensed addition to an existing category changes that category, to the point that similarity-based generalizations can give rise to wholly new categories. Thus, over time, the piecemeal addition of new members to a category can gradually work substantial changes. (De Smet 2014, 35)

For the reasons pointed out in Chapter 3, I claim that analogical thinking is the dominant mechanism in the development of BQs and that this mechanism interacts with conceptual persistence. In other words, in view of the refinements to the notion of CIP, I advocate a broad interpretation of analogy that not only includes semantic in addition to formal similarities (cf. also Anttila 2003, 426; De Smet 2013, 65; p.c.; Givón 1991, 258), but also acknowledges the possibility of syntagmatic relations to the source construction or, more precisely, recognizes both a syntagmatic and a paradigmatic dimension in analogy (cf. Anttila’s 2003 “Warp and Woof of Cognition”; De Smet 2013; Fischer 2011; Section 9.2.2.1) and incorporates the discourse level as an additional motivation for syntagmatic associations (Section 9.2.2.1).

9.2.2 The notion of analogy revisited 9.2.2.1 The usage-based model of analogy (De Smet 2013) De Smet’s (2013, 64–71) usage-based model of analogy is the starting point of our claim that the basis for analogy or similarity might be situated at the discourse level.

RELATION TO MODEL

BASIS OF SIMILARITY

Syntagmatic Meaning Form

Paradigmatic

Semantic analogy Formal analogy

Paradigmatic analogy

Distribution

Fig. 4: Types of analogy (copied from De Smet 2013, 69)

On the basis of two criteria, viz., the basis of the similarity and the relation to the model, De Smet explicitly distinguishes at least three subtypes of analogy (cf. Figure 4 which is copied in full from De Smet 2013, 69). The basis of analogy

486 | Theoretical reflection: the notion of CIP revisited

can be a similarity in meaning, form or distribution of the source and the target construction. As to the relation to the model, the similarity can hold either between the environments associated to the spreading pattern (syntagmatic relation) or between the spreading pattern and some paradigmatically related pattern (paradigmatic relation). According to De Smet (2013, 65), two major subtypes of analogy are to be distinguished, viz., ‘semantic analogy’ and ‘paradigmatic analogy’. The former is “a mechanism of analogical extension that allows a construction to extend its range of application on the basis of a semantic similarity between the source environment and the target environment” (2013, 65). For instance, at the end of the Late Modern period, the transitive pattern primarily associated with worth has extended to the semantically related adjective worthwhile, while the extension of the intransitive pattern (originally associated with worthwhile) to worth “has been triggered by the appearance of worthwhile in the same environment” (De Smet 2013, 68). Figure 5 aims to visualize the mechanism of syntagmatic analogy: before the extension takes place, pattern a is associated with environment X, while pattern b exclusively occurs in environment Y. Because of the similarity (of any kind) between both source environments Y and X, a may extend to Y and b may extend to X.246 SIMILARITY

Xenvironment _ apattern

X_ a extension

X_ b

Yenvironment _ bpattern

extension

Y_ a Y_ b

Fig. 5: Syntagmatic analogy

|| 246 The notion of syntagmatic analogy might be confusing, since the mechanism of analogy is considered to operate between two members of a paradigm, while the syntagmatic or horizontal dimension is associated with reanalysis in historical linguistics. However, De Smet (p.c.) assumes that analogy indeed always presupposes (and operates between) two members of a paradigm, but the similarity can hold both between the members of the paradigm (paradigmatic analogy) and between the environments these members are associated with (syntagmatic analogy).

Theoretical implications for the notion of analogy | 487

The second major subtype distinguished by De Smet (2013), viz., paradigmatic analogy, “involves the extension of a construction from one environment to another on the basis of a link between the spreading construction and some other paradigmatically related construction” (2013, 68). In other words, the range of occurrence of the former construction is extended by simply copying the distribution of the latter (De Smet 2013, 68). The example of morphosyntactic hypercorrection is an evident illustration. Paradigmatic analogy as defined by De Smet thus corresponds to the strict interpretation of analogy (restricted to structural similarities among members of a paradigm) more commonly used in historical linguistics. Figure 6 aims to visualize the mechanism of paradigmatic analogy: before the extension takes place, pattern a occurs both in environment X and environment Y, while pattern b exclusively occurs in environment X. Because of the similarity (of any kind) between pattern a and pattern b, pattern b may copy the distribution of pattern a and extend to environment Y as well. SIMILARITY

Xenvironment _ apattern

X_ a

Xenvironment _ bpattern

X_ b

extension

Y_ a

Yenvironment _ apattern Y_ b Fig. 6: Paradigmatic analogy

At first sight, the outcome of the extension by syntagmatic and paradigmatic analogy may look similar. However, the processes should not be equated at the theoretical level. De Smet (2013, 71; p.c.) argues that the distinction between syntagmatic and paradigmatic analogy is straightforward and particularly useful in theory, yet not always in practice since syntagmatic and paradigmatic motivations for analogy often co-occur. Likewise, the semantic, formal and distributional subtype of paradigmatic analogy are not distinguished in Figure 4 since “in practice they are mostly indistinguishable” (De Smet 2013, 69). The same observation holds for paradigmatic analogy and syntagmatic distributional analogy.

488 | Theoretical reflection: the notion of CIP revisited

In the following section, we will argue that from a usage-based point of view, the discourse integration of a specific pattern or construction can be a basis for similarity relations as well.

9.2.2.2 Discourse level as an additional motivation for analogy In view of the synchronic characterization of CIP as a two-way discourse phenomenon and De Smet’s (2013, 69) double warning that syntagmatic and paradigmatic motivations for analogy may often co-occur and that distinct subtypes of analogy are often indistinguishable, one has to cautiously handle the more traditional view that a specific source-construction grammaticalized via analogy to a target-construction, as if the accommodation would be unidirectional. I therefore propose to distinguish between two axes along with analogy and CIP interact in the functional organization and history of BQs and to take the discourse level into account as an additional basis for similarities. Paradigmatically, analogical thinking plays a crucial role in the extension of the N1- and the N2-paradigm. On the one hand, the quantifying reading associated to the binominal pattern extends from montón de ‘heap of’ – which appears to be the first QN to have developed a systematic quantifying use – to other QNs. On the other hand, individual QNs are shown to gradually extend their combinatorial pattern to new N2s. Crucially, in both cases of paradigmatic formal analogy (to new QNs, respectively new N2s), semantic (paradigmatic) motivations are at stake since the motivation for developing quantifying uses, and increasing the range of nominals over time, is rooted in the conceptual relation the noun bears to other nouns belonging to the same or a closely related cognitive domain (cf. the description of the lexical domain of QNs in Section 5.3.1 and the clustered structure of the N2-pattern per QN in Section 5.3.2). In other words, beyond structural analogy, local reanalysis is essentially driven by similarity relations that are conceptual in nature: the creative potential of the [N1 de N2] construction motivated by paradigmatic analogy is limited by conceptual similarity or contiguity relations. For instance, the extension of the BQ to new N2s will not be accepted if the new N2-combination is not somehow compatible with the QN’s original frame: new N2-collocates with aluvión de will either conform to one of the existing N2-clusters (e.g. ‘invader’, ‘reaction’, ‘discourse’) or highlight another facet of aluvión’s original frame. In other words, paradigmatically speaking, analogical thinking and conceptual persistence thus closely interact. Syntagmatically, analogical thinking refers to the conceptual similarities or mutual accommodation (i) between the QN and the N2 (construction-internal analogy) and (ii) between the BQ-construction and the surrounding discourse-

Theoretical implications for the notion of analogy | 489

pragmatic context (construction-external analogy). Synchronically, syntagmatic analogy or similarity relations between N1 and N2/the wider context can be seen as the driving force behind the choice of a particular QN. Diachronically, construction-internal syntagmatic analogy ensures the conceptual motivation or frame-based character of the N2-clusters. BQs are thus another linguistic domain where syntagmatic and paradigmatic motivations for analogy co-occur. The diachronic importance of construction-external analogy is less obvious. However, in building the refined constructional model, I have emphasized that the pragmatic inferences of quantity were at first always contextually embedded, at least for those QNs which developed a productive quantifying use before the end of the 19th century. In other words, while paradigmatic analogy may have been a motivation for the change into a quantifying expression, the driving force has been the discourse integration of the binominal pattern, i.e., the contextual and pragmatic elements that impose a quantifying inference. Since discourse integration is also the driving force behind the choice of an appropriate QN, the discourse level seems a crucial additional motivation for syntagmatic analogy.247 The interplay and apparent functional overlap between the ‘mechanisms’ of analogy and CIP call for a neat delineation of both phenomena. Basically, analogy endows the binominal syntagm with a creative potential, while conceptual persistence constrains or counterbalances ‘blind’ analogy, both syntagmatically and paradigmatically, and determines its conceptual shape. Since all QNs are considered to impose a particular, frame-based conceptualization on the N2entities, the BQ-construction provides the Spanish native speaker with a useful tool for expressive quantification. The fact that the pragmatic or communicative function of BQs hinges on the conceptual persistence, leads to the hypothesis that QNs are not likely to become completely bleached quantifiers like montón de. In other words, the motivation for change yielded by paradigmatic analogy is counter-balanced or constrained by syntagmatic analogy. With repetition and automation (Bybee 2007), as in the case of montón de, the syntagmatic constraints can eventually become less tight.

|| 247 As Fischer (2008; 2011), De Smet (2009; 2013) have repeatedly stated, the mechanism hypothesized to have caused a particular change gains in explanatory strength if it can be shown to be a deep-seated principle relevant in language processing and learning as well. Note that Itkonen (1994, 50) and Anttila (2003, 428) characterize analogization as an innate faculty or cognitive principle.

