From Latin to Romance: Morphosyntactic Typology and Change (Oxford Studies in Diachronic and Historical Linguistics) [Illustrated] 9780199584376, 0199584370

Citation preview

From Latin to Romance

OXFORD STUDIES IN HISTORICAL AND DIACHRONIC LINGUISTICS GENERAL EDITORS

Adam Ledgeway and Ian Roberts, University of Cambridge ADVISORY EDITORS

Cynthia Allen, Australian National University; Ricardo Bermúdez-Otero, Univerisity of Manchester; Theresa Biberauer, University of Cambridge; Charlotte Galves, State University of Campinas; Geoff Horrocks, University of Cambridge; Paul Kiparsky, Stanford University; Anthony Kroch, University of Pennsylvania; David Lightfoot, Georgetown University; Giuseppe Longobardi, University of Trieste; David Willis, University of Cambridge PUBLISHED

1 From Latin to Romance Adam Ledgeway 2 Parameter Theory and Linguistic Change Edited by Charlotte Galves, Sonia Cyrino, Ruth Lopes, Filomena Sândalo, and Juanito Avelar In preparation The Boundaries of Pure Morphology Diachronic and Synchronic Perspectives Edited by Silvio Cruschina, Martin Maiden, and J. C. Smith The History of Negation Vol. I: Case Studies; Vol. II: Patterns and Processes Edited by Anna Breitbart, Chris Lucas, and David Willis Gender from Latin to Romance Michele Loporcaro Vowel Quantity from Latin to Romance Michele Loporcaro Word Order in Old Italian Cecilia Poletto Syntactic Change and Stability Joel Wallenberg

From Latin to Romance Morphosyntactic Typology and Change

ADAM LEDGEWAY

1

3

Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York # Adam Ledgeway 2012 The moral rights of the author have been asserted Database right Oxford University Press (maker) First published 2012 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose the same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Library of Congress Control Number: 2011945220 Typeset by SPI Publisher Services, Pondicherry, India Printed in Great Britain on acid-free paper by MPG Books Group, Bodmin and King’s Lynn ISBN 978–0–19–958437–6 1 3 5 7 9 10 8 6 4 2

Contents Series preface List of tables List of abbreviations 1 From Latin to Romance: introduction 1.1 1.2 1.3 1.4

From Latin to Romance: the historical background Research questions Aims and objectives Acknowledgements

2 Syntheticity and analyticity 2.1 Traditional approach 2.2 Problems 2.2.1 Languages or constructions? 2.2.2 Absolute vs relative interpretations 2.2.3 Causal relations between analyticity and morphophonological erosion 2.2.4 Gradual change and competition 2.2.5 Explanatory power 2.2.6 Grammaticalization

ix xi xiii 1 1 2 4 8 10 10 12 12 16 21 23 24 28

3 Configurationality and the rise of constituent structure

30

3.1 Introduction 3.2 Nominal and verbal groups 3.2.1 Latin 3.2.1.1 Discontinuous structures 3.2.1.2 Conclusion 3.2.2 Romance 3.2.2.1 Adjectival positions 3.2.2.1.1 Restricted adjectival positions 3.2.2.2 Complements and adjuncts 3.3 The sentence 3.3.1 Classical Latin 3.3.1.1 Discontinuous structures 3.3.2 Late Latin and Romance

30 31 35 43 45 47 50 55 57 59 59 61 64

vi

Contents 3.4 Configurationality: concluding remarks 3.4.1 Degrees of configurationality

4 Configurationality and the rise of functional structure 4.1 Introduction 4.2 Determiner phrase (DP) 4.2.1 Indefinite article 4.2.2 Definite article 4.2.2.1 Late Latin 4.2.2.2 Romance 4.2.2.2.1 IPSE articles 4.2.2.2.2 Neuter articles 4.2.2.3 Conclusion 4.2.3 Other determiners 4.2.3.1 Romanian 4.2.3.1.1 Demonstrative article 4.2.3.1.2 Possessive article 4.3 Inflectional phrase (IP) 4.3.1 Romance auxiliaries 4.3.1.1 Semantic weakening 4.3.1.2 Morphosyntactic properties 4.3.1.3 Morphophonological specialization 4.3.1.4 Romance perfective auxiliary constructions 4.3.1.5 Romance synthetic future(-in-the-past) 4.3.2 Romance verb positions 4.3.2.1 Summary and conclusions 4.4 Complementizer phrase (CP) 4.4.1 Evidence for Latin CP structure 4.4.1.1 Archaic non-configurational pattern 4.4.2 Evidence for Romance CP structure 4.4.2.1 Topic and Focus Fields 4.4.2.1.1 Internal structure of Topic and Focus fields 4.4.2.1.1.1 Focus field 4.4.2.1.1.2 Topic field 4.4.2.2 Force and Finiteness 4.4.2.3 Other projections 4.5 Conclusion

71 77 81 81 82 82 89 89 96 100 105 107 110 113 113 115 119 119 121 124 127 130 134 140 146 150 150 156 158 159 162 162 166 169 176 179

Contents 5 From Latin to Romance: a configurational approach 5.1 Introduction 5.1.1 Early evidence for functional structure 5.1.2 Early evidence for configurationality 5.2 Head parameter: traditional observations 5.2.1 Some Romance counterexamples? 5.3 Changing directions: Latin 5.3.1 Early head-initial structures 5.3.1.1 Complementizers and adpositions 5.3.1.2 Comparatives 5.3.1.3 Relatives 5.3.1.4 Noun phrase 5.3.1.4.1 Adjectives 5.3.1.4.2 Genitives 5.3.1.4.3 Concluding remarks 5.3.1.5 Other categories 5.3.1.6 Summary 5.3.2 Verbal group 5.3.2.1 Auxiliary and dependent infinitive 5.3.3 Conclusion 5.4 Changing directions: from Latin to Romance 5.4.1 Head-last ) head-first: roll-up 5.4.2 Rise of head-initiality 5.4.2.1 Position of complement clauses 5.4.2.2 Position of nominal complements 5.4.2.3 Other patterns of harmonization 5.4.2.4 Clausal word order 5.4.3 Pragmatic variation: left-edge fronting 5.4.3.1 Identifying the left-edge: cola and left peripheries 5.4.3.2 Modifier fronting 5.4.3.2.1 Reanalysis: roll-up ) edge-fronting 5.4.3.3 Nominal fronting 5.4.3.4 From Latin to Romance 5.5 Conclusion 6 Head-marking and dependent-marking 6.1 Introduction 6.2 Variation in marking 6.3 Romance functional categories ) Romance head-marking 6.3.1 Head-marking on D 6.3.2 Head-marking on Infl

vii 181 181 183 185 196 198 202 204 205 207 209 210 210 213 218 219 224 225 234 235 235 236 238 242 249 252 255 258 259 262 269 270 277 281 284 284 286 289 290 292

viii

Contents

6.3.3 Head-marking on C 6.3.4 Extreme head-marking: the case of Ripano 6.3.4.1 Verbal domain 6.3.4.1.1 Subject-verb agreement 6.3.4.1.2 Subject-/object-verb agreement 6.3.4.1.3 Object-verb agreement 6.3.4.2 Contagious head-marking 6.4 Conclusion 7 The rise and fall of alignments 7.1 Introduction 7.2 Classical Latin 7.3 Late Latin and conservative Romance: active/stative syntax 7.3.1. Verbal group 7.3.1.1 Perfective auxiliary constructions 7.3.1.1.1 Latin background 7.3.1.1.2 Romance 7.3.1.2 Participle agreement 7.3.2 Nominal group 7.3.2.1 Extended and generalized accusative 7.3.2.2 Early Romance binary case systems 7.3.3 Sentence: word order 7.3.4 Other patterns 7.4 Innovative Romance: nominative/accusative syntax 7.4.1 Verbal group 7.4.1.1 Perfective auxiliary constructions 7.4.1.2 Participle agreement 7.4.2 Sentence: word order 7.5 Concluding remarks References Index

298 299 300 300 302 305 308 310 312 312 314 316 317 317 317 319 326 328 328 333 335 336 340 341 341 347 349 351 353 409

Series preface Modern diachronic linguistics has important contacts with other subdisciplines, notably first-language acquisition, learnability theory, computational linguistics, sociolinguistics, and the traditional philological study of texts. It is now recognized in the wider field that diachronic linguistics can make a novel contribution to linguistic theory, to historical linguistics, and arguably to cognitive science more widely. This series provides a forum for work in both diachronic and historical linguistics, including work on change in grammar, sound, and meaning within and across languages; synchronic studies of languages in the past; and descriptive histories of one or more languages. It is intended to reflect and encourage the links between these subjects and fields such as those mentioned above. The goal of the series is to publish high-quality monographs and collections of papers in diachronic linguistics generally, i.e. studies focussing on change in linguistic structure, and/or change in grammars, which are also intended to make a contribution to linguistic theory, by developing and adopting a current theoretical model, by raising wider questions concerning the nature of language change, or by developing theoretical connections with other areas of linguistics and cognitive science as listed above. There is no bias towards a particular language or language family, or towards a particular theoretical framework; work in all theoretical frameworks, and work based on the descriptive tradition of language typology, as well as quantitatively based work using theoretical ideas, also feature in the series. Adam Ledgeway and Ian Roberts University of Cambridge September 2011

This page intentionally left blank

List of tables Table 4.1: Forms of article with proper names in Catalan

100

Table 4.2: Catalan articles derived from IPSE

101

Table 4.3: Tonic/clitic possessive paradigms [MSG forms]

111

Table 4.4: Morphophonological specialization in HABERE paradigms

128

Table 4.5: Auxiliary/lexical paradigms of *volere, *andare, and DEBERE

129

Table 4.6: Reanalysis and grammaticalization of HABERE + PtP

133

Table 4.7: Romance auxiliary futures (cantar(e)/cantà/cânta ‘to sing’)

135

Table 4.8: Romance synthetic future

135

Table 4.9: Typology of Romance V(erb)-movement

146

Table 4.10: Dual complementizer systems in the dialects of southern Italy

170

Table 5.1: Distribution of GN / NG across different Latin authors/texts

214

Table 5.2: Incidence of verb-final positions in Latin according to Linde (1923)

226

Table 5.3: Distribution of OV / VO across different Latin authors/texts

228

Table 5.4: Placement of AcI and Finite complements in classical prose (Bolkestein 1989)

243

Table 5.5: Placement of AcI and QUOD-/QUIA-clauses in Christian authors (Herman 1989)

244

Table 5.6: Placement of AcI, QUOD-/QUIA-, and UT-clauses in 10th–12th-c. authors (Greco 2007)

245

Table 5.7: Some linearization patterns in classical Latin

252

Table 5.8: (P+) N/G and (P+) G/N linearizations in Caesar (B.G. 6) and Livy (36.1–19)

254

Table 6.1: Romance inflected infinitives (cant- ‘sing’)

293

Table 6.2: Romance inflected gerunds and participles (cant- ‘sing’)

294

Table 6.3: Ripano nominal agreement system

300

Table 6.4: Italian finite verbal agreement system (aprire ‘to open’)

300

Table 6.5: Ripano finite verbal agreement system (magnà ‘to eat’)

301

Table 6.6: Ripano matching agreement

302

Table 6.7: Ripano non-matching agreement

303

xii

List of tables

Table 6.8: Ripano object clitic agreement

306

Table 7.1: Typological alignments of A, S, and O

313

Table 7.2: Active/stative split in Romance perfective auxiliaries

321

Table 7.3: Romance triple auxiliation systems

325

List of abbreviations &P

coordination phrase

*

(i) unattested form or usage; (ii) recursive element or phrase

**

ungrammatical form or usage

?

dubious form or usage

=

cliticized to



null argument (subject or complement)

1

first person

2

second person

3

third person

A

(i) adjective position (head of AP); (ii) subject of a transitive clause

A.Test.

Antigo Testamento

Sam.

Samuel

A.Vales.

Anonymus Valesianus

ABL

ablative

Abr.

Abruzzese (dialect group of upper southern Italy)

ABS

absolute

Aca.

Acadian French (Canada)

ACC

accusative

Acf.

Acquafondata (Lazio, central Italy)

AcI

accusative and infinitive

ACT

active

ad. fil.

Praecepta ad filium (by Cato)

ADJ

adjective (category)

Adjn

adjunct

ADV

adverb (category)

AdvP

adverb(ial) phrase

AG

adjective-genitive order

Agn.

Agnonese (northern Molisan dialect of Agnone, upper southern Italy)

Agr(S/O)(P)

(subject/object) agreement (phrase)

Agr.

De agri cultura (by Cato)

Alg.

Alguerès (Catalan dialect of Alghero, northwest Sardinia)

xiv

List of abbreviations

Alt.

Altamurano (central Pugliese dialect of Altamura, upper southern Italy)

AN

adjective-noun order

And.

Andalusian (Spanish)

Anon. gen.

Anonimo genovese

Ant. Plac. Itin.

Antonini Placentini Itinerarium

AP

adjectival phrase

Apul.

Apuleius

Pysche

Pysche et cupido

Ara.

Aragonese (dialect group of northeast Spain)

Arl.

Ariellese (eastern Abruzzese dialect of Arielli, upper southern Italy)

ARo.

Aromanian (Romanian dialects spoken in Greece, Albania, Bulgaria, Serbia, and the Republic of Macedonia)

Ars gram.

Ars gramatica (Pompeius’ commentary of Donato’s Ars gramatica)

ART

article

Asp

Aspect (head position)

Ast.

Asturian (dialect group of northern Spain)

Atop

Aboutness topic

Auc. Nic.

Aucassin et Nicolette

Aug.

Augustine

Conf.

Confessiones

Evang. Iohan.

Tractatus in Evangelium Iohannis

AUG

augmentative

Auv.

Auvernhat (Occitan dialect group of Auvergne, central France)

AUX

auxiliary

AuxP

auxiliary phrase

Avi.

Aviglianese (northwestern Basilicatese dialect of Avigliano, upper southern Italy)

B.Afr.

Bellum Africum

B.Alex.

de Bello Alexandrino

B.Hisp.

de Bello Hispaniensi

Bad.

Badiot (Ræto-Romance dialect from Alta Badia, Dolomites of southern Tyrol)

Bal.

Balearic (Catalan)

Bar.

Barceloní (Catalan)

Bea.

Bearnés (Gascon dialect of Béarn, French Pyrenees, southeast France)

Beg.

Begurenc (northeastern Catalan dialect of Begur, Baix Empordà)

List of abbreviations

xv

Bel.

Bellunese (Venetan dialect of Belluno, northeast Italy)

Blt.

Bellantese (northeastern Abruzzese dialect of Bellante, upper southern Italy)

Bol.

Bolognese (dialect of city of Bologna, northern Italy)

BPt.

Brazilian Portuguese

Brg.

Barégeois (Occitan dialect of Barèges, Hautes-Pyrénées, southern France)

C

(i) central; (ii) complementizer position (head of CP); (iii) consonant

Cad.

Cadaquesenc (northeastern Catalan dialect of Cadaqués, Alta Empordà, Costa Brava)

Caes.

Caesar

B.C. B.G.

de Bello Ciuili de Bello Gallico

Cai.

Cairese (Ligurian dialect of Cairo Montenotte, northwest Italy)

Cal.

Calabrian (dialects of extreme south of Italy)

Car.

Cargeghese (dialect of Cargeghe, northwest Sardinia)

Cat.

Catalan

CDN

Charroi de Nîmes

CDP

Cronaca di Partenope

CDR

Chanson de Roland

CDT

Cronaca dei tumulti

CFoc

contrastive focus

Chi.

Chierese (central Piedmontese dialect of Chieri, northwest Italy)

Chron.

Chronicon/-a

Cic.

Cicero

Amic.

De amicitia

Arch.

Pro Archia

Att.

Epistulae ad Atticum

Balb.

Pro Balbo

Brut.

Brutus

Cael.

Oratio pro Caelio

Cat.

In Catilinam

Clu.

Pro Cluentio

De or.

De oratore

Deiot.

Pro rege Deiotaro

Diu.

De diuinatione

Diu. Caec.

Diuinatio in Caecilium

xvi

List of abbreviations

Dom.

De domo sua

Fam.

Epistulae ad familiaris

Fin.

De finibus

Flac.

Pro Flacco

Har. resp.

De haruspicum responso

Inv. rhet.

De inventione rhetorica

Leg.

De legibus

Leg. Man.

Pro Lege Manilia

Lig.

Pro Ligario

Marc.

Pro Marcello

Mil.

Pro Milone

Mur.

Pro Murena

Nat.D.

De natura deorum

Off.

De officiis

Orat.

Orator ad M.Brutum

Phil.

Orationes Philippicae

Q.F.

Epistulae ad Quintum Fratrem

Rab. Perd.

Pro Rabirio Perduellionis Reo

Rab. Post.

Pro Rabirio Postumo

Red. sen.

Post reditum in senatu

Rep.

De republica

S.Rosc.

Pro Sexto Roscio Amerino

Sen.

De senectute

Sest.

Pro Sestio

Sull.

Pro Sulla

Tull.

Pro Marco Tullio

Tusc.

Tusculanae disputationes

Vat.

In Vatinium

Verr.

In Verrem

CIL

Corpus Inscriptionum Latinarum

ClLD

clitic left-dislocation

circum.

circumstantial

Cmb.

Colombian (Spanish)

Cmp.

Campanian (dialects of upper southern Italy)

CNT

count

List of abbreviations Col. Rust.

xvii

Columella De re rustica

coll.

colloquial

COM

comitative

COMP/Comp

complementizer (category)

COMPL

complement

COMPR

comparative

ConcP

concessive phrase

COND

conditional

Cop(P)

copula(r phrase)

Cor.

Corsican

Cos.

Cosentino (northern Calabrian dialect of Cosenza, extreme South of Italy)

CP

complementizer phrase

Cpc.

Capcinese (northern Catalan dialect of Capcir, Pyrénées-Orientales, southern France)

Cpd.

Campidanese (dialects of southern Sardinian)

Crs.

Corese (southern Lazio dialect, central Italy)

CSPS

Condaghe di San Pietro di Silki

Csv.

Castrovillarese (northern Calabrian dialect of Castrovillari, extreme South of Italy)

CSV

Cartulario de San Vicente de Oviedo

Ctz.

Catanzarese (central-southern Calabrian dialect of Catanzaro, extreme South of Italy)

D

determiner position (head of DP)

D/P(.P)

prepositional determiner (phrase)

DAT

dative

DC

Disciplina clericalis (by Petrus Alphonsi)

DeclP

declarative phrase

Dem(P)

demonstrative (phrase)

DEM

demonstrative

DEP

deponent

DET

determiner (category)

DIM

diminutive

DO

direct object

DP

determiner phrase

DSG

Diálogos de S. Gregório

xviii

List of abbreviations

E

(i) east(ern); (ii) perfective auxiliary ESSE ‘be’

ECS

Era·m cossehlatz, senhor (by Bernatz de Ventador)

EDD

Existence de Dieu (by Fénelon)

Egd.

Engadinish (Romansh dialect, southeast Switzerland)

Eml.

Emilian (dialect of northeast Italy)

Enn.

Ennius

Ann. Trag.

Annales Tragoediae

EPP

Extended Projection Principle

EPt.

European Portuguese

ERG

ergative

ESp.

European Spanish

ExclP

exclamative phrase

Ext.

Extremaduran (dialect group, western Spain)

F

feminine

Fin

finiteness position (head of FinP)

FinP

finiteness phrase

Firm.

Firmicus

Err. prof. rel.

De errore profanarum religionum

FJ

Fuero Juzgo (Alfonso X El Sabio)

Flo.

Florentine

Foc(P)

focus (phrase)

FOFC

Final-over-Final Constraint

ForceP

(illocutionary) force phrase

Fr.

French

Frc.

Francavillese (Salentino dialect of Francavilla Fontana, southern Puglia, extreme South of Italy)

Fred. Chron.

Fredgar’s Chronicle

Frk.

Frankish

Frl.

Friulian (dialects spoken in northeast Italy)

FUM

Fazienda de Ultra Mar

FUT

future

GA

genitive-adjective order

Gai.

Gaius

Gal.

Gallese (northern Campanian dialect of Gallo Matese, upper southern Italy)

List of abbreviations GE

General Estoria (Alfonso X El Sabio)

Gell.

Aulus Gellius

Gen.

Genoese

xix

GEN

genitive

Ger.

German

Gest. Zen.

Gesta apud Zenophilum

G.l.

Grammatici latini

Glc.

Galician (northwest Spain)

GN

genitive-noun order

Grd.

Gardenese (Latin dialect of Val Gardena, Alto Adige, Dolomites of southern Tyrol)

GRD

gerund(ive)

Greg. Tur.

Gregorius Turensis (Historia Francorum)

Grz.

Grazzanisano (northwestern Campanian dialect of Grazzanise, upper southern Italy)

Gsc.

Gascon

H

perfective auxiliary HABERE ‘have’

HAS

Higher Adverb Space

Hist. Aug.

Historia Augusta

Hadr. Hor.

Hadrian Horace

Carm.

Carmina

Epist.

Epistles

HT

hanging topic

HypP

hypothetical phrase

I

(verb) inflection (head of IP)

IAM

In alto Mar (by De Jennaro)

IbR.

Ibero-Romance

Ibz.

Ibizan (Catalan)

IE

Indo-European

IFoc

informational focus

ILCV

Inscriptiones Latinae Christianae Veteres

IMP

imperative

IND

indicative

IndefQ

indefinite quantifier

INF

infinitive

xx

List of abbreviations

Infl

Inflection position (head of IP)

IntP

interrogative phrase

INTR

intransitive

IP

inflection phrase

IPFV

imperfect(ive)

IO

indirect object

Irs.

Irsinese (eastern Basilicatese dialect of Irsina, upper southern Italy)

It.

Italian

Itin. Hier.

Itineraria Hierosolymitana

Lad.

(Dolomitic) Ladin

LAS

Lower Adverb Space

Lat.

Latin

Laz.

Laziale (central Italian dialect area of Lazio)

LDA

Livro de Alveitaria

LDR

Loise De Rosa

LDSG

Libru de lu dialagu di sanctu Gregoriu

LDT

Libro de la destructione de Troya

Lec.

Leccese (southern Salentino dialect of Lecce, extreme south of Italy)

Leg. XII Tab.

Leges XII Tabularum

Leo.

Leonese (dialect group of western Iberia)

LFN

Los fueros de la Novenera

Lgd.

Lengadocien (Occitan dialects of Languedoc, southern France)

Lib. reg.

Libellus a regula sancti benedicti subtractus

Lig.

Ligurian (dialect group of northwest Italy)

lit.

literally

Liv. Per.

Livy (Ab urbe condita) Periochae

Lmb.

Lombard (dialect group of central northern Italy)

LOC

locative

Log.

Logudorese (dialects of northwest Sardinia)

LP

Lettera pisana

Lucr.

Lucretius

Lul.

Lulese (northeastern Sardinian dialect of Lula, province of Nuoro)

Lvl.

Livinallonghese (Ladin dialect of Livinallongo, northeast Italy)

LVV

Libro de’ vizî e delle virtudi (by Bono Giamboni)

List of abbreviations M

masculine

Mac.

Maceratese (central Marchigiano dialect of Macerata, central Italy)

Maj.

Majorcan (Catalan)

Manil.

Manilius

Mar.

Marsalese (northwestern Sicilian dialect of Marsala, southern Italy)

Mart.

Martial (Epigrammata)

Mdv.

Moldovan (Romanian)

Mel.

Melicucchese

Men.

Menorcan (Catalan)

Mex.

Mexican (Spanish)

xxi

Mil.

Milanese

MLuc.

Muro Lucano (eastern Basilicata, upper southern Italy)

mod.

modern

ModP

mood phrase

MRA

Mort le Roi Artu

MRK

marker

MRo.

Megleno-Romanian (Romanian dialects spoken in Greece, and the Republic of Macedonia)

Mtp.

Montpelhierenc (Occitan dialect of Montpellier, southern France)

Mtv.

Mantuan (southern Lombard dialect of Mantua, northern Italy)

Mul. Ch.

Mulomedicina Chironis

N

(i) north(ern); (ii) neuter; (iii) noun; (iv) noun position (head of NP)

NA

noun-adjective order

Nap.

Neapolitan

Neg(P)

negator (phrase)

Nem.

Nemorense (central Lazio dialect of Nemi, province of Rome)

Nep.

Nepos

Han.

Hannibal

Milt.

Miltiades

Timot.

Timotheus

NG

noun-genitive order

Niç.

Niçois (Occitan dialect of Nice, southeast France)

NOM

nominative

non-ref.

non-referential

Nov.

Novellino

xxii

List of abbreviations

NP

noun phrase

Nrs.

Nursino (southeastern Umbrian dialect of Norcia, central Italy)

Nuo.

Nuorese (dialects of Nuoro and area, northeast Sardinia)

O

(i) old; (ii) object

OBL

oblique (case)

Occ.

Occitan

Olo.

Olotí (northeastern Catalan dialect of Olot, Garrotxa)

Op

operator

Orib.

Oribasius

Oss.

Ossalais (Occitan dialect of the Vallée d’Ossau, Pyrénées-Atlantiques, southwest France)

Ov.

Ovid

Met.

Metamorphoses

R.A.

Remedia amoris

Tr.

Tristia

P

preposition

Pad.

Paduan (Venetan dialect of Padua, northeast Italy)

PADCE

Pro ai del chan essenhadors (by Jaufre Rudel)

Pal.

Palmero Spanish (Island of La Palma, Canary Islands)

PART

partitive

PASS

passive

PEJ

pejorative

Pel. Ars vet.

Pelagonius Ars veterinaria

Per. Aeth.

Peregrinatio Aetheriae

Pes.

Pescasserolese (western Abruzzese dialect of Pescasseroli, upper southern Italy)

Petr.

Petronius

Sat.

Satyricon

Cena Trim.

Cena Trimalchionis

PFV

perfective

Pgl.

Pugliese (southern Italy)

PIC

Phase Impenetrability Condition

Pie.

Piedmontese (dialect group of northwest Italy)

PIE

Proto-Indo-European

List of abbreviations Pl.

xxiii

Plautus Amph.

Amphitruo

Asin.

Asinaria

Aul.

Aulularia

Bacch.

Bacchides

Capt.

Captiui

Cist.

Cistellaria

Epid.

Epidicus

Men.

Menaechmi

Merc.

Mercator

Mil.

Miles Gloriosus

Poen.

Poenulus

Pseud.

Pseudolus

Rud.

Rudens

Stich.

Stichus

Trin.

Trinummus

Truc.

Truculentus

PL

plural

Plac. cap.

Placito capuano

Plf.

Palafrugellenc (northeastern Catalan dialect of Calella de Palafrugell, Costa Brava)

Plin.

Pliny (the Elder)

N.H. Plin. Ep.

Naturalis Historia Pliny (the Younger) Epistulae

PLPF

pluperfect

Plr.

Palmarino (Friulian dialect of Palmanova Udine, northeast Italy)

Poi.

Poirinese (Piedmontese dialect of Poirino, northwest Italy)

Pomp.

Pompeius

pop.

popular

POSS

possessive

PP

prepositional (or postpositional) phrase

PPtP

passive past participle

Prd.

Procidano (Campanian dialect of Island of Procida, Bay of Naples)

presup

presuppositional

xxiv

List of abbreviations

PRET

preterit

pro

null pronominal argument

PROG

progressive

PRS

present

Prv.

Provençal Occitan (southeast France)

PSPDI

Purgatorio di S. Patrizio Degli Innocenti

PST

past

Pt.

Portuguese

PTCP

participle

PtP

past participle

Q

question particle/marker

QP

quantifier phrase

QUANT

quantifier

quantif.

quantificational

Quint.

Quintilian

Decl. Mai.

Declamatio Maior

Inst.

Institutio oratoria

RADT

Le roman d’Apollonius de Tyr

RæR.

Ræto-Romance

Rgt.

Rouergat (Occitan dialect of Rouergue, Aveyron, southern France)

Rho.

Rhodanien (Occitan dialect, Rhône Valley, southeast France)

Rip.

Ripano (southern Marchigiano dialect of Ripatransone, central Italy)

Rmc.

Romanesco (now defunct dialect of Rome)

Ro.

Romanian

Rom.

Romance

Ros.

Rossellonès (Catalan dialect of Roussillon, Pyrénées-Orientales, southern France)

Rov.

Rovignese (Istriot coastal variety of Rovigno (Rovinj), Croatia)

S

(i) south(ern); (ii) subject

SA

intransitive subject of an unergative clause

SO

intransitive subject of an unaccusative clause

S.And.

Sant’Andrea (central-southern Calabria, extreme South of Italy)

Sal.

Salentino (dialect group of the Salento, southern Puglia, extreme south of Italy)

List of abbreviations Sall.

xxv

Sallust

Cat.

De Catilinae coniuratione

Iug.

Bellum Iugurthinum

Sas.

Sassoferratese (central Marchigiano dialect of Sassoferrato, central Italy)

SB

Sermone di Barsegapé (by Pietro da Barsegapé)

SBJV

subjunctive

SC

small clause

sc-set.

scene-setting (adverbials)

S.C. Bacch.

Senatus consultum de Bacchanalibus

SCL

subject clitic

SCSG

Statuto della Compagnia di S. Giorgio

SE

Syntactic Extraposition

SE

Santa Eulalia

Sen.

Seneca (the Elder)

Contr. Sen.

Controuersiae Seneca (the Younger)

Ep.

Epistulae

Tranq.

De tranquillitate animi

Serm. sub.

Sermoni subalpini

Sfl.

Sanfeliciano (southern Lazio dialect of San Felice Circeo, central Italy)

SG

singular

Sic.

Sicilian

SID

southern Italian dialect

S.Leo.

San Leonardo (Ræto-Romance, Badia, Dolomites of southern Tyrol)

S.Leu.

San Leucio del Sannio (northern Campania, upper southern Italy)

Sp.

Spanish

Spec

specifier position

Srd.

Sardinian

Srs.

Surselvan (Ræto-Romance dialect, southeast Switzerland)

SS

Sete Savis

Subj

subject

Suet.

Suetonius

Aug.

Diuus Augustus

Cal.

Gaius Caligula

xxvi

List of abbreviations

SUP.

supine

SUPERL

superlative

T

tense (position)

Tac.

Tacitus

Ann.

Annales

Germ.

Germania

Tar.

Tarantino (central Pugliese dialect of Taranto, upper southern Italy)

Tbs.

Trebisaccese (northeastern Calabrian dialect of Trebisacce, extreme South of Italy)

Ter.

Terence

Eun.

Eunchus

Hec.

Hecyra

Phorm.

Phormio

Terent.

(Claudius) Terentianus

Tert.

Tertullian

Apol.

Apologeticus

Top(P)

topic (phrase)

Tor.

Torinese (dialect of Turin, northwest Italy)

Tos.

Tossenc (northeastern Catalan dialect of Tossa de Mar, Costa Brava)

TP

tense phrase

TR

transitive

trans.

transitive (verb/clause)

Trb.

Tarbener (Catalan dialect of Tàrbena, Marina Baixa, southeast Valencia)

Trist. Ricc.

Tristano Riccardiano

Trn.

Trentino (northeast Italy)

Trs.

Triestino (northeast Italy)

Tsc.

Tuscan (central Italy)

Umb.

Umbrian (central Italy)

unacc.

unaccusative (intransitive verb/clause)

unerg.

unergative (intransitive verb/clause)

v(P)

light verb (phrase)

V

(i) verb; (ii) vowel

V.Vit.

Victor Vitensis

Hist. persecut.

Historia persecutionis africanae prouiciniae

List of abbreviations xxvii Val.

Valéian (southeastern Occitan dialect of the Vallée de l’Ubaye, Alpes-deHaute-Provence)

Var.

Varro

Ling. Rust.

De lingua Latina De re rustica

Ven.

Venetian

VFE

Vita e favole di Esopo (by Giovanni Brancati)

Vgl.

Vegliot (defunct Dalmatian dialect of Island of Veglia (Krk))

Vil.

Villalaghese (southwestern Abruzzese dialect of Villalago, upper southern Italy)

Virg.

Virgil

Aen. Ecl.

Aeneid Eclogues

Vlc.

Valencian (Catalan)

Vld.

Vallader (Romansh dialect of lower Engadine Valley, southeast Switzerland)

VMLR

Volgarizzamento della ‘Mascalcia’ di Lorenzo Rusio

VMNS

Venjansa de la mort de Nostre Senhor

VN

Vita nuova (by Dante)

VOC

vocative

vP

light verb phrase

VP

verb phrase

VSA

Vie de Saint Alexis

Vsg.

Valsuganotto (southeast Trentino dialect of Valsugana, northeast Italy)

Vto.

Veneto (dialect group of northeast Italy)

Vulg.

Vulgate

Mar.

Mark

Mat.

Matthew

V2

verb second syntax

W

west(ern)

Wal.

Walloon (French dialect of Wallonia, southern Belgium)

Wrl.

Warlpiri (indigenous language of Australia’s Northern Territory)

X

unspecified head element

XP

unspecified phrasal projection

This page intentionally left blank

1 From Latin to Romance: introduction 1.1 From Latin to Romance: the historical background With the expansion of Roman political and military domination outside of the historical homeland of the Latium vetus, a geographical area considerably smaller than the present-day Italian region of Lazio, Latin, a member of the ‘Italic’ branch of the centum Indo-European dialects,1 gradually established itself over the entire Italian territory and beyond, coming to be spoken in much of western and central Europe, the Mediterranean, and along the coast of North Africa. The survivors of these spoken varieties of Latin in the Iberian Peninsula, Gaul, Italy, Istria, and Dacia came to diverge quite significantly from each other, and especially from the classical Latin we know from literary texts and grammars. They began to emerge from at least the ninth century (Malkiel 1977; 1978; Wright 1983: 7; forthc.) as the early Romance vernaculars which we today associate with the standard Romance languages (Catalan, French, Italian, Portuguese, Romanian, Spanish) and the many non-standard languages and dialects of the Romània (see the various entries in Price 2000).2 Not only do we have extensive knowledge and rich (epigraphic and literary) textual documentation of Latin from as early as the sixth century BC, but we can also trace the evolution of the descendant Romance languages and dialects (some, such as Dalmatian, now defunct) through an abundance of medieval and modern Romance texts (Malkiel 1974; Kabatek forthc.; Pountain forthc.) and, since the end of the nineteenth century, the scientific observation and description of thousands of non-standard Romance dialects. The sum of these sources reveals an unparalleled

1

Meillet ([1928] 1977), Coleman (1987), Vineis (1998). For an original and in-depth consideration of the socio-historical and linguistic factors involved in the ‘making’ of the Romance languages, see the excellent treatment by Varvaro (forthc.; cf. also Banniard forthc.). The date of the early ninth century (more precisely 813 AD) conventionally associated with the ‘birth’ of the (western) Romance languages is that of the Council of Tours, in which it was ruled that the clergy should translate their sermons into the ‘rusticam Romanam linguam’ (‘rustic Roman(ce?) speech’) for the benefit of their illiterate parishioners. This view is disputed by Herman (1996a: 368–9) who claims that the emergence of the Romance vernaculars occurred, in some areas at least, as early as the first decades of the seventh century but in any case no later than 760–70 (cf. also Norberg 1943: 219; Mohrmann 1952; Wright 1982; 2002; Banniard 1992: 487–8). 2

From Latin to Romance

2

range of variation through time, space, and according to register and medium of expression, in short a fertile ground for formulating and testing new ideas about language structure, variation, change, and typology.

1.2 Research questions In what follows I shall not be concerned, at least not directly, with the thorny question, famously posed by Lot (1931), ‘À quelle époque a-t-on cessé de parler latin?’ (‘When did Latin cease to be spoken?’; see also Richter 1983), but with what I take to be a more fundamental question, namely ‘What were the changes that occurred in the morphosyntax of the speech of the populations of the formerly Latinspeaking regions which led to the wide typological variation witnessed in the Romance languages and dialects written and spoken today and in the past?’ Among Romance linguists of all theoretical persuasions,3 there is general recognition that, in the passage from Latin to Romance, the morphosyntax of the emerging languages underwent significant changes in three fundamental areas of the grammar involving: (i) the nominal group; (ii) the verbal group; and (iii) the sentence. The impact of such changes is most immediately observable in: (i) the gradual reduction (e.g. medieval Gallo-Romance, Romanian) and/or eventual loss (e.g. Ibero-Romance, central-southern Italo-Romance) of the Latin morphological case system (cf. }7.3.2.2) and the emergence of a series of so-called determiners (cf. }4.2) including the definite and indefinite articles (1a–b); (ii) the profusion of auxiliary verb structures (cf. }4.3.1) to mark such categories as tense (e.g. present perfect in 2a), aspect (e.g. continuous aspect in 2b), mood (e.g. epistemic modality in 2c), and voice (e.g. passive in 2d); and (iii) the gradual shift from an original unmarked (S)OV word order (3a) towards a fixed (S)VO (/V(S)O) order (3b; cf. }3.3, }5.4.2.4). (1)

3

a omnes ante uos consules senatui paruerunt (Lat., Cic. Red. all.NOM.PL before you consuls.NOM.PL senate.DAT obeyed sen. 17)4

See, for example, Harris (1978: 5–6), Bauer (1995: 5), La Fauci (1997: 11–12), Zamboni (1998: 128), Magni (2009: 247). 4 Here and throughout I follow the widely established practice of citing Latin forms, paradigms, and constructions in small capitals, but actual documented examples in lower-case font. Where I do depart, however, from current conventional practice is in the representation of the classical Latin high back vowel/ glide [w], which is today usually represented as ‘V’ in syllable onsets (e.g. VIVO ‘I live’) and U in all other positions (e.g. HABUIT ‘he had’) or, according to another school of thought, as ‘V’ when it appears in upper case and ‘u’ when in lower case (e.g. Viuo ‘I live’). By contrast, I have preferred to adopt U (lower case) / U (upper case) in all positions (hence, UIUO and HABUIT). One further departure from current typographical conventions concerns the decision to cite all non-attested forms, whether reconstructed for Latin or any other language (but in all cases preceded by a single asterisk) in phonetic transcription (e.g. *volere ‘to want’ replacing classical UELLE), and not in small capitals (e.g. *UOLERE) as is frequently the case in other works.

Introduction

3

b tous les consuls avant vous obéirent au sénat (Fr.) all the consuls before you obeyed to.the senate ‘all the consuls that ever existed before you were obedient to the senate’ (2)

a ai dormit (Occ.) I.have slept ‘I have slept’ b so kredende (Srd.) I.am believing ‘I believe’ c la pipa deu valer molt (Cat.) the pipe must be.worth.INF much ‘the pipe must be worth a lot’ d sînt invitaţi la un cocteil (Ro.) they.are invited to a cocktail.party ‘they are invited to a cocktail party’

(3)

a Caesar exercitum reduxit (Lat., Caes. B.G. 3.27.3) Caesar.NOM army.ACC led.back b (César) retiró (César) su ejército (Sp.) Caesar led.back Caesar his army ‘Caesar led his army back’

Moreover, in many historical treatments of Romance morphosyntax it is commonplace to interpret such changes as interrelated phenomena, rather than independent developments. For example, the loss of morphological case and the progressive establishment of (S)VO word order are frequently viewed, albeit erroneously,5 as complementary developments in the restructuring of the original system of argument marking: on the one hand the weakening of the Latin case system necessitates a more rigid word order to distinguish crucially between subject and object (Vennemann 1974; 1975; Bauer 1995: 5–6), while on the other hand a growing rigidification of word order renders the original case system increasingly redundant (Bourciez 1956; Zamboni 2000: 102). In a similar fashion, the gradual erosion of the case system, with its concomitant effects on word order, is also held to be responsible, according to one frequent view,6 for the increased use of prepositions (though see discussion in }2.2.3), part of a more general typological shift from synthetic to analytic structures also reflected in the frequent recourse to auxiliaries within the verbal group.7 5

For critical assessment, see Sasse (1977), Bichakjian (1987: 89), Bauer (1995: 7–8), La Fauci (1997: 41). See Grandgent (1907: 42–8), Muller and Taylor (1932: 65), Väänänen (1966: 115–9; 1982: 195–7), Lakoff (1972: 189), Adams (1976b: 49), Bauer (1995: 137–9), Molinelli (1998: 147). 7 Harris (1978: 15), Tekavčić (1980: 15), Schwegler (1990). 6

4

From Latin to Romance

Now, while the specific details of the complex morphosyntactic changes affecting the three key areas of the grammar hinted at above are relatively well known, scholars are still very much divided as to their correct interpretation, and how they are to be integrated within the overall typological change(s) witnessed in the passage from Latin to Romance. In what follows, I shall review a number of these competing approaches, comparing how individual developments (syntheticity vs analyticity; non-configurationality vs configurationality; (dis)harmonic linearization in terms head-final vs head-initial structures; head- vs dependent-marking; nominative-accusative vs active-stative alignments) can best be accounted for across different theories and approaches. In particular, I shall emphasize how a number of traditional ideas can be profitably reinterpreted in light of recent theoretical developments, highlighting what further insights, if any, they provide for our overall understanding of the nature of the broad typological and structural changes that differentiate the morphosyntax of the Romance languages from that of Latin. Perhaps surprisingly, it will be shown that many of the conclusions reached within one theory often find immediate parallels in competing and otherwise seemingly incompatible theories—a sure sign of the merits of a complementary and integrated approach to old questions.

1.3 Aims and objectives The present volume sets out to provide a comparative treatment of the morphosyntactic development and typology of the Romance languages which brings together in a single volume a comprehensive integrated and complementary analysis of the key issues presented in }1.2 above, in relation to a number of current competing theoretical and typological approaches. With the exception of massively detailed encyclopaedic works (e.g. Holtus, Metzeltin, and Schmitt 1988–96; Ernst, Gleßgen, Schmitt, and Schweickard 2003–6; Maiden, Smith, and Ledgeway 2011), current treatments of the Romance languages are restricted to comparative(-historical) synopses of the Romance languages in the traditional format of a ‘standard’ reference manual providing a structural overview of individual ‘languages’ and/or traditional themes (e.g. Diez 1836–8; Meyer-Lübke 1890–1902; Bourciez 1956; Lausberg 1956–62; Tagliavini 1972; Elcock 1975; Harris and Vincent 1988; Posner 1996; Alkire and Rosen 2010) or to structural overviews of individual languages or sub-branches of Romance (e.g. Brunot (and Bruneau) 1905–53; Sandfeld and Olsen 1936–62; Lombard 1974; Maiden 1995; 1998; Maiden and Parry 1997; Bosque and Demonte 1999; Zagona 2002; Azevedo 2005; Rowlett 2007). This book is not therefore intended to compete with or supersede the works mentioned above, but to complement them, by presenting a collection of fresh and original reflections on what I believe to be the principal questions and issues in the comparative-historical study of the morphosyntax of

Introduction

5

Latin and Romance, informed by contemporary thinking in both Classical Philology, Romance Linguistics, and general linguistic theory, and organized according to novel chapter divisions which reflect broader, overriding comparative and theoretical concerns and themes (generally neglected or left untackled in standard works), rather than those which are narrowly focused on individual languages or developments. Furthermore, while it is customary in many treatments of Romance historical morphosyntax to make, at most, passing comment on the relevant Latin background, the present treatment accords equal attention and space to the available Latin linguistic evidence in accordance with the expectations of the title of the present volume. While I make no claims to exhaustiveness in my treatment of Latin, especially as I am first and foremost a scholar of the Romance languages, I believe that a proper understanding of the Romance facts can only be achieved in the context of a thorough and robust understanding of the relevant Latin evidence. Equally well served in the following chapters, unlike many standard handbooks on Romance, are the numerous substandard regional varieties and dialects of Romance, whose rich array of morphosyntactic structures, often diverging quite radically from those familiar to us from the standard languages, ensures that they figure in the discussion just as prominently as the standard Romance languages. The present book therefore fills a significant gap in the current literature, aiming at originality in two key respects: the first in terms of an innovative and comprehensive approach to the historical morphosyntax of Latin and Romance, and the second in terms of its integration of contemporary thinking in general linguistic theory with the rich Latin and (especially non-standard) Romance empirical evidence to tell us new and interesting things about the structure of Latin and the Romance languages. Additionally, it is worth emphasizing that the approach to the historical comparative morphosyntax of Latin and Romance adopted in this book is deliberately an integrated one, in which we evaluate, compare, test, and, where possible, integrate the results and predictions of differing theories and approaches to the same data. This is one of the most original aspects of the approach of the book, which privileges the data above all else, rather than trying to force them into one particular favoured theory of the author. In short, while the book admittedly concedes an important role to linguistic theory and typology, it is fundamentally and above all to be read as a history of the Romance languages from their Latin roots, and not as a book on linguistic theory and typology which just happens to take Latin and Romance as its main field of inquiry. This, in part, explains the fact that in the following chapters I often make use of concepts taken from formal syntactic theories in apparently typological classifications. It is my view that typological distinctions such as head-/dependent-marking, (non-)configurationality, and active-stative alignments do not necessarily exclude formal syntactic models of language and can, and should, where possible, be explored

From Latin to Romance

6

in relation to these, an approach in any case already widely adopted by others.8 Moreover, given the variety of formal approaches pursued in each of the following chapters, the reader might legitimately question whether I endorse the claims, principles, and findings reported in individual chapters, which seem to contradict those reported in earlier or later chapters. Indeed, this apparent vagueness is deliberate, insofar as one of the main aims of the book is to present the reader with several competing analyses in order to compare and critically evaluate these to demonstrate the different predictions, strengths, and weaknesses of different formal approaches to the Latin and Romance data, not to defend at all costs one particular theory irrespective of its value for the data under discussion. Along the way, it will be highlighted how the perennially fertile and still under-utilized testing ground of the Romance languages has a central role to play in challenging linguistic orthodoxies and shaping and informing new ideas and perspectives about language change, structure, and variation. The present work is, then, aimed principally at scholars and researchers in the fields of Classics, Romance Linguistics, and General Linguistics, especially morphosyntactic theory, historical linguistics, and typology. It should appeal to an extensive international readership, and to ensure that the material is accessible to all, irrespective of their particular language or theoretical specializations, I have deliberately chosen to gloss and translate all non-English examples and simplify or omit detailed technical discussion to render the text accessible to a non-specialist audience. Consequently, in many places I have employed concepts and terms taken from formal syntactic theories to classify phenomena that are in fact dealt with from a typological point of view, hinting at formal concepts that are not explicitly described, but exemplified with specific constructions as part of an artifice to convey with greater simplicity the complex formal and theoretical ideas to an audience which might not necessarily be interested in the often finer technical details of a particular formal apparatus, but which is interested in some of the results of such an approach. Inevitably, in an attempt to appeal to such a wide audience which will have differing degrees of interest in empirical, linguistic, and theoretical issues, some compromises have had to be made. The Latin scholar may lament the lack of detailed philological attention paid to such factors as variation in accordance with author, period, and text type, while the Romance scholar may find the attention paid to the Latin evidence excessive for what might be conceived by some as essentially a morphosyntactic history of the Romance languages, and both might take issue with what they may perceive as an overformalization of the approach in some places. By contrast, the general linguist might find the formalization of the approach insufficient and excessively simplified in places, with too much attention paid to the empirical coverage at 8

See, for example, the proposals in Nordlinger (1998) to incorporate within the L(exical-)F(unctional) Grammar framework the typological distinction between head- and dependent-marking.

Introduction

7

the expense of the development of the theoretical conclusions. In short, to paraphrase a famous saying, while I recognize that I cannot please all of the intended readership all of the time, I can at least hope to please some of them all of the time and all of them some of the time. Finally, a brief remark must be made about the sources used for the Romance varieties exemplified in the following chapters. In addition to my own investigations and field notes, for exemplification from the different Romance varieties, I have drawn principally on the following sources: Aragonese (Alvar 1948; Saralegui 1992), Asturian (Cano González 1981; 1992; Academia de la Llingua Asturiana 1995; García Arias 2003), Catalan (Jordana 1933; Badia i Magarit 1951; 1962; 1991; 1995; Fabra 1956; Schlieben-Lange 1971; Yates 1975; Villangómez i Llobet 1978; Moll 1993; 1997; 2006; Solà 1993; 1994; Veny 1982; 1991; 2001; Blasco Ferrer 1984a; 1986; 1988; Wheeler 1988a; Wheeler, Yates, and Dols 1999; Hualde 1992; Bernat i Baltrons 2007; Solà, Lloret, Mascaró, and Pérez Saldayna 2008), Dalmatian and Istro-Romance dialects (Ive 1886; Bartoli 1906; Doria 1989; Ursini 1989), Extremeño (García Santos 1992), French (von Wartburg [1934] 1971; Pope 1952; Désirat and Hordé 1967; Price 1971; Ewert 1978; Battye and Hintze 1992; Jones 1996; Ayres-Bennett and Carruthers 2001; Fagyal, Kibbee, and Jenkins 2006; Rowlett 2007), Italian and the dialects of Italy (Alfonsi 1932; Rohlfs 1968; 1969; Tekavčić 1980; Durante 1981; Bruni 1984; Lepschy and Lepschy 1988; Renzi 1988; Renzi and Salvi 1991; Renzi, Salvi, and Cardinaletti 2001; Graffi 1994; Marazzini 1994; Avolio 1995; Maiden 1995; 1998; Maiden and Parry 1997; Ledgeway 2000; 2009a; Dalbera Stefanaggi 2001; Tortora 2003; Manzini and Savoia 2005; Maiden and Robustelli 2007; Salvi and Renzi 2010), Leonese (Alonso Garrote 1947; Born 1992), Occitan dialects (Grandgent 1905; Ronjat 1930–41; Roncaglia 1965; Bec 1967; 1973; Lafont 1967; 1991; Schlieben-Lange 1971; Alibèrt 1976; Wheeler 1988b), Gallego-Portuguese (Dias 1918; Huber 1933; Sten 1944; Thomas 1969; Mattoso Câmara 1972; Teyssier 1980; 1984; Álvarez, Monteagudo and Regueira 1986; Parkinson 1988; Nunes 1989; Costa 2000a; Azevedo 2005), so-called Ræto-Romance varieties, namely Romansh, Ladin, and Friulian (Ascoli 1873; Gartner 1883; 1910; Arquint 1964; Rohlfs 1975; Haiman 1988; Plangg 1989; Stimm and Linder 1989; Haiman and Benincà 1992), Romanian and dialects (Capidan 1932; Sandfeld and Olsen 1936–62; Agard 1958; Guţia 1967; Rosetti 1968; Lombard 1974; Mallinson 1986; 1988; Dahmen 1989; Stati 1989; Manoliu Manea 1989; Dobrovie-Sorin 1994; Cinque and Giusti 1995; Alboiu and Motapanyane 2000a; Daniliuc and Daniliuc 2000), Sardinian dialects (Wagner 1951; Blasco Ferrer 1984b; 1986; 1988; 1994; Jones 1988a; 1993; 1997), and Spanish (García de Diego 1951; Menéndez Pidal 1966; Zamora Vicente 1967; Lapesa 1980; Lloyd 1987; Penny 2000; 2002; Alarcos Llorach 1994; Butt and Benjamin 1994; Lipski 1994; Stewart 1999; Pountain 2001; D’Introno 2001; Hualde, Olarrea, and Escobar 2001; Zagona 2002). So-called Romance creoles are not considered in what follows since, as varieties which arose as a result of catastrophic social upheavals (for an overview, see

8

From Latin to Romance

Bachmann forthc.), they present a number of complications and unresolved issues, not least the vexata quaestio whether they can be legitimately classified as Romance. While the core lexicon of the creoles spoken in the Americas, Africa, and Asia is unmistakably Romance, their morphosyntactic features often diverge in numerous respects from those of their respective lexifiers. As a consequence, so-called Romance creoles arguably share greater structural affinities with other (e.g. English and Dutch) creoles than with Romance (cf. Green 1988b: 423–4, 470).

1.4 Acknowledgements The preparation of this book, and the original chapter (Ledgeway 2011) out of which many of the ideas of the present research grew, has been made possible thanks to the generous help of many people. Above all I should like to thank Martin Maiden, who not only has been an enormous inspiration to me over many years as an exemplary model of what outstanding empirical and theoretical scholarship can achieve within the Romance languages, but who also first suggested to me the idea of writing the current book after reading my chapter for the first volume of the Cambridge History of the Romance Languages (Maiden, Smith, and Ledgeway 2011). He has strongly encouraged and supported me ever since in this endeavour, especially through his unrelenting efforts and goodwill in providing regular and detailed critical feedback on earlier versions of the manuscript. Without him, this book would never have been written. In addition, I owe an enormous debt of gratitude to Jim Adams who has shown great interest in my work and who, through his meticulous reading of the manuscript, kindly offered me the considerable benefit of his extraordinary knowledge and command of the Latin evidence and of general linguistics. The book has also benefited enormously from the goodwill and kindness of a number of other colleagues and friends, who were generous enough to put at my disposal their knowledge and expertise in Latin, Romance, and Linguistics to offer me invaluable feedback through their thoughts, advice, and constructive criticisms on all or parts of an earlier version of the manuscript. Special thanks goes to Paola Benincà, Theresa Biberauer, Silvio Cruschina, Michele Loporcaro, Kate Mclachlan, Ian Roberts, and Giampaolo Salvi. My heartfelt thanks also go to a number of other colleagues who have contributed, either directly through discussions with me or indirectly through my reading of their work, to the formation and/or development of some of the ideas presented in the following chapters: Brigitte Bauer, Delia Bentley, Michela Cennamo, Roberta D’Alessandro, Federico Damonte, Nunzio La Fauci, David Langslow, Alessandra Lombardi, Guido Mensching, Renato Oniga, Mair Parry, Silvia Pieroni, Harm Pinkster, Cecilia Poletto, Eva-Maria Remberger, J.C. Smith, Rosanna Sornicola, Alberto Varvaro, and Nigel Vincent. The efforts of all those mentioned above have done much to improve the final manuscript, including the elimination of a number of embarrassing errors and oversights. Of course, it goes

Introduction

9

without saying that all views presented in the following chapters are my own and that any remaining infelicities or errors are entirely my own responsibility. I am also grateful to the members—too many to list—of the audiences of various seminars, invited talks, and conferences at which various parts of the material of the following chapters were presented, including at the Universities of Berlin, Bristol, Cambridge, Konstanz, Leiden, London, Manchester, Naples, Padua, Patras, Pescara, Potenza, Reading, Siena, and Zurich. Finally, I would also like to express especial thanks to John Davey and Julia Steer, Linguistics Editors at OUP, to Jennifer Lunsford, Production Editor for Linguistics at OUP, and to Jack Whitehead, who expertly copy-edited the present volume, whose generous help and support has proven invaluable in the preparation of the manuscript. On a personal note, I should like to thank my parents for their unfailing love, support, and encouragement and, not least, for regularly ‘stepping in’ whenever academic commitments take me away from home. Finally, my deepest gratitude goes to my wife Alessandra for her love, limitless patience, and understanding. It is to her and our three glorious children, Luca, Chiara, and Lorenzo, that this book is dedicated.

2 Syntheticity and analyticity 2.1 Traditional approach The principal differences in the morphosyntax of Latin and Romance have long been, albeit somewhat simplistically, viewed as representing two opposite poles of a syntheticity-analyticity continuum.1 This synthetic-analytic dichotomy, which goes back to the pioneering work of August Wilhelm von Schlegel (1818),2 points to a Sapirian ‘drift’ from the predominantly synthetic structures of Latin towards the characteristically analytic structures of Romance or, as Harris (1978: 15) succinctly puts it, ‘a tendency for syntax to take over a number of functions previously within the domain of morphology’ such that ‘an element of meaning previously conveyed by a stem and an affix is now expressed by a syntagm, that is, a combination of two or more elements that would traditionally be labelled words’. The examples are numerous and so well documented in the literature that they hardly need repeating here,3 suffice it to recall such classic examples as the replacement of: (i) the Latin suffixal comparative in -IOR ‘-er’ (1a) with a reflex of PLUS or 4 MAGIS ‘more’ followed by the ungraded adjective or adverb (1b); (ii) the Latin inflectional future (2a) with an auxiliary + infinitive construction (2b; cf. discussion in }4.3.1.5); and (iii) the accusative and infinitive (AcI) construction, in which the sole marker of subordination lies in the ‘exceptional’ accusative marking of the infinitival subject (3a),5 with a (non-)finite subordinate clause introduced by an overt complementizer (3b):

1 Meillet ([1937] 1964: 439), Bourciez (1956: 23), Harris (1978: 15–16), Tekavčić (1980: 15), Schwegler (1990), Posner (1996: 156–7). 2 For an historical overview of the use of the terms ‘synthetic’ and ‘analytic’, see Schwegler (1990: ch. 1). 3 See, for example, Schwegler (1990), Vincent (1997a: 102–3), Bauer (2006). 4 Wüest (1998: 92), Zamboni (2000: 121–2); cf. also }5.3.1.2. 5 The ‘exceptional’ accusative case of the infinitival subject is determined by the construction as a whole, as highlighted by the fact that CREDO ‘I believe’ canonically assigns dative to its complement (see }3.4.1, }4.4.1.1, }5.4.2.1).

Syntheticity and analyticity (1) a

ALTUS

tall

: ALT-IOR (Lat.) : tall-er

b mai/p(l)us more ‘tall; taller’ (2)

a

11

naut (Occ.) tall

PLU-ET (Lat.) rain-FUT.3SG

b at a próere (Srd.) it.has to rain.INF ‘it will rain’ (3)

a tacitum te silent.ACC you.ACC

dicere say.INF

credo / I.believe

‘Non not

reddes’ (Lat., Mart. 6.5.3–4) you.give.back

b criju ca ti sta diciennu ‘unn’ ’i I.believe that yourself= PROG.2SG saying not them= ‘I fancy you say to yourself: “You won’t repay them” ’

ripaghi’ (Cos.) you.repay

Contrasts like these, which ultimately serve to draw attention to a significant typological shift from the morphologically oriented structures of Latin to the increasingly syntactically oriented structures of Romance, are not just limited to the varieties and examples above, but are believed to surface across all Romance varieties and in all areas of the grammar, including the replacement of central case functions with specific prepositions (4a–b; see }2.2.3, }5.4.2) and the extension of the ESSE ‘be’ passive to the imperfective verbal paradigm (5a–b). (4)

a canis dog.NOM

nonne not.Q

similis similar.NOM

lup-o (Lat., Cic. Nat. D. 1.97) wolf-DAT

b el gos és semblant al the dog is similar to.the ‘are dogs similar to wolves?’ (5)

a larem Lar.ACC

corona crown.ABL

nostrum our.ACC

llop? (Cat.) wolf

decor-ar-i decorate-INF-PASS

uolo (Lat., Pl. Trin. 39) I.want

b je veux que notre Lare soit honoré I want that our Lar be.SBJV.3SG honoured d’ une couronne (Fr.) of a crown ‘I wish for our Household God Lar to be adorned with a crown’ Indeed, so deep-rooted is this transformation that Tekavčić (1980: 15) considers it ‘the guiding principle underlying the whole of Romance morphosyntax and undoubtedly its deepest and most important characteristic’.