490 | Theoretical reflection: the notion of CIP revisited

Diachronic

Synchronic

Paradigmatic

Frame-based realignment

Coercion

Syntagmatic

Frame-based host-class expansion

Mutual accommodation & textual coherence

Fig. 7: Conceptually-driven analogy or the interplay between analogy and CIP

The interaction between CIP and analogy is also called conceptually-driven analogy in Delbecque/Verveckken (2014). Figure 7 provides a simplified sketch of the diachronic and synchronic functioning of the mechanism, both paradigmatically and syntagmatically. Diachronically, paradigmatic conceptuallydriven analogy refers to the frame-based realignment of (new) concrete constructions to the more schematized model (which functions as analogical model or attracting force). Syntagmatic conceptually-driven analogy in change refers to the frame-based (construction internal and external) context expansion. Synchronically, conceptually-driven paradigmatic analogy is known as the phenomenon of coercion, whereby the construction imposes its image schema on the meaning of the lexical item. Conceptually-driven syntagmatic analogy refers to the isosemic relations or mutual accommodation between the grammaticalizing item and its discourse context. To conclude, my case-study on BQs can be read as additional evidence for De Smet’s (2013) claim that in analogy, paradigmatic and semantic motivations often co-occur and can be hard to tell apart. Further, the case-study highlights that the discourse level obviously functions as an additional motivation for analogical thinking. As to the development of BQs, paradigmatic analogy does not occur but motivated by semantic similarity or contiguity relations between the QN and some element(s) of its discourse context. As to the synchronic functional organization of BQs, the very same principle ensures textual coherence.

9.3 Conclusion This chapter set out to redefine the notion of persistence in the light of the diachronic and synchronic case-studies. I have shown that persistence is primarily conceptual in nature. Further, the case-studies have repeatedly illustrated that conceptual persistence is not a matter of all or nothing. On the contrary, it can best be defined as a gradual, unpredictable and two-way discourse phenomenon. Whether or not the distinction between different degrees of persistence will

Conclusion | 491

prove to be relevant for processes of GR in general, this chapter has shown that the interplay between CIP and analogy is not exclusive for GR, yet also applies to text organizing functions. In addition, my interpretation of CIP plausibly motivates those cases where GR ‘hesitates’ (De Smet/Ghesquière 2010). In view of the characterization of persistence as a two-way discourse phenomenon, I illustrated the usefulness of distinguishing between paradigmatic and syntagmatic analogy and added the discourse level as an additional motivation for analogical thinking. The claim that paradigmatic analogy enables constructional extension and is constrained by syntagmatic analogy is probably not restricted to changes that involve GR. Intuitively, I believe that the two-way functioning of analogy applies both to purely syntactic or purely semantic changes as to the emergence of new constructions. Since I am also reluctant to accept that speakers abruptly change their grammar or lexicon (see also De Smet 2009, 1751), I believe syntagmatic analogy with one element of the textual or discourse-pragmatic context can usually be observed, at least initially, regardless of whether this conceptual affinity eventually disappears.

| Part 4 Conclusions

10 Conclusion This investigation set out as a cognitive-functional analysis of QNs within the particular context of BQ-constructions. The research interest has been situated both in the development (diachrony) and the functional organization of BQs (synchrony). The analysis has been corpus-based and has included a dozen of selected QNs. The scrutiny of the dataset enabled me to propose, on the one hand, an alternative view on the boundaries between partitive, pseudopartitive and expressive binominal constructions, and, on the other hand, to contribute to theorizing on semantic change in GR, and to refine the notions of persistence and analogy in particular. In the present section, I will first line up the main tendencies observed and highlight once more the theoretical implications of these findings (Section 10.1). Since any empirical study includes weaknesses and leaves room for further investigation, I will conclude by indicating the issues most prominently calling for further elaboration (Section 10.2).

10.1 Concluding remarks The starting point of this investigation has been the claim that BQs engage in a GR process. This change results in two functional uses, in addition to the original lexical use that most often continues to coexist. In the quantifying use, the QN assesses the quantity of N2. In the (two-way) specifying use, the QN specifies the type of N2, and as a whole, the BQ categorizes a third entity. The point of departure of the case-studies has been presented in Chapters 2 and 3. The synchronic layering of uses can only be regarded as the outcome of GR if the functional uses are indeed always attested later than the lexical uses. In the literature on the BQ-construction in English (Brems 2011; Traugott 2008b), a fixed pathway of changes is proposed: (1) QNs emerge outside the context of the BQ-construction, i.e. they are first used as uncomplemented nouns (e.g. un montón ‘a heap’); (2) in a second step, they can be used literally within the BQconstruction (e.g. un montón de tierra ‘a heap of earth’); (3) via desemanticization, they can be used as a quantifier to N2 (e.g. un montón de palabras ‘a heap of words’); (4) after having acquired quantifying uses, QNs can start developing evaluative uses (e.g. no era más que un montón de piedras ‘it was no more than (just) a heap of stones’); and finally, (5) QNs may come to be

496 | Conclusion

used as free adjuncts (e.g. Sabe un montón ‘He knows a lot, he is very wise’).248 The validity of this pathway has been assessed against the case-studies. Chapter 4 presents a fine-grained analysis of the development of montón de, which is commonly considered to be the most general, typical and unrestricted QN in Spanish to the extent of competing with the absolute quantifier mucho/a(s). The pathway has shown to be valid for montón de, yet it generalizes over the cascade of pragmatic inferences and micro-changes that have led to the development of the functional uses. As shown in Chapter 3, in Present-Day Spanish, the head, the quantifying and the specifying uses display a stable collocational shape. Chapter 5 shows that the proposed GR cline does not apply to all QNs: some QNs stop halfway the cline, some do not start from a productive head use, some QNs do not even occur literally and some QNs simply invert the directionality of the cline. In view of the semantically clustered structure of the lexical field of QNs – which also remains stable over time – as well as the QN-related semantic clustering of the N2-collocates, I have proposed an alternative view on the ‘mechanisms’ of analogy and persistence: the BQ-construction is a locus of continuous interaction between analogical thinking and conceptual persistence. The diachronic case-studies point out that the GR of BQs is not a straightforward case of GR. Chapter 6 observes that, in addition to the tendency towards formal and conceptual persistence, BQ-constructions face us with an irregular frequency pattern which hosts high-frequency items, regularfrequency items and low-frequency items. More precisely, the causal relation between high frequency, desemanticization and GR only applies to two QNs, viz. montón de and mar de. A recent trend in GT frames the debate of the GR of low-frequency items in terms of analogy towards high-frequency items and, by extension, towards abstract constructions. I have elaborated a constructional network that accurately models both the unique and the partially shared pathways of change. I have drawn attention to the importance of carefully defining the analogical model which micro-constructions align to: I have shown that there is no need to invoke the influence of morphologically unrelated absolute quantifiers, nor to declare the existence of a macro-constructional model before the second half of the 19th century. The most schematic or macro-constructional level does not relate the binominal pattern to quantity assessment, yet to a continuum between size and type specification. More precisely, binominal quality

|| 248 Brems (2011) differentiates between quantifier and evaluative uses, while Traugott (2008b) describes both steps 3 and 4 as degree modifier uses.

Concluding remarks | 497

and quantity assessment builds on the coextensiveness of N1 and N2 and on the corresponding comparison construal. From the diachronic case-studies it appears that formal changes are conceptually driven. At stake in the synchronic case-studies is the conceptually motivated nature of the QN’s co-selection pattern. The contrastive semantic analysis in Chapter 7 has pictured the pragmatic utility of BQs in terms of the specific QN-related conceptual images. The GR of QNs is claimed to give way to a gradual abstraction of the QN’s source semantics. In the conceptual image each QN imposes on N2, up to three levels of schematicization can be observed. The conceptual images usually comprise a set of conceptual facets. In concrete occurrences, one or more specific facets are profiled in function of the precise discourse-integration of the BQ. The QN-related conceptual images obviously affect the combinatorial pattern of QNs, and reversely, specific contextual situations can invite the use of a particular QN. Chapter 8 focuses on the morphosyntactic behavior of the construction. The paradigmatic variation in four constructional slots – viz. the determiner, the premodifier, the N2 and, construction-externally, verb agreement – is shown to be conceptually driven and discourse-pragmatically motivated. To conclude, Chapter 9 is an attempt to fit the synchronic findings into the diachronic tendencies, and by doing so, proposes some refinements to the traditional views on persistence and analogy. More precisely, persistence is characterized as primarily conceptual in nature. Recall that formal persistence has been shown to be conceptually motivated. Further, conceptual persistence is claimed to be an unpredictable, gradual and two-way discourse phenomenon: N1 and N2 (and by extension the wider context) mutually accommodate to each other. On the one hand, N2 is the head of the construction and N1 is selected in function of the specific profile the speaker wishes to convey of N2. On the other hand, by imposing its conceptual image on N2, the QN automatically construes N2 as a multiplex referent compatible with (part of) the QN’s source frame. Provided that conceptual persistence and analogy closely interact in the GR of the BQ-construction, the reassessment of the notion of persistence entails possible modifications for the traditional view on analogy as well. More precisely, I have proposed to differentiate between paradigmatic and syntagmatic analogy, both from a synchronic and a diachronic perspective. Perhaps more thoughtprovoking is the claim that analogy endows the BQ-construction with a creative potential while conceptual persistence counterbalances blind structural analogy. The interplay between both lies at the center of the construction’s pragmatic value and the mechanism is called conceptually-driven analogy. In sum, the most valuable contributions of this study reside in the elaboration of a cognitive-functional model of the Spanish BQ-construction, i.e. a con-

498 | Conclusion

struction that is traditionally only dealt with in the margin of other research topics. I have focused on the construal operations activated by BQ-constructions and proposed an alternative view on the fuzzy borderlines between BQ-constructions and the so-called emphatic appositions or expressive binominal constructions. Further, in line with recent trends in theorizing on GR, the casestudies have brought about some striking observations: (1) it has been shown that the input to GR must not be semantically general (neither does the output); (2) functional shifts or reanalysis are not necessarily preceded by long periods of ambiguous uses (cf. De Smet 2009, recall that this has not even been the case for montón, which, among all QNs, most closely resembles genuine GR); (3) collocational restrictions do not trigger GR but rather constrain GR (cf. Ghesquière 2011; 2014); (4) I have toned down the modeling force of structurally unrelated analogical models and elaborated a constructional network model; (5) I have proposed an integrated approach of persistence and analogy which are traditionally only ascribed secondary status in semantic changes. In my view, the conceptually-driven analogy plays a major role in the internal and external expansion of the construction and explains why GR, or more carefully, desemanticization and context-expansion, are sometimes “on hold”.

10.2 Further research The present section points out some interesting research topics which remain untouched at the end of this investigation. They relate to the phenomena discussed, to the crosslinguistic generalization of the claims made and to possible further refinements of the frameworks subscribed to and to the semantics of quantification in general. It goes without saying that an expansion of the dataset would enable me to further refine the functional model of the BQ-construction in Spanish. First, it would be interesting to also include those QNs which designate abstract quantity notions yet also give way to subjectified uses, such as cantidad ‘quantity’, par ‘pair, couple’, mayoría ‘majority’, etc. Although the phenomenon is rare, the inclusion of QNs which preferably combine with the definite determiner or no determiner at all, e.g. la tira de ‘a lot of (lit. the strip of)’ (1) and cantidad ‘quantity’ (2), would shed light on the conceptual import of the determiner-slot to the conceptualization of the BQ-construction as a whole. The expanded dataset should also include those QNs which primarily depict small quantities, such as una pizca de ‘a pinch of, a little bit of’, un poquitín de ‘a little bit of’, una gota de ‘a drop of’, etc. Interestingly, there are usually unfrequent, some are restricted to contexts of negation (as in ni pizca de gracia ‘not a flicker (lit. pinch) of

Further research | 499

grace’) and some even lack binominal occurrences or are only marginally used with a PNP (no saber ni jota ‘not to know a thing (lit. jot)’, no le gusta ni pizca ‘he doesn’t like it at all (lit. not a pinch)’).249 Finally, a detailed analysis of the quantifying potential of expressive nouns and general type nouns (e.g. (un) tipo de ‘type of’, (una) suerte de ‘sort of’, (un) género de ‘genre of’, (una) especie de ‘sort of’) should be part of the extended dataset. Interestingly, especially the strings toda suerte de ‘all sort of’ and sin ningún género de ‘without any kind of’ seem to have developed productive quantifying uses. (1)

-De enfermedad, me han dicho siempre. Nunca he preguntado mucho. A ver, qué es lo que sabes. -A lo mejor no es verdad. Hace la tira de años que me lo contó mi padre. (1992, novel) ‘-Of illness, I was always told. I have never asked (too) much (about it). Let’s see, what is it that you know? -May be it is not true. It has been a lot of (lit. the strip of) years ago that my father told me.’

(2)

BERNARDO. – Es que ahora hay unas sectas que obligan a sus socios a reprimirse. ¡Uf!, cantidad de gente anda reprimida. PALOMA. – Claro. Por eso hay cantidad de gente desgraciada. (1990, theatre) ‘B. – The thing is that nowadays there are some sects that force their members to restrain themselves. Oof, a lot of (lit. Ø quantity of) people is repressed. P. – Of course. That’s why there is lot of (lit. Ø quantity of) unhappy people.’

Another interesting research line concerns the regiolectal and genre-related variation in the QN-paradigm. I have been told by native speakers that zarpada de ‘pounce of’ and manga de ‘bunch of’ are particularly productive in spoken Argentinian Spanish. Further, in some informal varieties of peninsular Spanish mazo ‘pack’ and taco ‘plug, mess’ as such can intensify adjectives, i.e. without determiner or linking preposition as in es mazo simpático ‘he is really/extremely nice’. In addition to extending the dataset onomasiologically, the analysis of the quantifying uses of QNs outside the BQ-construction is a promising line of research as well. Apparently, QNs do not need to be completed by N2s incompati|| 249 The latter observations might have an easy cognitive explanation: if the quantity depicted by the QN is already very small, it makes little sense to specify the constituting entities.

500 | Conclusion

ble with the literal reading to force the quantifying construal. In (3) and (4), one might wonder whether one is simply dealing with metaphorical exploitations of the QNs. One could argue that in (3) and (4), the multiplex referent quantified by the QN is implicit in the previous discourse respectively the postmodifying adjective. (3)

RAQUEL Pues la verdad es que no... INES (Al quite) ...es que no hemos parado, ¿verdad, Teresa? TERESA ¡Un aluvión! Violaciones, droga, abandonos de hogar... (1992, theatre) ‘R.– Well, the truth is that we did not… I. – …that we didn’t stop, isn’t it Teresa? T.– A flood! Violations, drugs, desertions…’

(4)

Hace ya mucho que nos encontramos expuestos al aluvión informativo, ese que impide procesar las noticias forzándonos a cumplir un triste papel receptivo (…). (1997, press) ‘A long time ago already we found ourselves exposed to the news flood (lit. informative flood), the one that prevents us from processing the news and forces us to fulfill a sad receptive role, (…).’

Further, spot checks in CREA have shown that at least two other constructional patterns are the locus of a quantifying exploitation of QNs, viz. on the one hand [a + QNplural] (e.g. a montones ‘lit. in heaps’, a mares ‘lit. in seas’, a espuertas ‘lit. in baskets’, etc. as in ganar dinero a espuertas ‘to earn lots of money’), and [QNplural y QNplural (de)] on the other (e.g. montones y montones ‘heaps and heaps’, pilas y pilas de manuscritos ‘piles and piles of manuscripts’). The reduced generalizability of these constructional patterns to other QNs raises the question whether these constructions emerge via GR or rather via lexicalization. However, if it can indeed be illustrated that specific QNs develop quantifying uses in linguistic contexts other than the BQ-construction, the crucial ‘triggering’ role traditionally ascribed to the specific context of a grammaticalizing item needs to be toned down. A complementary research line in the study of the semantics and pragmatics of QNs would be to analyze the type of metaphorizations typically evoked. The present investigation has enabled me to argue that specific N2s (or clusters of semantically related N2s) can preferably combine with specific N1-frames. The next step would be to investigate which metaphors are privileged and which factors determine the choice for a specific image. Blanco (2002) for instance observes five productive domains of metaphoric QNs: meteorology, geography, religion, military and architecture. First of all, I expect stereotypical

Further research | 501

QNs to proliferate in oral data. Second, intuitively, I assume that the metaphors evoked in BQs are every-day concepts and reflect the importance of specific culturally entrenched frames. For instance, while Spanish rosario ‘rosary’ is highly frequent, Dutch rozenkrans is marked, which may suggest that catholicity is more pervasive in the every-day life of Spanish speakers. Further, there seems to be a partial overlap with productive metaphors in verbal quantification (as in los problemas se amontonan ‘the problems are heaping up’, las críticas llueven de todas partes ‘criticisms are raining from everywhere’, nos bombardean con informaciones ‘they bombard us with information’, etc.). Finally, the alternative view I have proposed on the functional overlap between BQs and expressive binominal constructions would seriously gain in strength if it proofs to be crosslinguistically valid. The comparison to Germanic languages confronts us with an interesting challenge: both in English and in Dutch (and probably in other Germanic languages), the second constituent of expressive binominal constructions contains an indefinite determiner, as in a wonder of a woman or een wonder van een vrouw ‘a wonder of a woman’. This observation raises the question whether the comparison construal between N1 and N2 can be generalized to these languages. At first sight, since a and een in the above examples do not single out a specific individual but refer to the category of women as such, the hypothesis need not be immediately discarded. Equally interesting is the finding that the Dutch structural equivalent of the BQconstruction allows variation in the prepositional slot, viz. van, aan or no preposition, as in een stortvloed van kritiek ‘a flood of criticism’, een stortvloed aan informatie ‘a flood of (lit. to/at) information’, een stortvloed Ø actrices ‘a flood of (lit. Ø) actresses’, as well as compounds (een kaloriebom ‘a bomb of calories (lit. calorie bomb)’). From a cognitive-structural perspective, I tentatively venture that the constructional variants yield slightly different conceptualizations of the coextensive relationship involved. Likewise, it would be equally interesting to extend the onomasiological analysis to other languages and to verify whether the metaphorizations typically used in quantification correspond crosslinguistically. To conclude, the reassessment of persistence as a two-way analogical discourse pressure confronts us with a theoretical challenge, viz. the implementation of the (constraining role of the) discourse-component in theorizing on GR. So far, the role of discourse in GR has been confined to ‘pragmatic inferences’ which trigger functional readings in highly specific contexts. To my knowledge, the impact of the mutual accommodation between the context and the selection of the grammaticalized element has not yet been investigated in GT.