12

From Latin to Romance

2.2 Problems This traditional interpretation of the synthesis-analysis cycle proves, however, problematic on a number of accounts, as does the fundamental typological distinction on which it crucially rests.6 This was first noted by Humboldt (1836), who considered contrasts between Latin and Romance such as those exemplified in (1)–(5) above as purely superficial, external differences: though admittedly not as rich as Latin in their inflectional specifications, at a deeper, internal level the Romance languages are of the same inflectional type as the ancestor language, and hence ultimately constructed according to the same structural principles, allowing Humboldt to conclude (pp. 288–9) that, although individual inflectional forms were lost, inflection as a whole was not lost (‘Es sanken Formen, nicht aber die Form’). While this idea might be difficult to maintain in its strongest form in view of the undeniable typological differences between Latin and Romance to be discussed in the following chapters, it does highlight how the passage from Latin to Romance cannot be characterised tout court in terms of a shift from synthetic to analytic structures as long as Romance continues to display extensive syntheticity involving nominal and verbal inflections. Moreover, as Coseriu (1987: 58) observes, there is little justification for lumping all presumed Romance analytic developments into a single all-encompassing principle, for they represent a collection of extremely heterogeneous changes, the diatopic distribution, individual chronologies and details of which vary enormously from one construction to another.7 2.2.1 Languages or constructions? Among the various difficulties associated with the traditional synthesis-analysis dichotomy, one observes above all, with Schwegler (1990: 193), a ‘striking vagueness and ambiguity with which the terms ANALYTIC and SYNTHETIC, hence the concepts themselves, are used and understood in the literature.’ Exemplary in this respect is the erroneous tendency to define Latin and Romance in absolute terms as synthetic and analytic languages, respectively, for both languages clearly also display, albeit in smaller measure, tendencies in the opposite direction.8 For instance, among other things Latin boasts numerous prepositions (including AD ‘to(wards)’, EX ‘out of ’, CUM ‘with’, IN ‘in, on’, POST ‘after’, PROPTER ‘on account of ’, SUB ‘under’; see Hewson and Bubenik 2006: 248–55), independent markers of sentential negation (6a),9 a perfective 6

Schwegler (1990: 4–5), Bauer (1995: 10–11, 138, 166), Vincent (1997a: 99–100, 105). See, for example, the comparative discussion of HABERE ‘have’ in the formation of the perfect and future periphrases in Pinkster (1987) and Bauer (2006), and the discussion in }}4.3.1.4–5, }7.3.1.1, }7.4.1.1 below. 8 Coseriu ([1971] 1988: 210; 1987: 56), Schwegler (1990: 28), Vincent (1997a: 99). 9 Highly exceptional are cases such as NOLO ‘I do not want’, NESCIO ‘I do not know’, NEQUEO ‘I am not able’, and NEGO ‘I say no, deny,’ (< NE(C) ‘not’ + UOLO / SCIO / QUEO / AIO ‘I.want / I.know / I.am.able / I.say’), which belong to a small number of lexicalized cases in which the negator NE(C) has univerbized with the category it modifies (Ernout and Thomas [1953] 1993: 149–50). 7

Syntheticity and analyticity

13

passive periphrasis consisting of ESSE ‘be’ + PPtP (6b), and a number of overt markers of subordination, including the subjunctive complementizers UT/NE ‘that (/not)’ (6c): (6)

a quem amat, amat; quem non amat, non amat (Lat., Petr. Sat. 37.5) whom she.loves she.loves whom not she.loves not she.loves ‘Whom she likes, she likes; whom she does not like, she does not like’ b arma [ . . . ] sunt humi inuenta (Lat., Cic. Diu. 1.74) ground.LOC found.PASS.PFV.PCTP.N.PL weapons.N.PL are ‘the weapons . . . have been found on the ground’ c Pompeius suis praedixerat ut Caesaris impetum Pompey.NOM his.DAT.PL had.foretold that Caesar.GEN charge.ACC exciperent ne ue se loco mouerent (Lat., Caes. B.C. 3.92.1) they.received.SBJV that.not=or selves place.ABL they.moved.SBJV ‘Pompey had told his men beforehand to receive Caesar’s charge and not to move from their position’

In fact, Wüest (1998: 91) goes so far as to claim that early and classical Latin, despite systematically favouring synthetic forms, present, for instance, just as rich an array of future expressions as modern French. Thus alongside the synthetic paradigms in -BO (e.g. UOCA-BO ‘call-FUT.1SG’) and -AM (e.g. SCI-AM ‘know-FUT.1SG’), we also find the grammaticalized periphrasis built on ESSE ‘be’ and the future active participles (e.g. UOCATURUS ‘call.FUT.ACT.PTCP’ + SUM ‘I.am’: ‘I am about to call’), as well as various modal + infinitive constructions,10 not to mention the widely attested use of the present with future value (Bennett 1910: 18–22). Conversely, in Romance number and gender marking on nouns and adjectives is still typically suffixal (7a),11 as are person/number (7b) and temporal, aspectual, and modal categories (7c) on finite verbs: (7)

a o(s)/a(s) nos-o(s)/-a(s) veciñ-o(s)/-a(s) ruidos-o(s)/-a(s) (Glc.) the.M(PL)/the.F(PL) our-M(PL)/-F(PL) neighbour.M(PL)/F(PL) noisy- M(PL)/-F(PL) ‘our noisy neighbour(s)’

The modals with presumed future uses include INCIPERE ‘to begin’, POSSE ‘can’, UELLE ‘to want’, and ‘must’ (cf. Hofmann and Szantyr 1965: }175; Fleischman 1982; Haverling 2010: 395–9), although Pinkster (1989) finds no evidence for this claim. 11 In many Occitan varieties the original synthetic marking of plural number has optionally been reinforced with the creation of doubly marked inflectional plurals (Arnaud and Morin 1920: 273; Wheeler 1988b: 256) such as pial ‘hair’ > pial-s ‘hair-PL’ (> pials-es ‘hair.PL-PL’), amì ‘friend’ > amì-s ‘friend-PL’ (> amìss-es ‘friend.PL-PL’), molin ‘mill’ > molin-s ‘mill-PL’ (> moliss-es ‘mill.PL-PL’). Moreover, suffixal marking of feminine gender (viz. -a, predominantly realized as [ɔ], more rarely [a] (e.g. Auv, Niç., Mtp., Ros.) or [e] (e.g. Bea., EGsc.)) is so robust in Occitania that it is productively extended not just to professional titles (e.g. coconièr ‘egg.seller’ > coconièr-a ‘egg.seller-FSG’ (‘egg-seller’s wife; female egg-seller’), tabaton ‘tobacconist’ > tabaton-a ‘tobacconist-FSG’ (‘tobacconist’s wife; female tobacconist’)), but also to all masculine proper names (Arnaud and Morin 1920: 272) such as Paul-on ‘Paul-DIM’ > (la) Paulon-a ‘(the) Paul.DIM-FSG (= Paul’s wife)’, Guston ‘Auguste’ > (la) Guston-a ‘(the) Auguste-FSG (= Auguste’s wife)’. 10

DEBERE

From Latin to Romance

14

b partiv-i/-as/-a/-am/-atz/-an (Gsc.) leave.-IND.PST.IPFV.1sg/2SG/3SG/1PL/2PL/3PL ‘I/you etc. was/were leaving’ c parr-u/-ava/-ai/-assi (Sic.) speak-1SG.PRS.IND/1SG.PST.IPFV.IND/1SG.PST.PFV.IND/1SG.PST.SBJV (/COND) ‘I speak / was speaking / spoke / spoke (/would speak)’ To these we can add the wide-spread use of evaluative suffixes (e.g. It. mur-etto ‘wallDIM’, Lgd. pintair-às ‘drinker-PEJ (= drunkard)’, Pt. mulher-ão ‘woman-AUG’, Ro. frumuş-el ‘beautiful-DIM’, Ven. gat-on ‘cat-AUG’),12 superlative and residual comparative forms (e.g. Cat. alt-íssim ‘high-SUPERL.MSG’, fred-íssim ‘cold-SUPERL.MSG’, mill-or ‘good-COMPR.SG’, pitj-or ‘bad-COMPR.SG’, It. sicur-issimo ‘sure-SUPERL.MSG’, noios-issimo ‘boring-SUPERL.MSG’, super-iore ‘high-COMPR.SG’, infer-iore ‘low-COMPR.SG’, Pt. felicíssimo ‘happy-SUPERL’, mai-or ‘big-COMPR.SG’, men-or ‘small-COMPR.SG’), and the frequent retention of case distinctions in pronouns (e.g. MLuc. iə ‘1SG.NOM’, (a) mì ‘(P. ACC-DAT.MRK) 1SG.ACC-DAT’, (pə) mévə ‘(for) 1sg.OBL’, (cu) mìchə ‘(with) 1SG.COM’; Ro. eu ‘1SG.NOM’, (pe) mine ‘(P.ACC.MRK) 1SG.ACC’, mie ‘1SG.DAT’, meu ‘1SG.GEN’). If we consider furthermore that many of the Romance languages also display a whole host of innovative synthetic structures (Coseriu 1987: 59), notably the future and conditional, derived from an erstwhile infinitival periphrasis in conjunction with 13 HABERE ‘have’ (8a), sentence and manner adverbs in -men(t(e)) (8b),14 internal morphophonological alternations such as metaphony and related types of stressed vowel alternation to mark number and/or gender in nominal categories (8c–e),15 and inflected non-finite verb forms such as the old Neapolitan inflected infinitive (8f),16 the presumed inexorable driving force underlying Romance analyticity appears seriously undermined. (8) a

MONSTRARE

show.INF b

+

/ HABEBAM > Fr. (je) montrerai I.have / I.had (I) will.show.1SG HABEO

/ montrerais / would.show.1SG

+ MENTE > It. aperta-mente open.ABL mind.ABL open-ly ‘with (an) open mind’ > ‘openly’

APERITA

In modern French these are generally no longer productive (e.g. fill-ette ‘girl-DIM’, but not **chaussur(e)-ette ‘shoe-DIM’, which has to be expressed by the analytic petite chaussure ‘little shoe’). 13 See the discussion in }4.3.1.5, as well as the treatments in Valesio (1968; 1969), Coleman (1971), Harris (1978: ch. 6), Fleischman (1982), Green (1987), Pinkster (1987), Adams (1991), Maiden (1996), Loporcaro (1999), Nocentini (2001), Bourova (2005), La Fauci (2006; 2011b), Bourova and Tasmowski (2007). 14 See the discussion in }2.2.2 and }5.3.1.5, as well as Karlsson (1981), Bauer (2001b; 2003; 2006), Ricca (2010: 181–5). 15 Hilty (1991), Maiden (1991), Savoia and Maiden (1997), Chitoran (2002), Loporcaro (2011a: }1.2). 16 See the discussion in }6.3.2, as well as Maurer (1968), Loporcaro (1986), Raposo (1987), Jones (1992; 1993: 78–82), Vincent (1996; 1998b), Ledgeway (1998: 41–6; 2000: 109–14; 2007c; 2009a: 585–90), Mensching (2000), Pires (2002), Sitaridou (2002). 12

Syntheticity and analyticity c friddə cold\M

15

/ freddə (Pgl.) cold\F

d p[o]rc-o dirty\MSG-MSG

/ p[ɔ]rc-a(s)/-os (Pt.) dirty\-F(PL)/-MPL

e şcoal-ă school\FSG-FSG

/ şcol-i (Ro.) school\-FPL

f yà non credevano de may lo vede-re-no plu (ONap., LDT 66.33) already not they.believed of never him= see-INF-3PL anymore ‘they no longer believed that they would ever see him again’ In short, considerations such as these reveal that, if they are to be employed at all, the terms ‘synthetic’ and ‘analytic’ should not be predicated of languages but, at best, of individual constructions. Thus, to the extent that any generalizations can usefully be reached in relation to the synthetic and analytic typology, they must be made in relation to specific construction types, rather than individual languages. Exemplary in this respect is the often overlooked observation (though see Meyer-Lübke 1894: 2–3; Coseriu 1987: 58; La Fauci 1998: 524–5) that the fate of Latin inflection shows a differential treatment in the nominal and verbal domains: whereas all but a few isolated residues of the rich Latin nominal declension—early Gallo-Romance, Ræto-Romance, and Romanian aside (cf. }7.3.2.2)—have survived into Romance, the Latin verbal conjugation formally continues largely intact, even accommodating, in some cases, several new additions such as the future and the conditional (}4.3.1.5; see Maiden 2011a, b). This observation apparently explains why analytic developments occurring within the nominal domain typically assume a substitutive function replacing earlier synthetic structures (e.g. PAUL-O ‘Paul-DAT’ > Sp. a Pablo ‘to Paul’, NOU-IOR ‘new-COMPR.’ > Ro. mai nou ‘more new (= newer)’, INDE ‘thence’ > It. di là ‘from there’), unlike in the verbal domain where, rather than replacing earlier synthetic structures, they usually come to mark new functions that complement those already marked synthetically within the system.17 For example, Romance reflexes of FACTUM HABEO (‘do.PFV.PASS.PTCP I.have’) do not replace, at least not initially, the synthetic perfect FECI ‘I did, I have done’, but variously come to assume novel ancillary functions of the perfect (Harris 1982; Squartini and Bertinetto 2000), ranging from a present resultative (9a; Alfonzetti 1998) to a present perfect with marked iterative function (9b; Leal Cruz 2003: 132), before ultimately encroaching upon (some of) the functions proper of the original perfect, witness its canonical present perfective reading in standard peninsular Spanish (9c; Penny 2000: 159) and its full-fledged punctual perfective value in modern spoken French (9d; Harris 1970: 79–83). 17 Replacement of the synthetic imperfective passive (e.g. AMAT-UR ‘love.PRS.IND.3SG-PASS’) with the analytic (AMATUS EST ‘love.PFV.PASS.PTCP he.is’; cf. Fr. il est aimé ‘he is loved’) is not a counterexample to this generalization, since it represents the functional extension of an already existing periphrasis of the classical language.

16 (9)

From Latin to Romance a non m’ ha chiamatu (SCal.) not me= he.has called ‘I don’t know what’s happened to him (because he hasn’t rung me)’ b aquí también ha hecho frô (Pal.) here too it.has done cold ‘here too it continues to be cold’ c siempre la he escuchado con atención, pero nunca always her= I.have listened.to with attention but never ‘I have always listened to her attentively, but never again’ d la France a déclaré the France has declared ‘France declared war in 1939’

la the

guerre war

en in

más (ESp.) anymore

1939 (Fr.) 1939

2.2.2 Absolute vs relative interpretations Even assuming that the synthetic-analytic parameter must be interpreted in relation to specific constructions, as demonstrated above, we have to recognize that it is often erroneously applied in absolute terms, whereas individual constructions can ostensibly display varying degrees of syntheticity and analyticity (Vincent 1997a: 100). A case in point concerns what is called ‘mesoclisis’ in the (literary) European Portuguese future and conditional (e.g. falar-me-ão ‘they will speak to me’; see }4.3.1.5; Luís and Spencer 2005; Monachesi 2005: 152–8), where the possibility of separating the person/ number inflection (e.g. -ão ‘3PL’) from the future/conditional stem (e.g. falar- ‘speak. INF’) with an intervening object clitic (e.g. me ‘me.DAT’) casts some doubt on the simple suffixal nature of the former. Similar problems arise for diminutive forms like Pt. pãozinho ‘roll’ (< pão ‘bread’ + DIM -zinho), which in the plural are marked not only in the desinence of the diminutive, but also in the nasal vowel alternation on the nominal stem (namely, pãe-zinho-s ‘bread\PL-DIM.M-PL’). Another classic case is constituted by the Romance FACERE ‘make’ causative construction (10a; for the late Latin evidence, see Fruyt and Orlandini 2008), which, although clearly analytic inasmuch as it involves two separate lexemes (viz. reflex of FACERE + infinitive), mimics the syntax of a morphological causative (Comrie 1985) with both verbs behaving as a single verbal complex,18 licensing monoclausal properties such as clitic climbing (10b): (10)

a Gino fece distruggere i documenti alla Gino made destroy.inf the documents to.the ‘Gino made the secretary destroy the documents’

18

segretaria (It.) secretary

See Kayne (1975), Zubizarreta (1985), Burzio (1986), Alsina (1992), Guasti (1993).

Syntheticity and analyticity b Gino glieli fece Gino to.her=them= made ‘Gino had her destroy them’

distruggere destroy.INF

17

(**glieli) (It.) =to.her=them

Similarly, an often-cited case of greater Romance analyticity concerns the marking of sentential negation in Gallo-Romance, northern Catalan dialects, and Aragonese,19 which is frequently expressed discontinuously by a reflex of the original preverbal negator NON ‘not’ and a postverbal grammaticalized minimizer (e.g. form derived from PASSUM ‘step’, RES/REM ‘thing’, MICAM ‘crumb’, PUNCTUM ‘point; stitch’), in some cases leading to the loss of the preverbal negator (Stage II > Stage III in terms of Jespersen’s (1917) cycle): (11)

a je (ne) vous mentirais I NEG to.you= would.lie ‘I wouldn’t lie to you’

pas ((written/formal) Fr.) NEG

b l’ ostal que soi nascut n’ es pas riche ni paure (Rgt.) the house that I.am born NEG is NEG rich nor poor ‘The household into which I was born is neither rich nor poor’ c quel qu’ aima pas de dançar coneis pas lo that.one that loves NEG of dance.INF knows NEG the plaser de viure (Auv.) pleasure of live.INF ‘he who does not like dancing does not know the pleasure of living’ d al n’ è briza it NEG is NEG ‘it’s not a bad job’

un a

brott ugly

emstir (Bol.) trade

e capissu nɛŋ (Pie.) I.undertstand NEG ‘I don’t understand’ f l’ aiga és pas the water is NEG ‘the water is cloudy’ g no

faré yo I.will.do I ‘I will not do it’ NEG

clara (Ros.) clear

pas (Ara.) NEG

The greater autonomy of such negative structures is, however, only apparent. For example, limiting ourselves just to French, we must note that with finite verbs pas is 19

See, among others, Price (1962; 1986), Posner (1984; 1985a, b), Schwegler (1990), Zanuttini (1997: ch. 3), Parry (1997), Veny (1982: 51); see also the discussion in }5.2.1.

18

From Latin to Romance

always immediately postverbal and can never, for example, be separated from the finite verb by a nominal (12a), participial (12b), or infinitival (12c) complement. Indeed, the only other elements that may precede pas are certain other adverbs, themselves occurring in a fixed order (cf. Cinque 1999: ch. 1) such as peut-être ‘perhaps’ (12d). (12)

a il (ne) mange pas he NEG eats pas ‘he doesn’t eat meat’

la the

viande meat

(**pas) (Fr.) pas

b il (n’) a pas mangé he NEG has pas eaten ‘he hasn’t eaten the meat’

(**pas) pas

la the

c il (ne) veut pas manger (**pas) he NEG wants pas eat.INF pas ‘he doesn’t want to eat (the) meat’ d il (ne) mange peut-être he NEG eats perhaps ‘he perhaps doesn’t eat meat’

pas pas

la the

viande (Fr.) meat la the

viande (Fr.) meat

viande (Fr.) meat

This lack of autonomy manifested in the rigidly fixed positioning of pas is hardly what would be expected a priori of a supposedly analytic structure; on the contrary, pas and similar postverbal negators display a high degree of bonding with respect to their associated verb. Analogous issues arise from a consideration of the Romance sentential and manner adverbs in -men(t(e)) and the Romance compound perfect discussed in (8b) and (9b–d) above, respectively. Beginning with the former, we note, on the one hand, how the evidence of coordination from modern varieties such as French, Italian, and Occitan points to the synthesized suffixal nature of the original ablative nominal MENTE ‘with (a) mind’ (e.g. ModFr. humblement et doucement ‘humbly and sweetly’, ModIt. villanamente e aspramente ‘cruelly and bitterly’, Lgd. doçament e simplament ‘quietly and simply’), whereas the same coordination facts in modern Ibero-Romance, on the other hand, reveal how -ment(e) retains a greater degree of analyticity (e.g. Cat. dolçament i suau ‘sweetly and soft(ly)’, Pt. intensa e constantemente ‘intense(ly) and constantly’, Sp. impensada pero providencialmente ‘unexpected(ly) but providentially’), parallel to old French and old Tuscan (OFr./OTsc. humeles e dulcement / umile e dolcemente (CDR 1163 / Nov. 4) ‘humbly (lit. ‘humble’) and gently’). Even greater autonomy can be seen in old Occitan where not only is the reflex of MENTE (variously -men(s), -menz, -ment) usually attached to just one of the adjectives in coordinated structures (usually the first, e.g. cruelmen et amara ‘cruelly and bitter(ly)’, but also suau e bellament ‘soft(ly) and beautifully’; Diez 1874: 429), but

Syntheticity and analyticity

19

the two constituent parts of the adverbial structure can be also separated by intervening material (e.g. epsa . . . ment ‘identical . . . ly’; Grandgent 1909: }105). Turning now to the so-called analytic perfect, this is characterized by converse patterns of syntheticity and analyticity: whereas in modern French, Italian, and Occitan varieties the component parts continue to display considerable autonomy, witness the interpolation of such elements as adverbs of varying sizes (13a–c; Monachesi 2005: 134), in modern Ibero-Romance auxiliary and participle, despite continuing to be written as separate words, never allow interpolation of any sort (13d–f): (13)

a il a énormément he has enormously ‘he’s eaten a great deal’

mangé (Fr.) eaten

b avevo continuamente sbagliato (It.) I.had continuously erred ‘I had continuously made mistakes’ c avèm pas encara enviadas de we.have not yet sent of ‘we haven’t yet sent any postcards’ d jà han (**jà) already they.have already ‘they’ve already replied’

cartas cards

postalas (Lgd.) postal

contestat (Cat.) replied

e não tinha (**nunca) tocado nunca not he.had never played never ‘he had never played the saxophone’ f siempre habías (**siempre) always you.had always ‘you had always told lies’

o the

saxofone (Pt.) saxophone

mentido (Sp.) lied

Examples like these, coupled with the thorny problem of how one is to correctly measure the autonomy of linguistic units (Schwegler 1990: ch. 2), frequently obscured by conventional, yet non-systematic, orthographic representations of the ‘word’ (14a–e),20 lead Schwegler (1990: 193) to conclude, somewhat unsatisfactorily, that Cf. also Fr. est-ce que . . . ? [ɛs(ə)kə] lit. ‘is-this that’ (‘is it that . . . ?’), an erstwhile morphosyntactically complex interrogative cleft which, synchronically, has been reduced in normal speech to a morphosyntactically simplex polar interrogative particle [(ɛ)sk]: 20

(i)

est-ce qu’ Q

il he

‘Does he smoke?’

fume? smokes

[(ɛ)ski(l)fym]

20

From Latin to Romance

the labels synthetic and analytic can, at best, be understood as nothing more than ‘the rough measure of the morphemic interdependence of speech units’ [italics A.L.].21 (14)

a ¿por qué? (Sp.) / for what ‘why?’ vs ‘because’

per for

què? (Cat.) vs what

porque (Sp.) / for.what

perquè (Cat.)22 for.what

b da capo / daccapo (It.) from head / from.head ‘from the beginning’ c bien que (Fr./Sp.) vs well that ‘although’

benché (It.) well.that

d tal vez (Sp.) vs peut-être (Fr.) vs chissà (It.) / benlèu (Gsc.) / talvez (Pt.) who.knows well.soon such.time such time can-be.INF ‘perhaps’ e não trabalhando (mais) (Pt.) vs not work.GER (anymore) ‘not working (anymore)’

ne(mai)lucrând (Ro.) not.(anymore).work.GER

Yet even adopting a relativized interpretation of the traditional usage still fails to make any intuitive sense of many developments. For example, in the wake of Schwegler (1990: xv), Vincent (1997a: 99–100) proposes a scalar definition of the synthetic-analytic distinction in terms of the degree of phonological and morphosyntactic autonomy borne by the constituent grammatical properties of a given construction. On this view, however, one of the most important consequences of the presumed synthetic to analytic drift, manifested in the gradual replacement of an original ‘free’ word order with a ‘fixed’ (S)VO order (cf. }3.3.2, }5.4.2.4), must now, despite the obvious contradiction, be treated as a synthetic development. In particular, the remarkable syntactic autonomy and independence of the core constituents of the Latin sentence which could, in accordance with pragmatic principles, not only occur in all possible permutations (see }3.2.1, }3.3.1), but whose internal structure, when complex, could, in certain cases and in specific registers, be scattered discontinuously across the sentence (see }3.2.1.1, }3.3.1.1, }5.4.3), must be taken as an indication of greater analyticity. By the same token, the greatly reduced positional autonomy, coupled with the increased semantico-syntactic interdependence, of the 21 Indeed, such principles lead Bauer (2006: 287–8) to consider the Romanian definite article, an enclitic conventionally written as a single word together with its associated noun or adjective (e.g. Ro. câine=le ‘dog=the’), a synthetic development, whereas its proclitic counterpart in the other Romance varieties, where it happens to be conventionally written as a separate word, is described as an ‘analytic article’ (e.g. Fr. le chien ‘the= dog’). 22 Cf. also Fr. parce que ‘for.this which’ vs Occ. per ço que ‘for this which’ (both ‘because’).

Syntheticity and analyticity

21

core constituents of the Romance sentence which can now only be interpreted relative to each other, and whose constituent parts are cohesively bound together, are to be understood within the current approach as a synthetic development. 2.2.3 Causal relations between analyticity and morphophonological erosion Many scholars further identify a strict correlation between the rise and fall of analyticity and syntheticity, respectively, with processes of morphophonological weakening and erosion. A prime example concerns the phonetic erosion of the case system which is often considered the trigger for an increased use of prepositions.23 By the same line of argument, the weakening of the Latin case system is also believed to produce an increasingly rigidified word order able to unambiguously distinguish between subject and object,24 a development which, in turn, correspondingly renders the original case system increasingly redundant (Bourciez 1956; Zamboni 2000: 102). Yet, claims that such grammatical changes do not happen unless they are rendered necessary by concomitant changes in the phonology are simply not borne out by the Latin or Romance evidence.25 Beginning with the increased use of prepositions, on the basis of late Latin evidence Adams (forthc. a: }3.1.1.1) questions whether it is ‘convincing to see the change as having a neat single cause, such as attrition of case endings determined by phonetic developments’ since a ‘switch to prepositions is attested long before the phonetic changes usually cited in this connection (loss of final -m and -s, shortening of long vowels in final syllables) took place’. Indeed, on this point Pinkster (1990b: 195–6) observes that as early as the archaic period Latin prepositions had already developed very specific uses and characteristics of their own and were not simply analytic alternatives to the morphological case system, despite the documentation of the loss of final consonants in nominal forms from the early inscriptions onwards (pp. 198–9), including nominative M.SG -S from around 250 BC (e.g. Cornelio (CIL I2.8) for classical CORNELIUS), though restored by the classical period and generally retained even in late non-literary texts (Adams forthc. a: }3.3), and the somewhat later ACC.SG -M (15; cf. also Allen 1978: 30–1): (15) honc oino ploirume cosentiont duonoro optumo this.ACC one.ACC most.NOM agree.PRS.3PL good.GEN.PL best.ACC fuise uiro (Lat., CIL I2.9) be.INF.PRF man.ACC ‘Most people agree that this man has been the best of the good men’

23 Grandgent (1907: 42–8), Muller and Taylor (1932: 65), Löfstedt (1959: 126), Väänänen (1966: 115–19; 1982: 195–7), Lakoff (1972: 189), Adams (1976b: 49), Bauer (1995: 137–9), Molinelli (1998: 147). 24 Vennemann (1974; 1975), Bichakjian (1987: 89), Bauer (1995: 5–6). 25 Sasse (1977), Harris (1978: 9), Bauer (1995: 7–6), La Fauci (1997: 41), Adams (forthc. a: }3.1).

22

From Latin to Romance

Although a trace of final -M continues to surface in the form of nasalization and lengthening of the preceding vowel in the classical period, from around 150 AD (Beckmann 1963: 180–2) the presence of -M in texts represents nothing more than a conservative spelling convention (Pinkster 1990b: 199; 1993b: 240; Adams forthc. a: }3.3). Thus, we see that the loss of final consonants, even from an early period, apparently had no repercussions on the case system which continued unscathed and unchanged for centuries (Pinkster 1990b: 200).26 Indeed, even when in later texts we apparently come across neologistic uses of prepositions in place of traditional case forms, the complements of such prepositions always occur in a particular case form,27 often an indiscriminate ablative or accusative employed as a generalized prepositional case as in in eadem diem (A.Vales. 56) ‘in same.ACC day.ACC’ (‘on that same day’; cf. classical Lat. IN EO DIE ‘on that.ABL day.ABL’), an observation which incontrovertibly demonstrates that neologistic prepositional uses cannot be determined by phonetic erosion (Pinkster 1993b: 243). Rather, the growing use of prepositions, as Adams (forthc. a: }3.1) convincingly demonstrates, cannot be considered a single unified development, but instead covers a variety of heterogeneous cases differentiated by chronology and register. Whereas it is commonplace to retrace the Romance prepositional dative and genitive to early Latin uses (e.g. Plautus) of the prepositions AD ‘to(wards’) and DE ‘(down/away) from’, respectively,28 Adams’s examination of such cases in early, classical, and late Latin reveals that the resemblance between Latin and Romance is misleading and merely superficial. In particular, the evidence he adduces highlights how there was no abrupt move from syntheticity to analyticity, inasmuch as the Romance prepositional types emerged slowly over centuries from the gradual broadening of classical uses long before processes of phonetic erosion could have played a role in any such developments. Not dissimilar considerations apply to the presumed effects of phonetic erosion of the case system on the progressive establishment of (S)VO. For example, Molinelli (1998: 147, 162) maintains that the change from OV to VO is to be seen primarily as a consequence of the erosion of the case system, as does Salvi (2004: 59–60, 85–6, 97–8, 202, 204) who interprets the emergence of Romance VO order as a consequence of the weakening of the case system, such that complements are no longer required to raise to the preverbal position (> OV) to check morphological case, but can now remain in situ in the postverbal position (< VO). This view is, however, substantially weakened by the crosslinguistic observation that many Indo-European 26 This conclusion is further confirmed statistically by Pinkster (1993b: 242) who finds the proportion of NPs vs adpositional phrases in Caesar’s prose to be 85:15, whereas an examination of the first 306 nominal phrases of the fourth-century Peregrinatio revealed only a slight increase in the number of adpositional phrases, namely 77 (25%). 27 Adams (1977: 36–7), Bauer (1995: 138), Clackson and Horrocks (2007: 290). 28 See, for instance, Lindsay (1907: 20, 83), Bauer (1995: 137), Molinelli (1998).

Syntheticity and analyticity

23

and non-Indo-European languages have developed an SVO order whilst retaining distinct case markers,29 not to mention the fact that the emergence of SVO in IndoEuropean languages has been shown to precede the loss of case.30 A further complication with the theory that phonetic attrition of the classical Latin case system necessitated a fixed Romance SVO order to ambiguously encode grammatical relations is that it is simply not empirically true. Rather, as we shall see in more detail in }3.3.2 and }7.3.2.2, late Latin and early Romance retained at least a binary case system (nominative vs oblique) and were characterized by a Verb Second (V2) constraint, such that SVO was just one of many possible surface word orders. From this we can only conclude that there is no necessary causal relation between phonetic attrition, in this instance acting upon the case system, and the emergence of analytic structural changes. 2.2.4 Gradual change and competition The wide spread survival in Romance of synthetic forms, particularly in the verbal domain, observed above in }2.2.1, also poses a serious challenge to traditional accounts which identity phonetic erosion with the rise of analyticity (Wüest 1998: 94). Indeed, the survival of such forms highlights the fact that the emergence of analyticity in Romance does not involve a wholesale unitary move away from synthetic to analytic structures, but involves gradual changes and extensive periods of complementarity between competing synthetic and analytic structures (Bauer 2006). Latin comparative constructions, for instance, have since the earliest times displayed variation in the encoding of the standard of comparison, which could be marked synthetically through the ablative (e.g. SOLE CLARIOR ‘sun.ABL clear.COMPR’: ‘brighter than the sun’) or analytically through the particle construction introduced by the complementizer QUAM ‘than’ (e.g. CLARIOR QUAM SOL ‘clear.COMPR.NOM than sun. 31 NOM’). However, as has long been recognized, in most cases the two were not simple free variants but were generally in complementary distribution, the ablative construction typically occurring, for example, in proverbial expressions, negative structures, and rhetorical questions (see }5.3.1.2). Remaining with the comparative, we may also note that alongside the synthetic formation in -IOR (as well as the superlative in -ISSIMUS) analytic formations with PLUS and MAGIS ‘more’ (and MAXIME or PLURIMUM ‘most’ in the case of the superlative) are also attested since the earliest period (e.g. CLARIOR vs MAGIS/PLUS CLARUS, CLARISSIMUS vs MAXIME/PLURIMUM CLARUS; Lindsay 1907: 38; Wüest 1998: 92), and indeed were obligatory with those adjectives ending in -VUS (e.g. PLUS / MAXIME IDONEUS ‘fitter / fittest’). In both cases, we know that it is the

29 30 31

Adams (1976a: 98), Bichakjian (1987: 89), Roberts (1997). Koch (1974), Miller (1975), Sasse (1977), Bichakjian (1987: 89), Bauer (1995: 7), Magni (2009: 247). Bennett (1914: 292–7), Adams (1976a: 83–6), Bauer (2009a: 263), Magni (2009: 243–6).

24

From Latin to Romance

analytic variant that wins through in Romance, but this was not an immediate or foregone conclusion in the Latin period. Similar early synthetic-analytic variations are found in other areas, including the expression of the future and the present perfect, where from an early date the synthetic forms were, to some extent, in competition with periphrastic HABERE formations (Pinkster 1987; Bauer 2006: 289). In the case of the future it is the analytic structure (itself eventually synthesizing) which ultimately triumphs (}4.3.1.5), but in the case of the perfect the original synthetic construction lives on and the competition between the two forms is far from over (Harris 1982), with the division of labour between the two available paradigms resolved differently across the Romània (cf. discussion of examples (9a–d) above). The case of the future considered here also raises the related issue of competing analytic structures: assuming that there was a general drift towards analyticity, how are we then to explain the fact that the already existing periphrastic future forms of the classical period (cf. discussion in }2.2.1 of SUM ‘I am’ + future active and passive participle) were themselves replaced by the new 32 HABERE future periphrasis (Pinkster 1987: 221)? In short, we have observed that it is demonstrably false to portray the passage from Latin to Romance in terms of a simple and unified linear development from the synthetic to the analytic. In many cases, synthetic and analytic developments occurred in parallel and co-existed over long periods of time as contrasting and/or stylistic variants, in some cases up until the present-day. Where analytic structures have won through, these typically show a gradual development rather than a catastrophic or saltational change and co exist alongside many original synthetic structures that have survived into Romance from the ancestral language. Furthermore, in some cases these same analytic innovations have ousted earlier classical periphrases and/or have gone on themselves to become synthetic, developments which are patently incompatible with the traditional thesis of an inexorable shift away from syntheticity towards ever-greater analyticity. 2.2.5 Explanatory power Without doubt, however, the biggest problem for the traditional synthetic-analytic interpretation of the Latin to Romance development is that it offers no explanation whatsoever for the observed changes. In short, the predominant analytic patterns noted in Romance are nothing more than the partial surface reflex of a more deeprooted structural change, variously interpreted in the following chapters as the result of the emergence of full configurationality (chapter 3) and related functional structure (chapter 4), a change in the head/branching parameter (chapter 5), a move from 32

ADJ),

For discussion of the variation in the formation of adverbs (e.g. MODO/MENTE ‘manner.ABL/mind.ABL’ + see Wüest (1998: 93) and Bauer (2003; 2006: 296–300).

Syntheticity and analyticity

25

dependent- to head-marking (chapter 6), and the resolution of a centuries-old conflict between accusative-nominative and active-stative alignments in the nominal and verbal domains (chapter 7). By way of illustration, one only has to consider the parallel analytic developments in the nominal and verbal domains such as the use of prepositions and auxiliaries replacing earlier inflections: here the chief issue is not the replacement of synthetic forms with analytic ones, but, rather, a structural change in linearization involving the head or branching parameter that affects both inflectional morphology and syntax alike.33 Thus, the principal innovation in inflectional morphology has, according to one view, been the move away from structures in which grammatical modification (head) follows the lexical element (modifier) to structures in which the relevant grammatical modification (head) precedes the lexical element (modifier): (16)

a MARC-o > Fr. Marcus-DAT ‘to Marc(us)’ b

à to

Marc Marc

COGITA-UERAT > Cat. think-PRF.PST.3SG ‘he had thought’

pensat thought

havia he.had

In syntax too, verbal and nominal heads, once frequently preceded by such modifiers as direct objects/genitives and manner adverbs/adjectives, come instead to precede all such modifiers: (17)

METUIT a Lat.MORTEM death.ACC he.fears ‘he fears death’

b Lat.MORTIS METUS death.GEN fear.NOM ‘fear of death’ c Lat.LIBERE UIUIT freely he.lives ‘he lives freely’ d Lat.LIBER HOMO free man ‘a free man’ 33

>

>

>

>

It.teme he.fears It.il the

It.vive he.lives It.un a

la the

morte death

timore fear

della of.the

morte death

liberamente freely

uomo man

libero free

von Wartburg ([1934] 1971: 256), Harris (1978: 6), Bauer (1995: 10, 24, 166), Oniga (2004: 52, 75). The reversal in the head parameter is so pervasive that it equally surfaces in the area of derivational morphology (cf. Lat. SILUICOLA lit. ‘forest.inhabitant’ (‘forest dweller’) vs Cat. guardabosc lit. ‘watch.forest’ (‘forester’)), although admittedly at a later date than in the areas of inflectional morphology and syntax (Oniga 2004: 52–3).

26

From Latin to Romance

These latter examples, which clearly do not involve analyticity, therefore highlight that the relevant change in linear order in syntax is consistent with that observed in inflectional morphology, ultimately both derivable from a single integrated and comprehensive structural change.34 Perhaps the only genuinely convincing attempt to invest the traditional syntheticanalytic dichotomy with some explanatory power is to be found in Coseriu (1987), who, however, reinterprets the distinction as one between ‘internal’ and ‘external’ structures, respectively, an opposition ultimately going back to his earlier work (cf. Coseriu [1971] 1988: 211). According to this view, the Romance languages are not characterized by a tendency towards analyticity, but, rather, by a tendency to distinguish between external and internal determination and between relational and non-relational functions. Although it is undoubtedly true that the internal (viz. synthetic) determination of syntactic structures predominates in Latin, Coseriu makes the original observation that this is not the distinctive structural characteristic of Latin; rather what sets Latin apart from Romance is its failure to make any formal distinction between relational and non-relational functions. By contrast, Romance stands out, not by its admittedly steady increase in the number of externally determined (viz. analytic) structures, but, rather, by its consistent formal distinction between relational and non-relational functions, the former aligned with externally determined (= analytic) structures and the latter with internally determined (= synthetic) structures. Thus, the real change in the passage from Latin to Romance is not to be seen as a move from syntheticity to analyticity, but is to be identified with the emergence of a new formative principle of mapping relational and non-relational functions respectively onto externally and internally determined structures, as consistently and uniformly attested across a whole series of otherwise superficially puzzling and unrelated changes from Latin to Romance (cf. also Coseriu [1971] 1988: 224). It therefore follows, for example, that the internally determined (paradigmatic) functions of the Latin case system or comparative are replaced by externally determined (syntagmatic) expressions in Romance, since they both involve relational functions which establish a relation between a given term and one or more other terms (e.g. X is ‘of ’ Y, or X is ‘bigger than’ Y). By the same token, number and gender marking on nominal categories are argued to instantiate non-relational functions and thus continue to be marked paradigmatically by internally determined expressions. As for verbs, Coseriu (pp. 61–2) maintains that simplex verbs convey non-relational 34

Harris (1978: 16) too sees the emergence of a specified-specifier order as central to the developments in the syntax (e.g. SVO) and morphology (e.g. loss of inflection) of Romance. Yet, he does not try to subsume the shift from synthetic to analytic within this linear change but, rather, continues to treat it as an independent, albeit isolated, phenomenon, ultimately part of a general tendency towards more explicit structures (see also Bourciez 1956: 23).

Syntheticity and analyticity

27

functions in that they pick out a single moment in time or temporal space, and hence are internally determined structures, whereas compound verb forms establish a relation between two points in time or space, and hence display externally determined structures.35 A further revealing case concerns the fate of the Latin synthetic passive, which Coseriu (p. 63) claims to subsume three distinct functions: (i) the objective passive (18a); (ii) the middle (18b); and (iii) the impersonal passive (18c). (18)

a ab eis ita amantur (Lat., Cic. Amic. 27) by them.ABL thus they.love.PRS.PASS ‘they are thus loved by them’ b aut cum Fabricius, aut Aristides iustus nominatur (Lat., or when Fabricius.NOM or Aristides.NOM just.NOM call.PRS.PASS.3SG Cic. Off. 3.16) ‘or when Fabricius or Aristides is called “the just”’ c sic itur ad astra (Virg. Aen. 9.641) thus go.PRS.PASS.3SG to stars.ACC.PL ‘so it is man rises to the stars’

Significantly, of these three functions, it is only the first that expresses a relational function between a PATIENT and an AGENT, which gives rise to the Romance periphrastic construction (19a), while the other two non-relational functions are aligned with apparently internally determined structures (19b–c):36 (19)

a ils they

sont are

b Fabrice Fabrice c c’ this

est is

aimés loved

s’ self= ainsi thus

par by

appelle calls qu’ that

eux (Fr.) them le the

on one

juste (Fr.) just atteint attains

les the

astres (Fr.) stars

35 While this analysis might be true for many of the Romance languages, especially in earlier periods, it runs into difficulties when we consider those varieties that still express the present perfective, a relational function, through an internally determined (viz. synthetic) structure such as southern Calabrian (cf. also the meaning of compound form in (9a) above):

(i)

ancora non mi chiamau (SCal.) yet not me= he.called ‘he hasn’t rung me yet’

Equally problematic are examples such as (9d) from spoken French, where the compound form can now also mark the apparently non-relational function of the punctual perfective. 36 As pointed out to me by G. Salvi (p.c.), in reality examples (19b–c) appear to be just as analytic as example (19a).

From Latin to Romance

28

Coseriu ([1971] 1988: 219–20; 1987: 62) also applies this model to sentential word order, arguing that changes in this area cannot be appropriately characterized as a shift from a ‘circular’ (viz. SOV) to a ‘linear’ (viz. SVO) arrangement, but are to be viewed once again in terms of the emergence in Romance of a formal distinction between internal and external functions that was not systematically made in Latin. Specifically, functionally internal determinations such as adjectives and modals which modify exclusively the subject and the verb, respectively, come to be placed in Romance ‘inside’ the sentential core (understood here as the syntagm consisting of the subject and verb) (20a–b), whereas functionally external determinations such as complements come to be placed ‘outside’ of the sentential core (21a–b): (20) a

BONUS

HOMO

good.NOM

man.NOM

scribere write.INF

b el hombre bueno debe the man good must ‘the good man must write’ (21)

a

DEBET

(Lat.)

must escribir (Sp.) write.INF

HOMO

SUO

AMICO

LITTERAS

man.NOM

his.DAT

friend.DAT

letter.ACC.PL

b el hombre escribe una carta the man writes a letter ‘the man writes a letter to his friend’

a to

(Lat.) he.writes

SCRIBIT

su his

amigo (Sp.) friend

2.2.6 Grammaticalization By way of conclusion, we should also note that the postulation of an independent synthetic-analytic parameter is further undermined by the observation that all presumed cases of analytic development can otherwise be independently subsumed within the general theory of grammaticalization (Hopper and Traugott 1993: 17), integrating synthetic forms as those having achieved the highest degree of grammaticalization (Klausenburger 2000: 105, 152; Bauer 2006: 288).37 More precisely, the analytic developments witnessed in the history of Romance are clearly not in any way exclusive to the Romance family, but simply exemplify a crosslinguistic tendency for 37 By contrast, Bauer (2006: 301) argues that syntheticity and analyticity in themselves are not an indicator of the degree of grammaticalization of a given structure, insofar as the synthetic or analytic outcome of individual grammaticalization processes is determined by branching patterns: left-branching structures produce synthesis whereas right-branching structures yield analysis. Superficially, at least, there are, however, numerous counterexamples to this generalization including, for example, the Latin typically left-branching perfective passive (e.g. SCRIPTUS EST write.PRF.PASS.PTCP is: ‘it has been written’; cf. }5.3.1.5) and the (apparently) Romanian right-branching enclitic definite article (e.g. bărbatul ‘man=the’: ‘the man’; }4.2.2.2).

Syntheticity and analyticity

29

synthetic structures, once weakened through phonetic erosion or other forces within the system, to be progressively replaced by new competing structures which ‘given the nature of syntactic change, cannot help but be analytic’ (Vincent 1997a: 101).38 Once again, analyticity turns out to be a secondary or epiphenomenal development, ultimately the manifestation of a deeper change but not, significantly, its cause.

38

Thoroughly unconvincing is the position taken by Wüest (1998: 96–7), who argues that the largely analytic tendencies of Romance are the result of a dramatic morphosyntactic restructuring in the Latin period due to language contact, rather than spontaneous internal change. In particular, he argues that during the period of Romanization a sizeable section of the population of the Empire would have had only a very fragmentary knowledge of Latin, speaking a rudimentary code he terms an ‘inter-language’. Their speech, he presumes, would have been characterized by such features as the imperfect mastery of morphology, which, consequently, they would have tended to replace with periphrases that would subsequently become established in the speech of the entire population. Apart from the fact that there is no evidence, direct or indirect, for this presumed source of analyticity or the presumed inter-language itself (cf. Adams 2003), among the many other objections to this theory it will suffice here to recall that many of the original morphological structures of Latin (together with many of their irregularities), especially in the verbal domain (Maiden 2011a), have been retained in Romance (see }2.2.1), a situation which is clearly incompatible with the inter-language scenario postulated by Wüest.

3 Configurationality and the rise of constituent structure 3.1 Introduction Developments traditionally falling within the realm of the synthesis-analysis parameter, as well as many more far-reaching and significant syntactic changes, find a much more promising and comprehensive explanation in terms of the rise of configurationality and, in particular, the concomitant emergence of functional structure (see chapter 4). In this respect, it has long been noted, although the notion has not been formalized, that there is little evidence in Indo-European, and also in Latin, for fixed constituent structure: The Indo-European sentence was made up of autonomous words, each of which was able to express independent meaning and mark its own function within the sentence. When nouns reduced, then lost, their case inflections, the function of nouns within the sentence became marked by two new processes: 1) A grammaticalized word order. In French or English, the place of the noun is generally sufficient to indicate its function: le père aime le fils [‘the father loves the [= his] son’] indicates by word order what Latin indicates by inflection pater filium amat [‘father.NOM son.ACC loves’], filium pater amat [‘son.ACC father.NOM loves’], amat filium pater [‘loves son.ACC father. NOM’], etc., such that word order is the sole marker of grammatical differences. 2) Auxiliary words. The structure of the sentence was therefore changed [ . . . ], and led to an increase in the rigidification of word order and in the creation of auxiliary words. (Meillet [1937] 1964: 439)1

Similarly, discussing the nominal group Meillet ([1928] 1977: 156) observes that in Romance ‘what marks an attributive adjective, a noun in apposition and a complement as being semantically bound together is their juxtaposition, their formation of a group’, whereas in Indo-European ‘groups were not bound together in this way. Each of the constituent elements, which were independently inflected for their own particular function, could be separated from the others. While it was natural to say TOGAM NOUAM INDUE [‘toga.ACC.F new.ACC.FSG put.on.IMP’], nothing ruled out such

1

All quotes in languages other than English have been translated by the author.

The rise of constituent structure

31

sequences as NOUAM INDUE TOGAM [‘new.ACC.FSG put.on.IMP toga.ACC.F’], or TOGAM INDUE NOUAM [‘toga.ACC.F put.on.IMP new.ACC.FSG’]’. In a similar vein, Herman (2000: 84) goes so far as to claim that the establishment of fixed positions for the constituent parts of the nominal group represents ‘one symptom of a wider change in the nature of the grammar, a change that is indeed one of the most far-reaching in the transition from Latin to Romance’. Thus, in the development from Latin to Romance it is conventional to recognize a shift from a so-called free word order to a more rigid word order in which semantically related words are increasingly grouped together into syntagms (Bauer 2009a: 243). In light of such considerations, some researchers,2 though not without their opponents,3 have claimed that the most significant innovation characterizing the transition from Latin to Romance is to be sought in the move away from a non-configurational syntax, in which the relationships between individual linguistic items is signalled through the forms of the items themselves (case inflections, agreement), towards an increasingly configurational syntax, in which the relationships between related linguistic items is encoded by their fixed positions relative to each other (cf. Hewson and Bubenik 2006: viii). From this perspective, traditional generalizations about Latin’s prevalent recourse to synthetic, or rather morphological, strategies, in contrast to those of a predominantly analytic, or rather syntactic, nature in Romance (cf. }2.1), can now be subsumed naturally within the configurationality parameter as neatly summed up by Bresnan’s (2001: 6) slogan ‘[m]orphology competes with syntax’. Below we examine these developments in configurationality in relation to the three principal domains of the grammar: the nominal and verbal groups, and the sentence.

3.2 Nominal and verbal groups Quintilian’s (Inst. 1.4.19) oft-quoted observation ‘noster sermo articulos non desiderat’ (‘our language does not need articles’;4 cf. de Jonge 2006: 158) can be taken to mean not simply that Latin lacked articles, but, more fundamentally, that it lacked a dedicated position for articles and other types of determiner. Indeed, this implication is explicitly assumed by Lyons (1999: 155), who interprets the lack of definiteness marking in languages with non-configurational nominal syntax as the absence of a corresponding D(eterminer) position (see also Gil 1987). Analogously, in much recent work Bošković (2005a, b; 2008; cf. also Bošković and Gajewski forthc.) has demonstrated from a crosslinguistic perspective that in languages lacking articles,

2

Vincent (1988: 53–4, 62–3; 1997c: 149, 163; 1998a: 423–4), Ramat (1994: 261), Lyons (1999: 154–5, 305–6), Hewson and Bubenik (2006: 2), Ledgeway (2011: 387–434). 3 See, among others, Pinkster (1990a: 186) and Oniga (2004: 100–1), and, for a detailed configurational analysis of Latin, see also the discussion in chapter 5. 4 Cf. also Russell’s (2001: 115) translation ‘Our language does not feel its lack of articles.’

32

From Latin to Romance

including Latin, nominals do not project a D(eterminer) position (and associated D(eterminer)P(hrase)), as witnessed, among other things, by the fact they contravene the ban (so-called ‘Left Branch Condition’) on the extraction of adjectives (1a) and adjuncts (1b) out of NP (Noun Phrase), which otherwise holds in languages with articles (such as Romance) and hence a D(eterminer) position (see further the discussion in }5.4.3.4).5 (1)

a qui [summam] inter eos habet [summam auctoritatem] who highest.ACC.FSG among them.ACC has power.ACC.FSG (Lat., Caes B.G. 6.13.8) ‘who has the highest authority among them’ b [eiusdem generis] habent instituta [eiusdem generic sacrificia] same.GEN kind.GEN they.have established sacrifices.ACC.NPL (Lat., Caes. B.G. 6.16.1) ‘they have established sacrifices of that same kind’

Consequently, the absence of an article in Latin and its presence in Romance can be taken as compelling evidence for the emergence in Romance of a cohesively structured nominal group, namely NP (and ultimately DP) structure (for full discussion, see }4.2). This view is further supported by the observation that other determiner-like elements such as demonstratives (2) and possessives (3), which fill the same syntagmatic slot (or area) as the article (Vincent 1988: 53–4.), also typically come to fill the prenominal determiner position in Romance (cf. (a) examples below), whereas in Latin—questions of markedness aside for the moment—they could occur in either pre- or postnominal position just like adjectives (cf. (b)–(c) examples):6 (2)

a cette this

province province

(**cette) (Fr.) this

b Cuspius [ . . . ] mirifice quosdam homines ex Cuspius.NOM extraordinarily certain.ACC.PL men.ACC from ista prouincia tuetur et diligit (Lat., Cic. Fam. 13.6.2) this.ABL province.ABL protects and values ‘Cuspius [ . . . ] takes a surprising interest in the well-being of certain persons of your province’

5 In all following examples we indicate the underlying base (and intermediate transit) position(s) of all moved elements by representing them in strikethrough. 6 Marouzeau (1922: 133–7, 155–64), Devine and Stephens (2006: 361–2, 511–20), De Melo (2010). For limitations of space and expository reasons, we are deliberately oversimplifying some complex patterns here. For example, De Melo (2010: 76–7) finds that prenominal position of the possessive is preferred in Plautus, but that postnominal placement is more frequent with MEUS (1sg), TUUS (2sg), and SUUS (3) than with NOSTER (1pl) and VOSTER (2pl), and that of the first three SUUS shows a greater tendency to postposition than MEUS and TUUS. For detailed discussion of these and other patterns, the reader is referred to the cited sources.

The rise of constituent structure

33

c tuam iustitiam secutus tutissimum sibi portum safest.ACC self.DAT port.ACC your.ACC justice.ACC followed.NOM prouinciam istam duxit esse (Lat., Cic. Fam. 13.66.2) province.ACC this.ACC he.considers be.INF ‘in reliance on your sense of justice he has concluded your province to be his safest harbour of refuge’ (3)

a César retire ses troupes (**ses) (Fr.) Caesar withdraws his troops his b Caesar suas copias subducit (Lat., Caes. B.G. 1.22.3) Caesar.NOM his.ACC.FPL troops.ACC.FPL withdraws c copias suas Caesar [ . . . ] troops.ACC.FPL his.ACC.FPL Caesar.NOM ‘Caesar withdraws his troops’

subducit (Lat., Caes. B.G. 1.24.1) withdraws

The Romance verb phrase (VP) sees a parallel development in the emergence of a profusion of auxiliaries associated with a dedicated structural position situated to the immediate left of the VP (4a–c; see further discussion in }4.3).7 (4)

a l’ Enric va passar un any a Londres (Cat.; va < anar ‘to go’) the Henry PST.AUX.3SG spend.INF a year at London ‘Enric spent a year in London’ b vor da banii orfelinatului (Ro.; vor < a vrea ‘to want’) FUT.AUX.3PL give.INF money=the orphanage=the.DAT ‘they will give the money to the orphanage’ c acaban de descubrir la verdad (Sp.; acaban < acabar ‘to complete’) they.finish of discover.INF the truth ‘they’ve just discovered the truth’

Significantly, Latin, by contrast, had very few auxiliaries, with verb categories such as tense, aspect, and mood, as well as person and number, overwhelmingly marked inflectionally. The chief apparent exception was the perfective passive and deponent

7 Apparent exceptions are the Romance synthetic future and conditional paradigms (Valesio 1968; Coleman 1971; Fleischman 1982; Pinkster 1987; Vincent 1987; Maiden 1996; Nocentini 2001; La Fauci 2006), which continue a reduced form of the present/imperfect (or, NCIt. preterit) of HABERE ‘have’ suffixed to the infinitive (e.g. PERDER(E) ‘lose.INF’ + *ajo/*ia ‘I.have/I.had’ (+ *ɛbwi) > Occ. perdrai/perdriá (It. perderei) ‘I will/would lose’). While some have interpreted the postverbal position of the erstwhile auxiliary in these periphrases as an indicator of their early origin in line with an archaic OV typology, Adams (1991) provides convincing evidence to support their later emergence (see }4.3.1.5).

34

From Latin to Romance

auxiliary ESSE ‘be’ which, unlike its Romance counterparts, could occur both in preand postverbal position (5a–b):8 (5)

a puer quia clam te boy.NOM because secretly you.ACC/ABL ‘for unknown to you a boy was born’

est is

natus (Lat., Ter. Hec. 681) born.M.SG

b natus est nobis nepos (Lat., Ter. Hec. 639) born is to.us nephew ‘a nephew has been born to us’ Its presence within the verb system can be readily compared to that of demonstratives in the Latin nominal group: both marked grammatical categories but, in the absence of a VP/NP constituent, neither had yet been formalized by way of a dedicated functional position. This view is echoed by (Bauer 1995: 106) who underlines how ‘[t]he evolution of Latin shows not the creation of the auxiliary as such, but rather a change in the nature of the auxiliary element and in the place it occupied’, such that by the Romance period auxiliaries began to distinguish themselves from lexical verbs in a number of formal respects and, above all, in their fixed position.9 Significant in this respect are recent proposals by Bošković (2010: }5) who convincingly demonstrates that languages that lack a D(eterminer) position within the nominal group correspondingly lack a dedicated inflectional position for auxiliaries T(ense), and associated T(ense)P(hrase) projection, within the clause. Among the many arguments in support of this analysis which can also be applied to Latin, we may note that Latin verbal morphology, like that of many Slavonic languages which also lack articles, overwhelming privileges aspect over tense. From this one could conclude—though not entirely without problems—that in Latin there is no grammaticalized tense, in short no TP projection. By the same token, if Latin does not grammaticalize tense failing to project TP, this also explains the absence of overt expletive subjects (e.g.  TONAT ‘(it) thunders’) since expletives are only employed to satisfy the EPP,10 a property of the TP projection. It thus falls out without further stipulation that Latin, which lacks articles, should also lack auxiliaries, inasmuch as both properties now naturally fall together.

8 Marouzeau (1938: 23–7), Bauer (1995: 104–5; 2006: 294), Salvi (2004: 112 n. 62), Devine and Stephens (2006: 180–98), Vincent (2007: 65), Bailey (2008: }2), Spevak (2010: 150–1). The periphrastic futures constructed with the future active and passive participles and ESSE ‘be’ (more rarely, in the passive, with the supine and IRI ‘go.INF.PASS’) are relatively rare and were most probably never popular in the spoken language (Wüest 1998: 95). 9 This explains our claim above that Latin boasts very few auxiliaries, inasmuch as elements traditionally labelled as auxiliaries in the descriptive literature (e.g. UELLE ‘want’, POSSE ‘can’, DEBERE ‘must’), although semantically akin to auxiliaries in expressing modal, aspectual, and temporal categories, are not distinguished syntactically from other verbal predicates (cf. Nuñez 1991: 64; Haverling 2010: 208) and indeed display, among other things, a great deal of freedom in their placement (cf. the discussion of the variable position of HABERE ‘have’ and ESSE ‘be’ in Bourova 2005; 2007; 2008; Bourova and Tasmowski 2007). 10 The Extended Projection Principle (EPP) is the formal stipulation in generative syntax intended to capture the generalization that all clauses must contain a grammatical subject.