References Aarts, Bas, Binominal noun phrases in English, Transactions of the Philological Society 96:1 (1998), 117–158. Acuña-Fariña, Juan Carlos, The linguistics and psycholinguistics of agreement: a tutorial overview, Lingua 119 (2009), 389–424. Alarcos Llorach, Emilio, Estudios de gramática funcional del español, Madrid, Gredos, 31999. Alarcos Llorach, Emilio, Gramática de la lengua española, Madrid, Espasa-Calpe, 1995. Alcina Franch, Juan/Blecua, José Manuel, Gramática española, Barcelona, Ariel, 102001. Alonso, Amado/Henríquez Ureña, Pedro, Gramática castellana, Buenos Aires, El Ateneo, 1938. Alonso-Cortés, Ángel, Las construcciones exclamativas. La interjección y las expresiones vocativas, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática descriptiva de la lengua española, cap. 62, vol. 3, Madrid, Espasa-Calpe, 1999, 3993–4050. Andersen, Henning, Actualization and the (uni)directionality, in: Andersen, Henning (ed.), Actualization: Linguistic Change in Progress, Amsterdam/Philadelphia, Benjamins, 2001. Anttila, Raimo, Historical and comparative linguistics, Amsterdam, Benjamins, 1989. Anttila, Raimo, Analogy: the warp and woof of cognition, in: Joseph, Brian D./Janda, Richard D. (edd.), The handbook of historical linguistics, Oxford, Blackwell, 2003, 425–440. Athanasiadou, Angeliki/Canakis, Costas/Cornillie, Bert, Introduction to “Subjectification. Various Paths to Subjectivity”, in: Athanasiadou, Angeliki/Canakis, Costas/Cornillie, Bert (edd.), Subjectification. Various Paths to Subjectivity, Berlin/New York, Mouton de Gruyter, 2006, 1–13. Bach, Emmon, The algebra of events, Linguistics and Philosophy 9 (1986), 5–16. Bassols de Climent, Mariano, Sintaxis latina, Madrid, Consejo Superior de Investigaciones Científicas, 101992. Battus, Hugo, Opperlans! Taal-&letterkunde, Amsterdam, Querido, 2003. Beckwith, Christopher I., Phoronyms: classifiers, class nouns and the pseudopartitive construction, New York, Lang, 2007. Bello, Andrés, Gramática de la lengua castellana destinada al uso de los americanos, Notas de Rufino José Cuervo, Estudio crítico y edición de Ramón Trujillo, Madrid, Arco/Libros, 1988 (11847). Benninger, Céline, De la quantité aux substantifs quantificateurs, Paris, Klincksieck, 1999. Benveniste, Émile, De la subjectivité dans le langage, in: Benveniste, Émile (ed.), Problèmes de linguistique générale, Paris, Gallimard, 1966, 258–266. Blanco, Xavier, Les determinants figés, Langages 36:145 (2002), 61–81. Bock, Kathryn/Nicol, Janet/Cutting, J. Cooper, The ties that bind: creating number agreement in speech, Journal of Memory and Language 40 (1999), 330–346. Bock, Kathryn, et al., Number agreement in British and American English: disagreeing to agree collectively, Language 82:1 (2006), 64–113. Bosque, Ignacio, El nombre común, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática descriptiva de la lengua española, cap. 1, vol. 1, Madrid, Espasa-Calpe, 1999, 3–75. Bosque, Ignacio (ed.), REDES, Diccionario combinatorio del español contemporáneo, Madrid, Ediciones SM, 2004. Bosque, Ignacio, Procesos de abstracción en los paradigmas léxicos abiertos, Pandora: revue d’études hispaniques 7 (2007), 189–198.

504 | References

Bosque, Ignacio/Gutiérrez-Rexach, Javier, Fundamentos de sintaxis formal, Madrid, Akal, 2009. Boye, Kasper/Harder, Peter, A usage-based theory of grammatical status and grammaticalization, Language 88:1 (2012), 1–44. Brems, Lieselotte, Measure noun constructions: degrees of delexicalization and grammaticalization, in: Aijmer, Karin/Altenberg, Bengt (edd.), Advances in Corpus Linguistics: papers from the 22nd International Conference on English Language Research on Computerized Corpora, ICAME, Göteborg, 22–26 of May 2002, Amsterdam, Rodopi, 2004, 249–265. Brems, Lieselotte, The grammaticalization of small size nouns. Reconsidering frequency and analogy, Journal of English Linguistics 35:4 (2007), 293–324 (=2007a). Brems, Lieselotte, The synchronic layering of size noun and type noun constructions in English, Unpublished PhD thesis, KU Leuven, 2007 (=2007b). Brems, Lieselotte, Size Noun constructions as collocationally constrained constructions: lexical and grammaticalized uses, English Language and Linguistics 14:1 (2010), 83–109. Brems, Lieselotte, The layering of size noun and type noun constructions in English, Berlin, Mouton de Gruyter, 2011. Brinton, Laurel J./Traugott, Elisabeth Closs, Lexicalization and Language Change, Cambridge, Cambridge University Press, 2005. Brisard, Frank, Logic, subjectivity, and the semantics/pragmatics distinction, in: Athanasiadou, Angeliki/Canakis, Costas/Cornillie, Bert (edd.), Subjectification. Various Paths to Subjectivity, Berlin/New York, Mouton de Gruyter, 2006, 41–74. Brucart, José Maria, Concordancia ad sensum y partitividad en español, in: Almeida, Manuel/Dorta, Josefa (edd.), Contribuciones al estudio de la lingüística hispánica. Homenaje al profesor Ramón Trujillo, vol. 1, Tenerife, Montesinos, 1997, 157–183. Bublitz, Wolfram, Semantic prosody and cohesive company: “somewhat predictable”, Leuven Contributions in Linguistics and Philology 85 (1996), 1–32. Butt, John/Benjamin, Carmen, A new reference grammar of modern Spanish, London, Arnold, 1993. Bybee, Joan, Mechanisms of change in grammaticization: the role of frequency, in: Joseph, Brian D./Janda, Richard D. (edd.), The Handbook of Historical Linguistics, Oxford, Blackwell, 2003, 602–623. Bybee, Joan, From usage to grammar: the mind’s response to repetition, Language 82:4 (2006), 711–733. Bybee, Joan, Diachronic linguistics, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 945–987. Bybee, Joan/Eddington, David, A usage-based approach to Spanish verbs of “becoming”, Language 82:2 (2006), 323–355. Bybee, Joan/Perkins, Revere/Pagliuca, William, The evolution of grammar: tense, aspect and modality in the languages of the world, Chicago, University of Chicago Press, 1994. Cano Aguilar, Rafael/Echenique Elizondo, M. Teresa (edd.), Rafael Lapesa. Estudios de morfosintaxis histórica del español, vol. 1, Madrid, Gredos, 2000. Campbell, Lyle, What’s wrong with grammaticalization?, Language Sciences 23:2–3 (2001), 113–161. Campbell, Lyle/Janda, Richard, Introduction: conceptions of grammaticalization and their problems, Language Sciences 23 (2001), 93–112. Carlier, Anne/Goyens, Michèle/Lamiroy, Béatrice, “De”: a genitive marker in French? Its grammaticalization path from Latin to French, in: Carlier, Anne/Verstraete, Jean-Christophe

References | 505

(edd.), The Genitive: Case and gramatical Relations across Languages, Amsterdam, Benjamins, 2013, 141–216. Company Company, Concepción (ed.), Sintaxis histórica de la lengua española. Segunda parte: La frase nominal, 2 vol., México, Universidad nacional autónoma de México/Fondo de cultura económica, 2009. Corbett, Greville G., Agreement, Cambridge, Cambridge University Press, 2006. Corominas, Joan/Pascual, José Antonio, Diccionario crítico etimológico castellano e hispánico, Madrid, Gredos, 1991. Cornillie, Bert, Diachronic approaches to interaction. Interactional approaches to historical data. Challenges and pitfalls, Paper presented at the FEST research unit reunion, 11 March, Leuven, 2001. Croft, William, Radical construction grammar: Syntactic theory in typological perspective, Oxford, Oxford University Press, 2001. Croft, William, Construction grammar, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 463–508. Croft, William/Cruse, Alan, Cognitive Linguistics, Cambridge, Cambridge University Press, 2004. Cuenca, Maria Josep/Hilferty, Joseph, Introducción a la lingüística cognitiva, Barcelona, Ariel, 1999. Cuyckens, Hubert, Grammaticalization and the history of English, Leuven, Course presented at the LOT-school, June 2007. Database of Latin Dictionaries, Online database, Turnhout, Brepols, 2010 [last access: 21.09.2011]. Davies, Mark, Corpus del Español: 100 million words, 1200s–1900s, Online: http://www.corpusdelespañol.org, 2002 [accessed from February 2010 until December 2011]. Davidse, Kristin, The interaction of identification and quantification in English determiners, in: Achard, Michel/Kemmer, Suzanne (edd.), Language, Culture and Mind, Stanford, CSLI Publications, 2004, 507–533. De Bruyne, Jacques, Le suffixe “-ón” en espagnol moderne, Linguistica Antverpiensia 13 (1979), 7–53. De Cock, Barbara, A discourse-functional analysis of speech participant profiling in spoken Spanish, Unpublished PhD thesis, KU Leuven, 2010. De Cock, Barbara, Profiling Discourse Participants. Forms and functions in Spanish conversation and debates, Amsterdam, Benjamins, 2014. Delbecque, Nicole, Word order as a reflection of alternate conceptual construals in French and Spanish. Similarities and divergences in adjective position, Cognitive Linguistics 1:4 (1990), 349–416. Delbecque, Nicole, A construction grammar approach to transitivity in Spanish, in: Davidse, Kristin/Lamiroy, Béatrice, Case and Grammatical Relations across Languages, The Nominative & Accusative and their counterparts, vol. 4, Amsterdam/Philadelphia, Benjamins, 2002, 81–130 (=2002a). Delbecque, Nicole (ed.), Linguistique cognitive. Comprendre comment fonctionne le langage, Bruxelles, De Boeck-Duculot, 2002 (=2002b). Delbecque, Nicole, Los usos no situacionales de los demostrativos españoles: enfoque cognitivo, in: Sinner, Carsten/Wotjak, Gerd/Hernández Socas, Elia (edd.), Estudios de tiempo y espacio en la gramática española, Bern, Lang, 2011, 155–174.

506 | References

Delbecque, Nicole, “En el fondo”: polifuncionalidad y polifonía de la localización interna, in: Jiménez Juliá, Tomás, et al. (edd.), Cum corde et in nova grammatica. Estudios ofrecidos a Guillermo Rojo, Santiago de Compostela, Universidade de Santiago de Compostela, Servizo de Publicacións e Intercambio Científico, 2012, 247–263. Delbecque, Nicole/Cornillie, Bert (edd.), On Interpreting Construction Schemas. From Action and Motion to Transitivity and Causality, Berlin/New York, Mouton de Gruyter, 2007. Delbecque, Nicole/Verveckken, Katrien, La grille événementielle de mouvement et l’aspectualité. Le cas des verbes espagnols “entrar” et “salir”, in: Iliescu, Maria/SillerRunggaldier, Heidi M./Danler, Paul (edd.), Actes du XXVe Congrès International de Linguistique et de Philologie Romanes, Innsbruck, 3–8 septembre 2007, vol. 3, Berlin, Mouton de Gruyter, 2010, 69–84. Delbecque, Nicole/Verveckken, Katrien, Conceptually-driven analogy in the grammaticalization of Spanish binominal quantifiers, Linguistics 52:3 (2014), 637–684. De Mulder, Walter/Lamiroy, Béatrice, Gradualness of grammaticalization in Romance. The position of French, Spanish and Italian, in: Davidse, Kristin, et al. (edd.), Grammaticalization and Language Change: New reflections, Amsterdam, Benjamins, 2012, 199–226. Den Dikken, Marcel, Relators and Linkers. The Syntax of Predication, Predicate Inversion, and Copulas, Cambridge, MIT Press, 2006. De Smet, Hendrik, Analysing reanalysis, Lingua 119 (2009), 1728–1755. De Smet, Hendrik, The long way to normalcy: regularization in English “dare”, Paper presented at the FEST research unit reunion, 18 November, Leuven, 2011. De Smet, Hendrik, Spreading patterns: diffusional change in the English system of complementation, Oxford, Oxford University Press, 2013. De Smet, Hendrik, Does innovation need reanalysis?, in: Coussé, Evie/von Mengden, Ferdinand (edd.), Usage-Based Approaches to Language Change, Amsterdam/Philadelphia, Benjamins 2014, 23–48. De Smet, Hendrik/Ghesquière, Lobke, When grammaticalization hesitates, Paper presented at GRAMIS [Grammaticalization & (Inter)Subjectification], University of Brussels, 11–13 November, 2010. De Smet, Hendrik/Verstraete, Jean-Christophe, Coming into terms with subjectivity, Cognitive Linguistics 17:3 (2006), 365–392. Diewald, Gabriele, Context types in grammaticalization as constructions, Constructions SV19/2006, 2006. Dirven, René/Verspoor, Marjolein (edd.), Cognitive Exploration of Language and Linguistics, Amsterdam/Philadelphia, Benjamins, 1998. Di Tullio, Ángela/Kornfeld, Laura, Gramaticalización y cuantificadores en el habla coloquial del español rioplatense, Paper presented at the XV Congreso Internacional de ALFAL [Asociación de Lingüística y Filología de América Latina], Montevideo, Uruguay, 18–21 August, 2008. Doetjes, Jenny/Rooryck, Johan, Generalizing over quantity and quality constructions, in: Coene, Martine/D’Hulst, Yves (edd.), From DP to DP, The syntax and semantics of noun phrases, vol. 1, Amsterdam, Benjamins, 2003, 277–296. Eberenz, Rolf, La periodización de la historia morfosintáctica del español: propuestas y aportaciones recientes, Cahiers d’études hispaniques médiévales 32 (2009), 181–201. Eberhard, Kathleen M./Cutting, Cooper J./Bock, Kathryn, Making syntax of sense: number agreement in sentence production, Psychological Review 112:3 (2005), 531–559.

References | 507

Faarlund, Jan Terje, Pragmatics and syntactic change, in: Breivik, Leiv Egil/Jahr, Ernst Hakon (edd.), Language Change: Contribution to the Study of Its Causes, Trends in Linguistics 43, Berlin/New York, Mouton de Gruyter, 1989, 71–98. Fält, Gunnar, Tres problemas de concordancia verbal en el español moderno, Uppsala, Almqvist & Wiksell, 1972. Fauconnier, Gilles, Mental spaces: Aspects of meaning construction in natural language, Cambridge, MIT Press, 1985. Fauconnier, Gilles/Turner, Mark, Conceptual integration networks, Cognitive Science 22 (1998), 133–187. Fernández Leborans, María Jesús, El nombre propio, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 2, vol. 1, Madrid, Espasa-Calpe, 1999, 77–128 Fernández Ramírez, Salvador, Gramática española, 3.1. El nombre (Edición de José Polo), Madrid, Arco-Libros, 1986. Fillmore, Charles J., An alternative to checklist theories of meaning, Berkeley Linguistics Society 1 (1975), 123–131. Fillmore, Charles J., Frames and the semantics of understanding, Quaderni di Semantica 6:2 (1985), 222–254. Fillmore, Charles J., Double-decker definitions: the role of frames in meaning explanations, Sign Language Studies 3:3 (2003), 263–295. Fillmore, Charles J./Atkins, Beryl T.S., Towards a frame-based lexicon: The semantics of RISK and its neighbors, in: Lehrer, Adrienne/Kittay, Eva Feder (edd.), Frames, Fields and Contrasts: new essays in semantics and lexical organization, Hillsdale, Erlbaum, 1992, 75–102. Fillmore, Charles J./Kay, Paul/O’Connor, Mary Catherine, Regularity and idiomaticity in grammatical constructions: the case of “let alone”, Language 64 (1988), 501–538. Fischer, Olga, On the role played by iconicity in grammaticalization processes, in: Nänny, Max/Fischer, Olga (edd.), Form miming meaning, Iconicity in language and literature, Amsterdam/Philadelphia, Benjamins, 1999, 345–373. Fischer, Olga, Grammaticalisation: unidirectional, non-reversable? The case of to before the infinitive in English, in: Fischer, Olga/Rosenbach, Anette/Stein, Dieter (edd.), Pathways of change, Grammaticalization in English, Amsterdam/Philadelphia, Benjamins, 2000, 149– 169. Fischer, Olga, Morphosyntactic Change: Functional and Formal Perspectives, Oxford, Oxford University Press, 2007. Fischer, Olga, Grammaticalization as analogically driven change?, in: Narrog, Heiko/Heine, Bernd (edd.), The Oxford handbook of grammaticalization, Oxford, Oxford University Press, 2011, 31–42. Fischer, Olga, On mechanisms of language change: what role does analogy play?, Plenary paper at the 45th annual meeting of the Societas Linguistica Europaea, Stockholm, 29 August–1 September, 2012. García, Érica C, The Motivated Syntax of Arbitrary Signs. Cognitive constraints on Spanish clitic clustering, Amsterdam/Philadelphia, Benjamins, 2009. Geeraerts, Dirk, Where does prototypicality come from, in: Rudzka-Ostzyn, Brygida (ed.), Topics in Cognitive Linguistcs, Amsterdam/Philadelphia, Benjamins, 1988, 207–229. Geeraerts, Dirk, Vagueness’ puzzles, polysemy’s vagaries, Cognitive Linguistics 4 (1993), 223– 272.