The rise of constituent structure

35

The overall conclusion to be drawn from these preliminary observations is that marking of definiteness and various verb-related grammatical categories in Romance increasingly becomes associated with specific positions, namely the left edge of the NP (even in Romanian which has an enclitic definite article; see }4.2.2.3) and the left edge of the VP, whereas in Latin, in which the nominal and verbal groups were not configurationally structured, such categories were either not explicitly marked or had no fixed position (Vincent 1988: 52–3; 1993: 146–7). 3.2.1 Latin Now, if we take a closer look at the Latin nominal and verbal groups, we soon come to realize that not only did Latin lack dedicated positions for marking definiteness and various verb-related grammatical categories, but that it proves extremely difficult to justify the existence of any fixed positions for any constituent parts (Vincent 1988: 53, 60–1). Instructive in this respect are Herman’s observations about the nominal group, which apply with equal force to the verbal and sentential groups: In a language like Latin, there are two ways of indicating that different nominal elements belong to the same group [ . . . ] either through the appearance of an identical or equivalent case ending on each nominal element, namely ‘agreement’, or through the contiguous serialization of all such elements in the spoken or written chain. Of these only agreement is grammaticalized and obligatory serving to encode the precise confines of the syntagm, whereas positional contiguity represents nothing more than a frequent consequence, but in no way systematic, of the existence of a syntagm formed of various elements, which, at most, is a subsidiary side-effect. (Herman 1985: 346)

Such observations have given rise to the frequent, but only partially correct, claim that Latin word order is ‘free’, with all major constituents able to occur in apparently arbitrary locations within the sentence (Ross 1967: 42; Kessler 1995: 7). Yet, as Adams (1976a: 99) stresses, ‘the word order of Latin is not genuinely “free” in any useful sense of the term’. Instructive in this regard is the following observation: The word order of Latin is free, it is not indifferent. Free in the sense that, without exception, there are no obligatory, dedicated positions for each of the elements of the sentence. But not indifferent, because generally two different orders are not synonymous. (Marouzeau 1922: 1)

Marouzeau’s words highlight the now widely accepted view that Latin word order, while ‘grammatically’ free, is ‘pragmatically’ motivated,11 or as Meillet (1908: 330) put it, ‘word order had an expressive, and not a syntactic, value’, an observation which 11

Pinkster (1990a: 163–4), Salvi (2005: 436), Devine and Stephens (2006: 23), Hewson and Bubenik (2006: viii), Clackson and Horrocks (2007: 27), Baños Baños and Cabrillana (2009: 680), Bauer (2009a: 243), Spevak (2010: 27).

From Latin to Romance

36

goes back, at least, to Weil ([1844] 1887: 29–30) who discusses the three fictitious sentences in (6a–c): (6)

a

IDEM ROMULUS ROMAM same.NOM Romulus.NOM Rome.ACC ‘The same Romulus founded Rome’

b

HANC

c

CONDIDIT ROMAM ROMULUS (Lat.) founded.3SG Rome.ACC Romulus.NOM ‘Romulus founded Rome’

(Lat.) founded.3SG

CONDIDIT

ROMULUS (Lat.) this.ACC city.ACC founded.3SG Romulus.NOM ‘It was Romulus who founded this city’ URBEM

CONDIDIT

He notes that while the syntax is the same in each of three examples, insofar as in all three the subject is Romulus, the predicate founded, and the direct object Rome, ‘three different things are said in the three sentences’ because ‘the information that is to be imparted to another, the goal of the discourse, is different in the three forms of expression’ (p. 30; cf. also Devine and Stephens 2006: 3–4). For instance, the first would prove felicitous in recounting a series of events in the life of Romulus (in which Romulus is the established topic with the verb and object in wide focus), the second in showing a visitor around the city of Rome (in which Rome is the established topic with Romulus in narrow focus), and the third in talking about the foundation of various famous cities (with both Romulus and Rome in focus). In essence, this represents the view of pragmatic accounts such as Panhuis (1982: 2), who argues that ‘since the nominals and the verb are inflected, the order of the sentence constituents does not signal syntactic relations’ but ‘can be used to put the message in a particular communicative perspective’.12 Bearing in mind these pragmatic considerations, let us examine the construction of the nominal (7a–i) and verbal (8a–e) groups illustrated in the following Latin examples: (7)

a hic status rerum in Hispania erat / cum this.NOM state.NOM things.GEN in Iberia.ABL it.was when uidet quo sit in odio status hic he.sees which.ABL be.3SG in hatred.ABL state.NOM this.NOM rerum (Lat., Liv. 27.1.1 / Cic. Att. 2.22.1) things.GEN ‘this was the state of affairs in the Iberian Peninsula’ / ‘when he sees how fed up the people are with this state of affairs’

12 It has even been argued (see, for instance, de Jonge 2007: 225; Bailey 2008: }3.1) that such pragmatic notions as Topic and Comment were already present in Quintilian’s Institutio oratoria.

The rise of constituent structure

37

b nullum animal nisi exanimatum fulmine lightening.ABL no.NOM animal.NOM unless killed.NOM accenditur / animal nullum inueniri burns.PASS animal.NOM no.NOM find.INF.PASS potest can.3SG (Lat., Plin. N.H. 2.145 / Cic. Nat.D. 3.32) ‘no living creature can be burnt by lightening without being killed’ / ‘no animal can be found’ c Nempe is, qui et ipse [ . . . ] certe contra ipsum truly he.NOM who.NOM and self.NOM certainly against self.ACC Caesarem est congressus armatus / Caesar ipse, Caesar.ACC is entered.battle armed.NOM Caesar.NOM self.NOM qui illis fuerat iratissimus (Lat., Cic. Lig. 9 / Phil. 8.19) who.NOM them.DAT had.been most.angry.NOM ‘Why, the very man who himself [ . . . ] certainly was in arms and fought against Caesar himself ’ / ‘Caesar himself, who had been the most angry of all men with them’ d sed ego mea culpa [ . . . ]maceror / id non but I my.ABL guilt.ABL I.destroy.PASS this not est culpa mea (Lat., Pl. Cist. 76. / Merc. 774) is guilt.NOM my.NOM ‘I’m tormented by my guilt’ / ‘this is not my fault’ e in hac causa frumentaria cognoscenda / pecuniam in this.ABL case.ABL.F of.grain.ABL.FSG get.to.know.GER money.ACC.FSG frumentario nomine ereptam (Lat., Cic. Verr. 2.3.11 / 2.3.49) of.grain.ABL.NSG name.ABL.NSG seized.ACC.FSG ‘in investigating this case about grain’ / ‘money seized under the heading of the grain tax’ hominis nisi hominis uita f pro uita for life.ABL man.GEN unless man.GEN life.NOM reddatur (Lat., Caes. B.G. 6.16.2) is.returned ‘unless for the life of a man a man’s life be paid’ g Hunc, quem saepe uides intra penetralia this.ACC.MSG which.ACC.MSG often you.see among recesses nostrae Polladas [ . . . ] cum baculo peraque our.GEN Pallas.GEN with stick.ABL wallet.ABL=and senem / In Ueienti agro biceps old.man.ACC in Veii district two.headed.NOM natus puer [ . . . ] et Auximi puella born.NOM.MSG boy.NOM and Ariminum.LOC girl.NOM

38

From Latin to Romance

cum dentibus teeth.ABL (Lat., Mart. 4.53.1–3 / Liv. 41.21.12) with ‘this old man with a stick and wallet, whom you often see in the recesses of the temple of our Pallas’ / ‘In the district of Veii a boy was born with two heads [ . . . ] at Ariminum a girl was born with teeth’ h C. Uolusenus, tribunus / Omnium doctrinarum Gaius.NOM Volusenus.NOM tribune.NOM all.GEN.PL teachings.GEN inuentrices Athenas (Lat., Caes. B.G. 3.5.2 / Cic. De or. 1.4.13) inventors.ACC Athens.ACC ‘Gaius Volusenus, a tribune’ / ‘Athens, the inventor of all branches of learning’ (8)

a hic paulisper est pugnatum / ita uario here for.short.time is fought thus variable.ABL certamine pugnatum est (Lat., Caes. B.C. 3.67.5 / 1.46.3–4) contest.ABL fought is ‘there was fighting here for a short while’ / ‘thus the battle was fought with variable fortunes’ b Ad haec Caesar respondit / Quibus respondit to these.ACC.N Caesar.NOM replied which.ABL.PL replied Caesar se condiciones dare, non accipere Caesar.NOM self.ACC conditions.ACC give.INF not accept.INF consueuisse (Lat., Caes. B.G. 2.32.1 / B.Hisp. 13) be.accustomed.INF.PRF ‘To these things Caesar replied’ / ‘To this Caesar answered, “That it was his custom to give, not accept conditions”’ c liceatque populo Romano [ . . . ] bellum gerere / war.ACC wage.INF be.licit.SBJV.3SG=and people.DAT Roman.DAT neque illi gerere bellum possunt (Lat., Liv. 34.60.2 / 26.43.5) nor=and they wage.INF war.ACC they.can ‘and if Rome were free [ . . . ] to wage war’/ ‘and they cannot wage the war’ d bene qui latuit bene uixit / nec uixit male, qui nor lived badly who.NOM well who.NOM escaped.notice well lived natus moriensque fefellit (Lat., Ov. Tr. 3.4.25 / Hor. Epist. 1.17.10) born.NOM dying.NOM=and slipped.away ‘one who lives well, lives unnoticed’ / ‘nor has he lived ill, who from birth to death has passed unknown’

The rise of constituent structure

39

e tum, postquam ad te uenit, mensis agitur hic then after to you she.came month.NOM passes.PASS here iam septimus / tremo horreoque postquam already seventh I.tremble I.shiver=and after aspexi hanc (Lat., Ter. Hec. 394 / Eun. 84) I.saw her ‘then, this is but the seventh month since she came to you’ / ‘I quiver and shiver since I have seen her’ What the linear alternations in these examples demonstrate is that the various elements that make up the nominal and verbal groups—including demonstratives (7a), quantifiers (7b), intensifiers (7c), possessives (7d), adjectives (7e), genitives and other adnominal complements (7f), adjuncts (7g), and appositions (7h); auxiliaries (8a), subjects (8b), complements (8c), adverbs (8d), and adverbial adjuncts (8e)—may occur either to the left or the right of their associated nominal or verbal predicate. Admittedly some of these elements are reported to display an unmarked order, such that demonstratives, quantifiers, and intensifiers normally precede their associated nominals,13 whereas the postnominal position is favoured by possessives (though see De Melo 2010: 76), appositions, genitives, and other adnominal complements.14 Similarly, grammatical tradition has it that (classical) Latin preferred verb-final structures (Rubio [1972] 1982: 191–9), as noted by Quintilian (Inst. 9.4.26) who claimed that ‘it is far better to end a sentence with a verb if the composition so permits’ (cf. however Pinkster 1993b: 245), and ‘the older the text, the more regular this appears’ (Watkins 1964: 1039; cf. also Adams 1976a: 91–2; Bauer 2009a: 251). Thus, while there would appear to be general consensus that the unmarked position of the verb in classical Latin, particularly in the highest literary registers, is clausefinal (see }5.3.2),15 and hence typically preceded by all other elements of the verbal group, the verb may also occur in clause-medial position.16 As various scholars have

13 Gildersleeve and Lodge ([1895] 1997: 430 n. 1), Hale and Buck ([1903] 1994: }624), Fischer (1908), Marouzeau (1922: chs IV–VI), Muldowney (1937: 73), Feix (1934: 29–30.), Ernout and Thomas ([1953] 1993: 162), Bauer (1995: 80, 166), Devine and Stephens (2006: 492–520), Vincent (2007: }5), De la Villa (2010: 208–19). 14 Gildersleeve and Lodge ([1895] 1997: 430), Hale and Buck ([1903] 1994: }624), Ernout and Thomas ([1953] 1993: 162), Adams (1976a: 88), Vincent (1988: 54), de Jong (1983: 131–2.), Bauer (1995: 64–5; 79; 166), Devine and Stephens (2006: 361–8); Tarriño Ruiz (2009a: 260). 15 Kroll (1912), Kühner and Stegmann (1912–14 II: 611–2), Linde (1923), Perrochat (1926), Meillet ([1928] 1977: 120), Fankhänel (1938), Marouzeau (1938: 47), Ernout and Thomas ([1953]1993: 161), Adams (1976a), Elerick (1989a), Pinkster (1990a: 168), Bauer (1995: 90–2), Herman (2000: 86), Oniga (2004: 97). The clausefinal position is also reported to be the unmarked position of the finite verb in Proto-Indo-European (Delbrück 1900: 83; Watkins 1964: 1039–41; Hofmann and Szantyr 1965: 403–6; Lehmann 1974: 114; Konneker 1975: 367; Adams 1976a: 92; Bauer 1995: 88–9). 16 Ernout and Thomas ([1953] 1993: 161), Adams (1976a), Väänänen (1982: 259–60), Vincent (1988: 61), Oniga (2004: 98–9).

40

From Latin to Romance

noted,17 this latter option is most common when the postposed elements are complements of the verb.18 Crucially, though, in none of the cases considered above is the opposite, albeit marked, order ever excluded (Vincent 1988: 60), thereby providing significant proof for the absence of a predetermined positional template within the Latin nominal and verbal groups. This view finds further support in the observation that in some cases there is little agreement in the literature as to whether it is at all possible to exhaustively identify an unmarked position for some elements (cf. De Melo’s (2010) nuanced analysis of the position of possessives) and, if so, which of the two positions should be considered unmarked. This is hardly surprising considering that we are dealing with at least 1400 years of documented Latinity (viz. from 600BC, the date of the first Latin inscription, to 842, the date of the first Romance text to have survived to the present) during which, besides diachronic factors, choice of author and text type may also greatly influence the relevant patterns.19 Instructive in this respect is Elerick’s (1989a: 570) well-founded conclusion that ‘“Latin”, with no author, register, era or genre specification, is rarely a proper object of linguistic investigation’, a view shared by Cabrillana (1999: 327) who maintains that ‘we cannot generalize in an absolute way with regard to the basic arrangement of constituents for Classical Latin nor for that of a later period’, especially since ‘the transformation of the order of constituents is not homogenous; it does not take place at the same speed nor in the same way in all the different types of texts, authors, or geographic areas’.20 For example, despite the frequent claim reported above (together with associated references) that genitives tend to favour the postnominal position, Bauer (1995: 55–9; cf. also 2009a: 266) notes, chiefly on the strength of the evidence of Adams (1976a), that in classical Latin preposed and postposed nominal genitives occur in equal number (a view echoed by Elerick 1994: 70; Oniga 2004: 76; Crisma and Gianollo 2006: 73; Gianollo 2007: 66–7) and, indeed, that preposed pronominal genitives

17

Linde (1923), Feix (1934: 13–15), Marouzeau (1938: 87–8), Bauer (1995: 87–98). As G. Salvi (p.c.) points out, ‘medial position’ is often a misnomer, since in many such cases the verb is not actually in a clause-medial position, but continues to occur towards the end of the clause. In particular, while some constituents may occur in postverbal position (viz. SVX), other constituents continue to precede the verb (viz. SXXVX), thereby revealing the near-final position of the verb. 19 Amacker (1989: 493, 499), Elerick (1989a: 560), Pinkster (1990a: 186; 1991: 71; 1993a: 646; 2005b: 252). 20 This explains Devine and Stephens’s (2006) deliberate choice to restrict their analysis of Latin word order to an idealized and limited, although arguably coherent, textual sample: ‘While the general target of our work is the simple sentence in Classical prose [ . . . ], we have felt free to pick and choose both texts and lexical material suitable for our purposes. The primary aim in each case was to find data that might clarify the theoretically significant issues, not to attempt broad coverage of the subject matter, whether relative to periods, registers and styles or relative to the range of available lexical material. Consequently our analysis proceeds mainly on the basis of probes or tests, using just a few representative words in a restricted corpus of texts to build the elements of a general theory of Latin word order. [ . . . ] It seemed sensible at this stage to gravitate towards the clearest evidence and to leave for future research data involving a less transparent calculus of conditioning factors’ (p. 8). 18

The rise of constituent structure

41

outnumber their postposed equivalents (for medical Latin, see Langslow 2000: 242–53). If we also take chronological factors into consideration, then it appears that in the archaic period prenominal position was favoured whereas postposition came to predominate by the late period.21 Even greater uncertainty surrounds the question of adjective placement in Latin.22 For example, Herman (2000: 83) reports that ‘either order was possible in Classical Latin, and this remained the case in Late Latin too’, as well as in vulgar texts where ‘it is just as common to find the adjective first as it is to find the noun first’.23 Others, by contrast, variously identify the prenominal position (e.g. Gildersleeve and Lodge [1895] 1997: 430) or the postnominal position as unmarked.24 Yet others recognize a semantic opposition in the interpretation of the pre- and postnominal positions of the type in (9a–b), including such values as affective/qualifying vs intellectual/ distinguishing (Marouzeau 1922: 15 / 1953: 1), epithetical vs attributive (Ernout and Thomas [1953] 1993: 162), emphatic/subjective vs objective (Adams 1971: 12; 1976a: 88–9; Baños Baños and Cabrillana 2009: 703), qualifying vs determinative (Väänänen 1982: 260), focused vs non-focused (Pinkster 1990a: 185), descriptive vs distinctive (Bauer 1995: 67–72; 2006: 298), and intensional vs extensional (Fugier 1983; De Sutter 1986; Devine and Stephens 2006: 481–2), according to a distinction which, despite the terminological variation, Vincent (2007: 64–5) convincingly argues to be the precursor of the modern Romance situation. (9)

a dulces uoluptates (Lat., Cic. Fin. 2.30) sweet.NOM.PL desires.NOM. ‘sweet-flavoured pleasures’ b aqua dulcis (Lat., Cato Agr. 106.1) water.NOM fresh.NOM.SG ‘fresh water’

Similar ambiguity in opinions and analyses can be found in relation to the verbal group (for an overview, see Cabrillana 1993b). For instance, Panhuis (1982: 23) maintains that ‘the tendency of the verb to be final is true only for certain authors [ . . . ] and is not a general feature of the Latin language in most of its history’, a view confirmed by Pinkster (1990a: 168). Indeed, above we noted that the verb may also occur in clause-medial position, especially when it co-occurs with a complement. 21 Adams (1976a: 73–4; 1976b: 140; 1977: 70–1), Gianollo (2007: 76–7), Bauer (2009a: 265–8), Magni (2009: 238). 22 For an overview, see Giusti and Oniga (2007: 81–5), Spevak (2010: 224–38), Langslow (forthc.: }2). 23 See also Feix (1934: 27), Vincent (1988: 54), Oniga (2004: 95), Crisma and Gianollo (2006: 73–4), Gianollo (2007: 68–9). 24 Hale and Buck ([1903] 1994: }624), Adams (1976a: 88; 1977: 71), Pinkster (1990a: 185), Coleman (1991: 326), Lehmann (1991: 223), Bauer (2009a: 265), Magni (2009: 236–7), Tarriño Ruiz (2009a: 260–1), Spevak (2010: 225), Langslow (forthc.).

42

From Latin to Romance

Marouzeau (1938: 82), however, goes so far as to claim that the choice between verbfinal and verb-medial position was indiscriminately free, the latter frequently proving almost as common as the former in some texts, a view endorsed by Bauer (1995: 97) who concludes that the verb-medial position was never stylistically marked (cf. Ross 1970).25 Adams (1976a), on the other hand, prefers to see in classical Latin (S)OV a stylized order artificially preserved from an earlier OV stage of the language (see }5.3.2), as demonstrated by the evidence of subliterary texts of the first centuries AD where (S)VO order predominates (cf. also Clackson and Horrocks 2007: 29–30). Indeed, it has been frequently pointed out that,26 while the verb-final position prevails in main clauses in the prose of such authors as Caesar, accounting for around 80–90 percent of all cases,27 this percentage lowers to 59 percent in Plautus (Adams 1976a: 90–8), varies between 32 percent and 52 percent in Cicero, and is as low as 33 percent in Varro. Herman (2000: 86) goes even further, maintaining that the ‘categorization of the language as basically having SOV order is exaggerated, even as regards the Classical language; we can tell that in other genres of a less-elevated nature than historiography, such as in Cicero’s Dialogues, for example, the statistics are not the same as in Caesar: here, verb-final sentences are not the dominant type’. A similar degree of freedom is also reported for adverbial adjuncts,28 whose position in relation to the finite verb appears to be the least fixed. In view of these facts, it would not appear rash to interpret the Latin data as indicative of an early stage (see further }3.3, }5.3.1), inasmuch as the older IndoEuropean tendency to prepose nominal and verbal modifiers, as particularly evidenced by archaic/early Latin and comparative Oscan and Umbrian evidence,29 shows a progressive weakening through time. In particular, we witness a steady increase in postnominal and postverbal positioning of particular elements (especially dependent genitives and complements of the verb) of the nominal and verbal groups, whilst other elements (e.g. adjectives and adverbs) appear to freely oscillate between

25 The exception here is subordinate clauses (see }5.1.2, }5.4.2.4), where the verb-final position proves most resilient, continuing even into early Romance (Foulet 1923: 248, 268; Linde 1923; Bauer 1995: 91, 108; Oniga 2004: 99–100; Salvi 2011: 366–8). 26 Linde (1923: 154–5), Elerick (1989b: 1), Pinkster (1990a: 168–7), Bauer (1995: 90–1), Herman (2000: 86), Zamboni (2000: 102). 27 Indeed, Linde (1923: 154) describes Caesar as ‘fanatical about verb-final position’ (cf. also Pinkster 1996: 253). 28 Kühner and Stegmann (1912–14: 613), Adams (1976b: 139; 1977: 72), Pinkster (1990a: 168), Devine and Stephens (2006: 119–23), Bauer (2009a: 270), Danckaert (2010). 29 For the nominal group, see, among others, Gildersleeve and Lodge ([1895] 1997: 431), Delbrück (1900: 102), Rosenkranz (1933), Lehmann (1974: 74), Konneker (1975: 370), Bichakjian (1987), Vincent (1988: 56), Bauer (1995: 51–3), Magni (2009: 235–6, 238), De Melo (2010); and for the verbal group, Delbrück (1900: 83), Hofmann and Szantyr (1965: 397–410), Watkins (1964: 1039–41), Lehmann (1974: 114, 238), Konneker (1975: 367), Adams (1976a: 92), Panhuis (1982: 110–12), Bichakjian (1987), Bauer (1995: 88–9; 2009a: 250–2), Magni (2009: 227).

The rise of constituent structure

43

both positions (Pinkster 1990a: 168; Bauer 1995: 166). The result is a loosely defined nominal and verbal group with a remarkably free word order. 3.2.1.1 Discontinuous structures Indeed, the loose organization of the nominal and verbal groups is further substantiated by the existence of discontinuous structures,30 ‘one of the most distinctive features of Latin with regard to Romance’ (Väänänen 1982: 259), in which the expected logical contiguity between dependent elements (heads and modifiers) is interrupted. In reality, the discontinuous structures which go under the label of hyperbaton include several different types (Powell 2010: 168–71); for the moment, however, we limit our attention to the following representative examples of local disjunction, deferring a more detailed examination of these structures until }5.4.3: (10)

a infestam rei publicae dangerous.ACC.F thing.DAT.F public.DAT.FSG ‘a plague dangerous to the state’

pestem (Lat., Cic. Cat. 1.11) plague.ACC.F

b complures eiusdem amentiae scelerisque many.ACC.MPL same.GEN.SG madness.GEN crime.GEN=and socios (Lat., Cic. Cat. 1.8) allies.ACC.MPL ‘many associates in the same madness and crime’ In (10a) the adjective infestam ‘dangerous’, although directly modifying the noun pestem ‘plague’, is separated from the latter by its own dative complement rei publicae ‘to the state’. Similarly, in (10b) the quantifier complures ‘many’ is separated from the noun socios ‘associates’ over which it ranges by the intervening genitive modifier eiusdem amentiae scelerisque ‘of the same madness and crime’ (though see Powell 2010: 168). This same discontinuous pattern proves particularly common in sequences with nominals governed by prepositions (see 11a–c), in which an accompanying modifier (e.g. adjective, adverb, adnominal genitive) is placed before the preposition, thereby stranding its associated noun.31 More rarely, the noun itself can be placed before a governing preposition, stranding any accompanying modifiers (see 11d; }5.3.1.1), a usage principally limited to verse (Marouzeau 1953: 67; Bauer 1995: 136; 2009a: 287): (11) a

magno cum great.ABL.MSG with ‘with great grief ’

dolore (Lat., Cic. Q.F. 2.3) PAIN.ABL.F

30 Such discontinuous structures are generally termed hyperbaton or disjunction (cf. Powell 2010: 164) in the classical tradition, and scrambling in the generative literature (Corver and van Riemsdijk 1994a,b). 31 Gildersleeve and Lodge ([1895] 1997: 432), Hale and Buck ([1903] 1994: }627), Marouzeau (1953: 58–62, 68), Vincent (1988: 54), Bauer (1995: 131, 136), Kessler (1995: 5, 7), Oniga (2004: 102).

44

From Latin to Romance b pauca in uerba few.ACC.NPL in words.ACC.N ‘condense in a few words’ c quem which.ACC.MSG ‘to what end?’

ad to

confer (Lat., Pl. Rud. 661) condense.IMP.2SG

finem? (Lat., Cic. Cat. 1.1) end.ACC.MSG

d arbusta per timber.trees.ACC.N through ‘through tall timber-trees’

alta (Lat., Enn. Ann. 6.181) tall.ACC.NPL

Similar examples of discontinuous structures are found within the verbal domain. For instance, in (12a–b) the constituent parts of the prepositional complement in duas partes ‘in two parts’ and the direct object nostram inuidiam ‘our unpopularity’ are variously divided between the pre- and postverbal positions. It is also possible to find mixed structures, in which some elements of the verbal group precede the verb while others follow it. For example, in (12c) the verb follows the frequentative adverb saepe ‘often’ but precedes its complement omnia ‘all things’, and similarly in (12d) the verb again precedes its complement ueniam ‘pardon’ but follows the adverbial petentibus Aeduis ‘on the request of the Aedui’: (12)

a animaduerti [ . . . ] orationem in duas I.realized accusation.ACC.FSG in two.ACC.FPL diuisam esse partes (Lat., Cic. Clu. 1.1) divided.ACC.FSG be.INF parts.ACC.FPL ‘I realized [ . . . ] that the accusation is divided in two parts’ b nostram [ . . . ] ridebant inuidiam (Lat., Petr. Sat. 14) our.ACC.FSG they.laughed unpopularity.ACC.FSG ‘they mocked at our unpopularity’ c Dies intermissus aut nox interposita day.NOM.MSG suspended.NOM.MSG or night.NOM.FSG interrupted.NOM.FSG saepe perturbat omnia (Lat., Cic. Mur. 35) often disturbs all.things.ACC ‘Often a day or night’s delay disrupts everything’ d Caesar petentibus Aeduis dat Caesar.NOM requesting.ABL.PL Aedui.ABL.PL gives ueniam (Lat., Caes. B.G. 6.4.3) pardon.ACC ‘Caesar, upon the request of the Aedui, grants the pardon’

Now, such discontinuous structures within the nominal and verbal domains undoubtedly prove most frequent in early Latin and, in the classical period, in

The rise of constituent structure

45

deliberately archaic or artistic prose styles and poetry,32 leading Adams (1971: 1) to conclude that hyperbaton ‘was artistic rather than natural to ordinary speech’. Nonetheless, Oniga (2004: 101–2) and Bauer (2009a: 291) report that hyperbaton occurs in all linguistic registers of all periods and, according to Vincent (1988: 54), even ‘in mundane prose’ and hence ‘not unrepresentative of ordinary usage’.33 Indeed, Powell (2010: 179) has recently argued that hyperbaton ‘does not seem [ . . . ] to be particularly marked for register’ and must have been ‘natural to the language as spoken in Cicero’s time’, a view that De Melo (2010: 76, 79) equally substantiates for early Latin on the basis of possessive hyperbaton in Plautus. A similar conclusion is reached by Herman (2000: 82), who calculates that the contiguity of noun and associated adjective is frequently interrupted in classical prose texts (about 30 percent of cases in Caesar and 15–20 percent of cases in the philosophical works of Cicero).34 This view is all the more persuasive when one considers that, alongside its stylistic functions, disjunction is also generally recognized to frequently act as a pragmatic device,35 marking, for example, focalization through discontinuous fronting (}5.4.3), an aspect of language just as important to the spoken as the (formal) written language. Consequently, it would be incorrect to dismiss discontinuous structures as a purely literary artifice, especially since typologically their presence is entirely consistent with the other types of positional freedom in the nominal and verbal groups witnessed above. Indeed, identical discontinuous patterns are far from uncommon in many of the world’s non-configurational languages (cf. Powell 2010: 166–7), even those that lack a written tradition, such that ‘we must be on our guard against ruling out all poetic usage simply on the grounds that it is poetic’ (Vincent 1988: 28) since ‘[d]ecades of scholarship on poetic language, or “Dichtersprache”, cannot obscure the fact that Latin poetry is written in Latin’ (Habinek 1985a: 51). 3.2.1.2 Conclusion To conclude, we have demonstrated how, at least according to one view, it may be argued that within the nominal and verbal domains there is no evidence for fixed constituent structure: not only can the constituent parts of the nominal and verbal groups occur both to the left and to the right of their associated noun/verb, but that even such ‘relaxed’ contiguity is not a necessary condition,

32 Grandgent (1907: 30), Marouzeau (1949: 42; 1953: 62), Ernout and Thomas ([1953] 1993: 162), Adams (1971: 1–2), Pinkster (1990a: 186), Bauer (1995: 131–2). 33 Cf. also Fraenkel (1928: 76), Hofmann and Szantyr (1965: 689), Kessler (1995), De Melo (2010: 79), Spevak (2010: 23–4). 34 Even if such patterns represent a more archaic usage which was less and less typical of everyday speech in the classical and subsequent periods, this only pushes the apparent change in configurationality back to an even earlier stage, and which still requires an explanation in the Latin to Romance transition. 35 Marouzeau (1922: 216), Panhuis (1982: 80), Kessler (1995: 10), Pinkster (2005a), Clackson and Horrocks (2007: 227), Foubert (2007: 57, 60), Giusti and Oniga (2007: 93), Bauer (2009a: 286), Powell (2010: 176–7).

46

From Latin to Romance

insofar as semantically dependent constituent parts can be scattered discontinuously across the group. It is such observations which lead Lehmann (1991: 229) to conclude that the Latin nominal group ‘is not integrated tightly enough to be called a phrase’, inasmuch as ‘there is neither tight syntactic cohesion in such syntagms nor a pronounced hierarchical inequality between the substantival head and the various kinds of modifiers’, a conclusion readily applicable also to the verbal group.36 Naturally, such positional freedom is afforded by the rich case and agreement inflections of the Latin nominal and verbal domains, which ensures that dependencies between all constituent parts, whatever their position and whether contiguous or not, are readily identified by their morphological shape and not exclusively, if at all, by linear syntactic arrangement. For this reason, it has been claimed that the Latin nominal and verbal groups have a ‘flat’ or nonconfigurational structure, in which the various dependencies between the constituent parts of nominal and verbal structures such as ‘an author of a great book’ (namely, SCRIPTOR ‘author.NOM.M’, MAGNI ‘great.GEN.MSG’, LIBRI ‘book.GEN.M’) and ‘(he) wrote a great book’ (namely, SCRIPSIT ‘he.wrote’, MAGNUM ‘great.ACC.MSG’, LIBRUM ‘book.ACC.M’) are not signalled by their respective positions, witness the grammaticality of all possible permutations illustrated in (13a–d) and (14a–d). (13)

a SCRIPTOR MAGNI LIBRI (Lat.) b SCRIPTOR LIBRI MAGNI (Lat.) c MAGNI LIBRI SCRIPTOR (Lat.) d MAGNI SCRIPTOR LIBRI (Lat.) e LIBRI MAGNI SCRIPTOR (Lat.) f LIBRI SCRIPTOR MAGNI (Lat.)

(14)

a SCRIPSIT MAGNUM LIBRUM (Lat.) b SCRIPSIT LIBRUM MAGNUM (Lat.) c MAGNUM LIBRUM SCRIPSIT (Lat.) d MAGNUM SCRIPSIT LIBRUM (Lat.) e LIBRUM MAGNUM SCRIPSIT (Lat.) f LIBRUM SCRIPSIT MAGNUM (Lat.)

Adopting a tree-based representation, it is not possible therefore to assign to the Latin nominal and verbal groups a hierarchically organized constituent structure based on the notions of precedence and dominance (more technically, c-command). Rather, the only representations possible are the corresponding ‘flat’ structures in (15a–f), in which all constituent parts are assigned equal status:37

36

See also Vincent (1988: 53, 62; 1997c: 163), Devine and Stephens (2006: 119), Magni (2009: 235). For discussion and evidence in favour of assuming a flat(ter) constituent structure more generally, see Dryer (2009: }3). Of course, it is possible to maintain a configurational representation, if one assumes that 37

The rise of constituent structure (15)

a

N/V

9

b

N/V

9 MAGNI LIBRI MAGNUM LIBRUM

e

d

9 LIBRI MAGNI SCRIPTOR LIBRUM MAGNUM SCRIPSIT

9 N/V

9 MAGNI SCRIPTOR LIBRI MAGNUM SCRIPSIT LIBRUM

SCRIPTOR SCRIPSIT

N/V

N/V

SCRIPTOR LIBRI MAGNI SCRIPSIT LIBRUM MAGNUM

SCRIPTOR MAGNI LIBRI SCRIPSIT MAGNUM LIBRUM

c

47

f

N/V

9 LIBRI SCRIPTOR MAGNI LIBRUM SCRIPSIT MAGNUM

3.2.2 Romance In late Latin there is already considerable evidence that the more flexible linearizations of the classical period were rapidly giving way to a more fixed ordering of the internal components of the nominal and verbal groups,38 though somewhat more slowly in the latter case (Bauer 2009a: 268, 270). This transferral of functional load from the morphological inflections of the constituent parts of the nominal group to the relative positions in which they occur is increasingly betrayed in late Latin texts in the weakening of agreement relations, with the consequence that ‘[m]ore and more

the various surface orders of Latin are the result of a number of syntactically and semantically motivated movement and scrambling operations which disrupt an underlying English-style configurational structure—a possibility explored in chapter 5. There are also other configurational representations that we shall not discuss further here (though see Golumbia 2004) such as the various versions of the pronominal argument hypothesis (Jelinek 1984; Speas 1990; Hale 1994; Baker 2001; 2003), which essentially claim that the arguments of the verb are realized by (overt/covert) pronominal affixes or clitics on the verb, in turn coindexed with null argument positions within the VP (cf. also Devine and Stephens 2006: 119). It follows that the VP is configurationally structured, but all full NPs are freely generated in adjunct positions, hence omissible, where they stand in apposition to their associated (covert) pronominal affix/clitic, thereby giving the appearance of a flat VP structure. 38 For the nominal group, see Adams (1971: 10–12), Väänänen (1982: 260), Herman (2000: 81–4), Vincent (1988: 62–3), Bauer (1995: 59–62), and Gianollo (2007: 76–7), and for the verbal group, Linde (1923), Haida (1928), Hofmann and Szantyr (1965), Adams (1976a: 93; 1977), Väänänen (1982: 259–60), Vincent (1988: 62), Herman (2000: 86), Bauer (1995: 98–102).

From Latin to Romance

48

we notice the use of a nominative inflection in an adjective or in a noun in apposition, which should strictly have the same inflection as the noun to which it is allied’ (Herman 2000: 84; see also Herman 1985: 354–5; Väänänen 1982: 253–4). In the verbal domain, we witness a parallel growth in the generalized use of the accusative, the so-called extended accusative (see }7.3.2.1), to mark all nominals in addition to direct objects.39 In the transition to Romance this development is taken to its ultimate conclusion with the establishment of full-fledged NP and VP structures, in which there is a one-to-one isomorphic mapping between grammatical functions and dedicated syntactic positions.40 In particular, the basic structure of the Romance NP and VP can be summarized by way of the linear templates in (16a–b), where brackets indicate optional elements and asterisks indicate possible recursion: (16)

a NP: (DET) (QUANT) (*ADJ) N (*ADJ) (*PP) b VP: (AUX) V (*ADV) (*O) (*ADV)

However, the template is not simply linearly ordered, but is also subject to internal hierarchical ordering of its constituent parts. By way of example, consider the representative Catalan NP in (17a) and its representation in (17b) which reveals, in contrast to the flat structure of Latin in (13a–f), a layered configurational constituent structure: (17) a la vella senyora cansada de the.FSG old.FSG lady.FSG tired.FSG of ‘the tired old lady with black glasses’

les the.FPL

ulleres glasses.FPL

negres (Cat.) black.PL

39 Herman (1966), Pensado (1986), Gerola (1949–50), Väänänen (1966: 121), La Fauci (1988: 54–5), Zamboni (1998: 131–2), Bauer (1995: 138), Cennamo (2001a), Ledgeway (2011: 459–61). For a more cautious interpretation of the data, see Adams (forthc. e). 40 By way of illustration, one only has to compare the remarkable syntactic freedom of the Latin nominal and verbal groups observed in the permutations in (13a–f) and (14a–f) with the fixed order of the Romance NP and VP, as illustrated by the French strings in (i.a–f; un écrivain/il écrivit ‘a writer/he wrote’, de ‘of ’, grand ‘great(/big)’, livre ‘book’):

(i)

a b c d e f

Un écrivain d’un grand livre / Il écrivit un grand livre (Fr.) Un écrivain d’un livre grand / Il écrivit un livre grand (Fr.) **D’un grand livre un écrivain/ **Un grand livre il écrivit (Fr.) **Grand un écrivain d’un livre / **Grand il écrivit un livre (Fr.) **D’un livre grand un écrivain / **Un livre grand il écrivit (Fr.) **D’un livre un écrivain grand / Un livre il écrivit grand (Fr.)

Pragmatic and stylistic effects aside (including possible, though unreconstructable, intonational differences), the Latin strings in (13a–f) and (14a–f) all mean the same thing, whereas their exact French copies in (i.a–f) do not. Assuming register and intonation to be neutral, the only strings which prove grammatical are (i.a–b): (i.c, e) are ruled out because the complement (d’) un (grand) livre (grand) precedes rather than follows its head noun/verb écrivain/il écrivit, and (i.d, f) are excluded because the scope of the attributive adjective grand cannot be interpreted unless contiguous to its associated noun livre. Even in the case of the grammatical (i.a–b), the prenominal and postnominal orders of the adjective grand give rise to a semantic distinction, namely the figurative un grand livre ‘a great book’ and the literal un livre grand ‘a big book’.

The rise of constituent structure b

49

NP

3 DP

N'

5 la

3 N'

3 N'

PP

6

AP de les ulleres negres

3 5 AP

N cansada

5

!

vella

senyora

The nominal head senyora first combines directly with the prenominal adjectival phrase (AP) vella to form the intermediate nominal constituent, here labelled as N', namely [N' vella [N senyora]] (cf. discussion of compound nouns in }3.2.2.1 below). In turn, this newly created constituent combines with the postnominal AP cansada to form an even larger intermediate constituent [N' [N' vella [N senyora ]] cansada]. In turn, the prepositional phrase (PP) adjunct de les ulleres negres combines with this newly formed N' constituent to form another intermediate constituent (namely, [N' [N' [N' vella [N senyora]] cansada] [PP de les ulleres negres]]), which, once combined with the determiner phrase (DP) la ‘the’ which assigns a definite interpretation to the whole string, forms a semantically complete conceptual unit and hence an NP. Note that, in principle, it would be possible to have the nominal head first combine with the postnominal adjective and then, in turn, with the prenominal adjective (namely, **[N' vella [N' [N senyora] cansada]]), but this possibility is ruled out, among other things (see }3.2.2.1 below), for the reason that it would reverse the scope properties of both adjectives (incorrectly yielding the reading ‘the old, tired lady’, instead of ‘the tired, old lady’).41 It is also interesting to note at this point that, although the relevant 41 As G. Salvi (p.c.) points out, reversed scopal readings of this type are indeed available with preverbal subjective adjectives. For instance, whereas preverbal vecchia ‘old’ in (i.a) has scope only over the noun signora ‘lady’ with postnominal inglese ‘English’ taking wide scope over vecchia signora ‘old lady’ (viz. una [[vecchia [signora]] inglese] ‘an old lady who is English’), prenominal noiosa ‘annoying’ in (i.b) takes wide scope over both signora and inglese (viz. una [noiosa [[signora] inglese]] ‘an English lady who is annoying’). This leads us to conclude that there are in fact two prenominal adjectival positions, one an immediate sister to N for adjectives like vecchia and the other a more peripheral position adjoined to N' for subjective adjectives like noiosa (see further Cinque 2010: }}2.10.1–5).

(i) a una a

vecchia old

signora lady

inglese (It.) English

50

From Latin to Romance

adjectives agree in gender (feminine) and number (singular) with the nominal head, their differing degrees of syntactic cohesion with the latter are encoded, not by this agreement, but by their relative locality to the head. Consequently, we can say that the prenominal adjective vella is a ‘sister’ to the nominal head senyora, forming an intermediate N' constituent which dominates both AP and N, whereas the postnominal adjective cansada is a sister to this latter intermediate N' constituent. 3.2.2.1 Adjectival positions Significantly, the representation in (17b) highlights how in the passage from Latin to Romance the emergence of NP constituent structure gives rise to two dedicated adjectival positions, the prenominal position licensing given/non-contrastive readings and the postnominal position licensing new/contrastive readings.42 As noted in }3.2.1, although some scholars do not recognize such a semantic distinction between the pre- and postnominal positions for Latin since ‘often it is not possible to detect any salient variation in meaning when encountering minimal AN-NA pairs’ (Gianollo 2007: 68; see also Langslow forthc.), others already recognize such a semantic distinction between the two positions largely similar to that found in modern Romance (cf. Vincent 2007: 64–5), witness the contrasting literal and figurative interpretations of URBANUS ‘of the city (of Rome)’ in (18). (18)

URBANUS vs URBANUS PRAETOR (Lat., Marouzeau 1922: 14) magistrate of.city of.city magistrate ‘magistrate for Roman citizens’ vs ‘witty magistrate’

PRAETOR

However, recent work by Langslow (forthc.) has demonstrated that the contrastive readings of urbanus in (18) theorized by Marouzeau are not borne out by actual literary usage. In his corpus of prose writers he finds that prenominal urbanus with figurative meaning is very rare and is only licensed in this position when emphatically placed under focus, a position equally available to the literal reading of urbanus when focused. In short, the evidence adduced by Langslow points to an unmarked postnominal position for both the figurative and literal readings of urbanus, with the prenominal position serving as a marked position for the licensing of contextual focus. Whatever the correct interpretation, it is clear that, contrary to modern Romance, the contrastive and non-contrastive interpretations are not exclusively associated with

b una a

noiosa annoying

signora lady

inglese (It.) English

42 For alternative analyses in which distinct adjectival positions are reinterpreted as the surface reflex of the head noun (or NP) variously moving across a rigidly ordered series of different adjectival spaces and associated left-peripheral modifier/Spec(ifier) positions, see Crisma (1993; 1996), Cinque (1994; 2010: chs 4, 6), Longobardi (2001: }2.1), Laenzlinger (2005a,b), Rowlett (2007: 88–95), and the discussion in }3.2.2.1.1 below.

The rise of constituent structure

51

the post- and prenominal positions in Latin (19a),43 or even in early Romance (19b),44 a distinction which would only fully grammaticalize with the subsequent rise of full configurationality as demonstrated in the modern Spanish examples in (20a–b). (19) a animo aequo vs aequo disposition.ABL equal.ABL / equal.ABL ‘with favourable disposition’

animo (Lat., Nep. Dion 10.6 / 10.7) disposition.ABL

b li spangnoli soldati vs le compangnie spangnole (ONap., CDT 34–5 /65) the Spanish soldiers the companies Spanish ‘the Spanish soldiers’ vs ‘the Spanish companies (of soldiers)’ (20)

a una joven secretaria inteligente (Sp.) a young secretary intelligent ‘an intelligent young secretary’ b una inteligente secretaria an intelligent secretary ‘a young intelligent secretary’

joven (Sp.) young

Thus, in a sequence such as Spanish (20a), the secretary is identified by her intelligence and not her youthfulness, which is simply taken to be a known and, in this case, non-distinguishing quality of the individual concerned. Conversely, in the sequence (20b) the readings are reversed and we are talking about ‘an intelligent secretary’ about whom we want to convey that she is young. Oversimplifying somewhat, the properties of the prenominal and postnominal positions of the adjective can be summarized as follows.45 Contrary to what is claimed, at least in part (see discussion above in }3.2.1), for Latin, the prenominal position constitutes in Romance a marked position,46 not only in terms of the specialized readings it licenses, but also in terms of its greater frequency in formal registers. Putting such considerations aside, however, we note that adjectives in

43 Marouzeau (1922: 13–14), Gildersleeve and Lodge ([1895] 1997: 431), Herman (2000: 83), Pinkster (1990a: 185). De Melo (2010: 79) arrives at the same conclusion on the basis of possessive positions in Plautus, where he observes that ‘a possessive [ . . . ] can be focused whatever position it occurs in’. 44 Ledgeway (2007a; 2009a: 241–2), Vincent (2007), Thiella (2008: ch. 4), Giusti (2010a: 599–609). 45 See further Arnholdt (1916), Sandfeld and Olsen (1960: 98–114), Alisova (1967), Glatigny (1967), Guţia (1967: 151–4), Reiner (1968), Lapesa (1975), Vincent (1986; 2007: 57–61), Stati (1989: 123–4), Bernstein (1991; 1993), Giorgi and Longobardi (1991), Brito (1993), Crisma (1993), Badia i Margarit (1995: 433–6), Bosque (1996), Bosque and Picallo (1996), Berruto (1998), Pountain (1998), Demonte (1999), Abeillé and Godard (1999), Saltarelli (1999), Scarano (1999; 2005), D’Introno (2001: 418–21), Radatz (2001), Cinque (2004a), Gonzaga (2004), Ledgeway (2007a), Rowlett (2007: 84–95), Maiden and Robustelli (2007: 48–52). For discussion of some counterexamples in which adjectival interpretation is identical in both pre- and postnominal positions, see Cinque (2010: ch. 2). See also }4.2.3.1.1 for a discussion of the Romanian socalled demonstrative article in conjunction with postnominal adjectives. 46 Badia i Margarit (1962 I: 150), Stati (1989: 123), Ledgeway (2007a: 105), Vincent (2007: 58–9).

52

From Latin to Romance

prenominal/postnominal position typically correlate with the following respective interpretations (for a more nuanced overview, see Cinque 2010: ch. 2): (i) inherent/ non-inherent (21a–c); (ii) descriptive/distinguishing (22a–c); (iii) subjective/objective (23a–c); and (iv) figurative/literal (24a–c). (21)

a la blanche neige vs la voiture blanche (Fr.) the white snow the car white ‘white snow’ vs ‘the white car’ b l’ inglese eleganza vs la moneta inglese (It.) the English elegance the coin English ‘(typical) English elegance’ vs ‘English currency’ c su británica reserva vs la Embajada británica (Sp.) his British reserve the Embassy British ‘his British reserve’ vs ‘the British Embassy’

(22)

a una llarga conversació vs una conversació a long conversation a conversation ‘a long conversation’ vs ‘a conversation which is long’ b une courte lettre vs une jupe a short letter a skirt ‘a short letter’ vs ‘a skirt that is short’ c una vecchia scrivania vs una a old desk a ‘an old desk’ vs ‘a desk which is old’

(23) a (llunyanes) terres distant lands ‘(distant) lands’

llarga (Cat.) long

courte (Fr.) short scrivania desk

vecchia (It.) old

(llunyanes) (Cat.) distant

b jójna biála béstia vs un vestáit bil (Vgl.) a beautiful beast a dress beautiful ‘a beautiful animal’ vs ‘a fine dress’ c un (formidabil) om (formidabil) (Ro.) a tremendous person tremendous ‘a (tremendous) person’ (24)

a la grisa quotidianitat vs la camisa grisa (Cat.) the grey daily.routine the shirt grey ‘the grey (= dull) daily routine’ vs ‘the grey shirt’ b um antigo patrão meu vs Lisboa antiga (Pt.) a old boss my Lisbon old ‘an old (= former) boss of mine’ vs ‘old Lisbon’

The rise of constituent structure c dulcele vis vs o cafea sweet=the dream a coffee ‘the sweet dream’ vs ‘a sweet coffee’

53

dulce (Ro.) sweet

As formalized in (17b) above, the two positions are also differentiated by their respective degree of integration with the nominal head, which in turn correlates with their differing degrees of semantic autonomy: whereas postnominal adjectives are semantically autonomous and enter into a looser structural relation with their noun, witness their ability to license the same readings in predicative function (25a), prenominal adjectives enter into closer nexus with their associated noun, ultimately producing a marked reading only licensed in that particular configuration (25b). (25)

a a rapariga pobre (= the girl poor ‘the poor (= destitute) girl’ b a pobre rapariga (6¼ the poor girl ‘the poor (= wretched) girl’

a the

rapariga girl

é is

pobre) (Pt.) poor

a the

rapariga girl

é is

pobre) (Pt.) poor

It is for this reason that it has often been claimed that adjective + noun sequences behave like, and frequently correspond to, single lexemes or compound nouns:47 Cal. mala parola lit. ‘bad word’ = ‘swear word’ (cf. Cat./Occ. renec/renèc, Fr. juron, Sp. taco), Fr. grand-mère lit. ‘great mother’ = ‘grand-mother’ (cf. It. nonna, Occ. mameta, Ro. bunică, Sp. abuela), Fr. petit pain lit. ‘small bread’ = ‘(bread) roll’ (cf. Cat. panet, It. panino, Pt. pãozinho, Ro. chiflă), Nap./Sp. bona fémmena/buena mujer lit. ‘good woman’ = ‘prostitute’ (cf. Fr. putain, It. puttana, Sp. puta), Fr./Pt./Ro, petit déjeuner/ pequeno almoço/micul dejun lit. ‘small(.the) lunch’ = ‘breakfast’ (cf. Cat. esmorzar, It. colazione, Occ. dejunar, Sp. desayuno). Above this distinction was formally interpreted by treating prenominal adjectives on a par with complements insofar as they are generated as sisters to N (see also Sproat and Shih 1988; 1990; Lamarche 1991), whereas postnominal adjectives are treated as adjuncts and generated as sisters to N'. This conclusion is further supported by the observation that the greater cohesion between prenominal adjective and noun gives rise to a number of phonomorphological processes of a strictly local nature. For instance, in French the prenominal adjectival position is one of the contexts in which liaison (the pronunciation of an underlying word-final consonant immediately before a following vowel) is still productive even in colloquial usage:48

47 Gildersleeve and Lodge ([1895] 1997: 431), Radatz (2001: ch. 5), Ledgeway (2007a: 114–5), Vincent (2007: 59). 48 Lamarche (1991), Valois (1991), Battye and Hintze (1992: 140), Bouchard (1998; 2002), Fagyal, Kibbee, and Jenkins (2006: 67); cf., however, Cinque (2010: }4.1.2).

From Latin to Romance

54 (26)

a un ‘a b un ‘a

lége[ʀ] slight gro[z] fat

incident (cf. citation form léger [leʒe]) incident’ homme (cf. citation form gros [gʀo]) man’

Similarly, in Provençal (27a–b) and many upper southern Italian dialects (27c) it is only the prenominal adjective which displays overt (plural) agreement:49 (27)

a aquest[ej] polid[ej] raub[o] vs aquest[ej] raub[o] polid[o] (Prv.) these.PL pretty.PL dress.FSG these.PL dress.FSG pretty.FSG ‘these pretty dresses’ b li tèndri paraulo vs de pan tèndre (Rho.) the.PL tender.PL word.FSG of loaf.MSG tender.MSG ‘the tender words’ vs ‘some tender loaves’ c brutti[i] cos[ə] vs cos[ə] brutt[ə] (Nap.) horrible.PL thing thing horrible ‘horrible things’

Also relevant here are Romance reflexes of BELLUS ‘beautiful’, BONUS ‘good’, GRANDIS ‘big, great’, MALUS ‘evil’, and SANCTUS ‘saint, holy’ which commonly present apocopated forms in prenominal position (namely, Fr. (prevocalic position only), It. and Pt. bel, It./Sp. buon/buen, It. and Sp/Pt. gran/grã(o), Sp. mal ‘bad’, It. and Sp./Pt. san/ são), morphophonological reductions which can now be straightforwardly derived from their status as sisters to N:50

49 Koschwitz (1894: 75), Wheeler (1988b: 256–7), Lafont (1991: 10), Ledgeway (2007a: 105–7). As pointed out to me by G. Salvi (p.c.), paradoxically in some Romance varieties such as Gardenese Ladin (cf. Salvi 1997: 291–2) agreement only surfaces on the postnominal adjective (i.a–b), never on the prenominal adjective (ii. a–b). Facts like these would seem to suggest that the lack of agreement in cases such as (ii. a–b) follows form the strict syntactic bond between prenominal adjective and noun (cf. discussion of compounds above).

(i) a la the.FSG

muta girl.FSG

b la mutaŋs the.FSG girls.F ‘the curious girl/s’ (ii) a

la the.FSG

pitla small.FSG

b la pitla the.FSG small.FSG ‘the small girl/s’ 50

kurjæwza (Grd.) curious.FSG kurjæwzes (Grd.) curious.FPL muta (Grd.) girl.FSG mutaŋs girl.PL

In modern Portuguese bel and grã(o) are no longer the canonical prenominal forms, but are restricted to a small number of set (often compound) expressions (G. Salvi, p.c.).

The rise of constituent structure (28)

55

a un bel Espagnol (Fr.) ‘a handsome Spaniard’ b bel-prazer (Pt.) ‘fine-pleasure’ (= ‘pleasure’) c un buon ritorno (It.) ‘a good (= pleasant) return’ d grã pressa (Pt.) ‘great haste’ e un mal ingeniero (Sp.) ‘a bad engineer’ f

san Miguel (Sp.) ‘Saint Michael’

3.2.2.1.1 restricted adjectival positions In a number of, especially non-standard, Romance varieties including Asturian (ALlA 2001: 320), Occitan (29a),51 Sardinian (29b; Jones 1988a: 335; 1993: 42), and central and southern Italian dialects (29c),52 the prenominal adjectival position is extremely restricted and generally replaced by the postnominal position, which is neutral to the contrastive / non-contrastive distinction: (29)

a lo pònt vièlh the bridge old pont d’ Avignon (Fr.) bridge of Avignon ‘the old Avignon bridge’

d’ of

Avinhon (Lgd.) Avignon

vs

le the

vieux old

b na (**piccerella) patana piccerella (Nap.) a small potato small ‘a small potato’ c una (**nova) mákkina nova (Srd.) a new car new ‘a new car’

51

Nouvel (1975: 325), Wheeler (1988b: 268). Other striking Occitan examples, especially in comparison to their French equivalents (see Ronjat 1913: 29), include: (i) a l’ Escrituro Santo (Occ.) vs la Sainte Écriture (Fr.) the scripture holy the holy scripture (= ‘the Holy Scripture’) b segne/maire sire/mother 52

grand (Occ.) big

vs

grandbig

père/ father

-mère (Fr.) mother (=‘grandfather/-mother’)

Rohlfs (1969: 330), Saltarelli (1999), Maturi (2002: 157 n. 96), Ledgeway (2007a; 2009a: 232–6).

56

From Latin to Romance

It is not immediately obvious how these varieties can be readily accounted for in terms of the structure in (17b), apart from making the ad hoc and unsatisfactory assumption that the prenominal position is simply not available and hence not projected. However, these same facts find a much more convincing interpretation in those theories (cf. footnote 42 above) that assume a crosslinguistically fixed series of adjective positions immediately above the NP, across which the head noun may variously move in accordance with parametric variation (30): (30)

. . . (N) [AP1 Adj (N) [AP2 Adj (N) [AP3 Adj (N) [AP4 Adj . . . [NP N]]]]]

Assuming the much-simplified structure in (31a) in which AP1 and AP2 can be broadly understood as the ‘areas’ in which non-contrastive and contrastive adjectives, respectively, are generated, we can now formally capture in a highly simplified manner the differences between the non-standard varieties in (29a–c) on the one hand and standard Romance varieties like French, Italian, and Spanish on the other: in the former the nominal head typically raises to the highest available position above the highest adjectival projection (AP1), which hosts non-contrastive adjectives from where it precedes both non-contrastive and contrastive adjectives (31b), whereas in the latter the noun only targets the higher adjectival projection (AP1), from where it precedes non-contrastive adjectives but follows those with a contrastive reading (31c): (31)

a ...

(N)

[AP1 Adj1 (N)

[AP2 Adj2

[NP N ]]]

b . . . lo pònt [AP1 vièlh pònt

[AP2

[NP pònt ]]] (Lgd.)

c . . . le

[AP2

[NP pont ]]] (Fr.)

[AP1 vieux pont

By the same line of reasoning, we can explain the frequent prenominal position of contrastive adjectives in early Romance, as noted in (19b) above, by assuming that Nraising is only optional in the early varieties (32a–b), an archaic pattern still preserved to the present-day in Walloon where the nominal head barely moves at all (32c):53 (32)

a . . . li [AP1 [AP2 spangnoli [NP soldati ]]] (ONap.) ‘the Spanish soldiers’ b . . . le [AP1 compangnie [AP2 spangnole [NP compangnie ]]] (ONap.) the companies Spanish (= ‘the Spanish companies’) c . . . dès [AP1 [AP2 r’tchâfés [NP crompîres ]]] (Wal.) ‘some reheated potatoes’

This analysis can also readily explain why in other cases the prenominal and postnominal positions appear to be lexicalized, as in Catalan where, unlike mal

53

Bernstein (1991), Bouchard (2002), Rowlett (2007: 89), Cinque (2010: }6.1).

The rise of constituent structure

57

‘bad’ which occurs in prenominal position, the near-synonymous dolent (cf. also Pt. ruim ‘bad’) always occurs in postnominal position even when interpreted nonrestrictively. If we assume that mal lexicalizes a higher functional projection than dolent, then the observed positional (and concomitant subtle semantic) difference can be straightforwardly derived as in (33): (33)

. . . una [AP (mala) a bad ‘a bad suggestion’

proposta suggestion

[AP (dolenta) bad

[NP proposta ]]] (Cat.)

3.2.2.2 Complements and adjuncts This structural distinction between pre- and postnominal adjectives equally proves essential in distinguishing between complements and adjuncts. Consider, for example, the Portuguese nominal and verbal sequences in (34a–c): (34) a um estudante [PP de física] / vai estudando [NP a física] (Pt.) a student of physics he.goes studying the physics ‘a physics student’ / ‘he is studying physics’ b um estudante [PP de cabelo louro] / vai estudando [NP cada dia] (Pt.) a student of hair blond he.goes studying each day ‘a student with blond hair’ / ‘he is studying each day’ c um estudante [PP de Lisboa] / vai estudando [PP por causa a student of Lisbon he.goes studying on account do prêmio] (Pt.) of.the prize ‘a student from Lisbon’ / ‘he is studying because of the prize’ On the surface, the bracketed prepositional/nominal strings appear to be of equal status in that they all directly follow their nominal/verbal head. In (34a), however, the relevant PP/NP functions as the complement of the head estudante/estudando ‘student/studying’, whereas in (34b–c) the PPs/NPs are optional modifiers of the nominal/verbal head, and hence adjuncts. This difference can be seen in the grammaticality judgements associated with the strings in (35a–c), in which all three PPs and NPs are combined (and intonation is assumed to be neutral, excluding such marked processes as narrow focus and extraposition): (35)

a um estudante [de física] [de cabelo louro] [de Lisboa] / um estudante [de física] [de Lisboa] [de cabelo louro] (Pt.) a' vai estudando [a física] [cada dia] [por causa do prêmio] / vai estudando [a física] [por causa do prêmio] [cada dia] (Pt.) b **um estudante [de cabelo louro] [de física] [de Lisboa] / **um estudante [de Lisboa] [de física] [de cabelo louro] (Pt.)