508 | References

Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007. Geeraerts, Dirk/Grondelaers, Stefan/Bakema, Peter, The Structure of Lexical Variation, Meaning, Naming, and Context, Berlin/New York, Mouton de Gruyter, 1994. Ghesquière, Lobke, The directionality of (inter)subjectification processes in the English noun phrases: pathways of change, Unpublished PhD thesis, KU Leuven, 2011. Ghesquière, Lobke, The directionality of (inter)subjectification in the English noun phrase, Pathways of change, Berlin/Boston, Mouton de Gruyter, 2014. Gili Gaya, Samuel, Curso superior de la sintaxis española, Barcelona, Biblograf, 1993. Givón, Talmy, On understanding grammar, New York, Academic Press, 1979. Givón, Talmy, The evolution of dependent clause morpho-syntax in Biblical Hebrew, in: Traugott, Elisabeth Closs/Heine, Bernd (edd.), Approaches to Grammaticalization, vol. 2, Amsterdam, Benjamins, 1991, 257–310. Goldberg, Adele, Constructions: a Construction Grammar Approach to Argument Structure, Chicago, Chicago University Press, 1995. Goldberg, Adele, Constructions at work, the nature of generalization in language, Oxford, Oxford University Press, 2006. Goldberg, Adele, Coercion or the Partial Productivity of Constructions, Paper presented at the Third International AFLiCo Conference [Association Française de linguistique cognitive], Grammars in Construction(s), Grammars in Construction(s), University of Paris (Paris Ouest), 27–29 May, 2009 (=2009a). Goldberg, Adele, The nature of generalization in language, Cognitive Linguistics 20:1 (2009), 93–127 (=2009b). Grice, Herbert Paul, Logic and conversation, in: Cole, Peter/Morgan, Jerry (edd.), Syntax and semantics, Speech acts, vol. 3, New York, Academic Press, 1975, 41–58. Gutiérrez Rodríguez, Edita, Rasgos gramaticales de los cuantificadores débiles, Unpublished PhD thesis, Universidad Complutense de Madrid, 2008. Haiman, John, Natural syntax: Iconicity and erosion, Cambridge, Cambridge University Press, 1985. Haiman, John, Ritualization and the development of language, in: Pagliuca, William (ed.), Perspectives on grammaticalization, Amsterdam, Benjamins, 1994, 3–28. Halliday, Michael A.K./Matthiessen, Christian M.I.M., An introduction to functional grammar, London, Arnold, 32004. Harris, Alice C, Cross-linguistic perspectives on syntactic change, in: Joseph, Brian D./Janda, Richard D. (edd.), The handbook of historical linguistics, Oxford, Blackwell, 2003, 529– 551. Haspelmath, Martin, On directionality in language change with particular reference to grammaticalization, in: Fischer, Olga/Norde, Muriel/Perridon, Harry (edd.), Up and down the Cline, the Nature of Grammaticalization, Amsterdam/Philadelphia, Benjamins, 2004, 17– 44. Heine, Bernd, Cognitive Foundations of Grammar, Oxford/New York, Oxford University Press, 1997. Heine, Bernd, On the role of context in grammaticalization, in: Wischer, Ilse/Diewald, Gabriele (edd.), New Reflections on Grammaticalization, Amsterdam/Philadelphia, Benjamins, 2002, 83–101. Heine, Bernd, Grammaticalization, in: Joseph, Brian D./Janda, Richard D. (edd.), The Handbook of Historical Linguistics, Oxford, Blackwell, 2003, 575–601.

References | 509

Heine, Bernd/Claudi, Ulrike/Hünnemeyer, Friederike, Grammaticalization: a conceptual framework, Chicago, University of Chicago Press, 1991. Heine, Bernd/Narrog, Heiko, Grammaticalization and linguistic analysis, in: Heine, Bernd/Narrog, Heiko (edd.), The Oxford Handbook of Linguistic Analysis, Oxford, Oxford University Press, 2010, 401–424. Heine, Bernd/Reh, Mechtild, Grammaticalization and Reanalysis in African Languages, Hamburg, Buske, 1984. Hilpert, Martin, What can synchronic gradience tell us about reanalysis? Verb-first conditionals in written German and Swedisch, in: Traugott, Elisabeth Closs/Trousdale, Graeme (edd.), Gradience, Gradualness and Grammaticalization, Amsterdam/Philadelphia, Benjamins, 2010, 181–202. Himmelmann, Nikolaus P., Lexicalization and grammaticalization: opposite or orthogonal?, in: Bisang, Walter/Himmelmann, Nikolaus P./Wiemer, Björn (edd.), What Makes Grammaticalization: a Look from its Fringes and its Components, Berlin/New York, Mouton de Gruyter, 2004, 19–40. Hoffmann, Sebastian, Are low-frequency complex prepositions grammaticalized? On the limits of corpus data – and the importance of intuition, in: Lindquist, Hans/Mair, Christian (edd.), Corpus Approaches to Grammaticalization in English, Amsterdam/Philadelphia, Benjamins, 2004, 171–210. Hoffmann, Thomas/Trousdale, Graeme, Variation, change and constructions in English, Cognitive Linguistics 22:1 (2011), 1–23. Hopper, Paul J., On some principles of grammaticization, in: Traugott, Elisabeth Closs/Heine, Bernd (edd.), Approaches to Grammaticalization, Focus on theoretical and methodological issues, vol. 1, Amsterdam/Philadelphia, Benjamins, 1991, 17–36. Hopper, Paul/Traugott, Elisabeth Closs, Grammaticalization, Cambridge, Cambridge University Press, 2003. Huelva Unternbäumen, Enrique, Metáfora cognitiva y “blending” en el proceso de gramaticalización de construcciones nominales cuantificativas del español, Metaphorik 11 (2007), 1–43. Israel, Michael, The way constructions grow, in: Goldberg, Adele (ed.), Conceptual Structure, Discourse and Language, Stanford, Center for the Study of Language and Information, 1996, 217–231. Israel, Michael, The pragmatics of polarity, in: Horn, Laurence R./Ward, Gregory (edd.), The Handbook of Pragmatics, Oxford/Malden, Blackwell, 2004, 701–723. Itkonen, Esa, Iconicity, analogy, and universal grammar, Journal of Pragmatics 22 (1994), 37– 53. Kabatek, Johannes/Jakob, Daniel (edd.), Lengua medieval y tradiciones discursivas en la Península Ibérica: descripción gramatical – pragmática histórica – metodología, Frankfurt am Main/Madrid, Vervuert/Iberoamericana, 2002. Kabatek, Johannes, Tradiciones discursivas y cambio lingüístico, Lexis 29:2 (2005), 151–177. Keizer, M. Evelien, The English Noun Phrase: the Nature of Linguistic Categorization, Cambridge, Cambridge University Press, 2007. Keith, Allan, Classifiers, Language 53:2 (1977), 285–311. Kerbrat-Orecchioni, Catherine, De la sémantique lexicale à la sémantique de l’énonciation, Lille, Service de reproduction des thèses, 1979.

510 | References

Koptjevskaja-Tamm, Maria, A woman of sin, a man of duty, and a hell of a mess: nondeterminer genitives in Swedish, in: Plank, Frans (ed.), Noun Phrase Structure in the Languages of Europe, Berlin, Mouton de Gruyter, 2003, 515–558. Koptjevskaja-Tamm, Maria, “A lot of grammar with a good portion of lexicon”: towards a typology of partitive and pseudopartitive nominal constructions, in: Helmbrecht, Johannes, et al. (edd.), Form and Function in Language Research. Papers in Honour of Christian Lehmann, Berlin, Mouton de Gruyter, 2009, 329–346. Kurylowicz, Jerzy, The evolution of grammatical categories, Diogenes 13 (1965), 55–71. Labov, William, Principles of Linguistic Change. Internal Factors, vol. 1, London/New York, Blackwell, 1994. Labov, William, Principles of Linguistic Change. Social Factors, vol. 2, Oxford, Blackwell, 2001. Lakoff, George, Women, Fire, And Dangerous Things, Chicago, University of Chicago Press, 1987. Lakoff, George/Johnson, Mark, Metaphors we live by, Chicago, University of Chicago Press, 1980. Lamiroy, Béatrice, Les verbes de mouvement en français et en espagnol: étude comparée de leurs infinitives, Amsterdam, Benjamins, 1983. Lamiroy, Béatrice, La preposition en français et en espagnol: une question de grammaticalisation?, Langages 143 (2001), 91–105. Langacker, Ronald W., Foundations of cognitive grammar. Theoretical Prerequisites, vol. 1, Stanford, Stanford University Press, 1987. Langacker, Ronald W., Concept, image, and symbol: the cognitive basis of grammar, Berlin, Mouton de Gruyter, 1990. Langacker, Ronald W., Foundations of cognitive grammar. Descriptive application, vol. 2, Stanford, Stanford University Press, 1991. Langacker, Ronald W., Grammar and Conceptualization, Berlin/New York, Mouton De Gruyter, 2000. Langacker, Ronald W., Remarks on Nominal Grounding, Functions of Language 11 (2004), 77– 113. Langacker, Ronald W., Construction Grammars: cognitive, radical and less so, in: Ruiz de Mendoza Ibáñez, Francisco J./Peña Cervel, M. Sandra (edd.), Cognitive Linguistics, internal Dynamics and interdisciplinary Interaction, Berlin/New York, Mouton de Gruyter, 2005, 101–159. Langacker, Ronald W., Subjectification, grammaticization, and conceptual archetypes, in: Athanasiadou, Angeliki/Canakis, Costas/Cornillie, Bert (edd.), Subjectification: Various Paths to Subjectivity, Berlin/New York, Mouton de Gruyter, 2006, 17–40. Langacker, Ronald W., Cognitive Grammar, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 421– 462. Langacker, Ronald W., Cognitive Grammar. A basic Introduction, Oxford, Oxford University Press, 2008. Langacker, Ronald W., A constructional approach to grammaticization, in: Langacker, Ronald W., Investigations in Cognitive Grammar, Cognitive Linguistic Research 42, 60–80, Berlin/New York, Mouton de Gruyter, 2009 (=2009a). Langacker, Ronald W., Cognitive (construction) grammar, Cognitive Linguistics 20:1 (2009), 167–176 (=2009b).