From Latin to Romance

58

b' **vai estudando [cada dia] [a física] [por causa do prêmio] / **vai estudando [por causa do prêmio] [a física] [cada dia] (Pt.) c **um estudante [de cabelo louro] [de Lisboa] [de física] / **um estudante [de Lisboa] [de cabelo louro] [de física] (Pt.) c' **vai estudando [por causa do prêmio] [cada dia] [a física] / **vai estudando [cada dia] [por causa do prêmio] [a física] (Pt.) What these examples clearly illustrate is that complements like [de física] and [a física] enjoy a privileged position with respect to their selecting head, in that they combine directly with the latter to form an N'/V' constituent, hence the ungrammaticality of separating head and complement as in (35b–c). In contrast, adjuncts like [de cabelo louro], [de Lisboa] and [cada dia] and [por causa do prêmio] are less deeply embedded and as such are essentially unordered with respect to one another (35a,a'), attaching at the N'/V' level as illustrated in (36a–b): NP

(36) a

b

3 DP

VP

3

N'

4

AuxP

3 N'

um

PP

vai

3 6

N'

3 N

!

PP de Lisboa

6

PP de cabelo louro

6

estudante

V'

4

3 V'

3

V'

PP

6

PP por causa do prêmio

3 6 V

!

NP

cada dia

6

estudando

The same structural distinction also applies to adjectives like Catalan xinèsa ‘Chinese.FSG’ in the following contrasting pair: (37) a [NP l’ [N' [N invasió] [AP xinesa]]] (Cat.) the invasion Chinese ‘the Chinese invasion (= invasion of China)’ b [NP la [N' [N' [N cuina]] the cuisine ‘Chinese cuisine’

[AP xinesa]]] (Cat.) Chinese

In the former case, the adjective functions as a complement to the nominal head (hence a sister to N), whereas in the latter case it is an adjunct modifier (hence a sister to N').

The rise of constituent structure

59

3.3 The sentence 3.3.1 Classical Latin Traditional wisdom has it that in the transition from Latin to Romance sentential word order underwent a steady progression from SOV to SVO (Bauer 1995: 7; Vincent 1997c: 166; 2007: 65), ‘a process already well under way in Latin’ (Harris 1978: 5).54 On this view, word order in Latin is not absolutely free (pace Weil [1844] 1887: 53–4), although enjoying considerably greater freedom than in Romance,55 inasmuch as SOV order occurs with far greater statistical frequency than all other competing orders in most texts and most authors of the classical period.56 Indeed, SOV order would appear to be a conservative feature of Latin syntax directly retraceable to Indo-European.57 The overall conclusion then is that SOV is the most frequent, and hence unmarked, word order,58 but that all other orders are equally possible, although never entirely synonymous in that they convey some marked interpretation (Bauer 1995: 6; Devine and Stephens 2006: 79). This is essentially the view espoused by such scholars as Marouzeau (1922), Panhuis (1982), Pinkster (1990a: 181; 1991), Ostafin (1986), Vincent (1998a: 418–9), Oniga (2004: 97), Polo (2004), and Salvi (2004; 2005: 436), who see all orders other than SOV as the result of pragmatic factors and the organization of information structure (e.g. OSV would be derived from SOV via topicalization of the object; see also the discussion in }3.2.1 above).59 This view of Latin word order, though undoubtedly the most common, is not universally accepted (see Pinkster 1990a: 187–8; 1996: 254). A number of scholars have claimed that Latin word order was essentially unstable (Ramat 1980: 189), and that as early as Plautus SVO was already a frequent rival to SOV as the unmarked order (pace Pinkster 1991: 74; 1993b: 247), especially, though not exclusively, in lower and spoken registers (cf. }5.3.2).60, 61 Already we have seen above in }3.2.1 that, in the face 54 We use here O(bject) as a shorthand notation to indicate all types of verbal complement, including direct, indirect, and prepositional objects. 55 Pinkster (1990a: 163), Baños Baños and Cabrillana (2009: 680), Bauer (1995: 6). 56 Hale and Buck ([1903] 1994: }623), Linde (1923), Marouzeau (1938: 106), Tovar (1979), Pinkster (1990a: 180, 187), Kessler (1995: 1), Oniga (2004: 97), Bailey (2008: }4.3.3). 57 Ernout and Thomas ([1953] 1993: 161), Watkins (1964: 1039; 1998: 68), Lehmann (1972a; 1974: 114), Konneker (1975: 367), Adams (1976a: 92), Bichakjian (1987: 96), Bauer (1995: 86–9; 2009a: 250–2); for a more cautious view, see Clackson and Horrocks (2007: 28–9). 58 Though, see Charpin (1989) for a criticism of the purely superficial quantitative correlation between ‘most frequent’ and ‘neutral’ word order (cf. also Pinkster 1991: 70). 59 In contrast to Panhuis (1982), Del Vecchio (1989) demonstrates that word order in early authors such as Plautus is not always motivated pragmatically, but, rather, by stylistic reasons and, in particular, rhythm and metre, as well as by literary genre (cf. also Pinkster 1990a: 186). Particularly revealing in this regard are archaic legal texts such as the Senatus consultum de Bacchanalibus (6 BC) and Leges XII Tabularum (late 5 BC), where he recognizes (pp. 551–3) in the consistently clause-final position of the verb a mechanical tendency established by purely prescriptive considerations of style (cf. also Meillet [1928] 1977: 120). 60 Lakoff (1968), Ross (1970), Adams (1976a; 1976b: 137), Panhuis (1982), Vineis (1993: LII), Oniga (2004: 99), Clackson and Horrocks (2007: 27–30). 61 While Panhuis (1982) adopts a dynamic communicative perspective to Latin word order (cf. also Panchón 1986), according to which there is no such thing as an underlying neutral or basic word order

60

From Latin to Romance

of a frequent verb-medial position, many scholars have questioned the unmarked nature of the verb-final position (cf. Amacker 1989: 493), Herman (2000: 85) even arguing that the ‘categorization of the language as basically having SOV order is exaggerated, even as regards the Classical language’ (cf. also Pinkster 1991: 71). On this view, SOV as found in classical prose (notably in Caesar) and in legal and formulaic inscriptions (cf. Adams 1976a: 92; Clackson and Horrocks 2007: 28) is often considered a more conservative and stylistic order with little or no relation to spoken usage, such that it is appropriate to speak of grammars in competition (Kroch 1989; Polo 2004). In this respect, it is often noted that one of the probable sources for the steady growth of unmarked SVO, alongside of SOV, was complement clauses (Pinkster 1990a: 187; Oniga 2004: 99): on account of their ‘heavy’ nature and concomitant perceptual complexity (Vincent 1977), preverbal finite complement clauses already proved relatively infrequent in classical times (38a) and more frequently extraposed to a postverbal position (38b).62 Complement clauses thus provided a frequent context for SVO, readily reanalysed in time as a non-derived order, a topic we shall explore in considerably more detail, albeit from a different perspective with quite different conclusions, in }5.4.2.1. (38) a Ariouistus ex equis ut conloquerentur et praeter se Ariovistus from horses that they.confer.IMFV.SBJV and besides selves denos ad conloquium adducerent ten.by.ten.ACC to conference they.bring.IMFV.SBJV.3PL postulauit (Lat., Caes. B.G. 1.43.3) asked.3SG ‘Ariovistus then demanded that they should confer on horseback, and that, besides themselves, they should bring with them ten men each to the conference’

(p. 18; cf. also Panhuis 1984; though see Elerick 1989a: 570–1), insofar as linear arrangement serves to convey communicative effectiveness, he nonetheless finds that SVO predominates in early (and colloquial) authors such as Plautus (p. 125), and SOV, a ‘literary convention’ (p. 145), in classical authors such as Cicero (pp. 55, 115), as witnessed by backward gapping patterns such as (i) discussed in Panhuis (1980; see also }5.3.2): (i)

Gallos ab Gauls.ACC from

Aquitanis Aquitani.ABL

Garumna Garonne.NOM

flumen, river.NOM

a from

Belgis Belgians.ABL

Matrona Marne.NOM

et Sequana diuidit (Lat., Caes. B.G. 1.1.2) separates and Seine.NOM ‘The river Garonne separates the Gauls from the Aquitani, the Marne and the Seine (separate them) from the Belgians’ 62

The changing position of complement clauses and their concomitant effect on the position of nominal complements is explored in detail in }5.4.2.1.

The rise of constituent structure

61

b Ariouistus postulauit ne quem peditem ad conloquium Ariovistus asked.3SG lest any.ACC foot.soldier.ACC to conference Caesar adduceret (Lat., Caes. B.G. 1.42.5) Caesar.NOM bring.IMFV.SBJV.3SG ‘Ariovistus demanded that Caesar should not bring any foot soldier with him to the conference’ Despite the differences in the two principal positions outlined above, the upshot is that, although one (SOV) or more (SOV, SVO) unmarked orders can be recognized, word order in Latin was nonetheless considerably free, albeit conditioned by pragmatic considerations (see }3.2.1).63 From this it follows that grammatical functions within the Latin sentence could not invariably, nor necessarily, be read off surface linear order but, rather, were typically identified by the morphological form of individual items (Baños Baños and Cabrillana 2009: 680). Thus, given a simple transitive sentence such as ‘the boy calls the girl’ (PUER ‘boy.NOM’, UOCAT ‘calls’, PUELLAM ‘girl.ACC’), all six possible permutations are possible (Vineis 1993: LII; Herman 2000: 85), namely SVO (PUER PUELLAM UOCAT), OSV (PUELLAM PUER UOCAT), SVO (PUER UOCAT PUELLAM), OVS (PUELLAM UOCAT PUER), VSO (UOCAT PUER PUELLAM), and VOS (UOCAT PUELLAM PUER). Once again it emerges that even in the domain of the sentence we are apparently obliged to assume a flat structure, identical to that observed above in (15a–f) for the nominal and verbal groups: (39) a

S

9 PUER

d

b

PUELLAM UOCAT

S

9 PUELLAM UOCAT

PUER

9 PUELLAM PUER

e

c

S

S

UOCAT

9

UOCAT

9

S

PUER UOCAT

f

PUELLAM

S

9

PUER PUELLAM UOCAT PUELLAM

PUER

3.3.1.1 Discontinuous structures It is not always even possible to exhaustively identify one specific position within the sentence with a particular grammatical function. Not only can semantico-syntactic dependencies within the nominal and verbal groups be locally disrupted by placing logically contiguous elements in discontinuous 63 Moreover, we must not forget, as observed in }3.2.1 above, that we are dealing with at least 1400 years of documented Latinity, a time span during which the language of speakers and writers, including word order patterns, must have been subject to numerous changes (cf. also Elerick 1989a: 560). In this regard, one only has to think, for example, of the unmistakable grammatical differences between the early and modern stages of today’s European languages such as Italian (cf. Vincent 2005), for which no scholar would try to reconstruct, on the basis of 1000 years of documented evidence from a range of registers, styles, authors, and text types, a single word order as if early Tuscan and modern Italian represented a single unified code.

62

From Latin to Romance

positions within their group (e.g. MAGNAE UIR SAPIENTIAE ‘great.GEN.FSG man.NOM knowledge.GEN.FSG’ (‘a man of great knowledge’), BIBERE UOLO AQUAM ‘drink.INF I.want water.ACC’ (‘I want to drink some water’), but entire groups can be broken up and given discontinuous expression globally across the sentence.64 For example, in (40a) the subject of the sentence ‘the radiant Zephyrs’ does not form a single syntactic phrase, but is realized partly before the verb (candidi ‘radiant’) and partly following the verb (Fauonii ‘Zephyrs’), therefore apparently making it impossible to talk about a single or dedicated subject position as is usual in the Romance languages.65 Analogous considerations carry over to the examples in (40b–d), where the elements that make up the (in)direct object argument are scattered discontinuously across the sentence, including, in the case of (40c–d), at the beginning (meo, maxima) and at the end (seruo, animalia) of their clause. Particularly illustrative of the difficulty in identifying canonical argument positions are the two final examples, where the constituent parts of the subject and object arguments (40e) and the subject and locative arguments (40f) are intertwined and blended discontinuously among themselves (cf. Amacker 1989: 491). (40) a quem tibi candidi / primo restituent uere who.ACC.MSG you.DAT radiant.NOM.PL first.ABL.NSG return.3PL spring.ABL.N Fauonii / Thyna merce beatum (Lat., Hor. Carm. 3.7.1–3) Zehpyrs.NOM happy.ACC.MSG goods.ABL happy.MSG ‘whom the radiant Zephyrs will give back to you at the beginning of the spring, blessed with goods from Bithynia’ b hic optimus illis temporibus est patronus habitus (Lat., Cic. he best.NOM.MSG those.ABL times.ABL is lawyer.NOM had Brut. 106) ‘In those days he was considered the best lawyer’ c dedi [ . . . ] symbolum seruo tuo . . . epistulam [ . . . ]: I.gave symbol.ACC.N servant.DAT.M your.DAT.MSG letter.ACC.F meo tu epistulam dedisti seruo? (Lat., Pl. Pseud. 1200–3) my.DAT.MSG you letter.ACC.F you.gave servant.DAT.M ‘I gave [ . . . ] a token to your slave [ . . . ] a letter: To my slave you gave a letter?’ d maxima in India gignantur biggest.NOM.NPL in India are.created ‘The biggest animals grow in India’

animalia (Lat., Plin. N.H. 7.21) animals.NOM.NPL

64 Hale and Buck ([1903] 1994: }627), Grandgent (1907: 30), Ernout and Thomas ([1953] 1993: 162), Meillet ([1928] 1977: 156), Väänänen (1982: 259), Pinkster (1990a: 164, 185–6), Powell (2010). A more indepth discussion of discontinuous structures is undertaken in }5.4.3. 65 For a discussion of such distinctive word order patterns in Horace’s Odes, see Nisbet (2000).

The rise of constituent structure

63

e Amissos socios sermone longo lost.ACC long.ABL.FSG companions.ACC call.ABL.FSG requirunt (Lat., Virg. Aen. 1.217) they.search ‘they search for their lost companions with long calls’ f grandia per multos tenuantur great.NOM.NPL through many.ACC.MPL are.reduced riuos (Lat., Ov. R.A. 445) brooks.ACC ‘Great streams are channelled into many brooks’

flumina rivers.NOM.N

Strong empirical evidence such as this highlights how it makes little sense to posit the existence of phrasal lexical categories like NP, AP, AdvP, and VP in Latin, inasmuch as the elements that would conceptually make up these phrases need not cluster in cohesive groups but, rather, are free to occur individually in discontinuous strings across the sentence, ultimately bound together by their forms (inflection, agreement) rather than by means of word groups. This view is further substantiated by examples like those in (41a–d) below: (41) a huic ego me bello ducem profiteor (Lat., Cic. this.DAT.SG I myself.ACC war.DAT.N leader.ACC.MSG I.announce Cat. 2.11) ‘for this war I announce myself as leader’ b magno me metu great.ABL.MSG me.ACC fear.ABL.M ‘you will relieve me of great fear’

liberabis (Lat., Cic. Cat. 1.5) you.will.free

c per ego has lacrimas [ . . . ] te [ . . . ] oro (Lat., Virg. through I these.ACC.FPL tears.ACC.F you.ACC I.pray Aen. 4.314–19) ‘by these tears I beseech you’ d de ciuitatis enim iure [ . . . ] disceptamus (Lat., Cic. Balb. 29) law.ABL.F we.discuss regarding city.GEN.F for ‘for we are discussing civil law’ The underlined elements of the four sentences above illustrate the Wackernagel tendency of Latin weak pronouns and particular connectors (e.g. ENIM ‘for’, AUTEM ‘however’, UERO ‘indeed’) to occur in the second position of the clause,66 sometimes in groups as in (41a) and sometimes even in the third position if monosyllables precede Hale and Buck ([1903] 1994: }627), Ernout and Thomas ([1953] 1993: 161), Pinkster (1990a: 164), Adams (1994a), Vincent (1998a: 420–42), Salvi (2004; 2011: 363–5). 66

64

From Latin to Romance

(see 41d).67 Significantly, such sentences reveal that, in computing second position, the Wackernagel rule does not necessarily make reference to word groups,68 but simply to individual words. This explains why the relevant Wackernagel elements in (41a–d) can split up what would otherwise be a contiguous nominal or prepositional group, since the syntax of the language is essentially not sensitive to such groups.69 3.3.2 Late Latin and Romance Contrary to what was noted above for the nominal and verbal groups, which already display early yet significant signs of an emergent configurational structure in late Latin with the increasing contiguity and rigidification of their constituent parts, the sentence continues to exhibit considerable freedom even in the late Latin period (Herman 2000: 85–7), although verb-final orders are now very much diminished,70 accounting for as little as 30 percent, for example, in the late fourth-century Peregrinatio Aetheriae.71 Such continued freedom in the positioning of the subject and object was no doubt made possible by the survival of core case inflections which, even after the start of the decline of case morphology in the late Latin period, remained distinct in the nominative and accusative for a number of centuries (witness their survival in early Gallo-Romance; see }7.3.2.2).72 On this point, Herman (2000: 86) perceptively concludes that ‘[s]tatistically, the characteristic feature of late Latin texts seems to be to have the verb between the two noun phrases if two are there (including prepositional phrases)—that is, either SVO or OVS. Both these orders seem to have

67 G. Salvi (p.c.) claims that the splitting up of phrasal groups by Wackernagel pronouns (though not connectors), while common in early Latin (cf. the Cato example 41b) and classical poetry, tends to be avoided in classical prose (cf. Salvi 2004: ch. 4) where the internal structure of such phrases generally remains intact. For an alternative view, see however Adams (1994a: 133–41). 68 Although it can, witness the position of me following the NP in (i):

(i) amicorum litterae ad triumphum me, friends.GEN letters.NOM me.ACC to triumph ‘letters of my friends summon me to a triumph’

uocant (Lat., Cic. Att. 6.6.4) call.3PL

69 For a discussion of parallel structures in the non-configurational Australian language Warlpiri, in which the second-position AUX can both follow a complete NP or single constituent part of the NP, see Bresnan (2001: 6) and Mereu (2004: 120–1). 70 Muldowney (1937: 120–8), Adams (1976a: 93), Harris (1978: 19), Väänänen (1982: 259–60), Pinkster (1991: 72), Bauer (1995: 98–102; 2009a: 270), Cabrillana (1999: 321–2), Salvi (2004: 98), Clackson and Horrocks (2007: 291). 71 Linde (1923), Harris (1978: 19), Väänänen (1987: 106). 72 This empirical observation excludes a common view in the literature, also critically questioned by La Fauci (1988), that the change in word order from SOV to SVO is a consequence of the weakening and eventual loss of case distinctions, as predicted by Greenberg’s (1963) Universal 41 (Molinelli 1998: 147, 162; Magni 2000; Salvi 2004: 59–60, 85–6, 97–8). For instance, Oniga (2004: 96) hypothesizes that classical Latin SOV can be derived from the necessity of raising the subject and object out of the VP to SpecAgrSP and SpecAgrOP, respectively, to check their strong Case features (directly reflected in their rich morphological case forms). The rise and generalization of SVO in late Latin, by contrast, is related to a weakening and loss of case morphology, correlatively translated into a weak Case feature on AgrSP and AgrOP which can be checked covertly, allowing the subject and object to remain within the VP yielding the order SVO (see also the discussion in }2.2.3).

The rise of constituent structure

65

gained ground statistically since Classical times, and in some texts they form the clear majority.’ Significantly, it is precisely this predominant verb-medial order identified above by Herman for late Latin that, under the more usual label of Verb Second (V2), has been frequently claimed to constitute the transitional phase between an original Latin SOV order and the modern Romance SVO order.73 This V2 syntax is particularly well preserved in medieval varieties (though see Sornicola 2000), especially, though not exclusively, in Gallo-Romance and Ræto-Romance (where it survives to the present day in Swiss varieties and Ladin spoken in the province of Bolzano, where German contact influence might, of course, have played a role; Haiman 1988: 368–9; Haiman and Benincà 1992: 167–75).74 During this V2 stage sentences consist of two principal parts (42a), a sentential core with fixed S V O ADV order, and a left edge consisting of a C(omplementizer) position to which the finite verb is raised in root clauses (}4.4.1) where it is preceded by one or more elements fronted from the sentential core to be assigned a pragmatically salient (Foc(us), Top(ic)) reading. In embedded clauses, by contrast, the left edge generally hosts an overt COMP(lementizer) and the finite verb is consequently forced to remain within the sentential core, yielding the order S+V+O+ADV (42b). Thus, as the following representative early Romance examples demonstrate, alongside S+V+X (42c) we also frequently find in main clauses O+V(S) (42d), IO+V(S) (40e), OPP+V(S) (42f), and ADV+V(S) (42g):75,76 (42)

a [LeftPeripheryTop / Foc V b . . . [LeftPeripheryCOMP

[SententialCore S V O ADV]] [SententialCore S V O ADV]]

c [Lo cavaliere prese [lo cavaliere prese i marchi]] (OTsc., Nov. 4) ‘the knight took the marks’ d [Sa venoison fist [fist a l’ ostel porter sa venoison]] (OFr., CDN 30) his goods he.made to the hostel carry.INF ‘he had his goods carried to the hostel’

73 Harris (1978: 20–1), Renzi (1985: 267–75), Vincent (1988: 62; 1998a: 422–3), Salvi (2000; 2004: 107–11). Clackson and Horrocks (2007: 291–2) recognize a similar V2 pattern in the late fourth-century Peregrinatio Aetheriae, where they identify ‘an underlying order with the verb occupying the first position in the sentence, with an optional focus slot before it, which may be filled by a verbal argument (subject as default) or an adverbial phrase’ (p. 292). 74 See Price (1971: 259–60), Skårup (1975), Vanelli, Renzi, and Benincà (1985), Vanelli (1986; 1999), Adams (1987), Dupuis (1989), Herman (1990), Fontana (1993; 1997), Roberts (1993a), Benincà (1995; 2006), Lemieux and Dupuis (1995), Ribeiro (1995), Vance (1997), Kaiser (1999; 2002; 2002–3), Lombardi and Middleton (2004), Salvi (2004; 2005: 430–1), Ledgeway (2007b; 2008a). A clear exception to the general V2 nature of medieval Romance is old Sardinian, which appears to be a VSO language (Lombardi 2007). 75 Sornicola (2000: 104–6) argues that even in medieval Romance SVO is the predominant order. 76 In all examples that follow, topicalized constituents are underlined, contrastive foci appear in small capitals, and non-contrastive foci in bold. All other salient categories appear in italics.

66

From Latin to Romance e [A questo resposse [Iasone resposse a questo]] (ONap., LDT 60.12) to this replied Jason ‘Jason replied to this’ f [D’ ALGUÑAS COUSAS me calarei [me calarei d’alguñas cousas]] of some things myself= I.will.fall.silent (OPt., DSG 1.5.25) ‘I shall remain silent about certain matters’ g [Luenh es [lo castelhs e la tors es luenh]] (OOcc., Rudel PADCE 3.1) far is the castle and the tower ‘Far is the castle, far is the tower’

This same generalized raising of the finite verb to C(OMP) under V2 is also generally recognized to bear on the asymmetric distribution of null subjects in medieval Romance,77 a pattern more robustly represented in Gallo-Romance (French, Occitan, and northern Italian dialects) than either in Ibero-Romance or central-southern Italo-Romance (Salvi 2004: 30–1). In particular, when the verb raises to the vacant C(omplementizer) position, null subjects are freely licensed (43a–c), whereas in subordinate clauses, where the finite verb is forced to remain in situ within the sentential core, pronominal subjects must be phonologically expressed (44a–c), although not interpreted as emphatic or contrastively focused. Illustrative in this respect is the old Umbro-Tuscan example in (44c), where, despite the coreference of main and embedded clause subjects, the latter is overtly realized yielding a structure which would be judged ungrammatical, for example, in modern Italian. (43) a [SI errent [ errent si tant en so wandered.3PL (they) so.much in maniere qu’[ . . . ]]] (OFr., MRA 16.66) way that ‘they thus wandered so much in such a way that [ . . . ]’

tele such

b [Del cor sospir [ sospir del cor]] e [dels olhs plor of.the heart sigh.1SG (I) and of.the eyes cry.1SG (I) [ plor dels olhs]] (OPrv., ECS 19) ‘I sigh from the heart and cry from the eyes’ c [manifestamente l’hoe [ hoe veduto manifestamente manifestly it=have.1SG (I) seen nelle cose [ . . . ]]](OTsc., Nov. 3) in.the things ‘I have seen this clearly in those things [ . . . ]’

77

Adams (1987), Roberts (1993a: }3.2), Benincà (1994b; 2006; 2010: }3.2.1), Salvi (2004: 16–17, 26–31).

The rise of constituent structure

67

(44) a Endementiers [que [il regadoient les letres]] et la damoisele, while that they looked.at the letters and the young.lady [que [il plagnoient sa mescheance]] (OFr., MRA 71.47–8) that they regretted her misfortune ‘While that they looked at the letters and the young lady, [and that] they regretted her misfortune’ b Non es meravelha [s’ [eu chan / Melhs de nul autre chantador]] not is wonder if I sing better of no other singer (OPrv., ECS 1–2) ‘It is no wonder if I am a better singer than all others’ c Elli conosce certamente [ch’ [elli avea ucciso lo migliore cavaliere he knows certainly that he had killed the best knight del mondo]] (OUmb.-Tsc., Trist. Ricc. 238.6–7) of.the world ‘He certainly knows that he had killed the best knight in the land’ This asymmetrical distribution leads us to conclude that null subjects in medieval Romance were not licensed exclusively, if at all, by rich verb inflection for person and number, but, by a property which the finite verb uniquely acquires by raising to the vacant C(omplementizer) position, presumably the locus of finiteness in medieval Romance (cf. the discussion of the shift in the locus of finiteness and inflection in }3.4.1).78 78 Interestingly, this archaic pattern has been exceptionally retained in modern Corsican (Marchetti 1974: 25, 51, 85, 94, 119) which behaves as a canonical Romance null subject variety in root clauses (i.a–b), but which in embedded contexts usually requires referential (i.a–c) and non-referential (i.d) pronominal subjects to be overtly realized, albeit in reduced clitic form (cf. tonic/clitic eio/e, o (1SG), tù/tu (2SG), ellu/ella / (e)llu/(e)lla (3M/FSG), noi/no (1PL), voi/vo (2PL), elli/elle / (e)lli/(e)lle (3M/FPL)). This distribution of null and overt pronouns is however only superficially similar to medieval Romance, in that modern Corsican is not a V2 variety and the finite verb occurs in the sentential core in all cases, irrespective of the realization of the pronominal subject. However, when the finite verb does raise to the vacant C(omplementizer) position, as happens in polar interrogatives (i.e), realization of the clitic subject pronoun is blocked in a similar fashion to what was seen for medieval Romance (cf. 43a–c).

(i) a  avemu da fà from do.INF (we) have.1PL no truvemu (Cor.) we= find.1PL ‘we’ll pray to all the saints we meet’

e the

divuzioni devotions

b  e cumprerete à u scagnu (you) them= you.will.buy a the office ‘you buy them at the office when you come down’

quand’è when

à to

tutti all

vo you=

i the

santi saints

falate (Cor.) descend.2PL

c eo ogni volta ch’ o tornu in Cervioni, è ch’ I each time that I= return.1SG in Cervione and that o u sentu u nostru our I= it= hear.1SG the chjoccu mi mette sottusopra (Cor.) chiming me= it.puts under.over ‘each time I go back to Cervione and (that) hear our chiming bell I get overcome’

chè that

From Latin to Romance

68

Examples like these illustrate an early stage in the passage from Latin to postmedieval Romance: as in Latin, word order is free, though pragmatically conditioned, in the left edge, but is fixed, as in modern Romance, within the sentential core. Similarly, although pragmatic fronting to the left edge generally targets entire phrases as in (42c–g), a hallmark of a configurational Romance-style syntax (Poletto 2005: 210–11; 2006; Ledgeway 2007b: 126–8; 2009a: }21.1.2.2.2), it can also frequently target individual elements of a given phrase, a hallmark of a non-configurational Latin-style syntax, yielding discontinuous structures like those in (45a–c): (45)

a ma multo plu me reputo gloriuso (ONap., LDT 50.38) but much more myself= I.repute glorious ‘but I consider myself much more glorious’ b Grant á el gozo mio Çid con todos sos vassallos (OSp., Cid 803) big has the joy my Cid with all his vassals ‘Great is the joy that my Cid has with all his vassals’ c Tant a Boecis lo so.much has Boecis the ‘Boecis appears so dazzled’

vis face

esvanuit (OOcc., Boecis 28.4) dazzled

The emergence of a late Latin/early Romance V2 syntax clearly represents the outcome of an unmistakable compromise between two competing grammars, combining aspects of an earlier non-configurational syntax on the one hand and aspects of an innovative configurational syntax on the other. Eventually the new configurational pattern comes to prevail: on account of the high frequency of subject fronting to the sentential left edge as the default topic within the older V2 system, preverbal subjects are progressively reanalysed as occupying a non-derived position within the sentential core: (46)

[Left Periphery STopic Vfinite [Core S Vfinite X]] ) [Left Periphery  [Core S Vfinite X]]

The result is the unmarked SV(O) order of modern Romance, where all but UNDERGOER subjects are now associated with, and licensed in, a dedicated position within

d Ch’ ellu un ci sia troppu that it= not there= is too.much ‘Make sue that there isn’t too much noise’ e [QUANDU when

falate descend.2PL

[ (you)

‘When are you coming down?’

rumore (Cor.) noise

falate quandu]]? (Cor.)

The rise of constituent structure

69

the configurational structure of the sentential core (47a–b).79 In many GalloRomance varieties grammaticalization of this subject position has run its full course, such that even UNDERGOER subjects are now attracted to the preverbal position (48a–b), thereby erasing an earlier reflex of an active/stative distinction (see }7.4.2). (47)

a o João abriu a janela (Pt.) the João opened the window ‘João opened the window’ b abriu-se a janela (Pt.) opened=itself the window ‘the window opened’

(48) a Jean a ouvert la fenêtre (Fr.) ‘Jean has opened the window’ b la fenêtre s’ est ouverte (Fr.) the window self= is opened ‘the window opened’ Oversimplifying somewhat and putting aside some minor exceptions, Romance sentential word order can be said then to have converged in the modern languages towards a predominantly SVO order, in which the grammatical functions of subject and object are unambiguously marked by their respective positions to the left and the right of the verb, as illustrated by the following representative Aragonese example:80 (49) o pai de Chuan ha bendidas the father of Juan has sold ‘Juan’s father has sold the sheep’

as the

güellas (Ara.) sheep.PL

The principal exceptions here are Spanish and Romanian, where there is some controversy about the status of SVO (50a, 51a) as the unmarked order in view of the not uncommon occurrence of VSO (50b, 51b), though interpretations of the latter sequence, especially in Spanish, can vary considerably in accordance with diatopic and idiolectal variation.81 Particularly revealing in this respect is Vincent’s (1998a: 62) 79 However, some have argued that preverbal (definite) lexical subjects in null subject languages, especially Ibero-Romance, are invariably left-dislocated (see, among others, Contreras 1991; Moro 1993; Barbosa 1997; Alexiadou and Anagnostopoulou 1998); for an opposing view, see Cardinaletti (1997: }3; 2004: 148–9; 140–4). 80 See, among others, Padua (1960), Pontes (1987), Benincà (1988: 119), Berlinck (1989), Decat (1989), Doria (1989: 527), Costa (2000b), Ayres-Bennett and Carruthers (2001: ch. 9), Azevedo (2005: 168–9, 247–8). 81 Green (1977: 26), Harris (1978: 20), Stati (1989: 122), Zubizarreta (1998), Alboiu and Motapanyane (2000b: 24–5), Zagona (2002: ch. 5), Cardinaletti (2004: 117–8), Giurgea and Remberger (2009). There are also a number of traditional claims in the descriptive literature about the preference for V-final structures in Sardinian (i.a; Allières 2001: 157) and Sicilian (i.b; Rohlfs 1969: 323–4):

70

From Latin to Romance

observation that in modern Romance VSO, unlike VOS, proves extremely rare, inasmuch as it would involve interruption of the otherwise inseparable VP constituent. Therefore, when VSO obtains in Spanish and Romanian, it is generally argued to be a derived order involving raising of the verb out of the VP with subject in situ:82 (50) a el gato se comió un ratón (Sp.) the cat self= ate a rat b se comió el gato un ratón (Sp.) self= ate the cat a rat ‘the cat ate a rat’ (51)

a pisica a răsturnat vasul (Ro.) cat=the has overturned pot=the ba răsturnat pisica vasul (Ro.) has overturned cat=the pot=the ‘the cat knocked the pot over’

(i) a mandicatu as? (Srd.) eaten you.have ‘have you eaten?’ b iddu picciliddu he little ‘he’s a child’

è (Sic.) is

As highlighted by Loporcaro (2002), superficial observations such as these which propose an SOV word order for these varieties are not founded on a structural analysis of the data. Indeed, closer scrutiny of the relevant evidence shows that V-final position in such examples is a derived position produced by (informational/contrastive) focus-fronting of the other clausal material (Jones 1993: 332–5; Cruschina 2006; 2008; 2010; Bentley 2007; Mensching and Remberger 2010a, b; Zanini and Damonte 2010). 82

Zagona (2002: 214–16), Dobrovie-Sorin (1994), Motapanyane (1989). VSO order is also attested, albeit under considerably more restricted conditions, in other Romance languages such as Italian (though not for all speakers; G. Salvi p.c.) and Catalan (Solà 1992: 58 n. 13, 78; Cardinaletti 2004: 119). For instance, in Italian finite clauses focused pronominal subjects (i.a) and indefinite/negative quantified subjects (i.b–c) can occur in the so-called middle field with resultant VSO order: LUI/**GIANNI le chiavi (It.) (i) a di quel cassetto ha of that drawer has he/Gianni the keys ‘He/Gianni has got the keys to that draw’ b conoscerà qualcuno l’ indirizzo (It.) will.known somebody the address ‘Somebody will surely know the address’ c non mi toglierà nessuno questi diritti (It.) not me= will.remove nobody these rights ‘nobody will take these rights away from me’

In infinitival complements, by contrast, postverbal subjects are not necessarily focused, witness (ii): (ii)

CONTO, non Giorgio accettò di pagare lui IL Giorgio agreed of pay.INF he the bill not ‘Giorgio decided to pay the restaurant bill, not for the hotel’

l’ the

albergo (It.) hotel

The rise of constituent structure

71

To conclude, we have seen that, unlike the Latin sentence, the Romance sentence provides dedicated positions for the verb, its arguments, and any accompanying adjuncts, whose fundamental structure can be summarized according to the linear template in (52): (52)

S (AUX) V (*ADV) (O) (IO) (*ADV)

3.4 Configurationality: concluding remarks Although the existence or otherwise of non-configurationality is a controversial issue which still divides linguists, largely according to their choice of theoretical framework,83 we have seen that there is nonetheless an inescapable fundamental difference in the grammatical organizations of Latin and Romance syntax which simply cannot be overlooked: whereas in the former grammatical relations are encoded by the forms of words themselves through case and agreement morphology, so-called lexocentricity (Bresnan 2001: 109-12),84 in the latter grammatical relations are encoded through the syntactic context of individual words organized into distinct hierarchical phrase structure configurations. Indeed, as Vincent (1998a: 423–4) observes, (classical) Latin presents all of Hale’s (1983) classic tests for non-configurationality originally established on the evidence of Warlpiri (see also Mereu 2004: 119–22): (53) a word order determined by pragmatic, and not syntactic, properties (yielding so-called ‘free word order’) b discontinuous constructions c null anaphora85 83 For an overview of the various positions, see Hale (1981; 1982; 1983), Simpson and Bresnan (1983), Jelinek (1984), Williamson (1984), Ostafin (1986), Kiss (1987; 2008), Maracz and Muysken (1989), Speas (1990), Baker (1991; 1995; 2001; 2003), Kessler (1995: 2), Rögnvaldsson (1995), Austin and Bresnan (1996), Vincent (1999: 1112–13), Austin (2001), Bresnan (2001: 5–15, 109–14), Legate (2001; 2002), Frank (2003), Golumbia (2004), Mereu (2004: 119–79), Oniga (2004: 100–1), Pensalfini (2004), Polo (2004: 401), Devine and Stephens (2006: 83, 124), Hewson and Bubenik (2006), Aranovich (2007), Sugisaki (2007), Dryer (2009), Luraghi (2010a). 84 The idea of lexocentricity is not in itself a modern concept, but has precursors in many traditional analyses, witness the following remarks by Meillet:

In addition to the meaning expressed in the stem, inflection marks the role played by each word in the sentence; the word is therefore autonomous and is sufficient in itself to indicate its own meaning and role in the discourse. This complete autonomy of each word represents the fundamental property on which the whole structure of the Indo-European sentence rests. (Meillet 1908: 321; cf. also p. 324). 85 Hale (1983) considers the free occurrence of null anaphora, together with grammatically free word order and discontinuous structures, among the core defining properties of non-configurationality. Baker (2001) too considers the availability of null anaphora to lie at the heart of non-configurationality, a view contested by Pensalfini (2004: 393–402). Another issue which requires further investigation, which we cannot explore here, concerns the productivity of noun incorporation structures in Latin (e.g. ANIMUM ‘attention’ + ADUERTERE ‘direct.INF’ > ANIMADUERTERE ‘to realize’, CURA ‘care’ + AGERE ‘carry.out.INF’ > CURAGERE ‘to take care’) and their relevance, if any, for the non-configurational status of the language.

72

From Latin to Romance d rich case system to encode argument structure86 e rich agreement system f absence of expletive elements g absence of a VP constituent and subject-object asymmetries.

With the exception of (53c, f–g), all the other criteria have already been extensively described and richly exemplified in the preceding discussion. The validity of (c) and (f) for Latin are relatively easy to prove: not only did Latin lack expletive subjects with impersonal verbs (54a), but it also readily dropped referential subjects as well as objects (54b), a fact too often forgotten, when these could be readily recovered from the pragmatic context.87,88 (54)

a In uilla, cum  pluet, circumire  oportet, in farm when (it) rains surround.INF (it) is.necessary sicubi  perpluat, et signare carbone, cum  if.anywhere(it) rain.through.SBJV.3SG and mark.INF charcoal.ABL when (it) desierit pluere, uti tegula cease.FUT.PFV.3SG rain.INF so.that tile.NOM mutetur (Lat., Cato Agr. 155) exchange.SBJV.PASS.3SG ‘You should look around the farmstead while it is raining, and mark all leaks with charcoal, so that the tile can be replaced after the rain stops’

86 One important difference between Latin and Warlpiri concerns the distribution of case inflections in continuous and discontinuous structures (Hale 1983: 38–9; Lehmann 1985: 94–5; Austin and Bresnan 1996: 218). In Warlpiri agreeing case morphology on all the individual elements of a nominal group only proves obligatory when the elements are discontinuous (i.a–b), whereas in Latin such case-marking is obligatory whether linear continuity obtains or not (ii.a–b).

(i)

a maliki dog

wiribig-

ngki

-

ERG

PRF

b maliki- rli ERG PRF dog ‘the big dog bit me’ (ii) a CANIS dog.M.NOM b CANIS dog.M.NOM

ji 1SG.ABS

ji 1SG.ABS yalkubite

yalkubite rnu PST

MAGNUS

ME

MOMORDIT

big.M.NOM

me.ACC

bit

ME

MOMORDIT

me.ACC

bit

rnu (Wrl.) PST

wiribig

nki (Wrl.) ERG

(Lat.)

(Lat.) big.M.NOM

MAGNUS

The systematic case-marking of all elements of the Latin nominal group highlights how, contrary to Warlpiri, positional contiguity, even when it obtains, is unable to function as an alternative to lexocentricity which represents the sole exponent of grammatical relations. From this perspective, Warlpiri arguably displays a greater propensity towards configurationality than Latin. 87 Dressler (1971), Vincent (1988: 59; 2000: 38–40, 43–4), Mulder (1991), van de Wurff (1994; 1997), Luraghi (1997; 1998: 183–4; 2006), Sznajder (1998), Oniga (2004: 58–9), Pieroni (2010: 430–2), Spevak (2010: 99–106). 88 Significantly, Bošković (2010: }2.4, }4.5) relates this radical pro-drop property to the lack of a DP (determiner phrase) layer in the nominal group.

The rise of constituent structure

73

b si i j in iusk uocat j if (accuser) (accused) in law calls (accused) k ito. Ni j k it, i (there) go.FUT.IMP if.not (accused) (there) goes (accuser) antestamino. Igitur i emj call.wintess.FUT.IMP thus (accuser) him capito (Lat., Leg. XII Tab. 1) seize.FUT.IMP ‘If (the accuser) sues (the accused), (the accused) must attend (there). If (the accused) does not attend (there), (the accuser) should call witnesses. In that case (the accuser) should have him arrested’ Luraghi (2006: 237, 239–42) distinguishes between so-called pragmatically conditioned object-drop as in (54b), where the reference of the implied object is non-focal and highly topical and can be readily inferred from the discourse, and syntactically conditioned object-drop which is limited to specific structural environments, namely coordinated clauses (55a), conjunct participles (55b), and replies to polar questions (55c). (55) a Casesar exercitumi reduxit et [ . . . ] i in hibernis Caesar.NOM army.ACC led.back and (it) in winter.quarters.ABL conlocauit (Lat., Caes. B.G. 3.29.3) he.placed ‘Caesar led his army back and [ . . . ] lodged it in the winter camp’ b conuocatis suis clientibusi, i summoned.ABL.PL his.ABL.PL dependents.ABL (them) facile incendit (Lat., Caes. B.G. 7.4.1) easily he.excited ‘(Vercingetorix) summoned together his dependents, and easily excited them’ c nouistine hominemi? i (him) you.know=Q man.ACC ‘do you know the man? I do’

noui (Lat., Pl. Bacch. 837) I.know

Diachronically, Luraghi (2006: 245–8) highlights how in later Latinity pragmatically conditioned object-drop is the first to disappear, whereas syntactically conditioned null objects survive, albeit also increasingly rivalled by overt anaphoric pronouns (56a). Some residues of these syntactically conditioned (but not pragmatically conditioned) null objects are even found in early Romance (Luraghi 1998: 189–91), namely in coordinated clauses (56b) and in replies to polar interrogatives (56c).

74 (56)

From Latin to Romance a colligite alligate primum zizaniai et gather.IMP.2PL first darnel and bind.IMP.2PL in fasciculos (Lat., Vulg. Mat. 13.30) in bundles ‘First, gather up the darnel, and bind them in bundles’

eai them

b che tutta quanta si lacerò e  infranse (OTsc., LVV 59) which all how.much self= tore and (self) broke ‘which completely tore and broke’ c non avestù la torta? Messer sí,  not you.had the cake sir yes (it) ‘Did you not have that tart? Sir, yes, I had it’

ebbi (OTsc., Nov. 79) I.had

By contrast, the modern Romance languages are generally specified negatively in relation to these same criteria or, at most, present only partially positive specifications for some of them. For example, while most Romance languages license referential (57a) and expletive (57b) null subjects, others, especially Gallo-Romance, obligatorily realize both referential (57a') and expletive pronominal subjects (57b'). (57)

a  a' ils ‘(they) b  b' il ‘(it)

dorm (Ro.) dorment (Fr.) sleep’ plouă (Ro.) pleut (Fr.) rains’

Null objects, on the other hand, are even more restricted: fully referential null objects are principally limited to European and, especially, Brazilian Portuguese (though much rarer in more formal registers; cf. Bachmann 2011),89 where the construction appears to be an innovation and related to the increase in overt subject pronouns (58a; Galves 1993; Morais 2003), while generic null objects are more widely found, especially with animates (58b; Rizzi 1986), and even with inanimates in French (58c).90 (58)

a Você não trouxe [passaporte]i? Aí – é que está, eu trouxe i mas como you not brought passport there is that it.is I brought (it) but since não precisei mostrar i deixei i no hotel (BPt.) not I.needed show.INF (it) I.left (it) in.the hotel ‘Didn’t you bring your passport? – That’s the point, I did bring (it), but since I didn’t need to show it, I left it at the hotel’

89 See Thomas (1969: 98), Teyssier (1984: 88), Raposo (1986), Cyrino (1996; 1997a,b; 1998; 2001; 2003; 2006), Schwenter and Silva (2002), Azevedo (2005: 228–9, 234–7), Bachmann (2008). 90 Zribi-Hertz (1984), Schsler (2000), Rowlett (2007: 183).

The rise of constituent structure b Ciò induce  a sospettare that induces (people) to suspect.INF ‘This leads us/people to suspect that . . . ’

75

che . . . (It.) that

c Tu aimes [le porc]i? – Oui, j’ adore you likes the pork yes I adore ‘Do you like pork? – Yes, I really love it’

i (Fr.) (it)

Turning finally to (53g), linearization evidence was reviewed above to highlight the difficulty of identifying a VP constituent in Latin, a conclusion further substantiated by the apparent absence of any VP-fronting operations (Vincent 1998a: 423 n. 10).91 Further support for this view comes from the absence of subject-object asymmetries. In particular, given a configurational structure of the sentence, in which the subject in its preverbal position (SpecVP) hierarchically dominates (or in more technical terms, c-commands) the object in its more deeply embedded postverbal position (V,NP), there should follow a whole series of subject-object asymmetries which are often simply absent from non-configurational languages like Latin with its ‘flat’ organization of the verbal group (Lyons 1999: 154). For instance, in the surface configurational structures of Romance a reflexive anaphor can be an object whose reference is controlled by a preceding subject (59a), but it cannot be a subject referentially controlled by an object (59b). (59) a [Ana]i critica [a Ana criticizes ACC.MRK ‘Ana criticizes herself ’ b **[sí misma]i critica self same criticizes ‘herself criticizes Ana’

sí self

misma]i (Sp.) same

[a ACC.MRK

Ana]i (Sp.) Ana

In Latin, by contrast, which we have argued lacks hierarchical constituent structure, subject and object do not enter into an asymmetrical relation of dominance and precedence (viz. c-command) at the surface structural level. Consequently, as Vincent (1997c: 163) highlights following Bertocchi (1989), the Latin reflexive anaphor may precede its antecedent as in the following Plautine example:

It is not difficult, however, to find apparent counterexamples such as (i), if we assume that the underlying position of infinitival complements with accusative subjects is postverbal (but for a different view, see }5.4.2.1). 91

(i)

hasce] ait architectonem [VP laudauisse hasce] (Lat., Pl. Mostell. 760) [VP laudauisse praise.INF.PRFV these.ACC.PL he.says architect.ACC ‘he says that the architect has praised this (house)’

76 (60)

From Latin to Romance rex] [reginae]i placet (Lat., Pl. Stich. 133) [suusi her.NOM king.NOM queen.DAT pleases ‘[heri king] is pleasing to [the queen]i’

If we were to assume a configurational structure of this Latin sentence, it would apparently be impossible for the anaphor suus rex ‘her king’ from the higher subject position to be bound by its dative antecedent reginae ‘to the queen’ situated in its more deeply embedded object position. However, the example involves an unaccusative psych-predicate PLACERE ‘to please’, which, according to standard analyses,92 selects an EXPERIENCER argument reginae base-generated in a higher VP-internal position than the THEME argument suus rex in the complement position. At an underlying level, the dative argument reginae can therefore be argued to bind (under c-command) the anaphor suus rex, as well as at surface structure if we assume, with Belletti and Rizzi (1988), that the dative EXPERIENCER moves to the canonical preverbal subject position. Such an analysis is not, however, available in the following example with the nonpsych-verb PATERE ‘to be open’, in which, if we were to assume a configurational structure, we would not be able to explain how the topicalized dative argument Pompeio ‘to Pompeius’ successfully binds the nominative subject sua domus ‘his own house’, despite failing to c-command the latter in its base position.93 (61)

[Pompeio]i [sua domus]i patebit (Lat., Cic. Phil. 13.10) Pompeius.DAT his.NOM house.NOM will.be.open ‘his own house will be open to Pompeius’

Similar considerations apply to the long-distance control of reflexives, as in the following example discussed by Pieroni (2001): (62)

at Scipione [ . . . ] litterae sunt consecutae [a M. Fauonio]i, but Scipio.ACC letters are reached by M. Favonius Domitium cum legionibus adesse neque [sei] praesidium Domitius.ACC with legions approach.INF and.not self.ACC garrison tenere posse (Lat., Caes. B.C. 3.36.6–7) keep.INF be.able.INF ‘but dispatches from [Favonius]i reached Scipio, that Domitius was marching against him with his legions, and that [he]i could not maintain the garrison’

Significantly, Pieroni (p. 541) notes that one of Schachter’s (1977) reference-related criteria for subjecthood, namely the ability to control reflexives, is not invariably met

92

Postal (1971), Belletti and Rizzi (1988), Landau (2010). It is not clear whether PATERE is an unaccusative verb; if it were, then the example in (61) could be treated along the lines of (60) above. 93

The rise of constituent structure

77

in Latin, witness the example in (62) where the reflexive se ‘self ’ (functioning as accusative infinitival subject) is not bound by a nominative argument, but by the non-c-commanding adjunct PP a M.Fauonio ‘by M.Favonius’, a control relation which she explicitly acknowledges (p. 542) would prove ungrammatical in modern Italian. Evidence like this suggests that the correct interpretation of (Latin) anaphors cannot be simply read off surface syntactic structures, but follows from predication relations mapped between functional- and constituent-structures.94 In short, the examples reviewed in (60)–(62) highlight how in Latin the licensing of binding processes such as anaphora and control is ultimately not sensitive to c-command asymmetries, insofar as the surface arrangement of lexemes does not systematically correlate with the projection of argument structure from the lexicon. Rather, the interpretive properties of such expressions are predominantly licensed morphologically through Latin’s rich case and agreement systems.95 3.4.1 Degrees of configurationality Finally, although we have identified in this chapter a number of facts which point to a non-configurational organization of Latin syntax, this cannot be taken to imply that configurationality is entirely lacking in Latin (or in any language for that matter; Mereu 2004: 135–9). Rather, as will be shown in chapters 4 and 5, Latin displays unmistakable evidence for configurationality both within the clausal and prepositional domains,96 inasmuch as configurationality is not a discrete property, with individual languages ostensibly displaying differing degrees of (non-)configurationality. In the particular case of Latin, we have examined extensive evidence to illustrate the non-configurational, flat organization of the verbal, nominal, adjectival, and adverbial groups, while the same cannot be said so readily for the organization of the sentence and even less so for the prepositional phrase. As will be illustrated in considerable detail in }}4.4.1.–1 and }5.3.1.1, from as early as the archaic Latin period there is already considerable evidence for an incipient configurational C(omplementizer)P(hrase) structure, as witnessed in the existence of two conflicting patterns in 94

Hale (1994: 185), Bresnan (2001: 7–10), Mereu (2004: 137–9). A further indication of the emergence of subject-object asymmetries in the development of Romance comes from the way the deictics IPSE ‘self ’ and ILLE ‘that’ have become associated with subjects and objects. On this point, Vincent (1997c) demonstrates how the definite article (see further }4.2.2) was originally limited to marking subject NPs (or, more generally, external arguments; cf. also Pieroni 2010: 473) via the grammaticalization of topics (hence the sources of the Romance article ILLE, originally a firstmention cataphor, and IPSE a second-mention anaphoric topic marker), whereas object NPs frequently occurred without an article, even when fully individuated. On the other hand, objects (or, more generally, internal arguments) came to be marked by clitics, which, by virtue of being prosodically weak, were not suited to marking informationally prominent referents and incompatible with the topicmarking value of IPSE. 96 Latin can even be argued to be fully configurational in all areas of the grammar, a possibility explored in chapter 5. 95

78

From Latin to Romance

the marking of the clause. The first represents an archaic non-configurational pattern inherited from the Indo-European parent language, in which a number of core complementation structures without overt subordinators, notably the accusative and infinitive construction (AcI), manifestly do not involve a CP projection.97 The second constitutes an innovative configurational pattern, albeit attested since the archaic Latin period, in which a number of subordination types with overt complementizers, as well as an incipient V2 syntax, both frequently preceded by fronted topics and foci, make recourse to an articulated CP structure (Oniga 2004: 94; Devine and Stephens 2006: 118–9). In a largely similar vein, analogous claims have been made about the precocious development of configurationality in the Latin prepositional group, where the order head + complement had become established since earliest times following the reinterpretation of Indo-European adverbial particles in preverb function as adpositions.98 Indeed, non-configurational prepositional disjunction and postposition of the adposition (so-called ‘anastrophe’) were extremely uncommon in Latin (63a), occurring only in specific archaic styles and registers,99 while the firmly established configurational order preposition + complement prevailed in all other cases (63b; Vincent 1999: 1122). This configurational pattern is further substantiated by frequent structures such as (63c), where the focus-fronted adjectival modifier of the nominal complement can be taken to occupy the left-peripheral modifier/Spec(ifier) position of the PP (Bauer 1995: 137; Vincent 1999: 1127–8). (63)

a incedunt arbusta per they.advance timber.trees.ACC.N through ‘they march through tall timber-trees’ b qualis formicis [PP [P' per [COMPL like ants.DAT through Sen. Tranq. 12.3) ‘like ants climbing up the trees’

alta (Lat., Enn. Ann. 181) tall.ACC.NPL

arbusta]]] trees.ACC.PL

repentibus (Lat., crawling.DAT.PL

97 Palmer ([1954] 1990: 328), Haudry (1973), Bichakjian (1982), Calboli (1983b: 41–2), Lehmann (1984), Ramat (1994: 259–60, 264), Bauer (1995: 159–60; 2009a: 250, 258), Devine and Stephens (1999: 148). 98 Rubio ([1972] 1982: 199–200), Joseph (1991), Bauer (1995: 130–4; 2009a: 259–61), Cuzzolin (1995), Vincent (1998a: 424 n. 11; 1999), Oniga (2004: 94), Hewson and Bubenik (2006), Luraghi (2010b: 89–93). There are different views in the literature regarding whether the establishment of the Latin prepositional phrase represents a parametric shift from a left-branching postpositional syntax to a right-branching prepositional syntax (Marouzeau 1949: 44–57; 1953: 61–9; Bauer 1995: 137–9; 2009a: 260–1; Hewson and Bubenik 2006; Magni 2009: 229–33), or whether prepositional syntax is original with apparent cases of postpositions treated as Wackernagel enclitics (e.g. ME/TE=CUM ‘me./you.=with’ (‘with me/you’); Vincent 1999: 1128). See discussion in }5.3.1.1. 99 Leumann and Hofmann (1928: 495), Marouzeau (1949: 42; 1953: 62, 67), Bauer (1995: 131–2, 146–7; 2009a: 291), Penney (1999: 266–7, 249–68).

The rise of constituent structure c [PP [Spec MAGNO] [P' cum [COMPLmagno great.ABL with omnes ferebant (Lat., Caes. B.G. 7.15.2) all.NOM they.bore ‘all bore this with GREAT regret’

79

dolore]]] sorrow.ABL

We can therefore conclude that from its earliest attestations Latin presents extensive evidence of a non-configurational syntax built around otherwise isolated configurational CP and PP superstructures in the sentential and nominal domains (64a–b),100 which afford the language a fixed point of reference within an otherwise free, non-configurational syntax. (64)

a

CP 3 Spec C' Top/Foc 3 C S 9 C 1 C2 Cn

b

PP 3 Spec P' Top/Foc 3 P N 9 C 1 C2 C n

Perceptively, Vincent (1998a: 424) relates this limited configurationality of Latin to a similar limited configurationality in Warlpiri, where the only fixed position in the clause is the obligatory second-position AUX (with Topic and Focus positions to its immediate left; Swartz 1988; Shopen 2001), evidence which many have taken to be indicative of a fixed, configurational structure at the I(inflectional)P(hrase) level.101 Quite naturally, this confirms our conclusion above in relation to Latin that configurationality is not strictly a binary parameter, but rather a scalar property of language which can show varying degrees of (non-)configurationality in different areas

In (64a–b) ‘S’ and ‘N’ indicate ‘flat’ structure, standing for a non-projective exocentric verbal and nominal (small) clause, which lacks a categorial head dominating one or more distinct categories ‘C’ that do not bear the typical branching relations of endocentricity (Bresnan 2001: 110). 101 Hale (1994: 188), Bresnan (2001: 6–10), Austin and Bresnan (1996), Mereu (2004: 120, 123–6). By way of illustration, consider the following sentence pair where the pre-auxiliary position is occupied by a single syntactic constituent, variously wawirri ‘kangaroo’ (i.a) and wawirri yalumpu ‘that kangaroo’ (i.b). 100

(i) a wawirri kangaroo

kapi-rna AUX

pantispear

b wawirri yalumpu kapi-rna kangaroo that AUX ‘I will spear that kangaroo’

rni NON.PST

pantispear

yalumpu (Wrl.) that rni (Wrl.) NON.PST

Sequences of words that do not form a constituent cannot fill the pre-auxiliary position. Consequently, if wawirri yalumpu were not a constituent in (i.b), then such examples would constitute a violation to the otherwise generalized requirement that the auxiliary appear in second position (Hale 1983: 6–7; 1994: 188; Austin and Bresnan 1996: 217–18).

80

From Latin to Romance

of the grammar,102 variously mixing endocentrically organized functional projections with exocentrically organized lexical projections (Bresnan 2001: 113–14; Mereu 2004: 161).103 Thus, limiting ourselves to the clause, Latin displays a fixed hierarchical arrangement of discourse functions at the level of the clause (CP), but a nonconfigurational, ‘flat’ arrangement in verbal and noun phrase structure (65a), whereas in Warlpiri the core of configurationality is located at the level of the AUX (I(nflectional)P(hrase)), with verb and noun phrases displaying a flat structure (65b): (65)

a

CP 3 Spec C' Top/Foc 3 C S COMP/Vfinite 9 C 1 C 2 Cn

b

IP 3 Spec I' Top/Foc 3 Infl S AAUX 9 C1 C2 C n

At the appropriate level of abstraction, the difference between the Latin and Warlpiri clause is not whether one is more or less configurational than the other, but at what level configurationality, a minimum of which would appear to characterize all languages (Hale 1994; Mereu 2004: 134, 139), is located: CP (and PP) in Latin and IP in Warlpiri. This naturally explains why (late) Latin eventually develops a C(omplementizer)-oriented V2 syntax whereas Warlpiri displays an I(nflection)oriented V2 syntax, since the finite verb is attracted to C(omplementizer) in the first case and Infl(ection) in the second. At the same time, we can now formalize the passage from Latin to Romance in terms of a gradual top-down development of configurationality from CP downwards, such that, following the emergence of IP and concomitant configurational structure, the locus of verb inflection slowly but steadily shifts from C to Infl with the consequence that auxiliaries and subjects acquire dedicated positions in the latter’s associated head and left-peripheral modifier/ Spec(ifier) positions, respectively. Similar arguments carry over naturally to the PP, where we also witness a gradual top-down development of configurationality such that, following the emergence of DP (determiner phrase) and concomitant configurational structure, the locus of nominal inflection increasingly moves from P to D, yielding dedicated positions for articles and elements such as demonstratives in the latter’s associated head and left-peripheral modifier/Spec(ifier) positions, respectively.104 102

Hale (1994), Lyons (1999: 154), Mereu (2004: 134). Gianollo (2006; 2007) also assumes for Latin a non-configurational representation of double postnominal genitives and all postnominal adjectives which, in contrast to their prenominal configurationally structured counterparts, do not display any ordering restrictions. 104 It is significant that within the configurational approach to Latin presented in }5.4.2, the major structural changes in the passage from Latin to Romance are independently argued to involve a top-down reversal in the linearization of head-complement also beginning with CP and PP. 103

4 Configurationality and the rise of functional structure 4.1 Introduction As a concomitant of the emergence of hierarchical constituent structure in the nominal, verbal, and sentential domains examined in chapter 3, we have seen how the left edge of the phrase in all three cases provides for a dedicated position for functional elements, namely DET(erminers), AUX(iliaries), and COMP(lementizers), the latter also hosting the finite verb in V2 contexts. This reflects the traditional intuition popularized within the synthesis-analysis approach (cf. chapter 2) that highlights the emergence in Romance of articles and clitics, auxiliaries, and a whole host of finite and non-finite complementizers, all generally absent from Latin which together with ‘most of the old IE languages, was poor in functional categories such as complementizers, auxiliaries and determiners, which are by contrast well represented in Romance languages’ (Ramat 1994: 261). In current theory, grammatical elements of this type are generally considered to head their own functional projections, namely DP (determiner phrase), IP (Infl(ectional) phrase), and CP (complementizer phrase), which represent the locus of grammatical information relating to the nominal group, verbal group, and the clause/sentence, respectively. On this view, one of the most significant generalizations of the traditional synthesis-analysis approach can now be elegantly and simply rephrased in terms of the emergence of the functional categories DP, IP, and CP and eventual splits thereof,1 which, at least according to one view (though see }5.1.1), were either entirely absent from Latin (e.g. DP, IP) or only present in incipient form (e.g. CP; cf. Vincent 1993: 146–7). In what follows we shall review a selection of the diverse evidence for the emergence and existence of functional structure which has permeated Romance in all three domains.