References | 511

Langacker, Ronald W., A lot of quantifiers, in: Rice, Sally A./Newman, John (edd.), Empirical and Experimental Methods in Cognitive/Functional Research, Chicago, Center for the Study of Language and Information, 2011, 41–57. Langacker, Ronald W., Descriptive and Discursive Focusing in Complement Constructions, Paper presented at the CSDL-conference [Conceptual Structure, Discourse and Language], Santa Barbara, 4–6 November, 2014. Lázaro Mora, Fernando A., La derivación apreciativa, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 71, vol. 3, Madrid, EspasaCalpe, 1999, 4645–4682. Lehmann, Christian, Thoughts on grammaticalization: a programmatic sketch, Köln, Universität zu Köln, 1982. Lehmann, Christian, Thoughts on Grammaticalization, München/Newcastle, LINCOM Europa, 1995. Leonetti, Manuel, El artículo, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 12, vol. 1, Madrid, Espasa-Calpe, 1999, 891–928. Lewandowska-Tomaszczyk, Barbara, Polysemy, prototypes, and radial categories, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 139–169. Lightfoot, David W., Principles of Diachronic Syntax, Cambridge Studies in Linguistics 23, Cambridge, Cambridge University Press, 1979. Lipton, James, An Exaltation of Larks, New York, Penguin, 1991. Llyod, A. Kate/Nitti, John J. (edd.), Diccionario de la prosa castellana del Rey Alfonso X, 3 vol., New York, The Hispanic Seminary of Medieval Studies, 2002. Lyons, John, Semantics, 2 vol., Cambridge, Cambridge University Press, 1977. Malchukov, Andrei L., Dependency Reversal in noun-attributive constructions: towards a typology, München, LINCOM, 2000. Maldonado González, Concepción, Clave: diccionario de uso del español actual, Madrid, SM, Online: http://clave.librosvivos.net/, 31999 [last access: 03.06.2015]. Martí Girbau, Núria, The syntax of partitives, Unpublished PhD thesis, University of Barcelona, 2010. Martínez, José Antonio, La concordancia, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 42, vol. 2, Madrid, Espasa-Calpe, 1999, 2695–2786. Massini, Francesca, Binominal constructions in Italian of the N1-di- N2 type: towards a typology of Light Noun Constructions, in: Brems, Lieselotte/De Clerck, Bernard/Verveckken, Katrien (edd.), Binominal syntagms as a neglected locus of synchronic variation and diachronic change: towards a unified approach, Special issue in Language Sciences, Forthcoming. Meillet, Antoine, L’évolution des formes grammaticales, Scientia (Rivista di Scienza) 12:26 (1912) (reprinted in: Meillet, Antoine, Linguistique historique et linguistique générale, Paris, Champion, 1958, 130–148). Michaelis, Laura A., Type shifting in construction grammar: an integrated approach to aspectual coercion, Cognitive Linguistics 15 (2004), 1–67. Michaud Maturana, Daniel Augusto, Uso y cuantificación de “cifra de negocio” y “volumen de negocio”. Un análisis lingüístico cognitivo de cuantificadores y sintagmas de medida, Unpublished PhD thesis, KU Leuven, 2011.

512 | References

Mihatsch, Wiltrud, Wieso ist ein Kollektivum ein Kollektivum, Philologie im Netz 13 (2000), 39– 72, Online: http://web.fu-berlin.de/phin/phin13/p13t3.htm [last access: 10.04.2008]. Mihatsch, Wiltrud, Type-noun binominals in four Romance languages, in: Brems, Lieselotte/De Clerck, Bernard/Verveckken, Katrien (edd.), Binominal syntagms as a neglected locus of synchronic variation and diachronic change: towards a unified approach, Special issue in Language Sciences, Forthcoming. Milner, Jean-Claude, De la syntaxe à l’interprétation: quantités, insultes, exclamations, Paris, Le Seuil, 1987. Moliner, María, Diccionario de uso del español, Madrid, Gredos, 2008 (CD-Rom, third edition, 1966). Narrog, Heiko, From transitive to causative in Japanese, Diachronica 21:2 (2004), 351–392. Newmeyer, Frederick J., Language Form and Langue Function, Cambridge, MIT Press, 1998. Nieto Jiménez, Lidio/Alvar Esquerra, Manual, Nuevo tesoro lexicográfico del español (S. XIV1726), 11 vol., Madrid, Arco Libros/Real Academia Española, 2007. Noël, Dirk, Diachronic construction grammar vs. grammaticalization theory, Leuven, KU Leuven, Department of Linguistics, 2006. Noël, Dirk, Diachronic construction grammar and grammaticalization theory, Functions of Language, 14:2, 177–202. Norde, Muriel, Degrammaticalization, Oxford, Oxford University Press, 2009. Nuyts, Jan, Cognitive linguistics and functional linguistics, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 543–565. Oakly, Todd, Image schemas, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 214–235. Paradis, Carita, Metonymization: a key mechanism in semantic change, in: Barcelona, Antonio/Benczes, Réka/Ruiz de Mendoza Ibáñez, Francisco José (edd.), Defining Metonymy in Cognitive Linguistics. Towards a consensus view, Amsterdam/Philadelphia, Benjamins, 2011, 61–88. Peirce, Charles Sanders, Collected papers of Charles Sanders Peirce, Cambridge, Harvard University Press, 1931 (edd. Hartshorne, Charles/Weiss, Paul, 1974). Pottier, Bernard, L'isosémie lexico-grammaticale dans les textes, in: Luquet, Gilles (ed.), Travaux de linguistique hispanique, Paris, Presses de la Sorbonne Nouvelle, 1998, 401– 408. Pottier, Bernard, The typology of semantic affinities, in: Vanhove, Martine (ed.), From polysemy to semantic change, Amsterdam/Philadelphia, Benjamins, 2008, 93–106. Quirk, Randolph, et al., A comprehensive grammar of the English Language, London/New York, Longman, 1985. Radden, Günther/Dirven, René, Cognitive English Grammar, Amsterdam/Philadelphia, Benjamins, 2007. Rastier, François, Sémantique interprétative, Paris, Presses Universitaires de France, 1987. Real Academia Española, Corpus diacrónico del español (CORDE), Online: http://www.rae.es, [accessed from February 2010 until December 2011]. Real Academia Española, Corpus de referencia del española actual, Online: http://www.rae.es, [accessed from October 2007 until January 2010]. Real Academia Española, Esbozo de una nueva gramática de la lengua española, Madrid, Espasa-Calpe, 1973.

References | 513

Real Academia Española, Diccionario de la lengua española (DRAE), Vigésima segunda edición, Online: http://www.rae.es, 222001 (=2001a). Real Academia Española, Nuevo tesoro lexicográfico de la lengua española, Madrid, EspasaCalpe, DVD, 2001 (=2001b). Real Academia Española/Asociación de academias de la lengua española, Nueva gramática de la lengua española, Madrid, Espasa, 2009. Rhee, Seongha, Semantic changes of English prepositions against a grammaticalization perspective, Language Research 38 (2002), 563–583. Rice, Sally Ann, Towards a Cognitive Model of Transitivity, PhD thesis, University of California, San Diego, 1987. Rigau, Gemma, La estructura del sintagma nominal: los modificadores del nombre, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 5, vol. 1, Madrid, Espasa-Calpe, 1999, 311–393. Rodríguez Espiñeira, María José, Complementos en genitivo y lecturas diatélicas de los nominales de acción, Verba 37 (2010), 7–56. Rodríguez Espiñeira, María José/López Meirama, Belén, On the grammaticalization of the Spanish expression “puede que”, in: Seoane, Elena/López-Couso, María José (edd.), Theoretical and empirical Issues in Grammaticalization, Amsterdam/Philadelphia, Benjamins, 2008, 293–314. Rodríguez Espiñeira, María José/Pena Seijas, Jesús, Prefacio y resumen de los capítulos, in: Rodríguez Espiñeira, María José/Pena Seijas, Jesús (edd.), Categorización lingüística y límites intercategoriales, Special issue in Verba 61 (2008), 9–17. Rosch, Eleanor, Cognitive representations of semantic categories, Journal of Experimental Psychology: General 104:3 (1975), 192–233. Salvá, Vicente, Gramática de la lengua castellana según ahora se habla, Madrid, Arco/Libros, 1988. Sánchez López, Cristina, Los cuantificadores: clases de cuantificadores y estructuras cuantificativas, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 16, vol. 1, Madrid, Espasa-Calpe, 1999, 1025–1128. Schönefeld, Doris, Constructions, in: Doris Schönefeld (ed.), Introduction to Constructions all over: case studies and theoretical implications, Special volume of Constructions 1:9 (2006), Online: http://www.constructions-online.de, urn:nbn:de:0009-4-6860 [last access: 22.04.2009]. Seco, Manuel/Andrés, Olimpia/Ramos, Gabino, Diccionario del español actual, Madrid, Aguilar, 1999. Stickney, Helen, From pseudopartitive to partitive, in: Belikova, Alyona/Meroni, Luisa/Umeda, Mari (edd.), Proceedings of the 2nd Conference on Generative Approaches to Language Acquisition North America [GALANA], Somerville, MA, Cascadilla Proceedings Project, 2007, 406–415 Suñer, Margarita, Syntax and Semantics of Spanish Presentational Sentence Types, Washington, Georgetown University Press, 1982. Suñer Gratacós, Avel-Lina. La aposición y otras relaciones de predicación en el sintagma nominal, in: Bosque, Ignacio/Demonte, Violeta (edd.), Gramática Descriptiva de la Lengua Española, cap. 8, vol. 1, Madrid, Espasa-Calpe, 1999, 523–564. Talmy, Leonard, Toward a cognitive semantics: Concept structuring systems, vol. 1, Cambridge, MIT Press, 2000 (=2000a).