1

Cinque (1999; 2002; 2006), Lyons (1999: 322–3), Belletti (2004a), Rizzi (2004), Benincà and Munaro (2010).

82

From Latin to Romance

4.2 Determiner phrase (DP) 4.2.1 Indefinite article The clearest evidence for the rise of D(eterminer)P(hrase) structure in Romance comes from the universal appearance in all Romance varieties of the indefinite and definite articles. The former continues a weakened form of the Latin cardinal number UNUM/ -AM ‘one.MSG/-FSG’ (> e.g. Cat./It./Sp. un/una, Fr. un/une, Occ. u(n)/uno, Pt. um/uma, Ro. un/o), and in some varieties now formally contrasts with the numeral for ‘one’:2 (1) a nu/na guagliune/-a vs a.M/FSG youth.M/FSG ‘a boy/girl’ vs ‘one boy/girl’

unu/una one.M/FSG

guagliune/-a (Cal.) youth.M/FSG

b un vestitu vs (Quanti vestiti?) unu (Cor.; cf. una ‘a/one.FSG’) a.MSG dress.MSG how.many.M dresses.M one.MSG ‘a dress’ vs ‘How many dresses? – One’ c un băiat/ o fată vs (Cât¸i a.MSG boy a.FSG girl how.many.MPL fete?) unu/una (Ro.) girls one.M/FSG ‘a boy/girl’ vs ‘(How many boys/girls?) One’

băiet¸i?/ boys

Câte how.many.FPL

Plural forms from the accusative plural UNOS/-AS (see also Bauer 2011: }3.2.1), best considered indefinite quantifiers rather than plural articles since, unlike the corresponding singular articles, they generally prove optional (2a), are found in Ibero-Romance (2a–d) and, until the sixteenth century, also in French (3), where they were principally employed with collective plurals (Woledge 1956; Price 1971: 120).3

2 Exceptional is the situation found in many Ræto-Romance varieties (Haiman and Benincà 1992: 52) where the numeral and the indefinite article may co-occur:

øna na skwadra esoŋ pasa a trɛj (Bad.) (i) da from one a team are=we passed to three ‘from one team, we grew to three’ 3 Plural reflexes of UNUS are also found in a number of Italo-Romance dialects, where they are restricted to lexical pairings with specific quantifiers such as Cor. pochi/-e ‘few.M/F’, Sard. cantu de ‘few’ or numeral, and SID at(r)i ‘other.PL’ (Ledgeway 2009a: 186):

(i)

intensu une poche di nove canzone (Cor.) a ha he.has heard some few of new songs ‘he’s heard a few new songs’ comporau unas cantu ’e pinnas (Cpd.) b apu I.have bought some few of pens ‘I bought a few pens’ nn’ accattu n’ atri tria (Cos.) c mi myself= of.them= I.buy some other three ‘I’ll buy another three’

The rise of functional structure (2)

a vaig comprar (unes) I.go buy.INF some.FPL ‘I bought (some) oranges’

83

taronjes (Cat.) oranges

b había unas caxes enriba la it.had some boxes on the ‘there were some boxes on the table’

mesa (Ast.) table

c cantavam uns/umas senhores/senhoras (Pt.) sang.3PL some.M/FPL men/ladies ‘some men/women were singing’ d unos/unas socios/socias some.M/FPL partners.M/F ‘some very odd partners’ (3)

muy very

peculiares (Sp.) odd.PL

avoit unes grandes joes [ . . . ] et unes grans narines lées, he.had some big cheeks and some big nostrils wide et unes grosses levres [ . . . ] et uns grans dens gaunes and some large lips and some big teeth yellow et lais; et estoit cauciés d’ uns housiax et and ugly and he.was shod of some gaiters and d’ uns sollers de buef (OFr. AN 24) of some shoes of ox ‘he had great cheeks [ . . . ] and great wide nostrils, and thick lips [ . . . ] and great yellow ugly teeth; and he was shod in leggings and shoes of ox-hide’

It is these collective plurals (4a) that most frequently, though not exclusively (4b), also select the plural form of the indefinite article in Occitan (Wheeler 1988b: 260; Lafont 1991: 9), as well as in Ibero-Romance (4c–e): (4) a ausissiái una orresca clamada, unis gemècs, unis I.was.hearing a horrible roar some groans some renècs unis brams! (Lgd.) swear.words some shouts ‘I could hear a horrible roar, some groaning, some swearing, some shouting’ b fau unas còcas (Lgd.) I.do some cakes ‘I’m cooking some cakes’ c unes tisores / ulleres (Cat.) some scissors spectacles ‘(a pair of) scissors/spectacles’ d umas botas / meias (Pt.) some boots stockings ‘(a pair of) boots/tights’

84

From Latin to Romance e unos tejanos / unas jeans some.F some.M ‘(a pair of) jeans/curtains’

cortinas (Sp.) curtains

In Romanian, the plural form is most frequently used in the dative/genitive, namely UNORUM > unor ‘to/of some’, with a zero form or a lexical quantifier such as nişte, niscai(va) (coll.), or cât¸iva/câteva (M./F.) ‘some’ being employed in all other cases (5a). More rarely, in the nominative/accusative the plural forms unii/unele (‘some. MPL=the.MPL/ some.FPL=the.FPL’) are found (5b), although in such cases the determiner is stressed. (5)

a o

scriu (nişte) scrisori I.write some letters ‘I’ll write (some) letters to some friends’

FUT.MARK

să

SBJV.MRK

unor some.DAT

prieteni (Ro.) friends

b De ce unele femei sunt infidele sot¸ilor? – of what some.the.FPL women are unfaithful husbands.the.DAT.PL Pentru că unii bărbat¸i sunt avizi de putere for that some.the.mpl men are greedy of power şi averi (Ro.) and riches ‘Why are some women unfaithful to their husbands? – Because some men are only interested in power and wealth’ Significantly, most modern varieties (including Gallo-Romance, Ræto-Romance, and Italian) which lack reflexes of UNOS/-AS have grammaticalized another functional category in the determiner domain to mark indefinite plural N(oun)P(hrase)s, as well as singular mass indefinite NPs, by way of the partitive article de/di ‘of ’:4 (6) a Ugolin cultive des PART.the.MPL Ugolin grows ‘Ugolin grows (some) carnations’

œillets (Fr.) carnations

4 With the exception of early Romance (Renzi 2010: 346–7) and modern S.E. Occitan and Piedmontese, the partitive incorporates the definite article, although often omitted before adjectives and when under the scope of negation (Ronjat 1913: 46–7; Rohlfs 1968: 115–18; Renzi 1997: 163–4). In the past, the distribution of the partitive was much wider, including those varieties that also make use of a reflex of UNOS/-AS (Dias 1918: 98–9; Ledgeway 2009a: }5.1.3):

(i)

de fatica (ONap., LDR 51r.11) a aio I.have PART work ‘I’ve got (some) work’ do mell e do sall e do b deytem-lhe they.give.SBJV=him PART.the.MSG honey and PART.the.MSG salt and PART.the.MSG azeite (OPt., LDA 31) oil ‘may they give him some honey and some salt and some oil’

The rise of functional structure b d’

escagaròls amb de burre PART snails with PART butter ‘some snails with butter and parsley’

c mi a troef I SCL find ‘I’ll get some cheese’

del PART.the.MSG

d mi regalò dei to.me= he.gave PART.the.MPL ‘he gave me some flowers’

e and

de PART

85

jauverd (Lgd.) parsley

formag (Mil.) cheese fiori (It.) flowers

Contrary to the early emergence and grammaticalization of the definite article (}4.2.2), whose origins, albeit at first as a presumed ‘articloid’, have been traced to as early as between the third and eighth centuries AD (Lyons 1999: 333; though see }4.2.2.1), the indefinite article emerges much later in Romance and its usage does not become systematic in most varieties until around the fourteenth century.5 Before then the use of the indefinite article is usually reserved for particularized new referents, presumably a residue of its numeral origin, whereas a bare NP is still generally employed for non-particularized new referents.6 For example, in the eleventh-century French Vie de Saint Alexis, the article is not employed in the prayer of a desperate childless couple in their request to God (7a), since the meaning of the NP is ‘any child’. When, however, God blesses them with a child, this is reported with the use of the indefinite article (7b), since the NP now picks out a specific individual (although the article could still occasionally be omitted even in such cases of high individuation, witness its absence in (7c) just two lines later): (7) a Enfant nos done qui seit child us= give.IMP who be ‘Give us a child of your pleasing’ b Un fi lor a son to.them= ‘He gives them a son’

a to

ton your

talent (OFr., VSA 5.5) wish

donet (OFr., VSA 6.3) he.gives

c Bel nom li mistrent (OFr., VSA 6.5) fine name to.him= they.put ‘A fine name they gave him’ 5 Rohlfs (1968: 38–9), Pozas-Loyo (2008), Maiden (1995: 121; 1998: 131). By contrast, De la Villa (2010: 226–33) claims that in certain contexts some of the initial stages in the grammaticalization of the numeral for ‘1’ as the indefinite article, namely the loss of singulative meaning and the generalization of nonspecificity, are already evident in early Latin authors such as Plautus. 6 Price (1971: 118–19), Lapesa (1974: 453), Elvira (1994), Parry and Lombardi (2007: 91–2), Pozas-Loyo (2008).

86

From Latin to Romance

This early restriction on the distribution of the indefinite article is clearly revealed in the following examples, where the modern Romance translation or equivalent would require the use of the article: (8) a Buona pulcella fut good maiden was ‘Eulalia was a fine maiden’

Eulalia (OFr., SE 1) Eulalia

b infra spacio de cierti iuorni (ONap., LDT 251.11) within space of certain days ‘within a period of some days’ c Mal cosselh donet Pilat (OOcc., VMNS 106) bad advice gave Pilate ‘Pilate gave a bad piece of advice’ d Grande duelo avien las yentes great sorrow had the peoples ‘the Christian people felt a great sorrow’ e donami cavallo da give.IMP=me horse by ‘give me a horse to ride’

cristianas (OSp., Cid 29) Christian

cavalcare (OTsc., Nov. 4) ride.INF

Relics of this early non-usage of the indefinite article are still preserved today in Ibero-Romance in conjunction with reflexes of the quantifiers ALT(E)RUM ‘other’ (substandard in Catalan), CERTUM ‘certain’ (substandard in Catalan), MEDIUM ‘half’ (optionally also It. mezzo), TALEM ‘such’, and SIMIL(ANT)EM ‘similar’ (cf. also Fr./It. chose pareille/cosa simile ‘such (a) thing’):7 (9) a mañana será otro tomorrow it.will.be other altre dia (Cat.) other day ‘tomorrow is another day’

día (Sp.) day

/

demà tomorrow

serà it.will.be

(un) an

b Cada método emprega certo número de gases (Pt.) / apunten each method employs certain number of gases they.note (una) certa hostilitat (Cat.) a certain hostility ‘each method uses a certain number of gases’ / ‘they sense a certain hostility’

7 See Renzi (2010: 298, 325) for altro ‘other’ and certo ‘certain’ in early Tuscan. Modern Romanian alt ‘other’ (Lombard 1974: 215–17) fluctuates between both structures (i):

bărbat (Ro.) (i) iubeşte (un) alt she.loves an other man ‘she loves another man’

The rise of functional structure

87

c Compró medio kilo (Sp.) / ne bevo (un) mezzo bicchiere (It.)8 he.bought half kilo of.it= I.drink a half glass ‘he bought half a kilo’ / ‘I’ll drink half a glass’ d fará bem em evitar expor he.will.do well in avoid.INF expose.INF ‘he will be best advised not to reveal such an idea’ e al ver semejante paisaje se at.the see.INF similar countryside self= ‘he is taken aback at the sight of such a landscape’

tal such

idéia (Pt.) idea

maravilla (Sp.) he.wonders

More generalized relics of this early usage still abound in the modern languages in proverbs and fixed expressions such as Fr. rendre service ‘to do (a) favour’, prêter serment ‘to swear (an) oath’, à cheval ‘on (a) horse’, sous clef ‘under (a) lock and key’; It. portare giudizio ‘to pronounce (a) judgement’, a caval donato non si guarda in bocca ‘don’t look (a) gift horse in the mouth (lit. to horse gifted not one looks in mouth)’, in stato di ebrezza ‘in (a) state of drunkenness’, in carrozza ‘in (a) carriage’; Occ. donna responso ‘to give (a) reply’, cantar messa ‘to sing (a) mass’, trouva sousto ‘to find (a) shelter’, i’a bon lègo de camin ‘it’s (a) fair walk (lit. there is good league of walk)’; Pt. é melhor não cutucar onça com vara curta ‘let sleeping dogs lie (lit. it.is better not prod.INF (a) jaguar with (a) short staff)’, cozinhar em fogo baixo ‘to cook on (a) low light’, em coma ‘in (a) coma’; Ro. în stare lichidă ‘in (a) liquid state’, s-a dus bou şi s-a întors vacă ‘he went (as an) ox and returned (as a) cow (= he never achieved anything)’; Sp. sentirse como pez en el agua ‘to feel like (a) fish in water’, con mujer nueva ‘with (a) new wife’, querer por esposa ‘to want/take for (a) wife’, tener dolor de cabeza ‘to have (a) head ache’. In the modern languages, by contrast, indefinite NPs, whether particularized or not, witness the indicative/subjunctive alternation in (10), now generally require the article: (10)

busco una minyona que em neteja / netegi la casa (Cat.) I.seek a maid that me= cleans.INDIC cleans.SBJV the house ‘I am looking for a maid that cleans/to clean the house for me’

8 G. Salvi (p.c.) points out that the apparent optionality in the use of the indefinite article with mezzo ‘half’ in Italian correlates with a given/known (i.a) vs new (i.b) distinction, and hence its absence in conjunction with the clitic ne ‘of it’ in (9c) necessarily presupposes a known entity:

(i)

letto solo mezzo libro (It.) a ho I.have read only half book ‘I only read half of the book (which is already known to the interlocutors)’ letto solo un mezzo libro (It.) b ho I.have read only a half book ‘I only read half of a book (which is not known to the interlocutors)’

88

From Latin to Romance

Following Stark (2002), we should finally note that, beside the early opposition between zero and the indefinite article UNUM/-AM in introducing non-particularized and particularized rhematic NPs, respectively, early texts also display other rival quantifiers in these functions (see also Ledgeway 2009a: 187–8). Alongside zero, nonparticularized rhematic NPs could be introduced by a reflex of *al(i)kunʊ/-a ‘some’ (< ALIQUEM + UNUM/-AM ‘some’ + ‘one’), whereas particularized rhematic NPs could be introduced by a reflex of CERTUM/-AM ‘a certain’ besides the indefinite article UNUM/-AM. In terms of information structure, we find then that non-specific indefinite NPs marked by *al(i)kunʊ (or zero) tend to be backgrounded in the narrative discourse, occurring most frequently in subordinate clauses with irrealis or negated verbs (Stark 2002: 323–6), whereas indefinite NPs with specific interpretation marked by CERTUM (or UNUM/AM) are foregrounded, typically surfacing in main clauses with realis and affirmative verb forms, witness the following old Tuscan and Neapolitan examples:9 (11)

avvenne che de questa giovane se innamorò un it.happened that of this youg.lady self= fell.in.love a gentiluomo [ . . . ] il cui nome per alcuna buona accagione de gentleman the whose name for some good reason of tacere ho deliberato. [ . . . ] si pensò voler in self= he.thought want.INF in keep.secret.INF I.have decided ciò interponere l’ arte de una certa feminella that interpose.INF the art of a certain young.girl sua domestica (OTsc., Nov. 12, narr. 1) his servant ‘it happened that a noble young man fell in love with this young lady [ . . . ], whose name I decided not to tell for some good reason. [ . . . ] he thought of using the art of a certain girl, a servant of his’

9 Ronjat (1913: 38–9) reports in some Gascon varieties (Bearn, Bigorre, and Landes) a use of the adjective bet/bère ‘beautiful(M/F)’ similar to that of the indefinite article (though apparently limited to temporal expressions, hence presumably a case of lexicalization):

(i)

a bèt tems (Bea.) fine time ‘a long time ago’ plabut) (Bea.) b bèt dìe (qui abè fine day which it.had rained ‘one day (when it had rained)’ c bère noueit (Bea.) fine night ‘one night’ d a bèts cops (Bea.) to fine.PL times ‘sometimes’

The rise of functional structure

89

(12) a è pejo che serpente (ONap, Regimen 62) it.is worse than snake ‘it is more dangerous than a snake’ b Si alcune male happese if some bad had ‘If you had a skin condition’

la the

tua your

cute (ONap., Bagni 287) skin

c Quisto re Ylio fece fare in questa citate uno city a this king Ilyus made do.INF in this tiemplo (ONap., LDT 249.33) temple ‘This King Ilyus had a temple built in this city’ d de tutte le cose che eo ademanday appi certa of all the things that I asked I.had certain resposta (ONap., LDT 292.5–6 P) reply ‘for all the things I asked about I received a reply’ Interestingly, the last example highlights a common feature of this use of CERTUM, which due to its competition with the canonical indefinite article, tends to occur without reflexes of UNUM/-AM (Stark 2002: 323; Renzi 2010: 325), presumably because they were both exponents of the same category and position. Quite neatly, this explains the failure of the modern indefinite article to generalize, typically in IberoRomance, with quantifiers such as c(i)ert(o) observed in (9a–e) above. 4.2.2 Definite article 4.2.2.1 Late Latin Turning now to the definite article (see Vincent 1997c; 1998a; Zamboni 2000: 115–18), this continues a weakened form of the Latin distal demonstrative ILLE ‘that’ (> Cat./Sp. el/la, Fr. le/la, It. il/la, Ara./Cor./Glc./Pt./ SID o/a, NWCat. lo/la, Occ. lo(u) (/el)/la, Ro. -(u)l/-a) or the Latin intensifier IPSE ‘-self, very’, now with a limited areal distribution (> Bal./Costa Brava Cat. es/sa, Srd. su/sa) but in the past much more widely attested, including Gascony, Languedoc, the Alps, and large areas of southern Italy.10 It has long been claimed that in many late Latin texts both ILLE and IPSE, the latter especially in areas of the southern Romània (Vincent 1997c: 154), frequently occur in contexts in which their spatial deictic function is considerably weakened,11 and their principal 10

Aebischer (1948: 193), Ravier (1991: 89), Rohlfs (1968: 112), Naudeau (1979). Väänänen (1987), Renzi (1985: 144–7), Nocentini (1990), Vincent (1997c; 1998a). Recall that in many Romance varieties reflexes of IPSE came to mark addressee-related deixis (e.g. Sp./Pt. ese/êsse, SID ECCU+IPSU > chisso ‘this/that (near you)’; see Ledgeway 2004b). However, Pinkster (1996: 250), together with Rosén (1994: 131) and Bauer (2008: 46), claim that the origins of the article are (in part) to be sought precisely in 11

90

From Latin to Romance

role appears to be one of marking nothing more than anaphoricity and/or definiteness,12 a precursor to the modern article which Aebischer (1948) famously termed an ‘articloid’. Traditionally, then, the principal question has been whether the latter is a demonstrative with a much-increased frequency (Herman 2000: 84–5) or indeed an article, but with a still limited range of use (Lyons 1999: 333). Clearly, there are a priori elements of truth in both positions, which should not be seen as mutually exclusive solutions to the question, but simply as the start and the end points in an unresolved and ongoing process of grammaticalization. In this respect the evidence of the late fourth-century Peregrinatio Aetheriae in which IPSE and, to a lesser extent, ILLE figure heavily, has been much debated, although there is still no general agreement about their interpretation in the text. In terms of their distribution, Renzi (1976), Selig (1992; cf. however Pinkster 2006), Christol (1994), and Zamboni (2000: 116) argue that IPSE was predominantly used anaphorically in conjunction with second-mention items, and hence largely equivalent to ‘the aforementioned’, whereas ILLE could be used both anaphorically with second-mention items, as well as cataphorically with first-mention items including, for example, restrictive relatives (13). To this picture we can add, following Vincent (1997c), that IPSE, unlike ILLE, performed a topic-marking function (cf. also Fruyt 2003: 102), only picking out informationally prominent second-mention items (and hence an unsuitable candidate for the object clitic paradigm). (13) montes illi inter quos ibamus, aperiebant et and mounts.NOM those.NOM among which.ACC we.went opened faciebant uallem infinitam [ . . . ] Uallis autem ipsa ingens made valley.ACC endless.ACC valley.NOM but self.NOM huge.NOM est ualde (Lat., Per. Aeth. 1.1-2.1) is truly ‘th(os)e mountains, through which we were journeying, opened out and formed an endless valley [ . . . ] The (= aforementioned) valley is indeed truly huge’ the strong deictic function of ILLE in appositions like senex ille Caecilianus (Cic., S.Rosc. 46) ‘that old man in Caecilianus (lit. old that Caecilianus)’. 12 A particularly strong thesis along these lines is developed in Rosén (1994), who argues, principally on the basis of Augustine’s Confessiones, that the origins of the article function of ILLE are to be sought in Latin sentential word order patterns (cf. also Pieroni 2010: 473–5). Starting from the observation that the article first emerges with subject NPs in Romance, only spreading subsequently to object NPs and finally PPs (cf. Parry and Lombardi 2007; Bauer 2009b: 96), she claims that Latin preverbal subjects are prototypically topical and hence interpreted as definite in that position, whereas postverbal subjects are typically interpreted as rhematic and hence as indefinite in that position. Consequently, when word order is unable to encode marked definiteness patterns, as arises in the case of postverbal rhematic subjects with definite interpretation, she maintains that ILLE is pressed into service to mark the subject’s definiteness (viz. . . . V + ILLE + S), giving rise to an article-like distribution that would in time generalize. However, J. Adams (p.c.) points out that this presumed distributional pattern is simply not borne out by an examination of the Confessiones.

The rise of functional structure

91

However, Adams (forthc. b) has undertaken a critical review of such claims with a detailed investigation of the late Latin evidence, where he finds that: (i) genuine article-like instances of ILLE (though not IPSE) only occur, and then only occasionally, from the sixth century (14a); (ii) claims about the articloid status of IPSE in late Latin (14b), principally based on its presumed innovative anaphoric function (despite being widely attested in classical usage; Adams }4.1), are demonstrated to be completely unfounded (cf. also Fruyt 2003: 102). (14) a nam si in feruenti aqua missa fuerint, albumen for if in boiling.ABL water.ABL placed.NPL will.have.been egg.white.NOM coacolat et mediolum illut tarde sentit (Lat., Anthimus 17.5) coagulates and yolk.NOM that.NOM slowly feels ‘If they [the eggs] are put in boiling water, the white goes hard and the yolk scarcely feels the heat’ b intrat episcopus intra spelunca [ . . . ] et accipit codicem and he.receives code.ACC enters bishop.NOM inside cave.ABL euangelii, et stans ipse episcopus leget and standing.NOM self.NOM bishop.NOM reads gospel.GEN uerba Domini (Lat., Per. Aeth. 33.2) words.ACC Lord.GEN ‘the bishop enters the cave [ . . . ] and after receiving the book of the Gospel, he stands and himself reads the words of the Lord’ Perceptively, Adams (}6) notes that in rare examples like (14a) the demonstrative force of illut is much attenuated in that it introduces a noun, mediolum, which is itself contrastively focused with the preceding albumen. According to Adams, only in such examples where ILLE cannot be emphatic can it begin to lose its demonstrative function and approximate to what we might term an article-like function. Examples such as (14b), where a given referent, in this case episcopus, first enters the discourse in its bare form to return some lines later introduced by ipse, have traditionally been invoked as evidence for the anaphoric value of IPSE in its incipient article function (cf. Christol 1994: 149). Nonetheless, Adams’s close reading of the context (see also Fruyt 2003: 102) reveals that in this and similar examples IPSE simply assumes its classical emphatic reading, namely ‘the bishop himself’, singling out the ‘bishop’ from the other potential participants in such readings (presbyter ‘priest’, populus ‘people’). Among other failings, Adams also underlines how, at the cost of disregarding all other evidence, previous scholars have tended to limit their attention to the Peregrinatio, a text in which they have, moreover, simply failed to recognize the demonstrative functions, both classical and late, of ILLE and IPSE:

92

From Latin to Romance

(15) a ergo quarta peruenimus in summitatem illam montis Dei thus fourth.ABL we.arrived in summit.ACC that.ACC mount.GEN God.GEN sancti Syna, ubi data est lex (Lat., Per. Aeth. 3.2) Sinai where given is law.NOM saint.GEN ‘we arrived at the fourth hour, at that summit of Sinai, the holy mountain of God, where the law was given’ b cepimus iam et descendere ab ipsa summitate from very.ABL summit.ABL we.began already and descend.INF montis Dei, in qua ascenderamus (Lat., Per. Aeth. 4.1) mount.GEN God.GEN in which.ABL we.had.ascended ‘we began our descent from the very summit of the mount of God which we had ascended’ Superficially, (15a–b) appear very similar in that they both contain the nominal summitas ‘summit’ modified by the dependent genitive montis Dei ‘mount of God’ and a relative clause ubi/in qua . . . ‘where/in which . . . ’. Evidence like this has often led to the rash conclusion that ILLE and IPSE in late Latin should be considered as equivalent strategies in an emerging article function (cf. Aebischer 1948: 203; Christol 1994: 144). Yet, in both cases ILLE and IPSE are anything but equivalent and simply assume their classical functions: in (15a) the relative clause dependent on illam alludes to a well-known biblical passage assumed to be familiar to the reader, hence the implied demonstrative reading ‘that well-known summit’, a reading absent in (15b) where the relative clause simply refers to the earlier ascent and ipsa assumes its classical emphatic function ‘from the very summit’ (cf. also Trager 1932: 38). Indeed, Fruyt (2003: 103) argues that this latter intensifying value (‘précisément’ = ‘the very . . . ’) essentially underlies all uses of IPSE in the Peregrinatio. Consequently, we must reject, for instance, Nocentini’s (1990: 146) and Christol’s (1994) claims about a sharp functional distinction in the Peregrinatio between the cataphoric and anaphoric uses of ILLE and IPSE, respectively (cf. also Selig 1992: 165–70), especially since, as Adams (}7) highlights, they are also found in anaphoric and cataphoric functions within the same text. Equally misleading is Renzi’s (1976: 31–2) claim that a characteristic sign of the grammaticalization of IPSE as an article is to be found in its ‘new’ associative anaphoric function in sixth-century Latin (Lyons 1999: 3–4), where it is employed to introduce items not explicitly mentioned in the previous discourse. Yet Adams (}4.2) shows that such a usage is not a late innovation at all, but is robustly attested as early as Plautus. Moreover, in both of these cases the claimed uses of ILLE and IPSE, even if true, would not support their presumed articlelike status, since these same uses are also found with demonstratives and are certainly not defining features of the category of article. More generally, Adams points out (}4.1.1) that in late Latin IPSE became a synonym of IDEM ‘the same’, a development which is massively discussed and documented but disregarded in the (Romance)

The rise of functional structure

93

literature on the article, and it is this demonstrative meaning, together with its original contrastive reinforcing value ‘-self, very’ (cf. 14b, 15b), which has traditionally been overlooked in texts like the Peregrinatio: (16)

nam duxit nos statim ad ecclesiam, quae est at.once to church.ACC which.NOM is for he.led us.ACC foras ciuitatem in eo loco, ubi fuit outside city.ACC in that.ABL place.ABL where was domus sancti Abrahae, id est in ipsis house.NOM holy.GEN Abraham.GEN that is in same.ABL.PL fundamentis et de ipso lapide (Lat., Per. Aeth. 20.3) of same.ABL stone.ABL foundations.ABL and ‘He took us at once to the church, which is outside the city on the spot where stood the house of holy Abraham; it stands on the same foundations, and it is made of the same stone’

Finally, a further significant problem with previous analyses of late Latin IPSE underlined by Adams’s study (}4.1.1) is their failure to take proper account of the stylistic traditions of the genres in which the presumed article-like uses of IPSE frequently occur. Illustrative in this respect are the following extracts from the first four paragraphs of the Peregrinatio: (17)

montes illi, inter quos ibamus, aperiebant et mounts.NOM those through which.ACC we.went opened and infinitam [ . . . ] et trans uallem[2] faciebant uallem[1] made valley.ACC infinite.ACC and across valley.ACC apparebat mons sanctus Dei Sina [ . . . ] Habebat appeared mount.NOM holy.NOM God.GEN Sinai it.had autem de eo loco ad montem Dei forsitan but of that.ABL place.ABL to mount.ACC God.GEN perhaps illa, quam quattuor milia totum per ualle[3] four miles in.total through valley.ABL that.ABL which.ACC dixi ingens [ . . . ] Uallis[4] autem ipsa ingens I.said huge.NOM valley.NOM but self.NOM huge.NOM est ualde [ . . . ] Ipsam ergo uallem[5] nos trauersare valley.ACC we cross.INF is truly self.ACC so habebamus, ut possimus montem ingredi. [ . . . ] Haec est this.NOM is we.had that we.can mount.ACC enter.INF autem uallis[6], in qua factus est uitulus [ . . . ] Haec ergo but valley.NOM in which.ABL made is calf.NOM this.NOM so uallis[7] ipsa est, in cuius capite ille self.NOM is in whose head.ABL that.NOM valley.NOM

94

From Latin to Romance

locus est, ubi sanctus Moyses [ . . . ] quoniam nobis is where holy.NOM Moses.NOM because to.us place.NOM ita erat iter, ut [ . . . ] ad illud caput uallis[8] thus was route.NOM that to that.ACC head.ACC valley.GEN descenderemus [ . . . ] itaque ergo hoc placuit, ut [ . . . ] inde we.descended so thus this pleased that thence totum per mediam uallem[9] ipsam [ . . . ] rediremus we.returned in.total through middle.ACC valley.ACC self.ACC ad iter cum hominibus Dei, qui to route.ACC with men.ABL God.GEN who.NOM.PL nobis singula loca, quae scripta sunt, per to.us each.NPL place.NPL which.NOM.NPL written are through sic fuit, ut per ipsam uallem[10] ostendebant [ . . . ] iter route.NOM thus was that through self.ACC valley.ACC showed.3PL medium transuersaremus caput ipsius uallis[11] (Lat., Per. Aeth. 1.1–4.0) head.ACC self.GEN valley.GEN middle.ACC we.crossed ‘those mountains, through which we were journeying, opened out and formed an infinitely great valley[1] [ . . . ] and across the valley[2] appeared Sinai, the holy mountain of God. [ . . . ] The whole distance from that place to the mount of God was about four miles across the aforesaid great valley[3]. For the valley[4] is indeed very great [ . . . ] We had, therefore, to cross that same valley[5] in order to reach the mountain. [ . . . ] This moreover is the valley[6] in which that calf was sacrificed [ . . . ] This also is the very valley[7] at the head of which is the place where holy Moses [ . . . ] And as our route was [ . . . ] to descend to the head of the valley[8] [ . . . ] so we determined [ . . . ] to return on our journey throughout the whole length of the valley[9], together with the men of God, who showed us each place in the valley[10] which is mentioned in the Scriptures. [ . . . ] our route was to cross the middle of the head of that valley[11]’ In this passage the nominal UALLIS ‘valley’ occurs eleven times. In the first two instances, where the valley is being introduced, it occurs without modifiers and in its third occurrence, at which point it has now been established as the topic, it is modified by ILLE in its classical cataphoric demonstrative function anticipating the following relative clause (cf. Fruyt 2003: 109; Pieroni 2010: 471). In all but two of the following eight examples UALLIS appears to mechanically occur with IPSE. It is precisely passages like this in which IPSE recurs at short intervals with the same nominal that have given rise to the widespread, albeit erroneous, belief that IPSE had essentially

The rise of functional structure

95

become a pure anaphor (cf. Christol 1994: 152),13 already fulfilling the functions of an article in late texts such as the Peregrinatio (cf. though Vincent’s (1997c) claims about the topic-marking function of IPSE in such examples). Yet, as noted above, in some of the examples in (17) IPSE is certainly to be interpreted as a demonstrative in accordance with its classical emphatic reading (e.g. [7] = ‘the very valley’) or later acceptation as a synonym of IDEM ‘same’ (e.g. [5] = ‘that same valley’). In other cases in (17), however, the precise function of IPSE proves much more difficult to establish, with a distribution that appears somewhat mechanical and ultimately less semantically motivated. Nonetheless, as Fruyt (2003: 101), Sornicola (2009: 40–1), and Adams (}4.1.1) convincingly demonstrate, this does not imply that the recurrent use of IPSE here is indicative of an article-like value, but is simply a function of a technical specifying style (especially common in the language of law and technical disciplines) that can be traced back to early Latin, whereby nouns are redundantly specified by terms such as IS ‘this/that’, HIC ‘this’, ILLE ‘that’, IDEM ‘same’, IPSE ‘self-, very; same’, SUPRASCRIPTUS ‘aforementioned’, DICTUS ‘said’. Consequently, the recurrent use of IPSE in passages such as (17) should not be equated with that of an article, for it simply represents a feature of a redundant style of writing (termed copia dicendi) much imitated by Egeria, who also frequently repeats, among other things, antecedents within relative clauses: (18) duxit nos episcopus ad puteum illum, ubi to well.ACC that.ACC where led us bishop.NOM adaquauerat sanctus Iacob pecora [ . . . ] qui puteus had.watered holy.NOM Jacob.NOM flocks.ACC which.NOM well.NOM sexto miliario est a Charris; in cuius putei six.ABL miles.ABL ist from Charrae.ABL in whose well.GEN honorem fabricata est ibi iuxta sancta ecclesia [ . . . ] honour.ACC built is there close.by holy.NOM church.NOM Ad quem puteum cum uenissemus [ . . . ] (Lat., Per. Aeth. 21.1) to which.ACC well.ACC when we.had.come ‘the bishop took us to the well where holy Jacob had watered holy Rachel’s flocks [ . . . ] the well which is six miles from Charrae, and the well in the honour of which a very great and beautiful holy church has been built close by [ . . . ] To which well, when we reached it [ . . . ].’ 13 The widely proclaimed pure anaphoric function of IPSE in the Peregrinatio can be immediately dismissed, as noted by Fruyt (2003: 103), by examples such as (i) where it is placed together with a clearly anaphoric ILLE, highlighting the contrastive reading of IPSE:

ille medianus, in quo descendit maiestas Dei, tanto (i) ipse self.NOM that.NOM middle.NOM in which.ABL descended majesty God.GEN so.much altior est omnibus illis (Lat., Per. Aeth. 2.6) higher.NOM is all.ABL those.ABL ‘that central one [= mountain] itself, on which the Glory of God came down, is so much higher than them all’

96

From Latin to Romance

In short, Adams (}8) concludes that ‘the Latin evidence does not have much to tell us about the emergence of the definite article’, inasmuch as ‘[t]here is no sign of the obligatory use of ille or ipse in our texts’ (cf. also Trager 1932: 49; Calboli 2009: 118). Rather, we have to wait until at least the eighth century (e.g. The Rule of Chrodegang) to find a much-weakened, quasi-obligatory use of ILLE (cf. also Muller 1929: 84; Calboli 2009: 118–19),14 although Löfstedt (1956 II: 373–4) and Banniard (1995: 318) maintain that the frequent use of ILLE even in this and similar late texts is nothing more than a weakened demonstrative, and in any case certainly not an articloid or article (cf. also Herman 2000: 84–5). In a similar vein, Sornicola (2008; 2009: 40–4) fails to find any value associated with that of the article in the early medieval use of IPSE. In terms of the approach to the Latin–Romance transition developed in this and the previous chapter, this finding is fully expected, since we have already seen (}4.2.1) that the indefinite article is a Romance innovation with no recognized forerunners in the Latin of any period (pace De la Villa 2010: 226–33). Now, if on one level the passage from Latin to Romance can be characterized in terms of the emergence of functional categories including D(eterminers), then the emergence of articles, whether definite or indefinite, might be expected to proceed in parallel, such that the uncontentious recognition of the absence of an indefinite article, and hence a D(eterminer) position, in Latin implies the corresponding lack of a definite article in the same period. 4.2.2.2 Romance In view of the paucity of compelling evidence in late Latin, Adams’s conclusion regarding the probable late emergence of the definite article accords well with the earliest Romance evidence, where the use of the article proves anything but systematic. For example, The Strasbourg Oaths, the earliest extant Romance text (842) from modern-day France, does not contain any examples of the article (or the related category of object clitics), a situation replicated for the earliest texts of other areas of the Romània including the Indovinello veronese (early ninth century) and the Placiti cassinesi (960–3) from Italy, the Tomida femina (ninth/tenth century) from Occitania,

14 Varvaro (forthc.: }7) argues that the apparent late emergence of the article (and other typical Romance features) is simply an artefact of the Latin written tradition which deliberately suppressed such non-standard features. Starting from the observation that Romance varieties show a series of common features documented since the earliest written records, but which fail to surface in earlier Latin texts, Varvaro maintains that from the beginning of the late Empire there must have been a linguistic level even below that of substandard Latin which he terms ‘sub-substandard Latin’. These shared Romance features do not therefore represent a quite implausible series of common, albeit fortuitous, innovations in the individual Romance languages or the rapid dissemination of local innovations across the Romània under language contact in the course of the ninth–tenth centuries, but must have first emerged in this subsubstandard, though never coming to the surface in the written language until the advent of the written Romance vernaculars.

The rise of functional structure

97

and the Jurament feudal (1028–47) from Catalonia.15 Where, however, the definite article does occur in early Romance, it displays considerable attenuation of its original deictic force, in that reflexes of ILLE and IPSE in their article function no longer situate a referent negatively with regard to the deictic sphere of the speech act participants (ILLE) or positively with regard to the deictic sphere of the addressee(s) (IPSE), but increasingly come to mark shared cognition between speaker(s) and addressee(s). Nonetheless, the article still retained considerable identifying force, as witnessed by the fact that in early texts it is generally excluded with unique, abstract, and generic referents which, by definition, cannot be singled out:16 (19) a Paien unt tort e chrestïens unt pagens have wrong and christians have ‘Pagans are wrong and Christians are right’

dreit (OFr., CDR 1015) right

b leichatz estar ypocresie. Pocresie es tant a dire let be.INF hypocrisy hypocrisy is so.much to say.INF cum fengir de Diu amar (OGsc., DC 1v.32–3) as feign.INF of God love.INF ‘Leave hypocrisy well alone. Hypocrisy is paramount to pretending to Love God’ c giustizia mosse il mio justice moved the my ‘justice moved my lofty maker’

alto high

fattore (OTsc., Inferno 3.4) maker

d De qui crebanta camino del rey (OSp., LFN 210) of who rob road of.the king ‘Regarding those who commit assaults on the highway of the king’ As with the indefinite article, in a number of cases this early usage has been fossilized in the modern languages in proverbs and certain set expressions (Renzi 2010: 341–4), including Ast. (ta) en coche ‘(he is) in (the) car’, (ponlo) en suelu ‘(put it) on (the) floor’; Cat. carrer amunt/avall ‘up/down (the) street’, nedar dins mar ‘to swim in (the) sea’, parar/desparar taula ‘to lay/clear (the) table’; Fr. noblesse oblige, pauvreté n’est pas vice ‘poverty is not a vice’, blanc comme neige ‘white as (the) snow’, fermer boutique ‘to shut up (the) shop’, par terre ‘on (the) floor’; It. in nome di ‘in (the) name of’, in giardino ‘in (the) garden’, chiudere casa ‘to close up (the) house’; Occ. céu e terro ‘heaven and earth’, en carriero ‘in (the) street’; Pt. ter direito a ‘to have (the) right to’, traduzir de francês em português ‘to translate from (the) French into (the) Portuguese’, a pedido de ‘at (the) request of’, em proveito de ‘to (the) advantage of’; Varvaro (forthc.: }7) observes that the absence of the article in some of the earliest Romance texts can be attributed to their juridical nature, where the use of the article was felt to be too vulgar and hence avoided (cf. also Sornicola 2009: 34). He also provides a useful survey of a number of the earliest sporadic cases of the article in predominantly Latin and Romance texts across the Romània. 16 Banniard (1995: 373), Parry and Lombardi (2007: 83–4), Renzi (2010: 318–19, 329–30, 332–7). 15

From Latin to Romance

98

Sp. ausencias causan olvido ‘long absent, soon forgotten (lit. absences cause oblivion)’, en camino ‘on (the) road’, en dicho mes ‘in (the) said month’. In the modern languages, by contrast, shared cognition between speaker(s) and addressee(s) has come to assume increasing importance in the selection of the article, such that if a referent can be considered to form part of the interlocutors’ common universe of experience, then the article is employed.17 Thus, the article is now generally required with unique, abstract, and generic referents (20a–d), as well as with inalienable possessa (21a–c): (20) a la vida i the life and ‘life and death’

mort (Cat.) death

la the

b els gats i les rates the cats and the rats ‘cats and rats are animals’ c îmi plac me.DAT= please.3PL ‘I like flowers’

són are

animals (Cat.) animals

florile (Ro.) flowers=the

d dreptatea este lumina justice=the is light=the ‘justice is the light of life’ (21) a lo pel se me the hair self= me= ‘my hair stood on end’

quilhava raised

b je m’ étais cassé I me= was broken ‘I had broken my leg’ c Ida ha perso il Ida has lost the ‘Ida has lost her purse’

viet¸ii (Ro.) life=the.GEN

la the

sul on.the

cap (Lgd.) head

jambe (Fr.) leg

portafogli (It.) purse

17

An interesting counterexample is discussed by Marchetti (1974: 55), who reports that the article is not employed in Corsican before nouns indicating the inhabitants of localities and regions when definite (i.a), a property also displayed by seventeenth-century writers such as Paolo Sarpi (M. Maiden p.c.). Indeed, the use of the article in such contexts is reported to yield a partitive reading (i.b). (i)

a a

Balanini i cunnoscu (Cor.) Balanins them= I.know ‘I know the Balanins (= inhabitants of La Balagne) well’

MRK.ACC

vistu i Niulinchi (Cor.) b aghju I.have seen the Niolins ‘I’ve seen some Niolins (= inhabitants of Niolo)’

The rise of functional structure

99

In modern Romance, proper names, which are by definition intrinsically referential, often also take the article, including non-modified countries—though generally not in Spanish (e.g. España linda con Portugal y Francia ‘Spain shares borders with Portugal and France’) and, to varying degrees, in Catalan (El Regne Unit ‘the United Kingdom’ vs (la) Gran Bretanya ‘(the) Great Britain’; Wheeler, Yates, and Dols 1999: 57)—large islands, lakes, rivers,18 seas and oceans, deserts, and mountains (e.g. Pt. o Brasil (but Portugal without the article), a Madeira ‘(the island of) Madeira’, o Titicaca ‘lake Titicaca’, o Tejo ‘the Tagus’, o Báltico ‘the Baltic’, o Sahará ‘the Sahara’, os Andes ‘the Andes’), as well as, in some areas, first names (e.g. Cat. l’Artur, la Carme; CN.It. la Francesca (but with male names only in certain areas, e.g. Trn. It. il Francesco); Val. lou Jàque, la Mariòun (Ronjat 1913: 39; Arnaud 1920: 272); EPt. o Armando, a Marinha; Sal. ’u Francu, ’a Paola).19 Catalan, though not Valencian (cf. ha vingut (**la) Maria ‘(the) Maria has arrived’), has moved the furthest in this direction (Wheeler, Yates, and Dols 1999: 67–8), having developed a specialized paradigm (see Table 4.1) which, in the standard language, blends ILLE-derived forms (female names, vowel-initial male names) with a reflex of DOMINUS ‘master’ 18

The article is generally not employed with the names of rivers in Occitan (i.a–b; Ronjat 1913: 39; Arnaud 1920: 272), though counterexamples are not uncommon (i.c): (i)

a vau pescar à Tarn / à Lèira (Ldg.) at Loir I.go fish.INF at Tarn ‘I’m going fishing in (the) Tarn/(the) Loire’ b Avinhon, banhada pèr Ròse (Prv.) Avignon bathed by Rhone ‘Avignon, bathed by the Rhone’ que t’ escampèssi dins lo Ròse! (Lgd.) c meritariás you.would.deserve that you= I.threw in the Rhone ‘you deserve to be thrown into the Rhone’

19 Similar is the use of the article in vocative function (for an overview, see Ashdowne 2004) in French, Occitan, and Romanian; in the latter it occurs together with the vocative inflection -e in the masculine singular (i.c) and, at least in the standard language, assumes the genitive-dative case form in the plural (i.d):

(i)

a Salut, les gars! (Fr.) hello the guys ‘Hi there, guys!’ nos! (Lgd.) b Los amics, calem the friends silence.IMP.1PL =ourselves ‘Friends, let’s calm down’ c Ce glumă bună, profesorule! (Ro.) what joke good teacher=the.voc ‘What a good joke, sir!’ frat¸ilor şi surorilor! (Ro.) d Venit¸i, come.IMPL.PL brothers=the.GEN/DAT and sisters=the.GEN/DAT ‘Come, brothers and sisters!’

100

From Latin to Romance

> en (consonant-initial male names),20 whereas the colloquial language, especially in the north-western dialects, often extends ILLE forms to the whole masculine paradigm (Badia i Margarit 1962 I: 158; 1995: 446–7; Veny 1982: 36, 95). In Balearic varieties, by contrast, forms derived from DOMINUS/DOMINA ‘master/mistress’ are extended to the whole paradigm (Veny 1982: 67). TABLE 4.1 Forms of article with proper names in Catalan Standard Catalan

Substandard Catalan M

F

Balearic Catalan

M

F

M

F

+C

en Joan

la Joana

el Joan

la Joana

en Joan

na Joana

+V

l’Eduard

l’Alícia

l’Eduard

l’Alícia

n’Eduard

n’Alícia

4.2.2.2.1 ipse articles Catalan is also of interest in that the varieties spoken in the Balearics,21 as well as in some areas of the mainland (principally in and around Cadaqués, with some receding pockets to the south along the Costa Brava between the Ter and Tordera rivers),22 combine and productively contrast reflexes of both IPSE, the so-called article salat (PtP < salar ‘to use the IPSE-derived article’, see Perea 2005: 69–73; Busquet Isart 2010: 1–2), and ILLE (see Table 4.2).23 While both forms have definite reference, only the IPSE-derived forms have truly deictic force and are able to identify both anaphorically (22a) and cataphorically (22b) definite referents. Furthermore, in the Balearics the use of the article salat is still very productive and is regularly extended to neologisms (22c).

20 Parallels in the use of DOMINUS/DOMINA before first names are also found outside of Catalan, but do not have the status of articles, serving instead as honorifics variously used in conjunction with the aristocracy, clergy, and notable dignitaries including OOcc. en/na ‘Mr/Mrs’ (Grandgent 1909: 25); SID Donn’Antonio ‘father Antonio (priest)’, (d)onna Marcella ‘(Lady/Mistress) Marcella’; Sp. Doña Sofía ‘Her Royal Highness Queen Sofía (wife of King Juan of Spain)’, Don Carlos ‘(Lord) Carlos’; cf. also Fr. dom (e.g. Dom Pérignon) and dame ‘lady’ (e.g. Madame lit. ‘my.lady’). 21 Though not in the northern Majorcan costal locality of Pollença (Moll 1993: 40; 1997: 184; Veny 1982: 71; Forteza i Cortès 2008 I: 341), where only forms derived from ILLE are employed (e.g. eu/eus cap/-s ‘the head/-s’, l’/us homo/-s ‘the man/men’, la/les nina/-es ‘the girl/s’, amb los meus ulls ‘with (the) my eyes’). 22 For an in-depth analysis of its modern distribution along the Costa Brava, which has rapidly been receding since the 1950s, see Busquet Isart (2010) who identifies eight localities in whose speech it is still preserved to varying degrees, namely Cadaqués, Tossa de Mar, Lloret de Mar, Blans, Begur, Calella de Palafrugell, Regencós, and Sant Joan de Palamós. 23 Alcover (1905), Badia i Margarit (1962 I: 156; 1991: 141–2; 1995: 444–6), Villangómez i Llobet (1978: 65–7), Wheeler (1988a: 181), Moll (1993: 40–1, 69–71; 1997: 182–4; 2006: 182–3), Perea (2005), Veny (1982: 37–8, 61–3), Wheeler, Yates, and Dols (1999: 45–6), Bernat i Baltrons (2007: 109–10), Forteza i Cortès (2008 I: 333–41), Busquet Isart (2010).

The rise of functional structure

101

TABLE 4.2 Catalan articles derived from IPSE24 Costa Brava MSG

MPL

FSG

Balearic Islands FPL

MSG

MPL

FSG

FPL

+C

es

es

sa

ses

es

es

sa

ses

+V

s’

ses

s’

ses

s’

e(t)s

s’

ses

so

sos

sa

ses

amb ‘with’+

(22) a la fan pujar damunt her= they.make mount.INF on ‘they make her get on the mule’

sa the

mula (Maj.) mule

b tu éts sa dona que jo vaig tirar anita passada per you are the lady that I PST.1SG throw.INF this.night past for aquesta finestra? (Maj.) this window ‘Are you the lady that I threw out of this window last night?’ c s’ MP3, sa TDT (Maj.) the MP3 the TDT (= Televisión Digital Terrestre) ‘the MP3 player’, ‘the digital TV freeview box’ However, with the exception of Cadaqués, on the mainland the use of the article salat now appears in many varieties to be much more lexically restricted (Busquet Isart 2010: 95–9), typically limited to nouns belonging to such areas as the family (23a), seafaring (23b), and everyday life (23c), at times giving rise to idiosyncratic differences in usage among neighbouring areas (23d): (23) a sa meua dona (Beg.), es the my lady the ‘my wife’, ‘the kids’, ‘the parents’

nois (Tos.), kids

es the

pares (Tos.) fathers

b sa barca (Beg.), sa plaja (Beg.), es moll (Tos.), es pescadors (Tos.) ‘the boat’ ‘the beach’ ‘the peer’ ‘the fishermen’ c es pagesos (Plf.), s’ escriure (Plf.), sa porta (Tos.), es gats (Tos.) ‘the farmers’ ‘the writing’ ‘the door’ ‘the cats’

24 The morphological forms of the article salat present three major diatopic variations. First, the allomorphs given for Balearic varieties following the preposition amb ‘with’ are not used in Menorca and are often also replaced with es in Majorca and Ibiza (Busquet Isart 2010: 95). Second, in the Valencian dialects spoken around Tàrbena, and formerly in the Vall de Gallinera, the MPL form es is equally used before vowels. Third, in Ibiza and Formentera the MPL form is es, but is generally, though not invariably, realized ets in Majorca and Menorca before vowels (Villangómez i Llobet 1978: 65; Wheeler, Yates, and Dols 1999: 45; Forteza i Cortès 2008 I: 332 n. 103; Busquet Isart 2010: 94).

102

From Latin to Romance d la ‘the

roca/sa rock/the

popa (Beg.) vs stern’

sa ‘the

roca/la rock/the

popa (Tos.) stern’

The ILLE-derived forms, by contrast, are confined to marking unique, generic, and abstract referents,25 hence intrinsically referential and already fully identifiable (cf. Forteza i Cortès 2008 I: 338–41): Bal. el dimoni ‘the devil’, el Papa ‘the pope’, el Bon Jesús ‘the Christ child’, la Mare de Déu ‘the mother of God (= Virgin Mary)’, la Cúria ‘the Curia’, el cel ‘the sky; Heaven’, l’infern ‘Hell’, el purgatori ‘Purgatory’, la Seu ‘the Cathedral (of Palma)’ (cf. seu ‘seat’), la vera creu ‘the true cross’, el rosari ‘the rosary’, l’altar ‘the altar’, l’orgue ‘the church organ’, la mare superiora ‘the mother superior’ (cf. sa mare ‘(the) mother’), el rei ‘the king’, la reina ‘the queen’, el món ‘the world’, la mar ‘the sea’, la terra ‘the earth’, el nord/sud ‘(the) north/south’, l’est/oest ‘(the) east/west’, l’Europa ‘(the) Europe’, L’Havana ‘Havana’, la ciutat ‘the city’, l’amo ‘the owner, boss’, el jovent ‘(the) youth’, la gent ‘(the) people’, la pobresa ‘(the) poverty’, la vida ‘(the) life’, la mort ‘(the) death’, la justicia ‘(the) justice’, la música ‘(the) music’, la una, les dues . . . ‘(the) 1, 2 o’clock . . . ’, l’any/la setmana que ve ‘(the) next year/week’, l’endemà ‘the following day’, (a les 11) del dia/de la nit ‘(at 11 o’clock) during (lit. of the) the day/night’, el Mallorca ‘Mallorca FC’, la Rambla ‘La Rambla Avenue (in Palma)’ (but cf. es Born ‘the Passeig d’es Born (in Palma)’), l’Universitat ‘the University of Palma’, l’esquerra ‘the left’, la dreta ‘the right’; Beg. el cementiri ‘(the) cemetery’, l’extremunció ‘the last rites’, l’alcalde ‘the mayor’, la lluna ‘the moon’, la mar ‘the sea’, la nit ‘the night’, les bruixes ‘(the) witches’, rentar els plats ‘to wash the dishes’; Cad. la lluna ‘the moon’. In a number of contexts, it is therefore possible in such varieties to construct minimal pairs (especially by alternating singular with plural) based around this [deictic] contrast, including Maj. la Sala ‘the town hall’ vs sa sala ‘the large (public) room’, el Palau ‘the Royal Palace’ vs es Palau de S’Arròs ‘the Rice Palace (lit. the palace of the rice = name of restaurant in Palma)’, pensam en la mort ‘we’re thinking of death’ vs sa mort d’en Joan ‘Joan’s death (lit. the death of the Joan)’, l’Església ‘the (institution of the) Church’ vs s’església del poble ‘the village church’, la parròquia ‘the parish (of the interlocutors)’ vs sa parròquia de Sant Josep ‘the Parish of Saint Josep’, el cor ‘the (church) choir’ vs es cor ‘the heart’, el món ‘the world’ vs es meu món és la música ‘(the) my world is music’, el Bisbe ‘the Bishop (of Majorca)’ vs tots es bisbes de Mallorca ‘all the (past) bishops of Majorca’, el doctor/senyor/profesor + proper name ‘Doctor/Mister/Professor X’ vs es doctor/senyor/ profesor ‘the doctor/gentleman/professor’ (e.g. cal anar a buscar el doctor Bosch/es doctor ‘we must call out doctor Bosch/the (= a) doctor’), el pare Miquel ‘Father Miquel’ vs es pare d’en Miquel ‘Miquel’s father’; Ibz. dins la nit ‘in the night’ vs dins sa nit de Sant Joan

25 They have also been exceptionally retained in many set (adverbial) expressions such as Bal. com l’anell al dit ‘(to fit) like a glove (lit. like the ring at.the finger)’, blanc com la neu ‘white as (the) snow’, a les fosques ‘in the dark’, a la fresca ‘in the open’, a lo mariner ‘in the sailor fashion’. See further Forteza i Cortès (2008 I: 338, 340–1).

The rise of functional structure

103

‘during the night of the Saint Joan Festival’; Cad. la plaja ‘the beach’ vs ses plages ‘the beaches’; Trb. el port ‘the port’ vs es port d’Alacant ‘the port of Alicante’, el dia ‘the day’ vs es dia de Santa Bàrbara ‘Santa Barbara feast day’, la llei ‘the Law’ vs ses lleis ‘the laws’. As a general rule, the IPSE and ILLE articles can then be respectively characterized in terms of the featural specifications [+definite, +particularized, given] vs [+definite, particularized, +given]. It follows that the use of the ILLE-derived forms is comparable to the (redundant) use of the definite article with intrinsically referential entities such as countries in French or Italian (e.g. Fr. l’Islande ‘(the) Island’),26 and proper names in standard Catalan and Occitan (e.g. Gsc. l’òme de la Marineta ‘Marineta’s husband (lit. the man of the Marineta)’). In short, the ILLE-derived forms occur precisely in those contexts in which, we saw above, the definite article is typically absent in early Romance, namely in conjunction with unique, abstract, and generic referents which, by definition, cannot be singled out. The distinction between the two available articles is not, however, entirely robust in Balearic Catalan, as the following exceptions illustrate (e.g. Maj. es/**el sol ‘the sun’, sa/**la lluna ‘the moon’, es/**el Parlament ‘the (Balearic) Parliament’, es/**el Govern Balear ‘the Balearic Government’, l’amo ‘the owner’ (M) vs sa madona ‘the owner’ (F); Men. l’avi ‘the grandfather’ vs s’àvia ‘the grandmother’; Men./Maj. en el/es camp ‘in the country(side)’); neither is there complete agreement between the Balearics and the Catalan mainland (e.g. Maj./ Cad. la/sa setmana que ve ‘next week’, es/el sol ‘the sun’, sa/la lluna ‘the moon’, es/el capellà ‘the priest’, sa/la tardor ‘the autumn’, s’/l’hivern ‘the winter’, sa/la primavera ‘the spring’, s’/l’estiu ‘the summer’), the Balearics and Valencia (e.g. Maj./Trb. es/el port ‘the port’, s’/l’hivern ‘the winter’, sa/la nit ‘the night’), nor between Valencia and the Balearics/Catalonia (e.g. Trb./Bal. & Cad. a s’/l’esquerra ‘on the left’, a sa/la dreta ‘on the right’, s’/l’orgue ‘the organ’, sa/la parròquia ‘the parish’). Such data force us to admit a certain degree of idiosyncratic lexicalization in their synchronic distribution. Historically the IPSE-based forms were much more prevalent across the southern Romània (Aebischer 1948: 189–93), witness the presence of both articles in the following Old Spanish example: (24)

es dia es salido e la noch the day is exited and the night ‘the day has gone and the night has come’

entrada entered

es (OSp., Cid 1699) is

Also across most of the mainland Catalan-speaking territories the article salat proved relatively common until the fourteenth century (Rabella i Ribas 1998: 256–7; Moran 2007: 129), as still revealed today by its presence in numerous toponyms such as Sant Joan Despí (< d’es pí ‘of the pine’), Sant Martí Sarroca (< sa roca ‘the rock’), Sacosta 26 Indeed, in those cases when these Catalan varieties use the article before continents and countries, and cities, it is the ILLE-derived form, not the article salat, which is employed (e.g. Bal. l’Europa ‘(the) Europe’, l’Havana ‘Havana’; Busquet Isart 2010: 223).