514 | References

Talmy, Leonard, Toward a cognitive semantics: Typology and process in concept structuring, vol. 2, Cambridge, MIT Press, 2000 (=2000b). Taylor, John R., Cognitive Grammar, Oxford, Oxford University Press, 2002. Timberlake, Alan, Reanalysis and actualization in syntactic change, in: Li, Charles N. (ed.), Mechanisms of syntactic change, Austin, University of Texas Press, 1977, 141–177. Traugott, Elisabeth Closs, On the rise of epistemic meanings in English: an example of subjectification in semantic change, Language 65:1 (1989), 31–55. Traugott, Elisabeth Closs, The role of pragmatics in semantic change, in: Verscheuren, Jef (ed.), Pragmatics in 1998: selected papers from the 6th International Pragmatics Conference, Antwerpen, International Pragmatics Association, 1999, 93–102. Traugott, Elisabeth Closs, From subjectification to intersubjectification, in: Hickey, Raymond (ed.), Motives for Language Change, Cambridge, Cambridge University Press, 2003, 124– 139 (=2003a). Traugott, Elisabeth Closs, Constructions in grammaticalization, in: Joseph, Brian D./Janda, Richard D. (edd.), The Handbook of Historical Linguistics, Oxford, Blackwell, 2003, 624– 647 (=2003b). Traugott, Elisabeth Closs, The concepts of constructional mismatch and type-shifting from the perspective of grammaticalization, Cognitive Linguistics 18:4 (2007), 523–557. Traugott, Elisabeth Closs, Grammaticalization, constructions and the incremental development of language: suggestions from the development of degree modifiers in English, in: Eckardt, Regine/Jäger, Gerhard/Veenstra, Tonjes (edd.), Variation, Selection, Development, Probing the Evolutionary Model of Language Change, Berlin/New York, Mouton de Gruyter, 2008, 219–250 (=2008a). Traugott, Elisabeth Closs, The Grammaticalization of NP of NP patterns, in: Bergs, Alexander/Diewald, Gabriele (edd.), Constructions and Language Change, Berlin/New York, Mouton de Gruyter, 23–45 (=2008b). Traugott, Elisabeth Closs, (Inter)subjectivity and (inter)subjectification: a reassessment, in: Davidse, Kristin/Vandelanotte, Lieven/Cuyckens, Hubert (edd.), Subjectification, Intersubjectification and Grammaticalization, Berlin/New York, Mouton de Gruyter, 2010, 29–73. Traugott, Elisabeth Closs, Toward a coherent account of grammatical constructionalization, Plenary paper at the 44th Annual Meeting of the Societas Linguistica Europaea, Logroño, 8–11 September, 2011. Traugott, Elisabeth Closs, A constructional framework for rethinking models of grammaticalization and the role of analogy, Plenary paper at the CSDL-conference (Conceptual Structure, Discourse, and Language), Santa Barbara, 4–6 November, 2014. Traugott, Elisabeth Closs/Trousdale, Graeme, Gradience, gradualness and grammaticalization. How do they intersect?, in: Traugott, Elisabeth Closs/Trousdale, Graeme (edd.), Gradience, Gradualness and Grammaticalization, Typological Studies in Language 90, Amsterdam/Philadelphia, Benjamins, 2010, 19–43. Trousdale, Graeme, Constructions in grammaticalization and lexicalization. Evidence from the history of a composite predicate construction in English, in: Trousdale, Graeme/Gisborne, Nikolas (edd.), Constructional Approaches to English Grammar, Berlin/New York, Mouton de Gruyter, 2008, 33–67 (=2008a). Trousdale, Graeme, Grammaticalization, constructions and the grammaticalization of constructions, Plenary paper at the NRG4-conference [New Reflections on Grammaticalization 4], University of Leuven, 16–19 July, 2008 (=2008b).

References | 515

Trousdale, Graeme, Binominal constructions in non-standard dialects, Plenary paper at the workshop Binominal syntagms as a neglected locus of synchronic variation and diachronic change: towards a unified approach (Brems, Lieselotte/De Clerck, Bernard/Verveckken, Katrien), 43rd Annual Meeting of the Societas Linguistica Europaea, Vilnius University, 2–5 September, 2010. Tuggy, David, Schematicity, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 82–166. Ungerer, Friedrich/Schmid, Hans-Jörg, An Introduction to Cognitive Linguistics, London/New York, Longman, 1996. Väänänen, Veikko, Recherches et récréations latino-romanes, Napoli, Bibliopolis, 1981. Van der Auwera, Johan/Van Alsenoy, Lauren, Indefiniteness and the mass/count distinction, Paper at the 44th Annual Meeting of the Societas Linguistica Europaea, Logroño, 8–11 September, 2011. Vandewinkel, Sigi/Davidse, Kristin, The interlocking paths of development to emphasizer adjective pure, Journal of Historical Pragmatics 9:2 (2008), 255–287. Van de Velde, Freek, Left-peripheral expansion of the English NP, English Language and Linguistics 15:2, (2011), 387–415. Van de Velde, Mark, Dependency reversal in northern sub-saharan Africa, Research questions and preliminary results, Paper presented at the FEST research unit reunion, 18 February, KU Leuven, 2011. Van Doorslaer, Jessa, Concordancia de número, un análisis cognitivo-funcional de la concordancia entre verbo y regente [N1 + de + N2], Unpublished masterpaper, KU Leuven, 2009. Van Langendonck, Willy, Iconicity, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 394–418. Vázquez Rozas, Victoria, A usage-based approach to prototypical transitivity, in: Delbecque, Nicole/Cornillie, Bert (edd.), On Interpreting Construction Schemas. From Action and Motion to Transitivity and Causality, Berlin/New York, Mouton de Gruyter, 2007, 17–37. Verhagen, Arie, Construal and perspectivization, in: Geeraerts, Dirk/Cuyckens, Hubert (edd.), The Oxford Handbook of Cognitive Linguistics, Oxford, Oxford University Press, 2007, 48– 81. Verhagen, Arie, The conception of constructions as complex signs. Emergence of structure and reduction to usage, Constructions and Frames 1:1 (2009), 119–152. Vergara Wilson, Damián, From “remaining” to “becoming” in Spanish. The role of prefabs in the development of the construction quedar(se) + ADJECTIVE, in: Corrigan, Roberta L., et al. (edd.), Formulaic Language, distribution and historical change, vol. 1, Philadelphia, Benjamins, 2009, 273–295. Verveckken, Katrien, Grammaticization of Spanish Size Noun-Constructions: a cognitive perspective, Unpublished masterpaper, KU Leuven, 2007. Verveckken, Katrien, Towards a constructional account of high and low frequency binominal quantifiers in Spanish, Cognitive Linguistics 23:2 (2012), 421–478. Verveckken, Katrien, Partitive Construction, in: Niemeier, Susanne/Juchem-Grundmann, Constanze/Schönefeld, Doris (edd.), Cognitive Grammar, Wörterbücher zur Sprach- und Kommunikationswissenschaft (WSK), vol. 14, Berlin/Boston, De Gruyter, 2013. Verveckken, Katrien, Spanish binominal quantifiers: towards a reassessment of analogy, in: Brems, Lieselotte/De Clerck, Bernard/Verveckken, Katrien (edd.), Binominal syntagms as

516 | References

a neglected locus of synchronic variation and diachronic change: towards a unified approach, Special issue in Language Sciences, Forthcoming. Verveckken, Katrien/Cornillie, Bert, Un análisis cognitivo-funcional de la concordancia verbal con construcciones (pseudo)partitivas del tipo [N1 de N2], Revue Romane 47:2 (2012), 219–257. Verveckken, Katrien/Delbecque, Nicole, On the development of binominal quantifiers in Spanish: the notion of lexical persistence revisited, Cognitextes 11 (2014), Forthcoming. Vigliocco, Gabriella/Butterworth, Brian/Garrett, Merrill F., Subject-verb agreement in Spanish and English: Differences in the role of conceptual constraints, Cognition 61 (1996), 261– 298. Vos, Riet, A grammar of Partitive Constructions, Tilburg, PhD thesis, 1999. Vos, Riet, Las construcciones de cuantificador nominal en holandés y español, Foro Hispánico 21 (2002), 47–58. Wanner, Dieter, The power of analogy, an essay on historical linguistics, Berlin, Mouton de Gruyter, 2006. Wierzbicka, Anna, “Oats” and “wheat”: the fallacy of arbitrariness, in: Haiman, John (ed.), Iconicity in syntax, Amsterdam/Philadelphia, Benjamins, 1985, 311–342. Willemse, Peter, Indefinite possessive NPs and the distinction between determining and nondetermining genitives in English, English Language and Linguistics 11:3 (2007), 537–568. Wischer, Ilse, Grammaticalization versus lexicalization, “Methinks” there is some confusion, in: Fischer, Olga/Rosenbach, Anette/Stein, Dieter (edd.), Pathways of change. Grammaticalization in English, Amsterdam/Philadelphia, Benjamins, 2000, 355–370.