From Latin to Romance

104

(< ‘the slope’), Sescloses (lit. ‘the mountain passes’).27 Exceptional are the Valencian dialects spoken around Tàrbena and the Vall de Gallinera where, following a wave of Majorcan immigration in the seventeenth century, the article salat still remains robust (e.g. as cavall ‘the horse’, s’home ‘the man’; Veny 1982: 119), although Busquet Isart (2010: ch. 8) has now established its recent extinction in the latter locality. IPSE forms of the article are also found within Italy. All indigenous Sardinian dialects, though interestingly not the Catalan dialect of Alguerés, make exclusive use of IPSE-derived forms, namely M/FSG su/sa (e.g. su frate/sa die ‘the brother/day’; both > s’ + V: s’annu/ora ‘the year/hour’) and M/FPL sos/sas (sos frates/sas dies ‘the brothers/days’),28 but unlike the Catalan dialects above, do not formally contrast it with ILLE-derived forms which are simply absent from these dialects. Although IPSE-derived forms were formerly quite widespread in southern Italy, they are now entirely extinct, despite the existence of some apparently related forms. For instance, in the western Abruzzese dialect of Scanno (province of L’Aquila), intervocalic -LL- regularly yields a voiced alveolar affricate before high vowels (e.g. *bɛllʊ/-i > biezzə ‘beautiful’ (MSG/PL)), but was retained, at least originally, before other vowels (e.g. *bɛlla/-aj > bèlla/-ə ‘beautiful’ (FSG/PL); Rohlfs 1966: 332). This phonological development thus produces a distinction between the masculine articles with an initial affricate and the feminine articles with the original lateral, namely *illʊ/-i > zu/zi (pòirə/duluri) ‘the (pear trees/pains)’ vs *illa/*illaj > la/lə (tèrrə) ‘the (land/s)’, which, despite appearances, should not be retraced to different etyma (namely IPSE and ILLE, respectively). This interpretation is confirmed by the neighbouring dialect of Pescasseroli (Rohlfs 1968: 112; Saltarelli 2007), where, unlike the masculine singular and plural count article with initial sibilant (25a), the masculine mass article (cf. }4.2.2.2.2) displays an initial lateral (25b) on a par with the feminine forms of the article (26a–b). (25) a sə/ʃə the.MSG/MPL ‘the dog/-s’

canə (Pes.) dog

b lə vinə wine the.MSG ‘the wine/bread’ (26) a la casa house the.FSG ‘the house/-s’

/ panə (Pes.) / bread / lə the.FPL

casə (Pes.) houses

b la ssogna (Pes.) the pork.fat ‘pork fat’ 27 28

See, for example, Alcover (1905), Coromines (1989–97 VI: 457–60), Varvaro (forthc.: }7). In Campidanese dialects, the plural form is is for both genders (e.g. is frates/dies).

The rise of functional structure

105

This distribution of the sibilant and lateral confirms their common derivation from ILLE, inasmuch as the change -LL- > [-s] only occurred before [-ʊ] and [-i], but was inhibited before other vowels including [-o] of the neuter *illok, the apparent etymon of the mass article (Merlo 1906–7; 1917: 92; Loporcaro and Paciaroni forthc.: }4.4).29 In other neighbouring dialects, the change from lateral to sibilant has generalized to all prevocalic positions, thus giving the appearance of a series of IPSE-derived articles for nouns of all genders and numbers, though the actual forms all derive once again from ILLE (e.g. Vil. sə gallə ‘the rooster’, sa gallina ‘the hen’, sə puórchə ‘the pig(s)’; Rohlfs 1968: 112).

4.2.2.2.2 neuter articles The count vs mass distinction just considered for Pescasserolese is not just limited to this particular dialect, but is found in various guises in the determiner systems of various Ibero- and Italo-Romance varieties, where it is traditionally termed the neuter article (Hall 1968; Messing 1972). In the latter, it is widely found in a central-southern area extending from southern Umbria and the Marches to an area including the provinces of Naples, Matera, and Bari.30 In dialects of this area a lexically specified subclass of non-count masculine nouns, many of which were originally neuters in Latin (e.g. UINUM ‘wine’, MEL ‘honey’, LAC ‘milk’, FERRUM ‘iron’), occur with a distinct form of the article (and the demonstratives) originally derived from the neuter form *illok (and demonstratives *(ekku+)istok ‘this’, *(ekku+)ipsok ‘this/that’, *ekku+illok ‘that’); given the non-count reading of many of these original neuters, *illok-marking was eventually reanalysed as a non-count distinction and subsequently extended to many masculine mass nouns (including CASEUM ‘cheese’, PANEM ‘bread’, PISCEM ‘fish’). Formally, the distinction between reflexes of count illʊ and non-count *illok variously surfaces in the quality of the final vowel (27a), the lateral (27b)—or both (27c)—the absence/presence of initial-consonant lengthening, historically a case of sandhi assimilation of final -k (27d)—or even all three (27e): (27) a u the b ju the c ru the d ’o the

pratu / jóvitu / lópe vs o latte / pépe / granu (Nem.) meadow elbow wolf the milk pepper / corn cane dog

/ jalə rooster

chiuovə nail

vs

vs

lu the lə the

pane / fierro vs loaf clothes-iron

mèlə honey casə cheese

/ lardə pork.fat / pépə pepper

/ panə (Sfl.) bread / lattə (Gal.) milk

’o ppane / ffierro /ssale (Nap.) the bread iron salt

29 Cf. other similar alternations in the differential treatment of the original long lateral in the count and mass articles of central and southern Italian dialects (e.g. Nrs. ru cane ‘the dog’ vs lo mèle ‘the honey’; Rohlfs 1968: 108–10), apparently caused by the distinction between final [-ʊ] and [-o] in *illʊ/*illok (MSG/ NSG); cf. }4.2.2.2.2. 30 Rohlfs (1968: 108–10), Avolio (1996), Formentin (1998: 304–5), De Blasi (2006: 34–41), Ledgeway (2009a: 150–4), Loporcaro (2010), Loporcaro and Paciaroni (forthc.: }4.2); cf. also chapter 6, footnote 21.

106

From Latin to Romance e lu the

cuanə dog

vs

ppwanə (Avi.) bread

rə the

In these dialects, the neuter is a productive category and is regularly extended to all substantivized parts of speech such as adjectives (28a) and infinitives (28b), as well as neologisms (28c): (28) a ’o ttriste / nnuovo / nniro / cchiaro the sad new black clear ‘the sadness/newness/blackness/clarity/evilness’ b ’o mmagnà / ffummà / gghjastemmà the eat.INF smoke.INF blaspheme.INF ‘eating/smoking/blaspheming/stealing’ c ’o ‘the

rrap rap-music

/ rrock rock-music

/ bblues blues

/ mmale bad

(Nap.)

/ ffricà (Nap.) steal.INF

/ ffolk folk-music

/ ggasse (Nap.) gas’

Turning now to Ibero-Romance (Penny 2000: 102–3; ALlA 2001: 101–2), only in Asturian (29a) and Spanish (29b) is the neuter article formally distinct from the masculine singular count article (namely *illok > lo vs *illʊ > el), whereas in Catalan (29c) and Portuguese (29d) they are homophonous (namely Cat. el, Pt. o), except in the three grades of the demonstrative where Portuguese marks a formal distinction by way of the final vowel and the concomitant absence of metaphony (e.g. Pt. isto/ isso/aquilo (neuter) vs êste/êsse/aquêle (MSG count);31 cf. also Ast./Sp. esto/eso/aquello (neuter) vs ésti/éste, ési/ése, aquél (MSG count)): (29) a lo guapu d’ esti asuntu the pretty of this matter ‘the interesting thing about it all is . . . ’

ye . . . (Ast.)32 is

31 It is suprising to note, given the evidence of the other Ibero-Romance varieties, how final -o in Portuguese produces metaphony in the demonstrative (I thank M. Maiden for pointing this out to me). 32 In Asturian the neuter is further marked on masculine mass nouns through final non-metaphonizing inflectional (*illo(k)) > -o (i.a), an agreement extended to all adjectives modifying any noun with mass interpretation (i.b), including those of feminine gender (i.c), which formally contrasts with inflectional -u and -a (M/FSG) on adjectives modifying singular count nouns (Alonso 1958; Hall 1968; Penny 1970; Messing 1972; Posner 1996: 60; Arias Cabal 1999; ALlA 2001: 89–90, 222–7):

(i)

a pelo hair.MASS

vs

pilu, hair.CNT

queso cheese.MASS

vs

quisu (Ast.) cheese.CNT

‘(some) hair’ vs ‘(a single) hair’, ‘(some) cheese’ vs ‘(a single) cheese’ el arroz vs ta nigru el vasu (Ast.) b ta negro is black.M.CNT.SG the.MSG glass.MSG is black.MASS.SG the.MSG rice.MSG ‘the rice is black’ vs ‘the glass is black’ leche frío vs la casa fría (Ast.) c la the.FSG house.FSG cold.F.CNT.SG the.FSG milk.FSG cold.MASS.SG ‘the cold milk’ vs ‘the cold house’

The rise of functional structure

107

b siempre quiere lo imposible (Sp.) always he.wants the impossible ‘he always wants what he can’t have’ c el curiós és que . . . (Cat.) the curious is that ‘the strange thing is that . . . ’ d o ridículo da situação é que . . . (Pt.) the ridiculous of.the situation is that ‘the ridiculous thing about the present situation is . . . ’ 4.2.2.3 Conclusion To sum up, the D(eterminer)P(hrase) can be seen to provide the relevant (in)definiteness marking of its associated NP,33 as illustrated in the simplified representation of the Spanish example in (30a), and in some languages such as French (30b), where original final inflections for number and gender on nouns and adjectives have been drastically eroded, the accompanying determiner is not simply a spell-out of (in)definiteness, but is now also a quasi obligatory element of the nominal group as the sole exponent, in most cases, of number and gender:34 (30)

33

a

DP 3 D NP el/un 3 the/a N AP coche 5 car nuevo new

b

DP 3 D NP ! 6 le/la/les fiancé(e)(s) [f j nse] the.MSG/FS G/PL fiancé(e)(s)

An apparent exception to this generalization is the behaviour of prepositional phrases, which in certain fossilized contexts and expressions take a syntactically indefinite NP complement, although assuming definite reference (e.g. Cat. fora de casa ‘away from (the) home’; Fr. en Italie ‘in Italy’, par terre ‘on (the) floor’; It. in montagna ‘in (the) mountains’, da capo ‘from (the) beginning’; Pt. traduzir de português para italiano ‘to translate from (the) Portuguese into (the) Italian’; Sp. en parlamento ‘in (the) parliament’, en alta mar ‘on (the) high sea’). Although this usage has a particularly non-productive feel to it in most Romance varieties, presumably reflecting the original tendency for the article to surface predominantly in subject NPs but not in object NPs or PPs (Vincent 1997c: 162; Lyons 1999: 335; Parry and Lombardi 2007: 93), this archaic pattern is well preserved in Romanian, where the general rule (except for cu ‘with’) is that unmodified definite NPs occur without the article when governed by a preposition (e.g. după casă ‘behind (the) house’, sub pat ‘under (the) bed’, but după casa noastră ‘behind (the) our house’, sub patul din dormitor ‘under the bed in the bedroom’). In all such cases, it seems plausible to assume that the prepositional head exceptionally selects for an NP, and not a DP, complement, despite the definite interpretation of the latter. 34 For reasons of space, in the following tree representations we omit all intermediate X' categories, adopting a simple bare phrase structure representation (cf. Chomsky 1995).

108

From Latin to Romance

Even Romanian, which displays a suffixal definite article (31a–b), but a prenominal indefinite article (31c), readily fits into this same pattern:35 (31) a an + year

-ul/-ului (> anul/anului) =the.MSG.NOM-ACC/DAT-GEN

b casă + -a/-ei (> casa/casei), house =the.FSG.NOM-ACC/DAT-GEN c un a

an, year

ani + years case + houses

-i/-lor (> anii/anilor) (Ro.) =THE.MPL.NOM-ACC/DAT-GEN -le/-lor (> casele/caselor) (Ro.) =the.FPL.NOM-ACC/DAT-GEN

casă (Ro.) house

o a

More specifically, we can assume that, as a suffixal element, the Romanian definite article must incorporate with an appropriate head to form a well-formed word,36 be that a nominal head (vinul bun ‘wine=the good’; see 32a) or an adjectival head (bunul vin ‘good=the wine’; see 32b) raised from within the NP to left-adjoin to the D(eterminer) position: (32)

a

b

DP D N vin

DP

NP D N -ul vin

D A bun

A bun

NP D -ul

N vin

A bun

This same type of analysis is also frequently proposed for other Romance varieties (Longobardi 1994; 2001) to explain the intrinsic referential interpretation of nominals such as proper names whose definiteness is argued to be licensed, in some cases, through overt N(oun)-raising to a lexically null D(eterminer) position. In this way, we can straightforwardly explain minimal Romance-internal contrasts likes those in (33)–(34): (33) a [DP la the

[NP France]] (Fr.) France

b [DP Francia France

[NP Francia]] (Sp.)

35 With proper names other than female names in final -a, and in the colloquial language even with the latter, ILLE surfaces in proclitic position in the dative-genitive (on the demonstrative and possessive articles, see }}4.3.1.1–2):

lui Ion / Ana (Ro.) (i) pianul piano=the of.the= Ion Ana ‘Ion’s/Ana’s piano’ 36 See Grosu (1988; 1994), Cornilescu (1992), Giusti (1993; 1997: 102–6), Alboiu and Motapanyane (2000b: 3–4, 8). For an alternative analysis in which the article is treated as part of the noun’s inflectional structure that triggers raising of the [N=ART] complex to the D(eterminer) position, see DimitrovaVulchanova and Giusti (1998), Giusti (2002: 58–70).

The rise of functional structure (34) a [DP la/na the

109

[NP Joana]] (Bar./Bal.) Joana

b [DP Joana [NP Joana]] (Val.) Joana The definiteness reading of the proper noun in the (a) examples is licensed through lexicalization of the D(eterminer) position with a definite article, whereas in the varieties illustrated in the (b) examples no such article is available and the appropriate definiteness reading can only be achieved by overt raising of the nominal to that same position. That the nominal has indeed raised to the D(eterminer) position in such examples can be shown by contrasts like (35a–b) from Italian: [NP mio Gianni]] ((N)It.) (35) a [DP il the my Gianni b **[DP  [NP mio Gianni]] (It.) my Gianni c [DP Gianni [NP mio Gianni]] ((S)It.) Gianni my ‘my Gianni’ In the typically northern Italian (35a), the definiteness reading of masculine proper name Gianni is achieved through lexicalization of the D(eteminer) position with the definite article, such that Gianni remains in situ and follows the prenominal possessive. As (35b) illustrates, if the name remains in situ and the D(eterminer) position is left empty, then the former’s definiteness reading cannot be licensed and the derivation proves ungrammatical for all speakers. In southern regional Italian, however, it is far more usual in these cases for the definite article not to lexicalize the D(eterminer) position, but for the latter to be lexicalized by overt raising of the proper name, and hence the derived postnominal position of the possessive in (35c). By the same token, we can now also straightforwardly explain the diachronic differences in the use of the definite article with nouns indicating unique, abstract, and generic referents observed in }4.2.2.2 above. In particular, the non-use of the article in such cases in early texts can be interpreted as a case of generalized overt N(oun)-raising to an empty D(eterminer) position (36a), which has become more restricted in modern Romance where the D(eterminer) position is now generally lexicalized by an overt (albeit expletive) article (36b): (36) a [DP morte death

[NP morte]] è is

b [DP ’a [NP morte]] the death ‘death is natural’

è is

natural (ONap., De Jennaro IAM 14) natural naturale (mod.Nap.) natural

110

From Latin to Romance

4.2.3 Other determiners With the rise of the D(eterminer)P(hrase), other categories with determiner-like properties which in Latin had adjectival status are attracted to the D(eterminer) position (or better, the D area). The main categories involved here are two, namely demonstratives and possessives. In both cases we find that some Romance varieties preserve two complementary paradigms, one related to the adjective and the other to the determiner. For example, in Catalan, Occitan, Romanian, and Spanish demonstratives generally behave like determiners, lexicalizing the prenominal D(eterminer) position and hence in complementary distribution with the definite article. However, their original adjectival status has not been entirely jettisoned in these same varieties, inasmuch as they may still occur in the canonical postnominal adjectival position (typically with a pejorative reading in IberoRomance and Occitan, but unmarked (and potentially emphatic) in Romanian where the prenominal position takes on literary and formal overtones; Sandfeld and Olsen 1936: 157, 161; L. and M. Maiden p.c.),37 in which case the D(eterminer) position is filled with the definite article (37a–d). In other varieties such as Portuguese and Italian (37e), by contrast, only the determiner use of the demonstrative is attested: (37) a aquestes /aquelles opinions vs les opinions aquestes / aquelles (Cat.) these / those opinions the opinions these / those b aqueste / aquel brave òme vs this / that good man

lo the

brave good

òme man

aqueste this

/ aquel (Occ.) / that

c acest / acel program vs programul acesta / acela (Ro.)38 this / that programme programme=the this / that

37 The unmarked nature of the postnominal position in Romanian also explains why the shortened colloquial forms (e.g. acest(a) > ăst(a) ‘this.MSG’, acel(a) > ăl(a) ‘that.MSG’) prove more frequent in postnominal position than in the more formal prenominal position (Sandfeld and Olsen 1936: 157). Indeed, A. Stravinschi (p.c.) confirms that the shortened forms are restricted to the postnominal position. 38 It is interesting to note that in Romanian the two positions also correlate with a formal distinction in the shape of the demonstrative. In postnominal position the demonstrative assumes the form of the pronominal demonstrative, formally marked by final -a (cf. acesta ‘this.one.MSG’, acela ‘that.one.MSG’). On the assumption that pronominal demonstratives are simple D(eterminer) heads that select for a null NP complement, we may take this final -a to be underlyingly a suffixal article (just like the definite article (32a–b)) into which the adjectival demonstrative acest/acel ‘this/that’ incorporates (i.a). On this view, postnominal demonstrative structures such as (i.b) might be best analysed as a recursive DP structure containing the two suffixal articles -ul and -a (with the (potentially emphatic) underlying appositional reading ‘the programme(,) this/that one’):

(i)

[NP acest/acel]] (Ro.) a [DP acest/acel -a this =ART ‘this one’ [NP program [DP acest/acel-a [NP acest/acel]]]] (Ro.) b [DP program-ul programme=the this/that=ART ‘this/that programme’

The rise of functional structure d este this /ese / this(/that)

/ese /aquel país vs / this(/that) / that country /aquello (Sp.) /that

el the

111

país country

este this

e esta /essa /aquela mentira (Pt.) / questi /codesti /quei cappelli (It.) this / this / that lie these / these / those hats Similarly, the possessive in these same varieties shows a formal contrast between a tonic adjectival paradigm (postnominal in Spanish and Romanian, but pre- and postnominal in Occitan, and Catalan) and a clitic determiner paradigm generally limited to the singular persons (prenominal in all but Romanian; see Lyons 1986; Lombardi 2007):39 TABLE 4.3 Tonic/clitic possessive paradigms [MSG forms] Spanish

Occitan

Catalan

Romanian

Tonic

Clitic

Tonic

Clitic

Tonic

Clitic

Tonic

Clitic

1SG

mío

mi

mieu

mo(n)

meu

mon

meu

mu

2SG

tuyo

tu

tieu

to(n)

teu

ton

tău

tu

3SG

suyo

su

sieu

so(n)

seu

son

său

su

1PL

nuestro

nuestro

nòstre

–

nostre

–

nostru

–

2PL

vuestro

vuestro

vòstre

–

vostre

–

vostru

–

3PL

suyo

su

sieu

lor

–

so(n)

seu

son

In all cases, the clitic determiner forms are inherently definite on account of their lexicalization of the D(eterminer) position, whereas the tonic adjectival forms are underspecified for definiteness, and hence their associated noun occurs with a determiner: (38) a mi hija vs una / esta / ninguna my daughter a this no ‘my daughter’ vs ‘a/this/no daughter of mine’ b son gat vs lo / aquest (sieu) gat his cat the this his cat ‘his cat’ vs ‘his cat/this cat of his’

hija daughter

mía (Sp.) my

(sieu) (Occ.) his

c ton cosí vs un / aquell cosí your cousin a that cousin ‘your cousin’ vs ‘a/that cousin of yours’

teu (Cat.) your

39 This distinction is residually present in Aragonese where clitic forms are exceptionally preserved in just the first two persons in conjunction with kinship terms (e.g. (clitic) mi/tu pai ‘my/your father’ vs (tonic) o mío/tuyo can ‘(the) my/your dog’).

112

From Latin to Romance d mos txius my uncles ‘my uncles’

vs

los the

txius uncles

meus (Alg.) my

e socru-su vs socrul său / acest socru al său (Ro.) father.in.law=his father.in.law his this father.in.law the his ‘this father in law of his’ As with the definite article, the clitic determiner possessive in Romanian is also suffixal and hence attracts its associated nominal head to the D(eterminer) position. With the exception of Spanish and Occitan in which the clitic determiner forms are not lexically restricted, the determiner clitic forms in the other languages characteristically display a more restricted distribution in that they are limited to singular, unmodified kinship terms (e.g. Cat. la teva/**ta butxaca ‘your pocket’, but ta germana ‘your sister’), a distribution also found in the dialects of upper southern Italy (e.g. Nap. ’o vraccio tuoio ‘the arm your’ vs sòrata ‘sister=your’; Ledgeway 2009a: 252, 268–70). This distribution would suggest again that N(oun)-raising to the D(eterminer) position containing the proclitic possessive is only permitted with intrinsically referential nouns like kinship terms (cf. discussion of examples (33)–(36) above). Although this dual adjectival-determiner paradigm can be assumed to underlie all Romance varieties historically,40 it has been lost in most varieties in favour of the generalization of one of the two paradigms. For instance, in Asturian, Italian, and European Portuguese it is the adjectival paradigm which has prevailed (39a), whereas in French and Brazilian Portuguese (Thomas 1969: 80; Teyssier 1984: 105) the determiner paradigm has generalized (39b).41 However, in Asturian and Italian, and optionally in European Portuguese, the adjectival possessives exceptionally appear to function as determiners and lexicalize the D(eterminer) position whenever employed in conjunction with singular, unmodified kinship terms, since the determiner is excluded in such contexts (39c). (39) a los sos llibros (EAst.), la nostra città (It.), a minha loja (EPt.) the his books the our city the my shop ‘his books’, ‘our city’, ‘my shop’

40 Adams (forthc. c: }5.1), however, demonstrates that the Romance clitic forms find no immediate precursor in Latin, where ‘reduced’ forms (e.g. tus ‘your’, sus ‘his/her/their’) are simply the result of loss or contraction involving the two back vowels in hiatus (cf. TUUS ‘your’, SUUS ‘his/her/their). 41 This difference between European and Brazilian Portuguese varieties in relation to the possessive (presence vs absence of article) is not absolute, but represents an overwhelming tendency in the spoken varieties of both, which is often complicated by diamesic, diastratic, and diatopic factors. For example, the article is still employed in the spoken language of some parts of southeast Brazil (e.g. Rio de Janeiro, São Paulo), but is generally omitted in most other parts of the country.

The rise of functional structure b ses ‘his

enfants (Fr.),42 children’

c [DP (**lu) the ‘your dad’

to your

nosso ‘our

113

vizinho (BPt.) neighbour’

[NP pa]] (EAst.) dad

4.2.3.1 Romanian To conclude our discussion of Romance determiners, we must briefly discuss two other article types limited to Romanian, namely the so-called demonstrative article ECCE ‘behold’ + ILLE ‘that’ > cel (MPL cei, FSG cea, FPL cele) and the possessive article (AD ‘to’ >) al (MPL ai, FSG a, FPL ale). Both of these have received various interpretations in the literature,43 the technical details of which need not concern us here. Suffice it to note that, in essence, both articles can be considered expletives which serve as a last resort strategy to ‘fill’ an otherwise empty D(eterminer) position.

4.2.3.1.1 demonstrative article One of the principal uses of the demonstrative article is to substitute the canonical article when the NP occurs with a cardinal quantifier (40a) or when the cardinal selects for a null NP (40b). Given the strict complementary distribution of the demonstrative and canonical articles in this context (40c–d), it is logical to interpret the presence of the cardinal quantifier as the element that excludes the canonical article. (40) a cei the.MPL b cei the.MPL

trei three

trei [NP ] (Ro.) three

c **prietenii friends=the.MPL d **cei the.MPL

prieteni (Ro.) friends

trei (Ro.) three

prieteni (Ro.) friends

In the context of our analysis of the canonical Romanian article in (32a–b), this intuition finds an immediate answer. In particular, we can hypothesize that in examples like (40a) raising of the nominal head prieteni to the D(eterminer) position to incorporate with the suffixal article -i is not possible, since its passage to the D 42 The earlier adjectival paradigm of old French (e.g. cist meon fradre Karle (Strasbourg Oaths) lit. ‘this my brother Charles’ (= this brother of mine Charles)) has since been pressed into service as the pronominal paradigm in conjunction with the definite article (e.g. le mien ‘(the) mine’). 43 See, among others, for the demonstrative article cel, Manoliu Manea (1989: 105–6), Cornilescu (1992), Alboiu and Motapanyane (2000b: 3–4), and for the possessive article al, Dobrovie-Sorin (1987), Ştefănescu (1997), Alboiu and Motapanyane (2000b: 4–6), D’Hulst, Coene, and Tasmowski (2000).

114

From Latin to Romance

position is blocked by the intervening quantifier position lexicalized by the cardinal trei (a classic case of the Head Movement Constraint).44 To rescue the suffixal article -i, its empty nominal slot is lexicalized by directly inserting an expletive reflex of ECCE ‘behold’ (namely ce-), giving rise to the morphologically complex determiner [D [N ce]i] (41a). Similarly, in examples like (40b) where the NP contains a null nominal head, there can be no N(oun)-raising to the D(eterminer) position, since, as a suffix, the Romanian article must incorporate into an overt nominal head. Consequently, once again we see that whenever the movement strategy is blocked, the illicit stranding of the suffixal article -i under D is resolved by the last resort strategy of directly lexicalizing the empty N slot under D with the erstwhile presentative ECCE (41b), now an expletive nominal head.45 (41) a [DP [N ce+] -i [QP trei [NP prieteni]]] (Ro.) DEM =the.MPL three friends [QP trei [NP ]]] (Ro.) b [DP [N ce+] -i DEM =the.MPL three This same analysis can also account for the apparently optional use of the demonstrative article before postnominal adjectives: (42) a prietenii (cei) friends=the.MPL the.MPL ‘the intelligent friends’

inteligent¸i (Ro.) intelligent.MPL

b profesorul professor=the.MSG ‘the new teacher’

nou (Ro.) new.MSG

(cel) the.MSG

c demnitarul cel official=the.MSG the.MSG ‘the tall/**high official’

înalt (Ro.) tall.MSG

Rather than being optional, the use of the so-called demonstrative article in examples like those above serves to distinguish interpretively between a direct- and indirect-

44 This explanation presumably carries over to superlatives, where the head of the superlative AP mai ‘more’ + ADJ equally blocks raising of the nominal head to the suffixal article under the D(eterminer) position:

mai buni prieteni (Ro.) (i) cei the.MPL more good.MPL friends ‘the best friends’ 45 In current generative terms, the distinction between N(oun)-raising to D (yielding the canonical article) and direct lexicalization of the empty N slot associated with the suffixal determiner through an expletive nominal head (yielding the demonstrative article) is a case of Move vs Merge.

The rise of functional structure

115

modification reading of the adjective (Sproat and Shih 1988; 1990). In particular, the use of the article yields the indirect modification reading, such that the adjective is necessarily interpreted predicatively (cf. the Greek articulated predicative adjective), in essence understood as a reduced relative clause with a null copula, as shown by the obligatory predicative reading of înalt ‘tall’ (cf. the ungrammatical attributive reading **‘high’) in examples such as (42c).46 Consequently, with the demonstrative article pronounced in (42a–c), the implied reading is that of a restrictive relative (e.g. ‘the friends (that are) intelligent’). In terms of the analysis above, we can then treat the demonstrative article and following adjective as forming a D(eterminer)P(hrase) containing a null or elliptical NP that functions as the predicate of a reduced relative clause (43a). On a par with what was noted for (40b), the selectional requirements of the suffixal MPL article -i under the D(eterminer) position cannot be satisfied by N(oun)-raising, since the nominal head is null. Once again the derivation is rescued by the last resort strategy of directly inserting the expletive demonstrative ce- in the vacant N slot selected by the suffixal article (43b). (43) a [DP prieteni=i [CopP [DP [N_] -i [NP  inteligent¸i]]]] (Ro.) friends=the.MPL =the.MPL intelligent.MPL b [DP prieteni=i [CopP [DP [N ce-] -i [NP  DEM=the.MPL friends=the.MPL ‘the friends (who are) the intelligent (FRIENDS)’

inteligenți]]]] (Ro.) intelligent.MPL

4.2.3.1.2 possessive article Turning finally to the possessive article, we begin by considering the paradigm in (44a–c). In (44a) the possessee phrase câinele ‘dog=the’ and the genitive possessor phrase fetei ‘girl=of.the’ are superficially adjacent and as such the use of the MSG possessive article al is excluded. However, if the adjacency between the possessee and possessor is interrupted as in (44b), where the adjective frumos ‘pretty’ intervenes between possessee and possessor, then the MSG possessive article al agreeing with the head of the possessee phrase câine must be employed. However, as (44c) reveals, the presence of the possessive article cannot be determined simply by the lack of adjacency between the two phrases, since câine and fetei are otherwise linearly contiguous. Rather, what (44c) demonstrates is that the correct empirical generalization is that the possessive article is licensed whenever the possessor phrase does not stand immediately adjacent to the D(eterminer) position of the preceding possessee phrase: in (44c) the determiner position lexicalized by the free-standing indefinite article/proximal demonstrative (un/acest) is separated from the possessor phrase by the intervening nominal head câine. 46

Cornilescu (2006), Giusti (1993: 73–9; 2006), Cinque (2010: 102–4).

116

From Latin to Romance

(44) a [câinele] (*al) dog=the POSS.ART.MSG ‘the girl’s dog’

[fetei] (Ro.) girl=the.GEN.FSG

b [câinele frumos] al dog=the pretty POSS.ART.MSG ‘the girl’s pretty dog’

[fetei] (Ro.) girl=the.GEN.FSG

c [acest/un câine] al this/a dog POSS.ART.MSG ‘this/a dog of the girl’

[fetei] (Ro.) girl=the.GEN.FSG

Now, in the literature there have been various interpretations of these facts,47 the precise details of which need not concern us here. Instead, we shall simply interpret the possessive article, in line with its apparent etymology (AD ‘to, at’ > a ‘of ’ + ILLE ‘that’; cf. Ştefănescu 1997), as a sort of prepositional determiner (henceforth D/P) which, as an inflectional element, displays agreement in number and gender with the possessee (cf. Dobrovie-Sorin 1987), just as in Latin the possessive copular ESSE ‘be’ agrees with the THEME possessee promoted to subject in the verbal domain (45a). More specifically, we propose that the so-called possessive article corresponds to the inflectional equivalent within the nominal domain of the possessive copula of the clausal domain (in short, a sort of nominal copula), equally displaying agreement with the THEME possessee promoted to subject within the nominal proposition (45b): PULCHER]i / [CANES PULCHRI]j PUELLAE (45) a [CANIS dog.NOM pretty.NOM.MSG dogs.NOM pretty.NOM.MPL girl.DAT ESTi /SUNTj (Lat.) is are ‘The girl has a pretty dog/some pretty dogs’

b [câinele frumos]i / [câinii frumoşi]j ali / aij dog=the pretty.SG dogs=the pretty.PL POSS.DET.MSG POSS.DET.MPL fetei (Ro.) girl=the.GEN.FSG ‘the girl’s pretty dog / dogs’ We thus propose that the underlying structure of the possessive constructions in the verbal and nominal domains in (45a–b) is as in (46a–b), where the possessee-phrase ultimately raises from its base-position within a predicative small clause (SC) to the surface subject position (namely, the left-peripheral Spec(ifier) position of the possessive copula/article), whereupon it enters into a (Spec-head) number(/gender) agreement relation with the possessive copula/article: 47

For an overview, see Alboiu and Motapanyane (2000b: 4–6), D’Hulst, Coene, and Tasmowski (2000).

The rise of functional structure

117

(46) a [CopP [Subject ___] [Cop' [Cop ESSE [SC [Possessee] [Possessor] ]]]] b [D/P.P [Subject ___] [D/P' [D/P a- [SC [Possessee] [Possessor] ]]]] Now, in the case of strict adjacency such as (44a), we can assume that, in raising to the D(eterminer) position to adjoin to the suffixal article -le, the syntactically non-complex possessee câine (= N(oun) head) passes en route through the vacant D/P head, whose content it lexicalizes (cf. 47a). If, however, the possessee constitutes a syntactically complex nominal (= NP), as in (44b) where it is modified by an adjective, the whole possessee phrase raises to the subject position of D/P.P (cf. NP-subject raising), from where the nominal head alone can then raise to adjoin to the suffixal article under D (47b).48 Significantly, NP-raising to the subject position of D/P.P correlates with overt agreement on the D/P head, which inflects for number and gender with the NP in its associated subject position (cf. subject-verb agreement).49 (47)

a

DP

b

D D//P.P câine-le Subject D/P'

f

D/P câine

SC N/P câine

DP fetei

DP D D/P.P câine-le Subject D/P' NP D/P SC N AP al câine frumos NP DP câine frumos fetei

In short, we have two ways of licensing the nominal possessive construction and the D/P head: either the content of the head is licensed by N(oun)-raising of the

48

That the whole possessee phrase must raise when syntactically complex is substantiated by the observation that, if the adjectival head raises to adjoin to the suffixal article as in (i), the nominal head occurs to the left of the possessive article (viz. in the subject position of D/P.P), and not in its base position to the right of the possessive article (cf. **frumosul al câine fetei ‘pretty=the POSS.ART.MSG dog girl=of.the’). [SC câine frumos [DP fetei]]]]] (Ro.) (i) [DP frumos-ul [D/P.P[Subjectcâine frumos] [D/P' [D/Pal] pretty=the dog POSS.ART.MSG girl=the.GEN.FSG ‘the girl’s pretty dog’ 49

In most Daco-Romanian dialects, however, the possessive article assumes the invariant non-agreeing form a, as illustrated in the contrast between Moldovan dialect and standard Romanian in (i.a–b): ista a ɲew (Mdv.) (i) a fiʃoru boy=the this POSS.ART my.MSG meu (Ro.) b feciorul acesta al POSS.ART.MSG my.MSG boy=the this ‘this boy of mine’

From Latin to Romance

118

possessee through D/P en route to the D(eterminer) position, or its content is spelt out under a canonical (Spec-head) agreement configuration with a complex nominal raised to its phrasal subject position. This, in turn, explains the use of the possessive article in examples like (44c): N(oun)-raising of the syntactically non-complex possessee câine through D/P cannot apply since it cannot raise thereafter from D/P to D(eterminer), as this position is independently lexicalized by the demonstrative (acest) or the indefinite article (un). Consequently, the only way to license the D/P projection and the concomitant possessive construction is to raise the entire NP (still câine) to the phrasal subject position of D/PP, whereupon it licenses 3MSG agreement on the D/P head overtly manifested in the possessive article al (cf. 48): (48)

DP D un/acest

D/P.P Subject D/P' N/P câine D/P SC al NP câine

DP fetei

Interestingly, this analysis also provides us with a straightforward explanation of the use of the possessive article in conjunction with the pronominal possessive (49a). In particular, we can assume that the possessee is represented by a null pro(nominal) which raises to the subject position of D/PP, where under this (Spec-head) agreement configuration its features (= 3MSG) are spelt out overtly in the D/P head that realizes the possessive article al (49b). (49) a al POSS.ART.MSG

său (Ro.) her.MSG

‘her (one)’ b

D/P.P DP proi

D/P' D/P al i

SC DP pro

DP sau

The rise of functional structure

119

4.3 Inflectional phrase (IP) One of the most salient developments of the verb system in the passage from Latin to Romance has been the large-scale transferral of many verb-related inflectional categories to preverbal auxiliaries. As the overt realization of a functional category Infl(ection) spelling out grammatical information relating to the verb phrase (VP), it is this same category which, as the locus of verb agreement, licenses nominativemarked subjects in its associated left-peripheral Spec(ifier) position (SpecIP), the Romance dedicated preverbal subject position (see }7.3.3, }7.4.2). This is exemplified below in (50a–b) with a representative Catalan example, where the progressive auxiliary estava ‘was’ spells out the Infl head that selects for the VP constituent headed by the gerund aprenent ‘learning’: (50) a en Dominic estava aprenent the Dominic was learning ‘Dominic was learning the guitar’

b

la the

guitarra (Cat.) guitar

IP DP

I'

en Dominic Infl estava

VP

V DP ! aprenent la guitarra 4.3.1 Romance auxiliaries The emergence of an inflectional phrase (IP) projection thus correlates directly with the grammaticalization of a number of originally lexical verbs to produce a wide range of auxiliaries, a process whose effects are not uniformly mapped onto the semantic, phonological, morphological, and syntactic structures of the various Romance languages, which not only show considerable differences among themselves in relation to otherwise similar constructions,50 but which individually also display considerable variation from one auxiliary construction to another.51 Indeed, attempts to establish a pan-Romance definition of the linguistic category of auxiliary verb are notoriously riddled with difficulties, unlike the Germanic languages where it proves

50

Green (1982; 1987), Pountain (1982), Vincent (1987), Remberger (2006). Pottier (1961), Jones (1988b), Longo (1998), Longo and Souza Campos (2002), Alboiu and Motapanyane (2000b: 14–20). 51

120

From Latin to Romance

easier, though not entirely straightforward, to set up a number of formal criteria (such as the so-called NICE properties of English auxiliaries; see Steele et al. 1981; Harbert 2007: 285–92) to identify a common class of auxiliaries. For example, in English and German auxiliaries can consistently be identified, among other things, by their selection of an infinitival complement (rather than a past participle or gerund) and their incompatibility with the infinitival marker to/zu (51a–b), and their lack/ avoidance of a past participle (52a–b): (51) a I must b Ich I

muß must

(**to)

speak

(**zu) to

sprechen (Ger.) speak.INF

(52) a I have had to (**musted) speak Ich habe sprechen müssen I have speak.INF must.INF

(**gemußt) (Ger.) must.PTCP

In Romance, by contrast, all such generalizations present numerous exceptions. For example, just limiting ourselves to the reflexes of the modals *potere ‘can’, DEBERE/ HABERE DE-AB ‘must’, and *volere /QUAERERE in French, Italian, Neapolitan, Occitan, and Spanish, we can note, among other facts, that: (i) clitic climbing is obligatory in Neapolitan (53a), heavily preferred in Occitan (53b), possible, but not obligatory, in Italian and Spanish (53c–d), and excluded in modern French (53e): (53)

a ’e gghjammo b los anam c (li) andiamo d (los) vamos e nous allons we (them=) we.go ‘we are going to help them’

a (los) a a to

les (them=)

aiutà (Nap.) ajudar (Lgd.) aiutar(e)(li) (It.) ayudar(los) (Sp.) aider (Fr.) help.INF(=them)

(ii) in compound forms the auxiliary must be realized on the modal in French and Neapolitan (54a–b), on either the modal or the infinitive indifferently in Occitan and Spanish (54c–d), and on either the modal or the infinitive in accordance with a deontic/epistemic distinction in Italian (54e):52

52 In prescriptive varieties of Spanish the deontic/epistemic distinction is marked through the absence/ presence of the preposition de ‘of’ to introduce the infinitival complement of deber ‘must’:

(i)

debido salir (Sp.) a ha he.has had.to exit.INF ‘he had to leave’ debido de salir (Sp.) b ha he.has had.to of exit.INF ‘he must have gone out’

The rise of functional structure (54)

a Il a dû fumer (Fr.) b a avut’’a fummà (It.) c a degut fumar / deu d ha debido fumar / debe e ha dovuto fumare / deve he has had.to smoke.INF / he.must ‘he had to smoke / he must have been smoking’

aver haber aver have.INF

121

fumat (Lgd.) fumado (Sp.) fumato (It.) smoked

(iii) only in Italian and Occitan (55a–b), but not in French and Neapolitan (55c–d)— Spanish has long lost all traces of perfective ‘be’ (see }4.3.1.4, }7.3.1.1, }7.4.1.1)—is the auxiliary selection of the modal sensitive to the transitive/unaccusative nature of the embedded infinitive (cf. }4.3.1.4, }7.3.1.1): (55)

a b

non mi sono me soi not myself= am c je n’ ai d nu m’ aggiu I not myself= I.have ‘I couldn’t fall asleep’

potuto pogut been.able pas pu pututo not been.able

addormentare (It.) pas dormir (Lgd.) not fall.asleep.INF m’ endormir (Fr.) addurmentà (Nap.) myself= fall.asleep.INF

(iv) besides an infinitival complement, reflexes of *volere/QUAERERE, but not *potere and DEBERE, may take a finite complement and, in Neapolitan and Spanish, may also select a participial complement under particular conditions (56a–b): (56)

a quería que preparasen todo / preparado todo (Sp.) b vuleva ca preparassero tutta cosa / preparata tutta cosa (Nap.) he.wanted that they.prepared everything prepared everything

While acknowledging the absence of a discrete class of Romance auxiliaries, in what follows we shall review a number of general crosslinguistic properties or parameters of auxiliation (see Heine 1993), which characterize to varying degrees those Romance verbs which realize verb-related categories such as tense, aspect, mood, and voice (Ledgeway forthc. b). 4.3.1.1 Semantic weakening In the area of semantics, it proves a fairly easy task to recognize a number of Romance verbs which have undergone various degrees of semantic impoverishment, including such cases as IbR. seguir ‘to follow’, whose original lexical meaning is clearly still transparent, though weakened, in the continuous/iterative aspectual periphrasis with a following gerund (57a), and It. venire ‘to come’, which has been heavily desemanticized (or semantically bleached) in its dynamic passive auxiliary function with the participle (57b).53

53 See, however, Squartini (1999; 2003) who maintains that some residual semantic constraints do still condition the venire-passive.

122

From Latin to Romance

(57) a sigue estudando (Pt.) he.follows studying ‘he is still/goes on studying’ b le bozze venivano corrette (It.) the proofs came corrected ‘the proofs were being corrected’ In accordance with well-attested crosslinguistic pathways of auxiliation (Heine 1993: 45–8), the core Romance verb-related grammatical categories are thus derived from original lexical predicates indicating: (i) location (ESSE ‘be’, STARE ‘stand’, SEDERE ‘sit’): passive (58a), progressive/continuous aspect (58b–d), and present perfectivity (58e): (58) a el pantalón era/estaba planchado (Sp.) the trousers was/was ironed ‘The trousers were being (dynamic)/were (stative) ironed’ b vous êtes éternellement créant you are eternally creating qu’[ . . . ] (OFr., Fénelon EDD 2.5.3) that ‘you are continually creating all that [ . . . ]’ c era a se she.was to self= ‘she was taking a walk’

tout all

ce this

passejar (Occ.) walk.INF

d está a cantar / está he.is to sing.INF / he-is ‘he is singing’

cantando (E/BPt.) singing

e sono rimasti in montagna (It.) they.are remained in mountain ‘they have remained in the mountains’ (ii) motion (IRE ‘go’, UENIRE ‘come’, AMBULARE ‘go, walk’): iterative aspect (59a), progressive/protracted aspect (59b–c), retrospective aspect (59d–e), future time (59f–g), and past time (59h): (59) a l’ ai him= I.have ‘I saw him again’

tornat returned

a to

veire (Lgd.) see.INF

b es va posant bé (Cat.) self= she.goes placing well ‘she is (progressively) getting better’

The rise of functional structure

123

c há quatro anos que ando/vou/venho vendendo automóveis (BPt.) has four years that I.walk/I.go/I.come selling cars ‘I’ve been selling cars for four years’ d je viens de e

me

laver (Fr.)

venh de me lavar (Lgd.) I come from me= wash.INF ‘I have just washed’

f jeu vegnel I come ‘I shall wash’

a to

lavar (Srs.) wash.INF

g van a they.go to ‘they’ll sleep’

dormir (Sp.) sleep.INF

h vaig anar al mercat I.go go.INF to.the market ‘I went to the market yesterday’

ahir (Cat.) yesterday

(iii) possession (HABERE ‘have’, TENERE ‘hold’): iterative aspect (60a–b), resultative aspect (60c–d), and present perfect (60e): (60) a koza teŋs=tu fajt? (Pie.) what hold=you done ‘what have you been doing?’ b lo tenh de him= she.holds of ‘she keeps watching him’

velhat (Occ.) watched

c tinc preparat el sopar (Cat.) I.have prepared the dinner ‘I have got dinner ready/prepared’ d tengo pittate ’e I.have painted the ‘I’ve got the walls painted’

parete (Nap.) walls

e el ga invecià tanto (Ven.) he has aged a.lot ‘he has aged considerably’ (iv) volition (*volere): future time (61a–b) and deontic passive (61c):

124

From Latin to Romance

(61) a avionul va pleca dimineat¸a (Ro.; cf. mod.Ro. vrea ‘wants’) plane=the wants leave.INF morning=the ‘the plane will leave in the morning’ b voj I.want ‘I’ll leave’

parti (Frl.) leave.INF

c ’i lenzola vulianu cangiate (Cos.) the sheets wanted changed ‘the sheets had to be changed’ (v) obligation (DEBERE ‘must’, HABER DE/AD ‘have to’): future time (62a–d): (62) a je ne pense pas qu’ I not think not that ‘I don’t think she will come’

elle she

doive must.SBJV

venir (Fr.) come.INF

b sos óspites den éssere thuccatos prima de arrivare nois (Srd.) the guests must be.INF left before of arrive.INF we ‘The guests will have left before we arrive’ c prometo-lhe que hei-de recusá-lo (Pt.) I.promise=you that I.have-of refuse.INF=it ‘I promise you that I’ll refuse it’ d aju a turnari I.have to return.INF ‘I’ll come back at once’

subbitu (Sic.) at.once

4.3.1.2 Morphosyntactic properties In the area of morphosyntax, Romance auxiliation is clearly visible in the process of decategorialization, whereby the emergent auxiliary progressively jettisons the typical morphosyntactic properties of its erstwhile lexical verb status. For instance, the auxiliary typically loses the ability to select its own arguments, as shown by the different valencies of the lexical and auxiliary reflexes of DEBERE and *potere in (63)–(64), ultimately coming to inherit and govern the syntax of its lexical verbal complement (cf. Harris and Campbell’s (1995: 193) Heir-Apparent Principle; Ramat 1987: 16). Hence, unlike lexical verbs, auxiliaries impose, for example, no restrictions on the animacy or otherwise of their subject (65a–b), including reflexes of *volere/QUAERERE (65c–d): (63) a devo-te um I.owe=you a ‘I owe you a favour’

favor (Pt.) favour

The rise of functional structure b devo-te fazer um a I.must=you do.INF ‘I have to do you a favour’

125

favor (Pt.) favour

(64) a Dieu peut tout (Fr.) God has.power everything ‘God has authority over all’ b Dieu peut tout God can everything ‘God can do anything’ (65) a el alcalde ‘the mayor b el libro ‘the book

podría could

faire (Fr.) do.INF

dimitir (Sp.) resign’

podría costar could cost

c sembra che voglia it.seems that it.wants ‘it seems to want to rain’

poco (Sp.) little’ piovere (It.) rain.inf

d chere frittu (Nuo.) it.wants fried ‘it has to be fried’ Other reflexes of decategorialization include: (i) the emergence of gaps in the verb paradigm, such as the lack of an imperative for the reflex of perfective HABERE ‘have’ (66a) or the incompatibility of Italian progressive stare ‘stand’ + gerund with the preterit (66b), in contrast to Spanish estar (66c): (66) a **aie fini la tâche have.IMP.2SG finished the task ‘have the task finished by midday!’ b **stette he.was

avant before

midi! (Fr.) midday

studiando (It.) studying

c estuvo estudiando (Sp.) he.was studying ‘he was studying’ (ii) the inability to form passives, as witnessed by the infelicity of the Spanish progressive aspectual periphrasis ir ‘go’ + gerund (67a) when placed in the passive (67b):

126

From Latin to Romance

(67) a el tendero ha ido aumentando los precios (Sp.) the shopkeeper has gone increasing the prices ‘the shopkeeper has progressively been increasing the prices’ b **los precios han sido idos aumentando (Sp.) the prices have been gone increasing ‘the prices are progressively being increased’ (iii) the inability to take a nominal complement, as exemplified by Sp./Pt. perfective haber/haver ‘have’ (68a), now replaced by a reflex of TENERE ‘hold, keep’ (68b) in the possessive construction: (68) a **he/**hei I.have

tres three

casas (Sp./Pt.) houses

b tengo/tenho tres I.hold three ‘I have three houses

casas (Sp./Pt.) houses

(iv) the reduction and loss of verb inflection, as exemplified by the fossilization of Lat. UULT ‘want.3SG’ > o in the Romanian future construction o + să ‘that’ + subjunctive (69a), the southern Apulian progressive aspectual marker sta (< stare ‘stand’) now used in all six grammatical persons (69b; Ledgeway 2008b), and the wide-spread use of invariant va (< UADERE ‘go < walk’) in all but the 1/2PL in an andative asyndetic structure across the extreme south of Italy (69c–d).54 (69) a o să merg/ mergi/ meargă/ wants that.SBJV go.1SG go.2SG go.SBJV3SG meargă (Ro.) go.SBJV.3PL ‘I/you(SG)/(s)he/we/you(PL)/they will go’

mergem/ merget¸i/ go.1PL go.2PL

b sta + pperdu/ pperdi/ pperde/ pperdimu/ pperditi/ pperdenu (Lec.) stand lose.1SG lose.2SG lose.3SG lose.1PL lose.2PL lose.3PL ‘I am/you (SG) are/he is/we/you (PL)/they are losing’ c va + pperdu/ pperdi/ pperde/ pperdenu (Lec.) go lose.1SG lose.2SG lose.3SG lose.3PL ‘I am/you (SG) are/he is they are going to lose’ d va a + pighhiu/ pigghi/ pigghia/ pigghianu (Mar.) go and take.1SG take.2SG take.3SG take.3PL ‘I am/you (SG) are/he is/we/you (PL)/they are going to fetch’

54

Sornicola (1976), Ledgeway (1997: }4.1; 2008b), Cardinaletti and Giusti (2001; 2003), Sorrisi (2010).

The rise of functional structure

127

With the increased semantic integration and grammatical dependency between auxiliary and verbal complement (namely, [VP] + [VP] ) [AUX + VP]), the auxiliary construction comes to license a range of ‘local’ syntactic phenomena generally assumed to hold exclusively of monoclausal constructions, including, for example, the attraction of negators (70a–b) and clitic pronouns (71a–b) to the auxiliary and, in Ibero-Romance (including Catalan in this instance), the impossibility of intervening adverbs between perfective auxiliary and participle (72a–b):55 (70)

a o seu país non está the his country not is construir unha bomba nuclear (Glc.) build.INF a bomb nuclear ‘his country is not trying to build a nuclear bomb’ b nun habemos (**non) escaecer not we.have not forget.INF ‘we won’t forget the warning’

(71)

a mi ha me= he.has ‘he has seen me’

l’ the

(*non) not

buscando seeking

avisu (Ast.) warning

visto(**mi) (It.) seen =me

b cine mi-o poate(**mi-o) who me=it= can me=it= ‘Who can explain it to me?’ (72) a ja havia (**ja) already he.had already ‘he had already spoken’ b ela sempre me tinha she always me= had ‘she had always loved me’

explica? (Ro.) explain.INF

parlat (Cat.) spoken (**sempre) always

amado (Pt.) loved

4.3.1.3 Morphophonological specialization In many cases, this increased integration between auxiliary and dependent verb is translated morphophonologically in the

55 Cyrino (2010) points out some counterexamples in Brazilian Portuguese, where, given the lower position of verb movement in this variety, the non-finite verb form may host clitics (i.a), and negation (i.b):

(i)

a João tinha possivelmente me visto (BPt.) João had possibly me= seen ‘João had possible seen me’ não aprendido sua lição (BPt.) b Foxy Brown parece ter Foxy Brown seems have.INF not learnt her lesson ‘Foxy Brown seems not to have learnt her lesson’

128

From Latin to Romance

creation of morphologically specialized (and often synchronically irregular) auxiliary paradigms displaying phonologically reduced (typically clitic) forms, which, in certain cases, contrast with morphophonologically regular and full paradigms preserved for the original lexical meaning of the same verb (cf. UADO > Cal. ve (+ rapu) ‘I’m gonna (open up)’ vs vaju (+ a ru cinema) ‘I go (to the cinema)’). For example, although historically both derived from HABERE ‘have’, in the present tense Cat./Nap. perfective haver/avé differ from lexical haver/heure and avé ‘have, receive’ not only in exhibiting distinct morphologically reduced forms in specific persons (e.g. Cat. 1/2SG auxiliary he/has vs lexical hec/heus, Nap. 1/2PL auxiliary a(i)mmo/a(i)te vs lexical avimmo/avite; Ledgeway 2009a: 382–6), but also in the reduction of vowels to schwa in otherwise identical forms (e.g. Cat. 3SG lexical ha [a] (also heu) vs auxiliary ha [ə]). In other varieties such as Romanian and Corsican, for instance, the reduced auxiliary forms of HABERE are restricted to specific persons, with complete homophony with the corresponding lexical paradigm in the other persons. TABLE 4.4 Morphophonological specialization in HABERE paradigms Cat. haver(/heure)

Nap. avé

Ro. avea

Cor. avì

Lexical

Auxiliary

Lexical

Auxiliary

Lexical

Auxiliary

Lexical

Auxiliary

heig/hec

he

aggio

aggiu

am

am

aghju

aghju

has [as]

has [əs]

(h)aie

(h)ê

ai

ai

ai

ai

ha [a]

ha [ə]

ave

(h)a

are

a

hà

hà

havem

h(av)em

avimmo

a(i)mme/imme

avem

am

avemu

emu

haveu

h(av)eu

avite

a(i)te/ite

avet¸i

at¸i

avete

ate

han[an]

han [ən]

àveno

(h)anno

au

au

anu

anu

Also exemplary in this respect is Ro. (a) vrea ‘want’ (< *volere), which in its grammatical uses as a future auxiliary has developed specialized reduced forms in a number of persons (e.g. vrea o bere ‘he wants a beer’ vs va (< v(re)a) bea o bere ‘he’ll drink a beer’; Monachesi 2005: 133–4). Similarly, Catalan and Sardinian contrast a regular, full lexical paradigm for ‘go’ (anar) and ‘owe’ (dévere), respectively, with a morphophonologically reduced paradigm of the same now specialized as preterit and future auxiliaries (e.g. Cat. anem al mercat ‘we’re going to the market’ vs va(re)m anar al mercat ‘we went to the market’; Srd. mi devet meta vinu ‘he owes me a lot of wine’ vs det aer gana ‘he will be hungry’).56 The paradigms of all three verbs are given in Table 4.5 below: 56

Colon (1961), Jones (1988b; 1993: 89–92), Vallduví (1988), Pérez Saldanya and Hualde (2003), Juge (2006), Pisano (2009: 159–60).

The rise of functional structure

129

TABLE 4.5 Auxiliary/lexical Paradigms of *volere,57 *andare, and DEBERE Romanian vrea Lexical

Catalan anar

Sardinian dévere

Auxiliary

Lexical

Auxiliary

Lexical

Auxiliary

vreau

voi

vaig

và(re)ig

devo

de(v)o

vrei

vei

vas

va(re)s

deves

des

vrea

va

va

va

devet

det

vrem

vom

anem

và(re)m

devímus

demus

vret¸i

vet¸i

aneu

và(re)u

devítes

dedzis

vor

vor

van

va(re)n

deven

den

As a consequence of their typical clitic status, there follows a whole series of other typical auxiliary properties.58 For instance, Romance auxiliaries are generally incapable of bearing independent stress (73a), such that appropriate emphasis has to be achieved through other means such as lexical reduplication and topicalization (73b). (73) a **HO I.have

lavorato (It.) worked

ho lavorato (It.) b lavorare, work.INF I.have worked ‘I HAVE been working’ By the same token, Romance auxiliaries are also incapable of occurring in isolation through deletion (74a), preposing (74b), or clefting (74c) of their non-finite complement. (74) a **Florin a terminat serviciul, dar Doina Florin has finished work=the but Doina ‘Florin has finished his work, but Doina hasn’t’ b **pintado la puerta de verde, painted the door of green ‘painted the door green, no, he hasn’t’ c **c’ est payer maintenant this is pay.INF now ‘it is pay now that I must’

no, no

que that

nu not

no not je I

a (Ro.) has

la it=

ha (Sp.) he.has

dois (Fr.) must

57 Alongside the standard forms given in Table 4.5, colloquial Romanian allows even further morphophonological reduction of the auxiliary paradigm for vrea, including 1SG oi; 2SG ei, ăi, îi; 3SG a, o; 1PL om; 2PL et¸i, ăt¸i, ît¸i, ot¸i; 3PL or. 58 Jones (1988b), Ledgeway (2000: 155–61), Monachesi (2005: 140–52).

130

From Latin to Romance

4.3.1.4 Romance perfective auxiliary constructions Traditionally, the emergence of the HABERE ‘have’ + PtP periphrasis is retraced to the grammaticalization of an original resultative aspectual periphrasis:59 (75) a [VP [THEME [LOC

in ea prouincia] pecunias in that province money.ACC.FPL [AP collocatas]] habent] (Lat., Cic. Leg. Man. 18) placed.ACC.FP they.have ‘they have large sums invested in that province’

b [IP [VP [Loc

in ea prouincia] [Theme in that province magnas] collocatas] habent] big.ACC.FPL placed.ACC.FPL they.have ‘they have invested large sums in that province’

magnas big.ACC.FPL

pecunias money.ACC.FPL

The oft-cited Ciceronian sentence in (75a) is generally claimed to exemplify a resultative aspectual periphrasis (and not at this stage a temporal periphrasis), in that it foregrounds the present result of a backgrounded past action (Ernout and Thomas [1953] 1993: 223; Tarriño Ruiz 2009b: 481). Hence the meaning of (75a) is not that of a perfective past action ‘they have invested large sums in that province’, but that of a present state or condition ensuing from a previous past action ‘they have large sums invested in that province’. Consequently, in (75a) habent ‘(they) have’ is still a lexical predicate indicating possession which subcategorizes for a LOCATIVE subject (the possessor) encoded in the 3PL inflection -nt and a THEME direct object (the possessed), namely pecunias magnas collocatas ‘much money invested’. Note, in particular, that the verbal participle collocatas ‘placed, invested’ simply functions as an adjectival predicative modifier of the direct object magnas pecunias, hence its agreeing feminine plural form (in -as), which as part of its argument structure subcategorizes for a LOCATIVE complement (namely, in ea prouincia ‘in that province’) and an AGENT (‘the investor’), whose identity is not made explicit in this structure, though possibly recoverable from the extralinguistic context. Consequently, in (75a) the identity of the investor(s) can either coincide with the LOCATIVE subject of habent or is free to refer to another pragmatically salient, albeit implicit, individual or group of individuals. Given the nature of the resultative construction, it is thus only compatible with the participles of transitive predicates, hence the presence of an

59 Vincent (1982), Salvi (1982; 1987), Tuttle (1986), Pinkster (1987), Zamboni (2000: 127–8), Bentley (2006: 59–61, 64–9); though see the discussion in }7.3.1.1.1 for an alternative analysis. See now also Adams (forthc. d) for an excellent critical overview of previous analyses and for new interpretations of the distribution and development of HABERE + participle in Latin and in early Romance.

The rise of functional structure

131

implied AGENT or, more rarely, EXPERIENCER (e.g. ‘I have got something painted/ written/prepared/thought out’), but is clearly not compatible with intransitive predicates (e.g. **‘I have got cried/sung/gone/come’). Through time, however, resultative structures like (75a) became increasingly common and, given the frequent pragmatic inference that the LOCATIVE subject of HABERE and the implied AGENT or EXPERIENCER of the participle were one and the same—a pattern which apparently began with verbs of cognition and perception with EXPERIENCER subjects (76a), where it would have been impossible to interpret the implied EXPERIENCER of the participle and the LOCATIVE possessor of HABERE as disjoint in reference,60 especially if the direct object cannot be understood as the object of HABERE (76b–c)—this coreferential interpretation eventually became conventionalized, though not necessarily yielding a perfective reading in all cases.61 (76) a dicam de istis graecis [ . . . ] quid Athenis exquisitum I.will.say regarding these Greeks what Athens.LOC discovered habeam (Lat., Cato ad fil. frg. 1) I.have ‘I will say about those Greeks [ . . . ] what I have found out in Athens’ b cum cognitum habeas quod sit summi when known you.have what is supreme.GEN numen (Lat., Cic. Fin. 4.11) divine.will.ACC ‘when you realize the will of the supreme lord’

rectoris [ . . . ] ruler.GEN

c quamuis praeceptum habeamus, ut [ . . . ] (Lat., Col. Rust. 12.52.3) although instructed we.have that ‘though we have already given instruction that . . . ’ As a result of the conventionalization of this coreferential interpretation of the LOCATIVE of HABERE and the participial AGENT/EXPERIENCER (‘I have got much money invested in that province’ ) ‘I have invested much money in that province’), the

60 Thielmann (1885b: 509, 517–18), Pinkster (1987: 203–4), Vincent (1982: 84–5), González Rolán (1993: 526), Tarriño Ruiz (2009b: 481), Haverling (2010: 358–9), Adams (forthc. d: }2.2). 61 Scherer (1975: 72), Tarriño Ruiz (2009b: 481), Adams (forthc. d: }2.2). Pinkster (1987: 197) also convincingly hypothesizes that the reanalysis of the resultative construction as a perfect would initially have been favoured in contexts containing inalienable objects such as (i), where the participle cannot be readily omitted as is true of most other examples where the participle functions as a predicative modifier, for it would not mean anything to say ‘they had a head’.

cinctum habebant filo (Lat., Varro Ling. 5.84) (i) (Flamines) [ . . . ] caput Flamines head.ACC.N girt.ACC.N had thread.ABL ‘(the Flamines) had their hair girt with a woollen fillet’

132

From Latin to Romance

bond between participle and HABERE is closer than in the case of the predicative participial modifier, such that the original resultative periphrasis in (75a) is transformed into a present perfective periphrasis with concomitant reanalysis of its structure (Harris and Campbell 1995: 50–1), resulting in a modification in the underlying structure of the construction but with no overt surface manifestation (75b).62 In the new reanalysed structure in (75b), the identification of the implied AGENT (here the ‘investor(s)’) with the subject of habent entails that HABERE is no longer an independent lexical verb of possession, but now functions instead as an auxiliary referencing the person and number features of the participial subject and the tense and modal features of the entire construction. As such, auxiliary HABERE now simply inherits the argument structure, including the AGENT/EXPERIENCER subject, of its associated participle (see Harris and Campbell’s (1995: 193) Heir-Apparent Principle). Thus, collocatas no longer functions as an adjectival modifier of the nominal magnas pecunias, but now assumes full-fledged verbal status as the head of the VP, selecting the latter nominal as its NP object. As a result of this categorial change from adjective to verb, agreement of the participle with its direct object is progressively weakened (Väänänen 1982: 255), as precociously illustrated in examples such as the fourth-century extract from Oribasius in (77), since verbs in Romance canonically agree with subjects and not objects. (77) haec omnia probatum habemus (Lat., Orib. Synopsis 7.48)63 these.ACC.NPL all.ACC.NPL tried.ACC.NSG we.have ‘we have tried all these things’ Agreement disappears first in those contexts in which perceptual factors rendered its presence barely functionally relevant (e.g. with postverbal nominals), but resists with greater resilience in those contexts (e.g. with preverbal nominals and (3rd person) clitics) in which its presence plays a significant functional role in sentence-parsing (Smith 1993; 1995; 1999). Finally, as a genuine perfective construction no longer linked to the transitivity of lexical HABERE, the distribution of the periphrasis soon extends to include (intransitive) unergative participles (78), ultimately emerging as a generalized

62 Adams (forthc. d) demonstrates that the perfective interpretation of the periphrasis is ‘all but nonexistent’ in Plautus and early Latin, only emerging in classical Latin. However, his close reading of numerous classical and later examples underlines how, with the notable exception of participles expressing mental acquisition, the presumed perfective reading of the periphrasis frequently assumed in previous studies is far from certain in most cases, not to say often entirely unjustified, leading him to conclude (}3) that ‘even by the time of Gregory [of Tours] the perfect periphrasis was not fully grammaticalised’ (for a similar view, see Nuñez 1998; Tarriño Ruiz 2009b: 481; Haverling 2010: 356–9). 63 Although, as Adams (forthc. d: }4) observes, ‘[o]mnia is, however, a special case, as it tended to be fossilised as a collective and used in agreement with neuter singulars’.

The rise of functional structure

133

perfective periphrasis for all predicates with AGENT or EXPERIENCER subjects irrespective of their transitivity.64 (78) sicut parabolatum habuistis (7th-c. Lat., Formulae Salicae Merkelianae 260.7) thus spoken you.had ‘thus you had spoken’ In Table 4.6 below, we sum up the principal changes involved in the reanalysis and grammaticalization of the temporal perfective periphrasis: TABLE 4.6 Reanalysis and grammaticalization of HABERE + PtP HABERE

= lexical (+θ)

HABERE

= auxiliary (θ)

Resultative

Present perfective

2 subjects / biclausal

1 subject / monoclausal

Adjectival PtP

Verbal PtP

+ Participial Agr

 Participial Agr

Transitives

Transitives > Unergatives

With the rise of the perfective HABERE periphrasis, there remained a residue of morphologically active intransitive verbs with UNDERGOER subjects, namely unaccusatives (e.g. IRE/AMBULARE/UADERE ‘to go’, UENIRE ‘to come’, (DE)SCENDERE ‘to descend’, SALIRE ‘to leap’, ENTRARE ‘to enter’, CADERE ‘to fall’), which proved semantically incompatible with the HABERE periphrasis on account of their subcategorization of a subject of the UNDERGOER, rather than the AGENT/EXPERIENCER, type. This residue of unaccusatives was absorbed into the established ESSE ‘be’ perfective periphrasis for deponents and passives (79a–b), since they shared with the latter the property of subcategorizing for an UNDERGOER subject (79c). With the eventual loss of the middle morphology of the imperfectum,65 the original deponents (where they survive) and unaccusatives were no longer formally distinguished (80a), and continued to converge with the passive in the perfective paradigms (80b).

64 The shift in order from PtP + HABERE to HABERE + PtP must have been much later, since the former order, though not necessarily with the constituent parts in direct juxtaposition (Bauer 2009a: 292–3), predominates in the Latin of all periods, leading Bauer (2006: 294) to conclude that the change must have occurred in the early Romance period (cf. also }5.3.1.5). 65 In some cases subsequently marked by the reflexive marker SE ‘self-’ (Zamboni 2000: 125–6):

(i) Myrina quae Sebastopolim se uocat (Lat., Plin. N.H. 5.121; cf. deponent UOCATUR) Myrina which Sebastopol self calls ‘Myrina which is called Sebastopol’

134

From Latin to Romance

(79) a

b

SUM (Lat.) slipped I.am ‘I have slipped’

LAPSUS

SUM (Lat.) called I.am ‘I have been called’

UOCATUS

c *kadutʊ(s) sʊm (late Lat.) fallen I.am ‘I have fallen’ (80)

a

> *mori-re: CADE-RE; MORIT-UR > *mɔrit: CADIT (late Lat.) die-pass.INF die-ACT.INF fall-ACT.INF die.3SG-PASS die.3SG fall.3SG

MOR-I

b *mɔrtʊ sʊm: *βokatʊ sʊm: *kadutʊ sʊm (late Lat.) died I.am called I.am fallen I.am ‘I have died: I am/have been called: I have fallen’ The overall result of these developments is a split in the perfective forms of the verb between a periphrasis with auxiliary HABERE ‘have’ for transitive/unergative predicates with AGENT/EXPERIENCER subjects on the one hand, and a periphrasis with auxiliary ESSE ‘be’ for unaccusative and passive predicates with UNDERGOER subjects on the other. In short, the verb system of late Latin and, in turn, (early) Romance develops an active/ stative orientation (also often termed ‘split intransitivity’; cf. }7.3.1.1.2) in which intransitive subjects are formally distinguished according to their AGENT vs UNDERGOER characterization (La Fauci 1988: 51–2). 4.3.1.5 Romance synthetic future(-in-the-past) With the exception of a few isolated residues of ‘be’ in early French (e.g. ier, iers, ert, ermes, ertes, ierent < ERO, ERIS, ERIT, ERIMUS, ERITIS, ERUNT; Lausberg 1962: }918) and perhaps Sp. 2SG eres ‘you are’ (< ERIS ‘you will be’), the two paradigms of the Latin synthetic future in -BO (UOCA-BO ‘callFUT.1SG’, UIDE-BO ‘see-FUT.1SG’) and -AM (LEG-AM ‘read-FUT.1SG’, AUDI-AM ‘hear-FUT.1SG’), as well as the periphrases built on ESSE ‘be’ and the future active participle, left no trace in Romance.66 In their place, we find either the present, already widely attested in Latin (Bennett 1910: 18–22), or one or more infinitival periphrases variously involving, as illustrated in Table 4.7, the auxiliaries *volere ‘want’ (Romanian, Friulian, and southern Italian dialects), UENIRE AD ‘come to’ (Surselvan), DEBERE ‘must’ (Sardinian; cf. also the French subjunctive example in (62a) above), HABERE 66 Apart from the fact that Latin had two distinct morphological paradigms for the future (-BO in 1st/2nd conjugations and -AM in 3rd/4th conjugations) which produced a certain degree of redundancy in the system, witness hypercorrect forms such as RESPONDEAM ‘I shall reply’ and SCIBO ‘I shall know’ (for prescriptively correct RESPONDEBO and SCIAM), the loss of the Latin synthetic future paradigms is traditionally attributed, in part, to the formal ambiguity engendered by phonetic neutralization in the labials -U- [w] and -B- [b] > [β] (e.g. AMAUIT/AMABIT [amaβit] ‘he loved/will love’) and in short I and long E > [e] (e.g. UINCIS/UINCES [βinkes] ‘you win/will win’; Anderson 1979: 27; Pinkster 1987: 210; Herman 1996b: 60).

The rise of functional structure

135

‘have to/of’ (old Tuscan, Sardinian, southern Italian dialects, Ibero-Romance, and old Romanian), and IRE/AMBULARE/UADERE ‘go’ (French, Occitan, (Brazilian) Portuguese, and Spanish):67 AD/DE

TABLE 4.7 Romance auxiliary futures (cantar(e)/cantà/cânta ‘to sing’) Romanian

Surselvan

Sardinian

Abruzzese

Spanish

voi cânta

vegnel a cantar

de(v)o cantare

ajja cantà

voy a cantar

vei cânta

vegns a cantar

des cantare

hî da cantà

vas a cantar

va cânta

vegn a cantar

det cantare

a da cantà

va a cantar

vom cânta

vegnin a cantar

demus cantare

avem a cantà

vamos a cantar

vet¸i cânta

vegnis a cantar

dedzis cantare

avet a cantà

vais a cantar

vor cânta

vegnen a cantar

den cantare

anno da cantà

van a cantar

Alongside these periphrases, all Romance varieties, with the exception of Sardinian, Romanian, and Dalmatian, present a synthetic future (cf. Table 4.8) derived from an erstwhile periphrasis consisting of the infinitive followed by a weakened form of HABERE ‘have’ (e.g. CANTARE ‘sing.INF’ + *aio/as/a(t)/Vmo(s)/ete(s)/Vn(t) ‘I/you/(s)he/ we/you/they will sing’), forms which have now been largely jettisoned in the modern dialects of southern Italy (Loporcaro 1999).68 TABLE 4.8 Romance synthetic future Catalan

French

Italian

Occitan

Portuguese

Spanish

cantaré

chanterai

canterò

cantarai

cantarei

cantaré

cantaràs

chanteras

canterai

cantarás

cantarás

cantarás

cantarà

chantera

canterà

cantará

cantará

cantará

cantarem

chanterons

canteremo

cantarem

cantaremos

cantaremos

cantareu

chanterez

canterete

cantaretz

cantareis

canteréis

cantaran

chanteront

canteranno

cantarem

cantarão

cantarán

67 It is tempting to see in the future value of the Romance modals a usage which some (e.g. Hofmann and Szantyr 1965: }175; Fleischman 1982; López Fonseca 2009: 464; Haverling 2010: 395–9) have claimed can be traced back to the Latin usage of the modals DEBERE ‘must’ and UELLE ‘want’ (together with INCIPERE ‘begin’ and POSSE ‘can’). Pinkster (1989), however, finds no evidence for this claim in the Latin record. 68 There is a vast literature on the topic of the origins and developments of the Romance future and future-in-the-past/conditional including, among others, Thielmann (1885a), Valesio (1968; 1969), Coleman (1971), Lanly (1973), Harris (1978: ch. 6), Fleischman (1982), Green (1987), Pinkster (1987), Adams (1991), Roberts (1993b), Maiden (1996), Loporcaro (1999), Bentley (2000a, b), Nocentini (2001), D’Hulst (2004), Bourova (2005; 2007), La Fauci (2006), Bourova and Tasmowski (2007), Slobbe (2008), Parkinson (2009), Haverling (2010: 397–8).

From Latin to Romance

136

In a similar vein, the Romance synthetic future-in-the-past derived from the infinitive followed by a past tense form of HABERE, the perfect HABUI (> *ɛbwi ‘I.had’) in northern and central Italo-Romance (e.g. It. canterebbi > canterei ‘I would sing’) and the imperfect HABEBAM ‘I.had’ elsewhere (e.g. Sp. cantaría ‘I would sing’), has traditionally been considered a development parasitic on that of the future, such that once infinitive + HABEO had come to mark the future, this opened the way for infinitive + 69 HABUI/HABEBAM to mark future-in-the-past. In contrast, Benveniste (1968: 89–90) and Coleman (1971: 224) maintain that the documentary evidence shows the HABERE periphrasis to have first emerged as a future-in-the-past, with the future construction following later. There is, however, greater consensus among scholars that the irrealis conditional meaning represents a secondary development from an original future-inthe-past value in accordance with the crosslinguistic tendency for future forms to develop irrealis modal functions (Coleman 1971: 217; Fleischman 1982: 64). Traditional accounts trace the Romance synthetic future(-in-the-past) back to a substandard spoken Latin innovation (cf. Varvaro forthc.: }7) which was severely repressed and hence barely attested in the written language (Haverling 2010: 397–8), a view which finds some initial support in the characteristically Latinate postposition of the auxiliary typical of OV languages (Fleischman 1982: 119, 121): had the construction emerged in Romance, then we would have expected the reverse order in line with the emerging Romance VO typology (see }5.3.2.1). On this view (cf., for example, Slobbe 2008: 109–12), the future(-in-the-past) construction supposedly represents the gradual grammaticalization and concomitant desemanticization of HABERE from a full verb of possession, where it could often imply unrealized actions in conjunction with a transitive infinitive (81a), towards a deontic modal of obligation from the first century AD (81b),70 before emerging from around the third century AD (first in Tertullian, though see Pinkster 1987: 206), and now also in conjunction with intransitive and passive infinitives, as a simple future auxiliary (81c) and ultimately as a future inflection. This latter stage is famously represented in the late seventh-century Fredegar’s Chronicle from Gaul (81d), where the classical Latin future dabo is juxtaposed with the now synthesized HABERE periphrasis daras (< DAR(E) ‘give.INF’ + = *as ‘you.have’; Fleischman 1982: 68; Haverling 2010: 398), although the apparently more emphatic reading of the latter (‘you WILL (give me them)’) might indicate an early functional specialization of the two paradigms (Adams 1991: 160–1). However, there is an even earlier example by about a century preserved in the vulgar language of a tombstone inscription from Ledoix-Serrigny in central France (81e; Stimm 1977).

69

Bourciez (1956: 272), Harris (1971: 28), Fleischman (1982: 61), Vincent (1987: 246), Haverling (2010:

397). 70

Prior to its deontic interpretation as a modal of necessity and obligation from early imperial times, often functions as a modal of possibility in the Ciceronian period (Hofmann and Szantyr 1965: }175; Coleman 1971: 217; Adams 1991; Haverling 2010: 399). HABERE

The rise of functional structure

137

(81) a de re publica nihil habeo ad te about thing.ABL public.ABL nothing I.have to you scribere (Lat., Cic. Att. 2.22.6) write.INF ‘I don’t propose to say anything in this letter about the state’ b quid habui facere? (Lat., Sen. Contr. 1.1.19) what I.had do.INF ‘What did I have to/should I do?’ c tempestas illa tollere habet totam paleam storm.NOM that.NOM remove.INF had all.ACC straw.ACC de area (Lat., Aug. Evang. Iohan 4.12) from threshing.floor.ABL ‘the storm will carry off all the chaff from the threshing floor’ d ille respondebat: non dabo. Iustinianus he replied not I.give.FUT Justinian.NOM dicebat: daras (Lat., Fred. Chron. 2.62) said you.give.FUT ‘and he replied: I will not give [them to you]. And Justinian replied: you will give [me them]’ e landelinus ficit | numen | qui illa made divine.will.ACC who.NOM that.ACC Landelinus.NOM possideravit viva | vsqui annus mili will.possess live.SBJV until year.NOM thousand in d(eo) (Lat.; possideravit POSSUM DICERE ‘I.can say.INF’; Adams 1991: 134). What this suggests then is that there must be another explanation behind the exceptional establishment of this superficially archaic order in the future periphrasis. Indeed, Adams (1991) highlights some neglected evidence from Pompeius’ fifth– sixth-century commentary of Donato’s Ars gramatica to demonstrate that the constituent parts of the periphrasis could occur in both orders. In particular, he identifies how in colloquial registers from the early Empire onwards the original unmarked order infinitive + HABERE gives way to a new preferred order HABERE + infinitive, with the former order initially persisting as a marked order. Significantly, during this period the marked and unmarked orders were not however necessarily synonymous, but could encode a meaningful contrast. Adams (1991: 135–6) counts in Pompeius 55 instances of the periphrasis, 38 with the now unmarked order HABERE + infinitive and

71

Menéndez Pidal (1956: 377, 358), Rohlfs (1968: 334–5), Smith and Bergin (1984: 157), Roberts (1994), Luís and Spencer (2005), Monachesi (2005: 152–8).

The rise of functional structure

139

17 with the marked reverse order. In the predominant pre-infinitival order HABERE ambiguously expresses either ‘possibility’ (83a), an interpretation documented as early as Cicero (Thielmann 1885a: 84; Bulhart 1936–42: 245–6), or ‘obligation/necessity’ (83b), 11 examples of which are also conceivably, though not necessarily, compatible with an English future translation. In post-infinitival position, by contrast, no such interpretive ambiguity arises and HABERE can only be understood as a modal of ‘obligation/necessity’ (83c), 14 examples of which can also be readily interpreted with a simple future reading.72,73 (83) a qui bene habent metra who well have.3PL metres.ACC ‘those who can write metres’

scribere (Lat., Pomp. Ars gram. 125.15) write.INF

b ergo semiuocales ita se habent inchoare a uocali sono therefore semivowels.NOM thus self have.3PL begin.INF by vowel.ABL sound.ABL et desinere in naturalem sonum (Lat., Pomp. Ars gram. 101.11f) in natural.ACC sound.ACC and end.INF ‘therefore semivowels have to begin with a vowel sound and end in a natural sound’ c tolle inde istum modum et tollere habes remove.IMP thence this.ACC mood.ACC and remove.INF you.have promissionem (Lat., Pomp. Ars gram. 214.15) promise.ACC ‘take away from there that mood and you will take way the promise’ The evidence discussed by Adams reveals that while HABERE in post-infinitival position had not yet unambiguously grammaticalized as a future auxiliary, it was nonetheless this marked position alone, whose emphatic interpretation made it compatible only with the ‘obligation/necessity’ reading (Coleman 1971: 219), that was most readily associated with the future interpretation and which ultimately would give rise to the Romance future.74 Further evidence in this direction comes

72 Revealing in this respect is the observation that the ‘possibility’ reading was already associated with the pre-infinitival position as early as Cicero, when the classical language otherwise preferred the postinfinitival order of auxiliaries. 73 A similar picture emerges from Bourova and Tasmowski’s (2007) study, in which they find that the order of the two constituent parts in late Latin follows from the modal interpretation of the periphrasis. In particular, the order infinitive + HABERE, especially in clause-initial position, typically conveys an alethic modal value, whereas the reverse order tends to license a deontic modal reading. 74 Also relevant for understanding the post-infinitival position of the auxiliary is Benveniste’s (1968: 87–8) observation that most of the early examples of the future are ‘restricted to subordinate, chiefly relative, clauses’ (cf. also Adams 1976a: 99), where, as we shall see (cf. }5.1.2, }5.3.2, }5.4.2.4), a conservative SOV order was robustly retained as late as the early Romance period.

140

From Latin to Romance

from the observation (Adams 1991: 163) that in Pompeius post-infinitival HABERE occurs exclusively in the present, the tense behind the Romance future, whereas preinfinitival HABERE occurs in a variety of tenses. Similarly, HABERE is invariably juxtaposed to the infinitive in postposition (except for 2 cases where a Wackernagel clitic intervenes; cf. discussion of mesoclisis in (82a–c)), while in pre-infinitival position there are 5 cases of intervening stressed words or phrases, highlighting a higher degree of fusion between infinitive and auxiliary in the former case. 4.3.2 Romance verb positions In accordance with the assumption that in the passage from Latin to Romance there emerges a dedicated structural position Infl(ection) for auxiliaries, we can observe that the availability of such a position gives rise to a further dimension of variation across Romance. In particular, not only can the Infl position be lexicalized by distinct auxiliaries (cf. epistemic use of Catalan auxiliary deure ‘must’ to express supposition in (84a)) but, in the absence of the latter, may be overtly filled by the raised lexical verb where its finite inflectional features can be licensed (cf. epistemic use of future in substandard Catalan to express supposition in (84b); see Badia i Margarit 1962 I: 391). (84) a [Infl deu he.must b [Infl tindrá he.will.have ‘he must be right’

[VP tenir raó]] (Cat.) have.INF reason [VP tindrá raó]] (coll. Cat.) reason

Such an approach explains the observed differences in the (unmarked) position of the finite verb in languages like French and Italian (85a–b) on the one hand and Calabrian, Romanian, and Spanish (85c–d) on the other:75 (85)

75

a b c d e

Jean [I’ fumait Gianni [I’ fumava Gianni [I’  Ion [I’  Juan [I’  John (smoked) ‘John always smoked’

[VP [VP [VP [VP [VP

toujours sempre sempa mereu siempre always

fumait]] (Fr.) fumava]] (It.) fumava]] (Cal.) fuma]] (Ro.) fumaba]] (Sp.) (smoked)

On the different extent of Romance verb movement, see Lois (1989), Pollock (1989), Belletti (1990: 44–5), Kayne (1991), Cinque (1999: 152), Cornilescu (2000: 89–92), Alboiu and Motapanyane (2000b: 22–4), Tortora (2002; 2010), Zagona (2002: 162–24, 168–70), Ledgeway and Lombardi (2005: 103–6). For an indepth enlightening comparative investigation of French, Italian, and Spanish verb movement, see now also Schifano (2011).

The rise of functional structure

141

Exploiting the fixed positions of VP-adverbs like ‘always’ as a diagnostic indicator of the left edge of the VP,76 we can now straightforwardly distinguish between overt verbraising languages like French and Italian, where the finite verb raises to the Infl position to the left of VP-adverbs like ‘always’, and languages like Calabrian, Romanian, and Spanish, where the verb remains in situ to the right of such VP-adverbs and the Infl position is not overtly lexicalized in the syntax. This idea has been taken further in much recent work, where Infl is interpreted as a general label for the rich inflectional area of the clause (the I-domain) made up of a series of distinct inflectional projections dedicated to marking various temporal, aspectual, and modal categories which can also be identified by the semantically corresponding adverbial modifiers they host.77 Consequently, the semantically related adverb and auxiliary contrasts in examples like (86)– (88) can now be understood as distinct positional reflexes (namely, Spec(ifier) and head positions) of the same underlying functional projection. (86) a Gianni [AspHabitual [Spec solitamente]  Gianni usually b Juan [AspHabitual [Spec ] solía Juan was.wont ‘John used to smoke’ (87) a le the très very

[VP fumava]] (It.) smoked [VP fumar]] (Sp.) smoke.INF

tableau vaut [MoodEpistemic [Spec probablement]vaut [VP picture is.worth probably peu]] (Fr.) little

b le tableau [MoodEpistemic [Spec ] the picture très peu]] (Fr.) very little ‘the picture can’t be worth much’

vaut

doit [VP valoir must be.worth

(88) a [AspRepetitive [Spec ntorna]  sta [VP trona]] (Lec.) again PROG it.thunders b [AspRepetitive [Spec ] torna a [VP tronar]] (Cat.) it.returns to thunder.INF ‘it’s started thundering again’ Armed with these assumptions about a universal fixed hierarchy of adverb positions and corresponding functional projections, we can now construct a fine-grained typology of Romance varieties along the lines of (89):

76 77

Emonds (1978), Pollock (1989), Belletti (1990), Cinque (1999). Cinque (1999; 2002), Belletti (2004a), Rizzi (2004).

142

(89)

From Latin to Romance MoodIrrealis AspPerfect a tal vez siempre entiendo b forse capisco sempre capisco toujours compr. c je comprends peut-être compr. I understand perhaps understand always understand ‘maybe I always completely understand’

AspSgCompletive completamente completamente complètement completely

[VP . . . ] entiendo(Sp.) capisco (It.) compr. (Fr.)

Although in all three varieties exemplified in (89) the finite lexical verb invariably leaves its base position to vacate the verb phrase (VP), witness its position to the left of the completive adverb ‘completely’ immediately adjacent to the VP, it raises to different functional projections within the I(nfl)-domain as illustrated by its differential position with respect to different adverb classes. For example, in Spanish the finite verb raises to the head position of the continuative aspectual projection immediately below the adverb ‘still’, whereas in Italian it appears to raise slightly higher to the head position of the irrealis modal projection below the adverb ‘perhaps’, and in French it raises to the highest available position above all adverb classes. On the basis of evidence like this, let us assume that the various adverbs and their associated functional projections making up the inflectional core of the clause can be broadly divided into two ‘spaces’ termed the Lower Adverb Space (LAS) and the Higher Adverb Space (HAS), as sketched in (90): (90) ( . . . [C that) [HAS Subject Adv . . . [LAS Adv . . . [VP Verb Object]]](]) As illustrated from Italian and French in (91a–b), the HAS chiefly includes modal and temporal functional projections variously spelt out by pragmatic (‘frankly’), evaluative (‘(un)fortunately’), evidential (‘apparently’), epistemic (‘probably’), temporal (‘then’), irrealis (‘perhaps’), habitual (‘generally’), and subject-oriented (‘wisely’) adverbs, whereas the LAS principally comprises aspectual functional projections variously spelt out by such adverbial classes as (presuppositional) negation (‘not (even)’), anterior time (‘already’), terminative aspect (‘no longer’), perfective aspect (‘always’), completive aspect (‘completely’), and voice-related manner adverbs (‘well’). (91) a MoodSpeechAct . . . > sinceramente sincèrement sincerely >AspHabitual saggiamente (It.) sagement (Fr.) wisely

MoodEvidential . . . >T >MoodIrrealis >MoodVolitional apparentemente allora forse solitamente apparamment alors peut-être généralement apparently then perhaps generally

The rise of functional structure

143

b NegPresup >TAnterior >AspTerminative > AspPerfect >AspSgCompletive >Voice mica già più sempre completamente bene (It.) pas déjà plus toujours complètement bien (Fr.) not already anymore always completely well Assuming the clause to consist of a highly articulated functional structure, we can then interpret a number of surface differences across Romance in terms of the varying extent of verb movement around different adverb classes in these two adverbial spaces. For instance, finite auxiliaries in most Romance varieties typically target a variety of distinct head positions within the HAS (92a–e),78 although in southern Italian dialects, Romanian, and Spanish they also present the option of only raising as far as the LAS (93a–c): (92) a [HAS avìa purtroppu (avìa) I.had unfortunately I.had [LASvippitu troppu]] (Cos.) drunk too.much ‘I had unfortunately perhaps drunk too much’

forse perhaps

b [HAS j’ai sincèrement (j’ai) I’ve sincerely I’ve (j’ai) [LAS trop exagéré]] (Fr.) I’ve too.much exaggerated ‘to be honest I probably went over the top somewhat’ c [HAS ho francamente I’ve frankly stupidamente (ho) [LAS bevuto stupidly I’ve drunk ‘I’ve frankly perhaps drunk too much’

(ho) I’ve troppo]] (It.) too.much

probablement probably

forse perhaps

d [HAS pot probabil (pot) sigur [LAS deja they.can probably they.can surely already despre mine]](Ro.) about me ‘they probably can certainly already say this about me’ e [HAS puedo afortunadamente (puedo) I.can fortunately I.can [LAS ya descansarme]] (Sp.) already rest.INF=me ‘I fortunately can perhaps already take a rest’

78

(avìa) I.had

tal vez perhaps

(ho) I’ve

spune say.INF

asta this

(puedo) I.can

Cinque (1999: 49–50), Ledgeway and Lombardi (2005: 87), Monachesi (2005: 134, 208).

144

From Latin to Romance

(93) a [HAS sicuru (vo) [LAS ggià vo surely he.wants already he.wants ‘he undoubtedly already wants to finish’

finiscia]] (Cos.) finish.INF

pret¸urile (pot) [LAS adesea pot b [HAS indubitabil undoubtedly prices=the can often can considerabil]] (Ro.) considerably ‘undoubtedly prices may frequently diverge considerably’ c [HAS tal vez (quiere) [LAS todavía perhaps he.wants still ‘perhaps he still wants to sleep’

quiere he.wants

devia deviate.INF

dormir]] (Sp.) sleep.INF

Finite lexical verbs, on the other hand, show a more varied behaviour. In French they continue to raise to the highest position within the HAS (94a; Rowlett 2007: 106–7), whereas in Calabrian, Romanian, and Spanish they typically raise only as far as the LAS (94b–d).79 Italian represents an intermediate case (94e), where the finite lexical verb targets a clause-medial position sandwiched between the HAS and the LAS (Cinque 1999: 31, 110–11, 180 fn. 80; Ledgeway and Lombardi 2005: 87–8): (94)

a b c d e

[HAS . . . ] V-Medial elle connaît peut-être **connaît idda **canuscia forze ?canuscia ea **cunoaşte poate ?cunoaşte ella **conoce tal vez ?conoce lei **conosce forse conosce she knows perhaps knows ‘perhaps she already knows the receipe’

[LAS . . . ] [VP V . . . ] déjà **connaît la recette (Fr.) ggià canuscia ’a ricetta (Cos.) deja cunoaşte ret¸eta (Ro.) ya conoce la receta (Sp.) già ?conosce la ricetta (It.) already knows the receipe(=the)

These same assumptions about the fixed positions of adverbs also allow us to plot the differential position of non-finite verbs such as the active participle in the following examples:80

79 Lois (1989), Cinque (1999: 152), Ledgeway and Lombardi (2005: 86–9, 102 n. 12), Monachesi (2005: 178), D’Alessandro (2010: 35–6). According to Cyrino (2010), verb movement in Brazilian Portuguese also appears to target the LAS (in Cyrino’s analysis, the lower T2 position above adverbs like bem ‘well’). 80 Lois (1989: 34, 40), Cinque (1999: 45–9, 146–8), Abeillé and Godard (2003), Monachesi (2005: 134–6).

The rise of functional structure

145

(95) [LAS AspPerfect AspPlCompletive VoiceManner [VP . . . ]]] [HAS a j’ai **mangé toujours **mangé tout **mangé bien mangé (Fr.) b apo **mandigadu semper mandigadu tottu mandigadu bene **mandigadu (Srd.) c ho mangiato sempre mangiato tutto **mangiato bene **mangiato (It.) d aju mangiatu sempe **mangiatu tuttu **mangiatu buonu **mangiatu (Cos.) e am mâncat mereu **mâncat tot **mâncat bine **mâncat (Ro.) f he comido siempre **comido todo **comido bien **comido (Sp.) I’ve eaten always eaten all eaten well ‘I’ve always eaten everything properly’ In contrast to what was observed with the finite verb in (94a), the active participle moves least in French, staying very low in the clause within the LAS to the right of the manner adverb ‘well’ just above the VP (95a),81 whereas in Calabrian, Romanian, and Spanish the active participle, unlike the finite verb in (94b–d), moves to the highest available position in the HAS beyond the perfective aspectual projection spelt out by the ‘always’ adverb (95d–f).82 In Sardinian and Italian, on the other hand, the position of the active participle displays greater freedom, raising in Sardinian at least above ‘well’ but no higher than ‘always’ in the LAS (95b), and above ‘all’ and possibly above ‘always’ to the HAS in Italian (95c).83 Similar types of variation are also found in conjunction with the infinitive. On a par with the active participle the infinitive in French remains in a low position within the LAS (96a),84 a pattern also replicated in Calabrian, Romanian, and Spanish (96b–d; Ledgeway and Lombardi 2005: 89–91). Italian infinitives, by contrast, occur in the highest available position within the HAS (96e; Belletti 1990: 70–6; Cinque 1999: 143–6):

81 The eastern Abruzzese dialect of Ariellese behaves similarly (D’Alessandro 2010: 37), as shown by the position of the participle to the right of the measure adverb poche ‘a little’ in (i), which is argued by Cinque (1999: 11) to occupy the same position as manner adverbs like ‘well’:

le so ?capite poche capite (Arl.) it= I.am understood a.little understood ‘I understood it a little’

(i)

82 Romanian presents a handful of exceptions in the monosyllabic adverbs cam ‘somewhat’, mai ‘again’, prea ‘too much’, şi ‘also; already’, and tot ‘still’ (Dobrovie-Sorin 1994: 26, 77–8; Cinque 1999: 227 n. 11; Legendre 2000: 231–2; Ledgeway and Lombardi 2005: Monachesi 2005: 175–81), whose exceptional behaviour has been argued by some to derive from their clitic status which requires them to attach to a verbal head:

au mai/ şi/ cam/ tot dansat (Ro.) they.have again also somewhat still danced ‘they (also/still) danced (again/a little)’

(i)

83

For the interpretive differences produced by the variable position of the Italian participle, see the discussion surrounding examples (98a–b) below. 84 Engver (1972), Belletti (2005b), Rowlett (2007: 108–10).

From Latin to Romance

146

(96) a Il faut loger bien (loger) [LAS le it-is=necessary him= accommodate.INF well accommodate.INF [VP loger . . . ]] (Fr.) ‘he must be given proper accommodation’ b signu stangu ’i [LAS (?pulizzà) sempe pulizzà I.am tired of clean.INF always clean.INF [VP pulizzà ’a casa]] (Cos.) the house ‘I’m fed up with always cleaning the house’ c promisiunea de a [LAS nu (vorbi) mai vorbi promise=the of to not talk.INF more talk.INF cu el]] (Ro.) with him ‘the promise of never speaking with him again’

niciodată [VP vorbi never

d está cansada de [LAS (hacer) siempre hacer she.is tired of do.INF always do.INF [VP hacer los mismos papeles]] (Sp.) the same roles ‘she is tired of always playing the same roles’ e mi dispiace [HAS parlargli francamente me= it.displeases speak.INF=him frankly (**parlargli) [VP parlargli così]] (It.) thus speak.INF=him ‘I regret speaking to him frankly like that’ 4.3.2.1 Summary and conclusions In light of the discussion above, Table 4.9 captures the essential surface differences across Romance in a highly simple way in terms of the possible clausal positions/spaces targeted by different verb forms:

TABLE 4.9 Typology of Romance V(erb)-movement Higher Adverb Space Fr.

It.

VAux

+

Vlexical

+

VActivePtP



VInfinitive



Clause-Medial Position Cos., Ro., Sp.

Lower Adverb Space

Cos., Ro., Sp.

Fr.

It.

Fr.

It.

Cos., Ro., Sp.

+

+





+





+







+

(+)



(+)

+

+

+



+



+

+



+









+



+

The rise of functional structure

147

In the literature there is no general consensus regarding the correct interpretation of V(erb)-movement to different clausal positions (for an overview, see Schifano 2011), although traditionally there have been many attempts to relate the extent of movement to the richness or otherwise of the inflectional Agr(eement) of the verb,85 witness Baker’s (1985; 1988: 13) Mirror Principle and Bobaljik’s (2002) Rich Agreement Hypothesis. In essence, approaches of this type attempt to drive syntactic operations from crosslinguistic morphological differences in individual languages. Admittedly, this view finds some initial support in the Romance data where we have observed that auxiliary verbs, presumably the richest inflectional forms of all, may raise to the HAS in all Romance varieties. However, a brief glimpse of the results in Table 4.9 suffices to dispell such an approach, inasmuch as all the Romance varieties we have examined are what may be termed inflectionally rich languages, yet they display some quite marked differences in the extent of finite lexical and non-finite V-movement. This conclusion is further substantiated by the observation that much of the rich inflection of the modern French verb, unlike that of the other varieties examined, is predominantly orthographic, yet it shows higher verb movement of finite lexical verbs to the HAS than all other varieties. Neither does the distribution of Romance verb placement lend itself to an interpretation in terms of finiteness, according to which overt/high V-movement can be understood as a reflex of the verb’s need to check its strong finiteness (person/ number, tense) features against corresponding functional projections situated higher in the clause. While French appears to provide a neat and consistent confirmation of this pattern inasmuch as finite and non-finite verbs invariably target the HAS and LAS, respectively, other Romance varieties directly undermine any such approach. For instance, on account of their exclusively grammatical functions auxiliary verbs can be considered among the most finite which, while explaining the tendency for Romance auxiliaries to occur in the HAS, fails to predict the equally possible lower position of the Calabrian, Romanian, and Spanish auxiliary within the LAS (cf. 93a–c). By the same token, Italian finite lexical verbs were shown to target a clause-medial position, yet non-finite verbs were seen, counterintuitively, to target the highest available position within the HAS. Finally, in Calabrian, Romanian, and Spanish we have observed that both finite and non-finite verbs tend to occur in a low clausal position within the LAS, highlighting the absence of any finiteness pattern whatsoever. Similar problems surround the novel analysis recently advanced by Biberauer and Roberts (2008), who propose that V-movement should be understood, not in terms of richness of Agr(eement) marking (witness French where all finite verbs move to the highest available position but fail to license pro-drop), but, rather, in terms of the

85

Roberts (1985), Lightfoot and Hornstein (1994), DeGraff (1997), Rohrbacher (1997), Vikner (1997).

148

From Latin to Romance

richness of synthetic T(ense)-marking. Superficially, their approach would seem to capture the difference between high(er) V-movement in Romance and low(er) V-movement in Germanic, inasmuch as the former (97a) tends to present a greater number of synthetic temporal paradigms than the latter (97b). However, we have seen that a variety like Spanish generally displays low verb movement to the LAS, despite presenting as many as 8 synthetic temporal paradigms (97c). (97) a Italian 1SG of parlare ‘to speak’ (7): parlo (PRS), parlerò (FUT), parlerei (COND), parlavo (IPFV.PST), parlai (PRET), parli (PRS.SBJV), parlassi (PST.SBJV) b German 1SG sprechen ‘to speak’ (3): spreche (PRS.IND/SBJV), sprach (PST), spräche (PST.SBJV) c Spanish 1SG hablar ‘to speak’ (8): hablo (PRS), hablaré (FUT), hablaría (COND), hablaba (IPFV.PST), hablé (PRET), hable (PRS.SBJV), hablase (PST.SBJV I), hablara (PST.SBJV II) Rather, a more viable explanation for the variable placement of Romance verb forms is to be sought in the licensing of their distinct temporal, aspectual, modal, and voice interpretations in relation to the different functions and interpretations that a given morphological verb form can convey. Compelling evidence for this view has recently been identified by Fedele (2010) in her comparison of a number of northern and southern regional Italian varieties and dialects, where a single verb form is shown to occur in distinct positions in accordance with its differing aspectual, modal, and temporal values. For instance, following Cinque (1999: 184 n. 8), Fedele notes that in Italian compound perfective paradigms the position of the participle with respect to the perfective adverb sempre ‘always’ correlates with significant interpretive distinctions: when the participle follows the adverb (98a) it yields a non-perfective reading, but licenses both a perfective and non-perfective reading when it precedes (98b):86 (98) a Gianni ha sempre fumato (It.) Gianni has always smoked ‘Gianni has always smoked (and still does)’ b Gianni ha fumato sempre (It.) Gianni has smoked always ‘Gianni has always smoked’ Similarly, in Friulian Fedele (2010: 13–32) notes that so-called past imperfective verb forms raise to different positions in accordance with their distinct modal interpretations. Under the future-in-the-past reading, they can raise to the irrealis modal position, as demonstrated in (99a) where the verb sits between the irrealis adverb 86 See also the discussion of differential verb positions in Cosentino with respect to the adverb mancu (Ledgeway and Lombardi 2005: 99), which is interpreted as a presuppositional negator before the verb and as an emphatic negator ‘not even’ following the verb.

The rise of functional structure

149

forsit ‘perhaps’ and the modal necessity adverb necessariamentri ‘necessarily’. However, under the ludic (99b) and counterfactual (99c) readings the verb is forced to raise even higher to the epistemic modal projection, as indicated by its position to the left of the epistemic adverb probabilmentri ‘probably’ and to the right of the deictic temporal adverbs ore/che volte ‘now/then’: (99) a a desevin che Gianni forsit al comprave necessariamentri il libri (Frl.) SCL they.said that Gianni perhaps SCL bought necessarily the book ‘they said that perhaps Gianni would necessarily buy the book’ b Gianni probabilmentri al ere ore il dotor . . . (Frl.) Gianni probably SCL was now the doctor ‘(Let’s pretend then that) Gianni is probably now the doctor . . . ’ c Se Gianni probabilmentri al comprave che volte if Gianni probably SCL bought those times libri, al studiave (Frl.) book SCL studied ‘If Gianni had probably bought the book then, he would have studied’

il the

Additional evidence for this semantically driven view of V(erb)-movement is also to be found in Italian contrasts like (100a–b), where the active interpretation of the participle correlates with a higher position than that of its passive counterpart.87 However, temporal interpretation is also relevant here: whereas the lower position of the passive participle proves grammatical in generic temporal contexts such as (100b), it is excluded in examples like (100c) where the specific temporal reference of the clause licenses and requires a higher position of the passive participle.88 (100) a hanno [LAS sempre accolto bene they.have always received well suo spettacolo]] (It.) his performance ‘they always received his performance well’

(**accolto) received

b il suo spettacolo è stato [LAS sempre the his performance is been always (accolto) [VP . . . ]] (It.) received ‘his performance was always well received’ 87

(accolto) received

[VPil the

bene well

Cinque (1999: 102–3, 147–8), Belletti (2005b), D’Alessandro and Roberts (2008: 481–2). See, however, Schifano (2011) for convincing arguments why Romance V(erb)-movement cannot be considered semantically driven. She maintains that the extent of Romance V(erb)-movement should instead be related to the varying formalization of the Mood, Tense, and Aspect functional fields of the clause in different Romance varieties. 88

150

From Latin to Romance c ieri il suo spettacolo era yesterday the his performance was bene (**accolto) [VP . . . ]] (It.) well received ‘his performance was received well yesterday’

stato been

[LAS accolto received

4.4 Complementizer phrase (CP) In the same way that the head positions D(eterminer) and Infl(ection) constitute the spell-out of grammatical categories related to their associated NP and VP complements, the sentential core too, now formally represented by inflectional phrase (IP), can be considered to be embedded within a further left-peripheral layer of functional structure termed CP (complementizer phrase). In accordance with parametric variation, the left periphery may spell out fundamental clausal distinctions such as finiteness, illocutionary force, and other discourse-related categories (e.g. Topic, Focus), as well indirectly replicating information encoded within the inflectional domain (IP) of the sentential core (see Rizzi 1997). Unlike DP and IP, however, there is already extensive evidence in early Latin for the structure of CP (Vincent 1998a), which, together with P(repositional)P(hrases) (cf. }3.4.1, }5.3.1.1), exceptionally constitute a significant precursor to the later widespread extension of configurational and functional structure to other areas of the grammar (see }4.5 below). 4.4.1 Evidence for Latin CP structure In }3.3.1 we noted that unmarked word order in Latin is predominantly considered to be SOV and that all other permutations, rather than being ‘free’, are pragmatically determined orders derived from underlying SOV.89 On this view, V-initial orders (101a) involve fronting of the verb to the left periphery, and XVS(X) orders (101b) involve the additional step of fronting some other pragmatically salient element to the left periphery (cf. also (42a) in chapter 3) under topicalization or focalization.90 (101) a [Miserat enim [ei Pharnaces coronam auream miserat]] had.sent indeed him.DAT Pharnaces.NOM crown.ACC.F golden.ACC.FSG (Lat., B.Alex. 70) ‘Pharnaces had indeed sent him a golden crown’

89 Marouzeau (1922), Pinkster (1990a: 181; 1991), Ostafin (1986), Vincent (1998a: 418–19), Oniga (2004: 97), Polo (2004) and Salvi (2004; 2005: 436; 2011: 356–8), Baños Baños and Cabrillana (2009: 680). 90 Pinkster (1990a: 182; 1993b: 246), Bolkestein (2001: 249), Oniga (2004: 97–8), Salvi (2005: 436–41), Spevak (2008a: 363), Danckaert (2010).

The rise of functional structure b [Idem facit [Caesar same.ACC.N did Caesar.NOM ‘The same does Caesar’

151

idem facit]] (Lat., Caes. B.G. 1.15.1)

More specifically, we take fronting of the verb to target a vacant C(omplementizer) position and fronting of any accompanying topicalized or focalized element to target its associated left-peripheral modifier/Spec(ifier) position, as illustrated in (102): (102)

CP Spec Ø idem

C' C

IP... (idem) miserat facit

V Miserat facit Preposed V orders still constitute a marked word order in classical Latin, often claimed to be a stylistic and/or pragmatic device serving to emphasize the verb of the utterance and licensing narrative functions,91 such as introducing description and marking progress of action of narration (103a) and marking thetic sentences with a presentative or eventive structure (103b): (103) a Transfigitur scutum Pulloni [ . . . ] Auertit hic casus removes this.NOM case.NOM is.pierced shield.NOM Pullo.GEN uaginam [ . . . ] Succurrit inimicus illi Uorenus (Lat., Caes. B.G. scabbard.ACC runs enemy.NOM to.DAT Vorenus.NOM 5.44.7–9) ‘Pullo’s shield is pierced through [ . . . ] This event turns his scabbard aside [ . . . ] His enemy Vorenus runs up to him’ b legantur tamen in Africam maiores appoint.PASS.3PL however in Africa greater.NOM.PL nobiles (Lat., Sall. Iug. 25.4) known.NOM.PL ‘nevertheless men of years and rank were sent to Africa’

natu birth.ABL

91 Linde (1923: 159), Marouzeau (1938: 81–2), Ernout and Thomas ([1953] 1993: 161), Pinkster (1993a: 649; 1993b: 246), de Jong (1994: 92), Bauer (1995: 93–5; 2009a: 276–9), Salvi (2004: 50–1), Devine and Stephens (2006: 145–52).

152

From Latin to Romance

Among its pragmatic values, verb-fronting has also frequently been associated with marked (typically subjunctive) modal interpretations,92 including optatives (104a), jussives (104b), concessives (104c), assertives (104d), polar interrogatives (104e), and imperatives (104f). Crosslinguistically, all these modal values have been linked in the literature with verb-fronting to a vacant C(omplementizer) position.93 (104) a stet haec urbs (Lat., Cic. Mil. 34.93) stand.SBJV3SG this.NOM city.NOM ‘may this city continue to stand!’ b caperes aut fustem aut you.take.IPFV.SBJV.2SG or club or ‘you should have taken a club or a stone!’

lapidem (Lat., Pl. Rud. 842) stone

c fuerint cupidi, fuerint irati, fuerint be.PFV.SBJV.3PL eager.NOM.PL be.PFV.SBJV.3PL angry.NOM.PL be.PFV.SBJV.3PL pertinaces (Lat., Cic. Lig. 18) firm.NOM.PL ‘granting that they were ambitious, that they were angry, that they were obstinate’ d uictores [ . . . ] trucidarent quos pellere non poterant. victors.NOM slaughtered who.ACC.PL drive.INF not they.could Pepulerunt tamen iam paucos they.drove yet now few.ACC superantes (Lat., Liv. 22.49.4) survivors.ACC ‘the victors slaughtered those that they could not drive away. Nevertheless they did drive off the few still surviving’ e dixitne tandem causam C. Fidiculanius Falcula [ . . . ]? (Lat., Cic. Clu. 103) said=Q at.last trial.ACC C. Fidiculanius Falcula ‘Did C. Fidiculanius Falcula stand trial after all?’ f libera rem publicam free.IMP.2SG thing public ‘free the republic from fear!’

metu (Lat., Cic. Cat. 1.20) fear.ABL

In addition to these pragmatic interpretations, there is also clear evidence that Latin verb-fronting is often syntactically motivated. In particular, it has been observed that

92 Schneider (1912: }13–33), Linde (1923: 159, 161), Möbitz (1924: 118–19), Rosén (1998), Devine and Stephens (2006: 144–50), Polo (2004: 381), Bauer (2009a: 276). 93 Rooryck (1992), Rivero (1994a, b), Graffi (1996), Zanuttini (1997: ch. 4).

The rise of functional structure

153

fronting of the verb in main clauses frequently occurs whenever preceded by a subordinate clause (especially conditional and temporal types), an ablative absolute, a negation, an adverb, or an adverbial phrase,94 as illustrated in the following representative examples: si resilierit, destinaui illum (105) a Quod because if he.is.restless I.determined him.ACC artificii docere (Lat., Petr. Sat. 46.7) trade.GEN learn.INF ‘if he is restless, I have determined that he will learn a trade’ b et puer iacentem sustulisset, animaduertit and boy.NOM prostrate.ACC picked.up noticed Trimalchio (Lat., Petr. Sat. 34.2) Trimlachio.NOM ‘and the boy picked it up from the ground, Trimalchio noticed him’ c non respuit condicionem (Lat., Caes. B.G. 1.42.2) not he.rejected proposal.ACC ‘he did not reject the proposal’ In short, there can be no doubt that these syntactically determined contexts of verbfronting represent an unmistakable precursor to the full-fledged V2 syntax of late Latin/early Romance outlined in }3.3.2, a conclusion further confirmed by the observation that verb-fronting in Latin rarely occurs in subordinate clauses (Bauer 1995: 96; Salvi 2004: 102), since in these cases the C(omplementizer) position would typically already be lexicalized by an overt complementizer (e.g. UT ‘so that’, NE ‘so that . . . not’, SI ‘if’, CUM ‘when’, QUOD/QUIA ‘that’) and hence unavailable to host the fronted verb (cf. also discussion in }5.4.2.4). In the fullness of time, this incipient V2 pattern would generalize as the unmarked word order replacing earlier SOV, as confirmed by its greater frequency in lower registers (including in authors such as Petronius and Tertullian and in later texts like the Peregrinatio Aetheriae) from where it eventually passed into the Romance vernacular (Salvi 2004: 102; Bauer 2009a: 276). Indeed, in their examination of word order in the late fourth-century Peregrinatio Clackson and Horrocks (2007: 292) recognize ‘an underlying order with the verb occupying the first position in the sentence, with an optional focus slot before it, which may be filled by a verbal argument (subject as default) or an adverbial phrase’. In a similar vein, Vincent (1998a: 418–23) finds convincing evidence for the presence and structure of CP outlined above not only in late Latin texts such as the Peregrinatio (106a), but also in Golden Age authors such as Cicero (106b) and even in early, non-literary Latin authors such as Cato (106c), where the finite verb and complementizer compete for the C position and the associated left-peripheral 94

Kroll (1918), Orinsky (1923: 93), Möbitz (1924: 120–1), Marouzeau (1938: 80), Adams (1976b: 137), Bauer (1995: 95–6; 2009a: 275–6), Polo (2004: 399–400).

154

From Latin to Romance

modifier/Spec(ifier) position (SpecC) hosts topicalized or focalized elements (see also Vincent 1997c: 169 n. 17):95 SpecC

C

IP . . .

(106) a trans uallem apparebat mons sanctus across valley appeared mount.NOM holy.NOM trans uallem apparebat (Lat., Per. Aeth. 1.1) ‘across the valley there appeared the holy mount’ b domus ut domus propugnacula et house.NOM that defences.ACC and praesidium habeat (Lat., Cic. Fam. 14.18) protection.ACC has.SBJV ‘in order that the house may have defences and guards’ c Ad uillam cum to farm when ‘when you reach the farmstead’

uenies you.come

ad uillam (Lat., Cato Agr. 1)

From examples like these, Vincent (1998a: 422–3) concludes that, despite an otherwise non-configurational syntax in which word order is constrained only by pragmatic principles, Latin exceptionally provides for a configurational superstructure at the level of the clause. The latter provides for two fixed positions, C(omplementizer) and SpecC(omplementizer), the latter filled by topicalized and focalized elements and the former increasingly targeted not just by complementizers in embedded clauses, but also by the finite verb in main clauses according to a pattern which would generalize in time producing a V2 syntax. Thus, although the Latin clause provides for a fixed configurational CP structure with respective left-peripheral modifier/ Spec(ifier) and head positions, the non-configurationality of the other areas of the grammar is revealed by the observation that SpecC can host indifferently both whole constituents (107a) and individual words of would-be constituents such as the noun pratum ‘meadow’ in (107b) which is stranded from its associated predicative adjectival modifier inrigiuum ‘irrigated’ (Vincent 1998a: 420, 422): (107) a [CP [Spec Pater familias] [C' ubi [IP pater familias when father family.GEN uillam uenit]]] (Lat., Cato Agr. 2) farm comes ‘When the master reaches the farmstead’

ad at

G. Salvi (p.c.) notes that in early examples such as (106c) subordinators such as UT ‘so that’, SI ‘if’, and ‘when’ could often behave as weak ‘Wackernagel’ elements attracted to the second position in the clause, a consideration which might explain the post-focal position in examples such as (106b). However, in examples such as (106c) and (107a) the subordinator follows not the first element of the clause (viz. ad ‘to’, pater ‘father’), but the whole syntagm suggesting a structural effect. 95

CUM/UBI

The rise of functional structure

155

b [CP [Spec pratum] [C' si [IP pratum inrigiuum meadow if irrigated habebis]]] (Lat., Cato Agr. 8) you.will.have ‘an irrigated meadowland if you have it’ Salvi (2004: 55–6, 94–8, 101–7; 2011: 356–8), on the other hand, offers a more nuanced view of the Latin data above. Specifically, he proposes that in early and classical Latin overt lexicalization of the C(omplementizer) position and its associated leftperipheral modifier/Spec(ifier) position with a focalized element were in strict complementary distribution. Consequently, if a single constituent was narrowly focused, it could be fronted to clause-initial position in the left periphery to occupy the Spec(ifier)C(omplementizer) position (Salvi 2005: 438–41; Spevak 2007),96 but the verb would remain within the sentential core (108a). When however the scope of focus did not range over a single constituent, but over the entire event (thetic sentences; 103a–b) or over the illocutionary force of the clause (optatives, jussives, concessives, assertives, interrogatives, imperatives; 104a–f), the verb could raise to the vacant C position to license the relevant marked pragmatic effect. In such cases, the associated left-peripheral modifier/Spec(ifier) position is argued to host a null/ abstract operator variously associated with a locative, temporal, or causal interpretation (narrative, thetic sentences; 108b) or with a particular illocutionary force (optative, jussive, concessive, assertive, interrogative, imperative; 108c), the content of which is licensed and made explicit by the verb raising to the C position (cf. also Polo 2004: 402; Devine and Stephens 2006: 157–72). (108) a [CP [Spec

MAGNAM] [C'  [IP

haec res Caesari magnam big this.NOM thing.NOM Caesar.DAT difficultatem ad consilium capiendum seizing difficulty.ACC to plan.ACC adferebat]]] (Lat., Caes. B.G. 7.10.1) caused ‘this matter caused Caesar great difficulty in forming his plan of campaign’

b [CP [Spec -OpTemporal] [C' conclamatur call.together.PRS.PASS.3SG arma conclamatur]]] (Lat., Caes. B.C. 1.69.4) arms ‘there is a call to arms’

[IP ad to

96

Latin also had a postverbal focus position (Pinkster 1991: 69–70; 1993b: 244; Bolkestein 2001: 249; Salvi 2004: 47; Spevak 2007), witness (i): turris altitudo perducta est AD CONTABULATIONEM (Lat., Caes. B.C. 2.4.5) (i) ubi when tower.GEN height.NOM conducted is to flooring ‘when the turret was raised to the height for flooring’

156

From Latin to Romance c [CP [Spec -OpOptative] [C' ualeant be.strong.PRS.SBJV.3PL mei ualeant!]]] (Lat., Cic. Mil. 93) my.NOM.MPL ‘may my fellow citizens fare well!’

[IP ciues citizens.NOM

Only in lower registers (from the second century AD) and in the later Latin period in the transition to Romance does the marked process of verb-fronting to C(omplementizer) become generalized in root clauses (Salvi 2004: 96–7, 107–11, 111), thereby integrating even narrow focus constructions into this new unmarked order such that the former complementarity between V(erb)-raising to C and overt raising to the latter’s left-peripheral modifier/Spec(ifier) position (cf. 108a) is now lost with simultaneous lexicalization of both positions (109a).97 At first, however, the left-peripheral modifier/Spec(ifier) position could only be occupied by fronted rhematic/focused constituents (cf. 109a), but from the sixth (possibly fourth) century onwards we also begin to find thematicized elements in preverbal position (109b), a fact which Salvi (2004: 110–11) interprets as strong evidence for the independent development of focus- and theme-fronting under V2.98 (109) a [CP [Spec LITEM] [C' habuit [IPPtolemes pater meus litem quarrel had.3SG Ptolemy.NOM father.NOM my.NOM sopera uestimenta mea habuit]]] (Lat., Terent. 254.20) over clothes.ACC my.ACC ‘my father Ptolemy quarrelled about my clothes’ b [CP [Spec purpurius Purpurius.NOM [IP purpurius episcopum

episcopus] bishop.NOM centum hundred.ACC

[C' tulit carried

folles tulit]]] (Lat., Gest. Zen. 194.31) purses.ACC ‘Bishop Purpurius received one hundred bags of silver’ 4.4.1.1 Archaic non-configurational pattern Although the early signs of the emergence of a CP projection in Latin are undeniable, the language does however

97 The rise of generalized verb-fronting in conjunction with fronted focused constituents in root clauses finds a natural explanation in terms of the analysis of ESSE ‘be’ fronting developed in }5.4.2.4, where it is shown how the distributional pattern of ESSE which, as a Wackernagel element, precisely seeks out a focal host to which to attach, could subsequently generalize to all other verbs. 98 Indeed, Salvi (2004: 67–8 n. 1) and Ledgeway (2008a: 449–51) discuss evidence for assuming that focus- and theme-fronting in early Romance target distinct left-peripheral Spec(ifier) positions.

The rise of functional structure

157

still preserve evidence of an earlier archaic stage predating the emergence of CP structure. In Indo-European, complex sentences did not involve subordination, namely a CP structure with an overt subordinator, but, rather, were expressed by paratactic, correlative, or participial constructions.99 Indeed, evidence of the archaic pattern is still evident in Latin,100 where alongside hypotactic subordination structures with overt subordinators like UT(I) ‘(so) that’ (110a), we still find correlative (110b) and paratactic (110c) patterns involving simple juxtaposition of two clauses. (110) a uolo I.want

uti that

mihi me.DAT

respondeas (Lat., Cic. Vat. 7.17) you.reply

b cui testimonium defuerit, is tertiis witness.NOM lack.FUT.PFV.3SG he third.ABL.PL who.DAT diebus ob portum days.ABL before doorway.ACC obuagulatum ito (Lat., Leg. XII Tab. 2.2) wail.SUP. will.go.FUT.IMP ‘he whose witness has failed to appear may summon him by loud calls before his house every third day’ c uolo etiam exquiras (Lat., Cic. Att. 8.12.6) I.want also you.search ‘I want you also to investigate’ Similarly, the accusative and infinitive (AcI) construction, arguably the most important complementation pattern of the classical language, demonstrably does not involve an embedded CP structure (cf. Calboli 2009: 122). For example, in a sentence such as (111a), the infinitival clause (laudauisse hasce . . . architectonem) and the finite clause (ait) are not only juxtaposed, but the latter breaks up the contiguity of the former, thereby destroying any continuous linear identification of the complement clause. Consequently, the only marker of subordination in such examples appears indirectly on the accusative-marked infinitival subject (e.g. architectonem). More specifically, the accusative marking of the infinitival subject cannot be determined by the matrix predicate, since the construction is found in conjunction with predicates such as CREDERE ‘to believe’ which canonically assigns dative to its complement

99 Palmer ([1954] 1990: 328), Haudry (1973), Bichakjian (1982), Calboli (1983b: 41–2), Lehmann (1984), Ramat (1994: 259–60, 264), Bauer (1995: 159–60; 2009a: 250, 258), Devine and Stephens (1999: 148). For a more cautious interpretation of the Indo-European evidence, see Clackson (2007: 173–6). 100 Bennett (1910: 208–9, 244–5), Ernout and Thomas ([1953] 1993: 300–1), Oniga (2004: 134–5).

158

From Latin to Romance

(Vincent 1994). Rather, the accusative case of the infinitival subject must be seen as a global property of the construction, ultimately the sole marker of the logical relationship between both clauses. But even here too we see from the earliest documents the slow and progressive emergence of configurationality in the marking of subordination at the level of the left periphery in the gradual expansion of rival finite complement clauses introduced by the overt complementizers QUOD/QUIA ‘because (> that)’ (111b).101 (111) a laudauisse hasce ait architectonem (Lat., Pl. Mostell. 760) praise.INF.PRFV these.ACC.PL he.says master.builder.ACC ‘he says that the master-builder has praised that (house)’ b scis enim quod epulum dedi (Lat., Petr. Sat. 71) you.know for that feast.ACC I.gave ‘for you remember that I gave a public banquet’ To conclude, from as early as the archaic Latin period there is extensive evidence of two conflicting patterns in the marking of the clause. The first represents an archaic non-configurational pattern inherited from the Indo-European parent language, in which a number of core complementation structures without overt subordinators, notably the AcI construction, manifestly do not involve a CP projection. The second constitutes an innovative configurational pattern, albeit attested since the archaic Latin period, in which a number of subordination types with overt complementizers (e.g. UT ‘so that’, QUOD/QUIA ‘because; that’), as well as an incipient V2 syntax, both frequently preceded by fronted topics and foci, make recourse to an articulated CP structure. 4.4.2 Evidence for Romance CP structure The CP structure reviewed above for Latin is further reinforced and extended in the transition to Romance, coming to permeate all structures of the emergent languages. In the first instance, as already outlined above in }3.3.2 and }4.4.1, this development most noticeably surfaces in the generalization of verb-fronting to the vacant C(omplementizer) position in main clauses as part of the late Latin/early Romance V2 syntax, generally accompanied, in turn, by fronting of one or more pragmatically salient constituents to the left periphery. However, it also surfaces indirectly in the loss of the AcI construction, one of the most notable causalities of the widespread development of CP structure. From an early date among non-literary authors

101 Perrochat (1932), Adams (1976b: 94; 1977: 61; forthc. a: }3.5), Bolkestein (1989), Calboli (1989b), Herman (1989), Cuzzolin (1994a, b), Bauer (1995: 165), Karlsen (2001), Ferraresi and Goldbach (2003), Greco (2007). The competition between the AcI construction and QUOD-/QUID-clauses is explored in detail in }5.3.1.1, }5.4.2.1.

The rise of functional structure

159

(Perrochat 1932), but not until the postclassical period in other text types, especially among Christian writers, the AcI construction was commonly replaced by a finite complement clause introduced by the complementizers QUOD and QUIA, a usage finally consolidated as the core complementation pattern in vulgar texts after the fall of the Empire.102 Clearly, there was no place in an emerging linguistic system with full configurational structure for a non-configurational complementation pattern such as the AcI, hence its eventual demise (for a different interpretation, see }5.4.2). A further area highlighting the consolidation of the CP projection is evidenced by the emergence in Romance of non-finite complementizers derived from the prepositions DE and AD to introduce infinitival clauses, which, as we shall see in }4.4.2.2, generally occupy a lower position within the left periphery than the finite complementizers derived from QUOD/QUID and QUIA. Evidence like this has led many researchers investigating the structure of the Romance left periphery, traditionally defined in terms of CP and its associated left-peripheral modifier/Spec(ifier) and head positions (Chomsky 1986: }1), to analyse it as a split domain, hierarchically articulated into several fields and associated projections.103 In what follows we shall review some of the evidence in support of this richly articulated representation of the C-domain, although limitations of space allow us to consider here only a small selection of the available evidence. 4.4.2.1 Topic and Focus fields The traditional assumption of a simple CP layer immediately above the sentential core forces us to assume that topicalized or focused elements target the same position, namely the left-peripheral modifier/Spec(ifier) of the complementizer phrase (SpecC). This assumption, however, runs into a number of empirical difficulties. For instance, it incorrectly predicts that fronted topicalized and focused constituents should occur in complementary distribution given the availability of a single position. Yet one does not need to look far to find evidence to the contrary. Already we saw in late Latin how in conjunction with generalized V(erb)-raising to the C(omplementizer) position constituent fronting under V2 was at first limited to focused elements and only subsequently extended to thematicized elements (109a–b), an unexpected asymmetry if both elements targeted the same position. Equally revealing in this respect is the first documented attestation of the 102 Ernout and Thomas ([1953] 1993: 304–5), Rohlfs (1969: 189), Väänänen (1982: 273), Bauer (1995: 165), Herman (1989; 2000: 88–9), Zamboni (2000: 119–20). 103 Benincà (1988; 1996; 2001; 2006; 2010), Campos and Zampini (1990), Dobrovie-Sorin (1994: 93–111), Duarte (1996), Rizzi (1997; 2001; 2004), Alboiu and Motapanyane (2000b: }4.2) Poletto (2000; 2001; 2005; 2006), Roehrs and Labelle (2003), Hill (2002; 2007), Munaro (2002; 2003), Paoli (2002; 2003a, b; 2005; 2007), Zagona (2002: 208–29, ch. 6), Chinellato and Garzonio (2002), Poletto and Zanuttini (2003), Belletti (2004a), Benincà and Poletto (2004), Garzonio (2004), Ledgeway (2004a; 2005; 2006; 2007b; 2008a; 2009b; 2010a, b; forthc. a), Salvi (2004), Azevedo (2005: 248–9), Damonte (2005; 2010), Morin (2005), Cruschina (2006; 2008; 2010), Demonte and Fernández Soriano (2007; 2009), Rowlett (2007: ch. 5), González i Planas (2009), López (2009), Benincà and Munaro (2010), D’Alessandro and Ledgeway (2010a).

160

From Latin to Romance

vernacular from the Italian peninsula, the Placito capuano, containing a brief, formulaic, sworn declaration dating from March 960 (see also Benincà 2002: 241; Ledgeway forthc. e):104 (112)

Sao ko kelle terre, per kelle fini que ki I.know that those lands for those confines which here possette parte sancti contene, TRENTA ANNI le contains thirty years them= possessed party of.saint Benedicti (OCmp., Plac. cap.) Benedict ‘I know that, those lands, within those borders which are contained here [in the document/map], have belonged for thirty years to the part [= monastery] of St Benedict [of Montecassino]’

Although an extremely brief glimpse of the early vernacular, it nonetheless contains for our purposes invaluable early evidence of the fine structure of the C-domain and, in particular, incontrovertible proof of the existence of at least two left-peripheral positions. Even within a theory in which multiple left-peripheral modifier/Spec(ifier) positions are allowed (Chomsky 1995; 2000; 2001), it is not immediately clear how the rigid Topic + Focus ordering in (112) is to be captured. Moreover, this example, in which the contrastively focused constituent TRENTA ANNI ‘thirty years’ is preceded by the two topicalized constituents per kelle fini que ki contene ‘within those borders that are contained here’ and kelle terre ‘those lands’, highlights that even the postulation of two left-peripheral positions is not sufficient.105 Rather, the relevant positions must be reconceived as distinct pragmatico-syntactic spaces along the lines of Benincà and Poletto (2004), according to which we can identify from left to right at least two fields termed Topic and Focus, respectively. Not only is this demarcation between Topic and Focus justified at a pragmatico-semantic level, in that elements appearing in the Topic field are generally interpreted as ‘old’ or ‘given’ information whereas the Focus field is typically associated with informationally ‘new’ elements, but it also finds considerable confirmation at the syntactic level. For instance, in contrast to elements appearing within the Topic field, which often call

104 In all examples that follow, topicalized constituents are underlined, contrastive foci appear in small capitals and non-contrastive foci in bold. All other salient categories appear in italics. 105 Of course, we are dealing with written texts, the pragmatico-semantic interpretation of which is notoriously difficult to reconstruct and open to subjective interpretation. For example, G. Salvi (p.c.) notes that trenta anni ‘thirty years’ in (112) could be a case of informational, rather than contrastive, focus. We, however, take quantifiers like trenta ‘thirty’ and other elements with a clear scalar interpretation (e.g. grande ‘big’) to invole an implicit contrast with all other relevant values (e.g. venti ‘twenty’, piccolo ‘small’).

The rise of functional structure

161

for a resumptive pronominal clitic where available (cf. 113a),106 those appearing within Focus (cf. 113b) typically prove incompatible with a pronominal copy (Benincà 2001: 43–4): (113) a Mario, de so sorela, *(el) ghe ne Mario of his sister SCL= there= of.her= ‘Mario is always talking about his sister’ b alte studii nu (*le) a other studies not them= he.has ‘he wasn’t able to study anything else’

parla speaks putut been.able

sempre (Pad.) always face (Ro.) do.INF

Additional evidence for this strict structural demarcation between the Topic and Focus fields comes from the distribution of clitic placement in medieval Romance,107 which generally display enclisis following topicalized constituents (cf. 114a) and proclisis in conjunction with constituents fronted to the Focus field (cf. 114b): (114) a [Topic depois que o moço foi criado after that the child was raised [Focus  [CP levou- [IP levou o Ana carried =him Hannah a Sylo]]]] (OPt., A.Test. Sam. 1.1.24, in Kaiser 1999: 254) to Shiloh ‘after she had weaned the child, Hannah took him to Shiloh’ b [Focus Entom [CP lhe disse [IP nostro then to.him= said our Senhor lhe disse entom]]] (OPt., A.Test Sam. 1.3.11, in Kaiser 1999: 254). ‘Then our Lord said to him’ Lord These facts find a straightforward explanation in terms of the traditional ToblerMussafia Law,108 one of the principal generalizations of which states that enclisis obtains whenever the verb occurs in clause-initial position. Thus, in the case of focus fronting in (114b), proclisis invariably obtains since the verb (raised to the lower C(omplementizer) head CFin under V2) occurs in second position preceded by a fronted constituent in the Focus field. However, whenever the topicalization space hosts a hanging topic or a left-dislocated constituent and the Focus field remains empty (cf. 106 Also relevant here is the use of the resumptive clitic le ‘them’ referencing the topic kelle terre ‘those lands’ in (112) above. 107 Benincà (1994c: 228–38; 2002: 243–4; 2006: 67–8; 2010: 54–9), Salvi (2004: 152–6), Poletto (2005: 226), Ledgeway (2007b: 131–4; 2008a: 443). 108 Ultimately a Romance-specific reanalysis of the Wackernagel Law (see Benincà 1995; Wanner 1996; Vincent 1998a: 422).

162

From Latin to Romance

114a), only enclisis is possible because the verb now raised to CFin technically occurs in clause-initial position, inasmuch as elements contained within the Topic space are extra-sentential and hence invisible in the computation of the Tobler-Mussafia generalization. In short, we interpret the observed proclisis-enclisis alternation as a side effect of V2 fed by verb raising to CFin, which creates either a V1 structure and enclisis with no fronting to the Focus field (114a) or a V2 structure and proclisis with fronting to the Focus field (114b). Robust evidence like this demonstrates that topicalized and focused constituents indeed target distinct spaces within the left periphery, forcing us to recognize a representation of the C-domain along the lines of (115): (115)

[CP Comp [TopP Top [FocP Foc [C-FinP Comp/V2 [IP . . . ]]]]]

4.4.2.1.1 internal structure of topic and focus fields 4.4.2.1.1.1 Focus field The evidence of Romance suggests the existence of, at most, a single focus position specialized in licensing contrastively focused interpretations (typically correcting a previous assertion; 116a), although some varieties such as French and Turinese (Lambrecht 1994; Paoli 2003a) appear not to license any left-peripheral focus position (116b–c). (116) a LA CAMISA BLANCA quería (no la the shirt white I.wanted not the ‘I wanted the white shirt (not the red one)’

roja) (Sp.) red

b **LA CLÉ j’ ai perdue/ c’est la clé que j’ ai perdue (Fr.) the key I have lost it.is the key that I have lost ‘it’s the key that I lost’ c (**IL GELATO) a l’ ha catà IL GELATO, nen la torta (Tor.) the ice-cream SCL= has bought the ice.cream not the cake ‘It is the ice-cream that he bought, not the cake’ In the medieval period, however, the Focus field was much more productive across Romance (Skårup 1975; Vanelli 1986; 1999), licensing during the V2 phase not just contrastive focus (117a), but also informationally non-contrastive focused interpretations (117b), a structure still frequently attested today in varieties such as Romanian, Sardinian, Sicilian, and Triestino (117c–d).109 109 Jones (1993: 332–45), Cruschina (2006; 2008; 2010), Bentley (2007), Mensching and Remberger (2010a), Paoli (2010). The complex structure of the Focus field is further substantiated by Romanian which, exceptionally among the Romance languages, allows multiple wh-fronting in root interrogatives on a par with Slavonic languages (Rudin 1988; Alboiu and Motapanyane 2000b: 29–30):

spusese? (Ro.) (i) Cine ce who what had.said ‘Who had said what?’

The rise of functional structure (117) a AUTRE CHOSE ne pot li roi other thing not could the king ‘the king could not find anything else’

trouver (OFr., MRA 79.24) find.INF

b Com tanta paceença sofria ela with so.much patience suffered she enfermidade (OPt., DSG 4.13.13) illness ‘She suffered this illness with great endurance’ c Ce iei? – O bere what you.take a beer ‘What will you have? – I’ll have a beer’

163

esta this

iau (Ro.) I.take

d A cu i dasti i chiavi? A Salvu i detti (Sic.) to who them= you.gave the keys to Salvu them= I.gave ‘Who did you give the keys to? I gave them to Salvu’ Although the contrastive and informational foci in (117a-d) might appear to move to the same left-peripheral position, as suggested by the fact that they can never cooccur, there are good reasons to believe that they target distinct positions. More specifically, we can view the Focus space as a hierarchically structured field, articulated from left to right into the subfields of Contrastive Focus (CFoc) and Informational Focus (IFoc) which provide dedicated positions for contrastively and informationally focused constituents.110 This distinction is supported, among other things, by the observation that in those varieties that display both types of foci only contrastive focus, but not informational focus, is compatible with embedded contexts (Cruschina 2010: 255–6; Paoli 2010: 281–3), witness the Triestino contrast in (118a–b) from Paoli (2010: 282–3). (118) a (DA at i SCL.3PL=

MONTI) Monti gavessi had

se self= meio better

pensava thought roba, stuff

che that no not

DA

at al’ at.the

MONTI) Monti Emporio (Trs.) Emporio

‘It is AT MONTI’S that we thought they had better things, not at the Emporio’ b Una pelicia me preocupa che (?una pelicia) a fur me= worries that a fur se cioghi (Trs.) self= buys ‘I’m worried that she will buy herself a fur coat’ 110

la SCL.3FSG=

Benincà (2002: 238–9), Rizzi (1997), Kiss (1998), Belletti (2001a; 2004a), Benincà and Poletto (2004), Cruschina (2008: ch. 3).

164

From Latin to Romance

Similarly, in these same varieties strict adjacency between the verb and the focused constituent is only required in the case of the lower informational focus position,111 as revealed by the following Sardinian contrast (taken from Cruschina 2008: ch. 3): (119) a

DURCHES, a su pitzinnu appo the sweets to the child I.have puliches (Srd.) fleas ‘I bought sweets for the child, not fleas’

comporadu, Bought

SOS

b **Retzidu dae Predu as received from Predu you.have ‘Did you receive the gift from Predu?’

su the

no not

sos the

regalu? (Srd.) gift

This same lower focus position also appears to be involved in hosting (non-D-linked) wh-interrogatives, since these too require strict adjacency with their associated inflected verb (Munaro 2010: 293–5): (120)

Al marcà quanti libri (*al marcà) at.the market how.many books at.the market comprà? (Bel.) bought? ‘At the market, how many books did you buy?’

avé-o have.2PL=2PL.SCL

As subfields, however, even CFoc and IFoc can be further dissected to reveal additional positions within these spaces. For example, following Poletto and Zanuttini (2000), Benincà and Poletto (2004: 61) note in the Ræto-Romance V2 dialect of S. Leonardo a differential licensing of contrastively focalized constituents in embedded contexts. More specifically, in complements to verbs of low assertion (viz. non-bridge verbs) only circumstantial and quantificational adverbs such as da trai ‘sometimes’ can be fronted under contrastive focus (121a), whereas other adverbial types such as d sigy ‘for sure’ (121b) and other categories such as objects like l giat ‘the cat’ prove ungrammatical (121c): (121) a al s DA TRAI l cruzie c he self= worries that sometimes him= ‘he is worried because he saw him sometimes’

a-al odù (S.Leo.) has=he seen

b **al s cruzie c D SIGY mang-ela a ciasa (S.Leo.) he self= worries that for sure eats=she at home ‘he is worried because she is going to eat at home for sure’

111

Cruschina (2010: 252–5), Mensching and Remberger (2010a: 271–3), Paoli (2010: 288–9).

The rise of functional structure

c **al s cruzie c L GIAT a-al he self= worries that the cat has=he ‘he is worried that it was the cat he has seen’

165

odù (S.Leo.) seen

Under the usual assumption that non-bridge verbs do not select a full CP layer, the contrast in (121a) vs (121b–c) follows straightforwardly: the CP space is pruned below the focus projection(s) which license the contrastive reading of particular adverb classes and objects, but retains the lower focus projection(s) responsible for licensing the contrastive reading of circumstantial and quantificational adverbs. A similar split within the IFoc space is evidenced by the behaviour of indefinite quantifiers (IndefQ; Benincà and Poletto 2004: 62–3; Cruschina 2008: ch. 3, }3.5.1). Although not all Romance varieties permit fronting of informationally new, noncontrastive constituents (cf. 122a), many varieties do readily allow fronting of indefinite quantifiers even in the absence of a contrastive reading (cf. 122b): (122) a ¿Qué te quieres comer? – what yourself= you.want eat.INF comeré una pera (Sp.) I.will.eat a pear ‘What do you want to eat? – I’ll eat a pear’ b Algo me something myself= ‘I must eat something’

tengo I.must

que that

(**Una a

pera) pear

me myself=

comer (Sp.) eat.INF

c creo que algo podrá hacer (Sp.) I.believe that something he.will.be.able do.INF ‘I think that he will be able to do something’ Just like other foci, fronted indefinite quantifiers such as algo in (122b) prove incompatible with clitic resumption and on a par with informational focus, but unlike contrastive focus, must stand strictly adjacent to the verb. However, the contrast in grammaticality of informational focus-fronting and quantifier-fronting in varieties like Spanish evidenced in (122a–b)—widely attested in many other varieties, including northern Italian dialects where bare quantifiers in subject function always target the Focus field—suggests that non-contrastive focus should be dissected into at least two distinct positions. Indeed, this splitting of the noncontrastive focus space is further substantiated by examples like (122c), which demonstrate that fronted indefinite quantifiers, unlike canonical informational foci (cf. 118b), can be fronted even in embedded contexts. We might tentatively interpret this contrast as indicative of a lower position for indefinite quantifiers which is not pruned in embedded contexts, as is the higher position dedicated to canonical informational foci.

166

From Latin to Romance

At this point in our discussion, we can thus sketch the following extended representation of the focus field within the left periphery (curly brackets indicate fields): (123) [CP Comp [TopP Top {Focus [CFocP1 Obj/Adv [CFocP2 Advcircum./quantif. [IFocP1 IFoc [IFocP2 IndefQ [C-FinP Comp/V2 [IP . . . ]]]]]]}]]

4.4.2.1.1.2 Topic field One does not need to look far to find evidence for the complex internal structure of the Topic field. In addition to early examples like (112) above, modern Romance abounds in structures with multiple topics: (124) a Henri, au ciné, il y Henri to.the cinema he= there= ‘Henri doesn’t go the cinema anymore’

va goes

plus (Fr.) no.more

b mie banii mi-i aduce poştaşul to.me money=the me=them= brings postman=the ‘the postman brings me the money to my house’ c E tui sa fà pappàda and you the bean eaten ‘And have you eaten the beans?’

ti yourself=

dd’ it=

acasă (Ro.) to.home

hasi? (Srd.) you-have

fijə mia lo spumante a capodanno dI the kids my the spumante at New.Year’s.Eve biudo (Sas.) drunk ‘my children drunk spumante on New Year’s Eve’

l’ it=

a has

Evidence such as this has led many to suggest a number of further subdivisions within the Topic field, the most significant of which is that between the Frame and Theme subfields (Benincà and Poletto 2004: }3.1; Benincà 2006: 54–8). In pragmaticosemantic terms, these two subfields differ in that the former defines the ‘frame’ to which the sentence refers, including its spatio-temporal coordinates, while the latter defines the entities which the sentence is about, including the theme of predication and other anaphoric constituents taken to express shared knowledge (Chafe 1976: 50). The pragmatico-semantic primacy of Frame with respect to Theme is also reflected at the syntactic level in the obligatory ordering Frame + Theme (125a). Within the former we can recognize hanging topics (HT) and scene-setting adverbials (Advsc.-set.), with hanging topics situated above scene-setting adverbs (125b). Besides their left-most position, hanging topics are distinguished from other topic elements in being restricted to a single occurrence per sentence and in invariably surfacing as DPs (Benincà 2001: 43), their syntactic function being obligatorily signalled by a resumptive pronoun or epithet within the sentential core.

The rise of functional structure

167

(125) a La Maria d’ aquest tema no en parlis amb ella (Cat.) the Maria of this subject not of.it= talk.SBJV2SG with her ‘Don’t talk to Maria about this subject’ b io ’a quanno è muorto pàteto I from when is died father=your staie ntussecanno ’a vita mia! (Nap.) you.are poisoning the life my ‘Since your dad died, you’ve been poisoning my life!’

me me=

As for scene-setting adverbs, their position within the higher portion of the left periphery finds support in the V2 variety of S. Leonardo considered in (121) above (Benincà and Poletto 2004: 66). Whereas in root clauses a scene-setting adverb such as the temporal duman ‘tomorrow’ proves entirely grammatical when fronted under topicalization or contrastive focus (cf. 126a), this is not the case for the topicalized reading in embedded clauses (cf. 126b), even when selected by a bridge verb, an observation which suggests that the top ‘frame’ layer of the CP has been pruned in these cases since its semantics is fundamentally incompatible with the informational structure of subordination: (126) a duman / tomorrow

DUMAN TOMORROW

va-al goes=he

a to

Venezia (S.Leo.) Venice

b Al m a dit c *duman / DUMAN he me= has told that tomorrow TOMORROW a Venezia (S.Leo.) to Venice ‘(He told me that) he is going to Venice tomorrow’

va-al goes=he

Turning now to the Theme subfield, here too we need to recognize several subtypes of thematicized constituent, the number of which, unlike topic elements occurring in the Frame subfield, is in principle unrestricted, although subject to the pragmatic intentions of the speaker to repeat or re-establish particular anaphoric constituents of the previous discourse or to introduce cognitively accessible referents considered to form part of the interlocutors’ shared knowledge. Syntactically, all elements occurring in the Theme subfield are generally referenced by a resumptive clitic pronoun, where available, constituting a case of clitic left-dislocation (ClLD; Cinque 1990): (127)

ieu, sus la plaja, i passe pas jamai lei vacanças (Prv.) I on the beach there= I.spend not ever the holidays ‘I never spend my holidays on the beach’

168

From Latin to Romance

As for the topic types occurring in Theme, there is no general consensus as to the number of distinct positions involved and their precise pragmatic interpretations.112 A broad distinction between aboutness topics (ATop) and all other types of (clitic) left-dislocated topics (sometimes termed referential or familiarity topics) is, however, widely recognized. The former represent what the (categorical) sentence is about, thus standardly equated with the subject of predication and hence limited to a single occurrence per clause and the preverbal position, whereas the latter re-establish contextually given referents belonging to the previous discourse, hence optional and unlimited in number and occurring in both pre- and postverbal position. As a general principle, then, aboutness topics precede all other topic elements in the Theme field, as witnessed by the order of clitic left-dislocated subject and locative complement in the Provençal example in (127). In many varieties, these different topic categories are distinctly marked. For instance, Ledgeway (2010b) demonstrates that, as part of a topic-announcing or topic-shifting strategy, the aboutness topic in Campanian dialects is encoded by a doubling distal demonstrative chillo ‘that.one’ (128a), whereas Cruschina (2006; 2008: ch. 1, }2.1) proposes for Sicilian a strict principle of Syntactic Extraposition (SE), which requires all [-focus] constituents to be obligatorily dislocated to dedicated functional positions (128b): San Pietro, ’o (128) a Chillo, him= that.one.M Saint Peter scianche, a Gesù (Grz.) hips, to Jesus ‘Saint Peter tapped Jesus on his hip’

tuzzuliaie tapped

bA Maria ci dissi ca pitrusinu to Maria to.her= I.said that parsley jardinu un ci nn’ avi a garden not there= of.it= she.has to ‘I told Maria not to plant parsley in the garden’

rint’ in

’e the

n’ u in the chiantari (Sic.) plant.INF

Putting together the results of our discussion so far, the richly articulated functional structure of the left periphery can be represented schematically as in (129): (129) [CP Comp {Topic [FrameP1 HT [FrameP2 Advsc.-set. [ThemeP1 ATop [ThemeP2 ClLD {Focus [CFocP1 Obj/Adv [CFocP2 Advcircum./quant. [IFocP1 IFoc [IFocP2 IndefQ [C-FinP Comp [IP . . . ]]]]]]}]]]]}]

112

For an overview, see Benincà and Poletto (2004: 64–70), Frascarelli and Hinterhölzl (2007), Cruschina (2008: ch. 1, }1.5).

The rise of functional structure

169

4.4.2.2 Force and Finiteness A further area highlighting the consolidation of the CP projection is evidenced by the emergence in Romance of non-finite complementizers derived from the prepositions DE ‘of, from’ and AD ‘to’ to introduce infinitival clauses, which to all intents and purposes parallel the use of finite complementizers derived from QUOD/QUID and QUIA ‘(because >) that’ to introduce tensed clauses (130a–c): (130) a Digues-li [CP [C' que [IP vingui]]] / tell=him that he.comes.SBJV ‘Tell him that he come/to come’

[CP [C' de [IP venir]]] (Cat.) of come.INF

che [IP doveva Ugo [CP [C' Ugo that he.had.to a[IP tornare]]](It.) to return.INF ‘I convinced Ugo that he should return/to return’

b Convinci I.convinced /[CP [C'

c Elle contrôle le courrier avant she checks the post before de [IP sortir]]] (Fr.) sorte]]] / [CP [C' exit of exit.INF ‘she checks her mail before I leave/leaving’

tornare]]] return.INF

[CP [C' que [IP je that I

ne not

However, despite appearances, the presumed parallelism between finite and nonfinite complementizers is not perfect, as revealed by their respective positions in relation to topics: (131) a So I.know

che, that,

la the

data, date,

l’ it=

b So, la data, di averla I.know, the date, of have.INF=it ‘I know (that), the date, I got (it) wrong’

ho I.have

sbagliata (It.) mistaken

sbagliata (It.) mistaken

Evidence like this forces us to assume that the Topic and Focus fields outlined above are, in turn, closed off upwards by a complementizer position Force marking the illocutionary force of the clause, hosting such items as the Italian finite declarative complementizer che ‘that’ (131a), and downwards by a complementizer position Fin(iteness) specifying the modality and/or finiteness of the clause (Rizzi 1997), hence also the position targeted by the finite verb under V2 (cf. 114a–b; Ledgeway 2007b; 2008a), hosting such items as the Italian infinitival complementizer di ‘of ’ (131b), as schematicized in (132). (132) [ForceP che [TopP Top [FocP Foc [FinP di [IP . . . ]]]]] Indeed, some Romance varieties present dual finite complementizer systems which appear to exploit both the higher and lower complementizer positions within the left

170

From Latin to Romance

periphery. Such is the case in Romanian and many southern Italian dialects,113 which contrast an indicative/declarative complementizer (QU(I)A ‘because’ >) că/ca that lexicalizes the higher Force position, and therefore precedes topics and foci (133a, 134a), and a subjunctive/irrealis complementizer (SI ‘if ’ >) să and (QUOD ‘because’ >) cu that lexicalizes the lower Fin(iteness) position, and therefore follows topics and foci (133b, 134b): (133)

Ion a sunt sigură că pe I.am sure that ACC.MRK Ion de anul trecut (Ro.) of year=the past ‘I’m sure that you saw Ion last year’

lhim=

b vreau MÂINE să meargă la I.want tomorrow that he.go to ‘I want him to go to the match tomorrow’ (134)

a aggiu tittu ca I.have said that ‘I said that Lia is coming’

la the

b oyyu lu libbru cu lu I.want the book that it= ‘I want Mario to buy the book’

Lia Lia

ai you.have

văzut seen

meci (Ro.) match ene (Sal.) comes

ccatta buys

lu the

Maryu (Sal.) Mario

Such dual complementizer systems are not simply restricted to Romanian and Salentino, but are widely found in the dialects of southern Italy, as illustrated in Table 4.10 (Rohlfs 1969: Ledgeway 2004a; 2005; 2006; 2009b; forthc. a): TABLE 4.10 Dual complementizer systems in the dialects of southern Italy Italian

penso che verrà

voglio che lui mangi

‘I think that he will come’

‘I want that he should eat’

Sicily

pensu ca vèni

vògghiu chi mmanciassi

Sicily (prov. of Messina)

critu ca vèni

ògghiu mi mancia

Southern Calabria

pensu ca vèni

vogghiu mu (mi) mangia

Northern Calabria

criju ca vèni

vuogliu chi mmangia

Salento

crisciu ca vène

ogghiu cu mmancia

Naples

pènsə ca vènə

vògliə chə mmangə

Northern Apulia

pènsə ca vènə

vògghiə chə mmangə

Abruzzo

pènsə ca vènə

vòjjə chə mmangə

113 For Romanian, see (Dobrovie-Sorin 1994: 93–111), Alboiu and Motapanyane (2000b: 3–5, }4.2), and for southern Italy Calabrese (1993), Lombardi (1997; 1998), Ledgeway (1998; 2004a; 2005; 2006; 2007c), Damonte (2002; 2005; 2006a, b; 2010), Roberts and Roussou (2003: 88–97), Manzini and Savoia (2005 I: 455–501, 650–76), Vecchio (2010).

The rise of functional structure

171

Further compelling evidence for these two complementizer positions comes from those varieties which allow the simultaneous lexicalization of both positions around a fronted topic or focus constituent, including many early Romance varieties (135a–d),114 and a number of modern varieties including Abruzzese (135e; D’Alessandro and Ledgeway 2010a), Ligurian (135–6; Paoli 2002; 2003a, b; 2005), and modern Ibero-Romance varieties (135g–h).115 Furthermore, Vincent, Bentley, and Samu (2004) and Vincent (2006: 12) show on the basis of early Sardinian evidence that these so-called recomplementation structures are not limited to finite contexts, but are equally found in cases of infinitival complementation with the non-finite complementizer de ‘of’ (135i). (135) a je te adjure par le vray Dieu que ta fille I you= beseech by the true God that your daughter ne la donnes a mariage a autre Tarsienne, que tu Tarsienne that you not her= give to marriage to other que a moy (OFr., RADT f48b) than to me ‘I beseech you before God that you may give your daughter in marriage to me alone’ b ti faci meravilla ched io ti iscrisi che yourself= you.do wonder that I you= wrote that se tu avei pió denari che me if you had more money that me= li mandasi (OTsc., LP 33–4) them you.send ‘you are surprised that I wrote to you that, if you had any more money, you should send me it’ c Onde dize Josepho que en casa de so padre where says Josepho that in house of his father que le llamaron primiera mientre Ciro (OSp., GE 177r.2.6) that him= they.called firstly -ly Ciro ‘Whereupon Josepho said that in his father’s house he was called originally Ciro’ dÈ da sape(re) ch(e) is from know.INF that diligentem(en)te custodito et diligently cared.for and 114 115

lu cavallo b(e)n the horse well a(m)modato cavalcato, properly ridden

Wanner (1998), Paoli (2003a), Ledgeway (2004a: }4.3.2.2; 2005: 380–9), Vincent (2006). Demonte and Fernández Soriano (2009), Gupton (2010: 227–34), Villa-García (2010).

et and così thus

172

From Latin to Romance

como se (con)vè, ch(e) illo no(n) sia fatigato de as self= is.right that he not be tired of grande et sup(er)flua travaglia (OLaz., VMLR 158.27-9) big and superfluous work ‘It is to be noted that, a horse (which is) well and attentively cared for and properly ridden in accordance with good practice, it should not be overburdened with too much unnecessary work’ a Urtone, gni lu zie, e so ditte ca dumane, I.am said that tomorrow to Ortona with the uncle ca nin gi da’ ji (Arl.) that not there= he.must go.INF ‘I told him that tomorrow he shouldn’t go to Ortona with his uncle’ f A Teeja a credda che a Maria SCL believes that the Mary the Teresa parta duman (Lig.) leaves tomorrow ‘Teresa believes that Mary will leave tomorrow’

ch’ that

a SCL

g dijeron que en esa discoteca que they.said that in this disco that Juan lo vieron ayer (Sp.) Juan him= they.saw yesterday ‘they said that they saw Juan in that disco yesterday’

a

irmán de Iago que h Din que o they.say that the brother of Iago that xogar moi ben ao futbolín (Glc.) play.INF very well to.the football ‘they say that Iago’s brother plays football very well’

sabe knows

ACC.MRK

i Conporait Istephane Unkinu a Barusone de Nurki sa bought Istephane Unkinu to Baruscone of Nurki the parte sua dessu salto de Conca maiore, part his of.the wood of Conca Maggiore et ego deiuili .ij. uaccas in .viij. sollos, and I gave=him 2 cows in 8 coins in fine de si lu perdea custu, de in aim of if it= he.lost this of torraremi saltu (OLog., CSPS, Bonazzi 1997: 182) wood return.INF=me

The rise of functional structure

173

‘Stefano Unchino bought from Barusone of Nurki his share of the woody terrain of Conca Maggiore, and I gave him 2 cows worth 8 coins for this, so that if I were to lose this, he would give me the woody terrain in return’ A not too dissimilar distribution of the two complementizers is found in many modern Salentino varieties (Damonte 2006a; 2010; Vecchio 2010), where the lower irrealis complementizer cu is replaced by the higher complementizer ca whenever the left periphery hosts a fronted constituent, as in the following Salentino examples: (136) a vogghiu I.want

(*Carlu) Carlu

cu that

(*Ccarlu) Carlu

b vogghiu (*Carlu,) ca Carlu, I.want Carlu that Carlu ‘I want Carlo to come with us’

vveni comes veni comes

cu with cu with

nnui, us

Carlu (Frc.) Carlu

nnui (Frc.) us

The postulation of the higher and lower complementizer positions also provides an elegant explanation for cases of embedded V2, which in the simple CP model were difficult to accommodate without reference to ad hoc assumptions such as the reinterpretation of V2 as V-to-I(nfl) movement (Santorini 1995; Vikner 1995: }4.2.1) or CP recursion (Authier 1992; Vickner 1995; Vance 1997: ch. 4). Now within the split C-model the co-occurrence of an overt complementizer and the raised finite verb can be viewed as simultaneous lexicalizations of the Force and Fin heads, respectively (Ledgeway 2007b: 139–40; 2008a: 458–61), as in the Old Neapolitan example in (137): che [TopP lo (137) resoltande certa speranza [ForceP emerged=therefrom certain hope that the Priamo [FinP poterrà [IP nde tpoterrà recoperare re king Priamus will.be.able =therefrom recover.INF la soro soa]]]] (ONap., LDT 102.26) the sister his ‘therefrom has come certain hope that King Priamus will be able to rescue his sister from there’ Finally, unique within Romance is the situation encountered in Gascon since around the sixteenth century, where the [+finite] feature of affirmative root clauses is exceptionally spelt out in the systematic lexicalization of the lower complementizer position (Fin) through the complementizer que ‘that’.116 Firm proof that que spells 116 Rohlfs (1970: 205–7), Wheeler (1988b: 272–4), Sauzet (1989), Joseph (1992), Lafont (1991: 16–17), Ravier (1991: 90–1), Bec (1967: 47–8). Some Gascon dialects also generalize the use of enunciative que to negated root clauses (Rohlfs 1970: 208):

174

From Latin to Romance

out the lower complementizer head is provided by the observation that, apart from object clitics, nothing can intervene between que and the finite verb such that all preverbal lexical subjects must occur to the left of que. This latter observation highlights the fact that, unlike in other Romance varieties, there is apparently no preverbal subject position within the sentential core in Gascon, such that all subjects have to be syntactically fronted to a topicalized (typically if old and definite; 138a) or focalized (typically if new and indefinite; 138b) position within the left periphery. Presumably, null subjects (pro) also target a left-peripheral topicalized position for these clauses are also introduced by que (138c), as are clauses with non-referential null subjects which apparently lack any subject position whatsoever, be it in the sentential core or the left periphery (138d). These facts find an immediate explanation in the assumption that que lexicalizes the Fin position, since finiteness is standardly assumed to license nominative Case, hence Gascon would appear to have grammaticalized the locus of finiteness and, by implication, the licensing of nominative subjects within the left periphery (as happens in V2 varieties), rather than in the sentential core. (138) a [TopP ta pay [FinP qu’ your father that ‘your dad’s arrived’

[IP ey is

b [FocP quauque trufandèr [FinP que whatever joker that ‘any joker will tell you . . . ’ c [TopP pro

[FinP que [IP t that you ‘we’ll leave you alone’

[IP vos dirà . . . ]]] (Gsc.) you= will.say

dechàm we.leave

d [FinP que [IP calè que that it.was.necessary that ‘they had to embark’

arribat]]] (Gsc.) arrived

soul]]] (Gsc.) alone

s’ selves=

embarquèssen]] (Gsc.) they.embarked

Generally, the lower que is not lexicalized in embedded clauses (139a–b), therefore making it impossible to tell whether preverbal subjects are situated in the left periphery, as in root clauses, or whether we need to assume that the sentential core exceptionally provides for a subject position in embedded contexts. However, (i)

a mes arrrés que n’ ou ne dabon (Oss.) but nobody that not him= of.it= gave ‘but nobody would give him any of it’ que nou plòy (Brg.) b oe today that not it.rains ‘it won’t rain today’

The rise of functional structure

175

embedded examples with lexicalization of the lower que can be found if the topic or focus fields host one or more fronted constituents (139c–d), giving rise to a pattern reminiscent of the recomplementation structure observed in (135a–i). The existence of examples like (139c–d) might therefore be taken to suggest that even in examples such as (139a–b) the preverbal subject is always topicalized or focused. (139)

a Que vos i hè un hred qui pela that you= there= makes a cold which peals [FinP/IP lo vent boha]] (Gsc.) the wind blows ‘It gets frightingly cold there whenever the wind picks up’ b [CP Com [FinP/IP lo mètge n’ avè pas as the doctor not had not guarir eth medish . . . (Gsc.) cure.INF he self ‘As the doctor was unable to cure himself, . . . ’

[CP quan when

poscut been.able

se self=

c Que ’m cau préner l’ autà, [ForceP per’mor since that me is.necessary take.INF the car [TopP Montpelhèr [FinP qu’ [IP ei lonh]]]] (Gsc.) Montpellier that is far ‘I must take the car, as Montpellier is far away’ d Los senhors Occitans que volèn ganhar Prètz, the men Occitan that wanted win.INF Prètz [TopP los Francés [ForceP mentre que while that the French ac volèn préner tot]]]] (Gsc.) [FinP que [IP s’ that selves= thus wanted take.INF all ‘The Occitan people wanted to achieve Prètz [= self-worth], whereas the French wanted to take everything’ Research still needs to be done to establish the origin of this typically Gascon usage, although one possible line of explanation (cf. Rohlfs 1970: 207–8) is to trace it to an original explicative or exclamative use of the Romance finite complementizer, where it often serves to mark strong illocutionary force (typically assertion, negation of a previous presupposition, or evidentiality; D’Alessandro and Ledgeway 2010a: 6 n. 6; D’Alessandro and Di Felice 2010), common to many other Romance varieties and especially Ibero-Romance:117 117 Some Corsican varieties mark a formal distinction in the complementizer employed to introduce declarative and adhortative/exclamative clauses: chì in the former case (i.a), generated in the higher Force position, and chè in the latter case (i.b), generated in the lower Fin(iteness) position:

176

From Latin to Romance

(140)

a ¡Echala, tozudo! – ¡Que no la throw=it stubborn that no it= ‘Don’t be stubborn, throw it way! – ‘No, I won’t!’

echo! (Ara.) I.throw

b Ca nin chischə! / Chi nin chischə! (Arl.) that not you.fall that not you.fall ‘(Don’t worry,) you won’t fall!’ / ‘(Watch out,) you might fall!’ c Senyer, mon fiyll que s’ lord my son that self= ‘Sir, my son has injured himself ’

es is

naffratz (OCat., SS v.1267) injured

d Iu, parola mia d’ onore, I word my of honour sacciu spiegà ch’ è successu (Cos.) I.know explain.INF what is happened ‘I give you my word, I can’t explain what happened’ e ¡Que no quiero that not I.want ‘I don’t want wealth!’

ca that

nun not

mi me=

riqueza! (Sp.) riches

4.4.2.3 Other projections In the literature, many other projections and positions have been proposed in investigations of the left periphery of Romance. In this respect, mention should be made of clause typing projections such as those discussed in Munaro (2004), who reconceives Force as a field consisting of as many as four distinct clause typing projections sandwiched between the Frame and Theme subfields, namely Conc(essive)P > Hyp(othetical)P > Excl(amative)P > Int(errogative) P, the order of which is held to mirror from right to left an increasing degree of assertive force. On the evidence of Ræto-Romance, northern Italian dialects, and Italian, Munaro demonstrates that the clausal adjuncts of concessive and conditional clauses raise to the left-peripheral modifier/Spec(ifier) position of a ConcP and HypP in the matrix C-domain, where they enter into a local relation with that particular Force projection:

(i)

a Dì sì Diu vole, saremu in casa (Cor.) a Caccara chì, à ott’ ore tell.IMP2SG to Grandma that at eight hours if God wants we.shall.be in house ‘tell Grandma that, God willing, we shall be home by eight o’clock’ un caschete! (Cor.) b Chè vo that you= not fall ‘Watch you don’t fall’

The rise of functional structure (141)

177

a [ConcP [Spec sedi-al rivat o no sedi-al rivat]i [HypP . . . be=SCL arrived or not be=SCL arrived voi vie istés sedi-al . . . ]]]! (Frl.) [IP jo o I SCL go away same ‘Whether he has arrived or not, I am leaving all the same’ b {Force [HypP [Spec

fùsselo vegnùo anca Mario] . . . were=he come also Mario [IP gavaressimo podùo dìrghelo fùsselo . . . ]]}! (Pad.) we.would.have been.able say.INF=to-him=it ‘If Mario had come as well, we would have been able to tell him’ (Pad.)

Optimal candidates for the lexicalization of ExclP in Romance include finite complementizers (142a–b) and 3rd person tonic pronouns such as Paduan lu (142c; Benincà 1996) and Calabrian and Sicilian iddu (142d; Ledgeway 2003b: }2.3). In (142c), it is necessary to assume remnant movement of the core sentence to the left-peripheral modifier/Spec(ifier) position ExclP (or perhaps to the specifier of some higher position), whereas in the Catanzarese example in (142d) the highest Force head (here labelled Decl(arative)P(hrase)) is simultaneously lexicalized by the declarative complementizer ca ‘that’. (142) a {Force . . . [ExclP

¡Que . . . [IP that

se self=

ha it.has

estrechao!]]} (Ara.) narrowed

‘it has shrunk!’ b {Force . . . [ExclP

¡Que . . . [IP that ‘I don’t want it, okay!’

c {Force . . . [ExclP [Spec

a SCL

xe is

no not beo] nice

lo it=

quiero, I.want

vaya!]]} (Sp.) go.IMP

lu . . . [IP a xe beo ]]}! (Pad.) he

‘It’s really nice!’ d {Force [DeclP ca . . . [ExclP iddu that he ‘You’ll be frightened!’ (Ctz.)

[IP ti spagni!]]]} (Ctz.) yourself= you.frighten

Finally, interrogative force is licensed in the lowest of the series of Force projections (presumably to be identified with Rizzi’s (2001) Int(errogative)P(hrase)), the head of which is variously lexicalized by Ibero-Romance and Occitan que ‘that/what’ (143a–b), Florentine o(cche) (143c; Garzonio 2004), and central-southern Italian dialect che/chi/ce ‘that/what’ (143d–e).118 Rohlfs (1969: 157-9), Manzini and Savoia (2005 I: }3.8.2), Cruschina (2008: ch. 5), Ledgeway (2009a: 811–13). 118

178

From Latin to Romance

(143)

a ¿Que

no benibas con not you.came with ‘Did you not come with them?’

els? (Ara.) them

Q

b Que

bos bié? (Bea.) you.want come.INF ‘Do you want to come?’ Q

c O

a casa quando tu ci Q to home when you there ‘Home, when do you intend to return?’

vài want

tornare? (Flo.) return.INF

d c’

aspiette a quaccheduno? (Nap.) Q you.wait ACC.MRK. somebody ‘are you waiting for somebody?’

e Ce Q

sta PROG

cchiovi? (Sal.) it.rains

‘Is it raining?’ In many Gallo-Romance varieties, by contrast, interrogative force is licensed by FocP (Munaro 2010), either through V(erb)-raising to Foc (144a–b) or through lexicalization of the same with the finite complementizer che/cha (144c–d). A not too dissimilar situation is found in Sardinian (144f), where in polar interrogatives the head of FocP is lexicalized by the particle (AUT ‘or’) a.119 (144) a [FocP[Spec

Où] travailles where work ‘Where do you work?’

b [FocP [Spec Se] anwhat have ‘What did they do?’

[IP -tu you [IP o =SCL.3PL

c [FocP [Spec Où] que where that ‘Where do you work?’

[IP tu you

d [FocP [Spec

[IP r’

Cosa] cha what that ‘What has he done?’

travailles où]]? (Fr.)

fat done

se?]] (Plr.)

travailles où]]? (coll.Fr.) work

SCL.3SG=

ha fait cosa?]] (Poi.) has done

e [FocP A [IP kere vénnere a domo Q wants come.INF to house ‘Do you want to come to my house?’ 119

an

mea?]] (Srd.) my

Jones (1993: 244–7), Manzini and Savoia (2005 I: 610–11), Mensching and Remberger (2010a, b).

The rise of functional structure

179

To conclude, we give below in (145) the full structural representation of the left periphery of the Romance clause in accordance with the entire range of data reviewed above. A non-trivial consequence of this interpretation of the C-domain is that it predicts the availability of a series of dedicated head positions spelling out specific pragmatico-semantic interpretations and grammatical categories, as well as associated left-peripheral modifier/Spec(ifier) positions to which appropriate constituents may raise to license particular interpretive, discourse, or scope effects at the interface. In this respect, a simple CP model which offers only a single head position but multiple leftperipheral modifier/Spec(ifier) positions seems less apt to capture the relevant interpretations and empirical generalizations, particularly in those cases in which both the head and left-peripheral specifier positions of two or more projections are simultaneously lexicalized under the Spec-Head agreement configuration. (145)

[DeclP que/că/ca [FrameP1 HT [FrameP2 Advsc.-set. [ConcP whether-clause [HypP si/ se-clause [ExclP que/lu/iddu [ThemeP1 ATop [ThemeP2 ClLD [IntP o(cche)/que/ c(h)e [CFocP1 Obj/Adv [CFocP2 Advcircum./quant. [IFocP1 IFoc [IFocP2 IndefQ [FinP să/cu/de/a/V2 [IP . . . ]]]]]]]]]]]]]]]

4.5 Conclusion To sum up, we have seen that, in contrast to Latin, the Romance languages present abundant evidence for the widespread existence of functional structure and associated functional categories in the left edge of the nominal, verbal, and clausal groups. Within the context of the discussion of the rise of configurationality in chapter 3, these facts now find an immediate and natural explanation. In particular, we can link the rise of Romance functional structure directly to the emergence of configurationality: as long as Latin had a flat, non-configurational structure (146a), it also lacked functional structure, which only began to emerge when the language started to develop configurationality (146b), first at the level of the clause (CP) and in the prepositional group (PP), and subsequently within the verbal (IP) and nominal (DP) groups.120 (146)

a

X 9 X X X

b Þ

XP 2 X X 2 X ...

120 A case in point is Vincent’s (1999) study of the rise of prepositions from Indo-European to Romance, where he shows that the gradual emergence of this functional category is accompanied by the progressive instantiation of configurational structure (cf. }5.3.1.1).

180

From Latin to Romance

In short, when in the passage from Latin to Romance the language began to project configurational phrase structure according to the universally available X-bar schema, the functional projections CP, PP, IP, and DP (and eventual splits thereof) came at once ‘for free’ (Vincent 1993: 146–7): (147)

CP Comp I(nfl)P que/de/a Aux VP (h)aver(e)/avoir esse(re)/être V P de/di/a

PP DP Det le/il/el/la un/una/une

NP

This is not to say that, semantically, Latin lacked entirely any functional elements; already we have seen and argued above and in chapter 3 that Latin possessed, among other categories, possessives, quantifiers, and demonstratives in the nominal domain, and a perfective passive/deponent copula ESSE ‘be’ and, increasingly, a number of verbs with growing temporal, aspectual, and modal values (including HABERE ‘have’ underlying the later Romance perfective and future periphrases) in the verbal domain. Significantly, though, all such elements are lexical in Latin, in that they do not differ syntactically or formally from canonical adjectives and verbs, respectively. Among other things, we have seen that they fail to occupy a fixed position within the phrase or clause, whereas in Romance they come to occupy unique and specialized dedicated positions that are unavailable to lexical adjectives and verbs, a difference which we interpret here as a consequence of the emergence of configurationality and the concomitant projection of functional structure and associated fixed positions. Thus, what we witness in the passage from Latin to Romance is not necessarily the creation of new grammatical categories, although there are, as we have seen, many such cases, but a change in the formal characterization of such categories. Increasingly, they come to be associated with structural positions (and their concomitant syntactic properties) in the newly available functional structure which, once available within the system, come to attract and host a number of new lexical items which, through their conventionalization with said positions, progressively distance themselves from their original and corresponding lexical uses to develop as grammaticalized elements, ultimately giving rise to the superficial effects of greater analyticity.

5 From Latin to Romance: a configurational approach 5.1 Introduction In contrast to chapters 3 and 4, which develop an approach to the changes in word order and sentence structure from Latin to Romance in terms of the progressive rise of configurationality and concomitant functional structure, the present chapter attempts to show how the same empirical generalizations can be captured within an approach that assumes the presence of both configurational and functional structure already in Latin. On this view, the unmistakable differences between (different stages of) Latin and Romance, most notably observable in the replacement of an essentially pragmatically determined word order with an increasingly grammatically determined word order and the concomitant emergence of functional categories (determiners, auxiliaries, and complementizers), can now be explained by formal changes in the directionality parameter and the differential role of functional structure in the two varieties. For example, the gradual rigidification of word order according to grammatical principles in the passage from classical Latin to Romance can now be explained in terms of a progressive reversal of the directionality parameter. Assuming the ordering of heads (in bold) and complements (underlined) in the development from (Indo-European/) archaic Latin to Romance to have undergone a shift from one harmonic principle of linear organization to another (viz. head-last ) head-first)—witness the consistently typical head-final and head-initial positions of parisuma and pari illustrated for archaic Latin and Romance in (1a–b), respectively—the greater freedom of word order traditionally recognized for (classical) literary Latin can now been seen as a result of its occupying a (perhaps artificially sustained) intermediate position in this change, resulting in mixed (dis)harmonic linearizations such as those in (2a–b): (1) a quoius forma uirtutei parisuma fuit (archaic Lat., CIL 12.7) whose beauty.NOM valour.DAT most.equal.NOM was

182

From Latin to Romance cui bellezza fu pari al b la the whose beauty was equal to.the ‘whose beauty was fully equal to his valour’

valore (It.) valour

(2) a constantibus hominibus par erat (Lat., Cic. Diu. 2.113) resolute.ABL.PL men.ABL equal.NOM it.was ‘[our apprehension] was equal to that of men of strong character’ b illa erat uita [ . . . ] libertate esse parem ceteris (Lat., Cic. Phil. 1.34) that.NOM was life.NOM freedom.ABL be.INF equal.ACC rest.DAT.PL ‘What he considered life [ . . . ] was the being equal to the rest of the citizens in freedom’ Besides this fluctuation at the syntactic level between a conservative head-final and an innovative head-initial structural organization, pragmatics is also widely recognized to play a significant role in determining Latin word order. As will be illustrated in detail below, this aspect of Latin sentential organization, largely absent in Romance, can be captured by assuming the greater accessibility of topic- and focus-fronting to left-peripheral positions situated in the left edge of individual functional projections. In Romance we have seen (cf. chapter 4) that functional structure is readily exploited and made visible through the lexicalization of head positions with functional categories such as determiners and auxiliaries (3a) and through operations such as N(oun)- (cf. }3.2.2.1.1, }4.2.3.1) or V(erb)-raising (cf. }4.3.2) to these same head positions within the functional structure (3b). By contrast, Latin lacks such functional categories and N-/V-raising but, nonetheless, displays ubiquitous evidence for the presence of functional structure through its extensive exploitation of topic- and focus-fronting to the left edge of these same functional projections (the so-called Spec(ifier) positions immediately to the left of the empty head positions), as represented schematically in (3c): (3) a [IP/DP [Spec ] [I'/D' XAux/Det [I'/D' V/N b [IP/DP [Spec ] c [IP/DP [Spec XPTopic/Focus] [I'/D' 

[VP/NP V/N XP]]] (Romance) [VP/NP V/N XP]]] (Romance) [VP/NP V/N XP]]] (Latin)

Informally, then, we can say that at the level of functional structure Latin privileges an XP-type syntax and Romance an X-type syntax, with generalized overt exploitation of full-phrasal left-peripheral modifier/Spec(ifier) positions in the former and head positions in the latter. By way of illustration, consider the nominal examples in (4), where the left-peripheral modifier/Spec(ifier) position in the left edge of the D(eterminer)P(hrase) is overtly signalled by the raised focused adjectival modifier summo ‘highest’ in the Latin example (4a), whereas in its Spanish translation (4b) it is the head position of the DP that is overtly spelt out by the presence of the indefinite article un ‘a’ and the adjectival modifier del más alto ‘highest’ remains in situ. In

A configurational approach

183

short, there emerges a strict complementarity between pragmatic fronting and the availability of functional categories like determiners and auxiliaries, such that we may hypothesize that the pragmatically determined word order of Latin follows from the availability of fronting to left-peripheral positions within the left edge, a fronting operation blocked in Romance by the overt lexicalization of the corresponding head position (4c; cf. discussion in }5.4.3.4). Indeed, once local fronting to these leftperipheral positions becomes available, these same positions may function as escape hatches, feeding further fronting operations to more remote and pragmatically more salient left-peripheral positions within the nominal and clausal superstructure, witness the extraction of summam to the clause-initial position in (4d) violating the socalled Left Branch Condition (cf. }3.2). (4) a [DP [Spec summo] [D'[NP homo [summo ingenio]]]] (Lat., Cic. De or.1.104) man.NOM talent.ABL highest.ABL b [DP [Spec ] [D'un [NP hombre [del más alto talento]]]] (Sp.) a man of.the more high talent c **[DP [Spec del más alto talento][D'un [NP hombre [del más alto talento]]]] (Sp.) of.the more high talent a man ‘a man of the highest ability’ d qui [CP [Spec summam] [C' inter eos habet [DP [Spec summam] who.NOM highest.ACC among them.ACC has [NP auctoritatem [summam]]]]] (Lat., Caes B.G. 6.13.8) power.ACC ‘who has the highest authority among them’ 5.1.1 Early evidence for functional structure While these ideas will be developed and justified in greater detail below, it is worth highlighting at this point a number of more general considerations which substantiate the assumption of full configurationality and the presence of functional structure already in Latin. Indeed, we have already seen at }3.4.1 how even under a largely non-configurational approach to Latin it is nonetheless necessary to assume the presence of some functional structure and configurational organization at the level of the complementizer (CP) and the preposition (PP). Moreover, to assume the progressive emergence of functional structure in the passage from Latin to Romance raises some non-trivial issues about the nature of functional structure which, according to standard interpretations,1 is considered to be a primitive property of the computational component, and hence a core design feature of the linguistic

1

Cinque (1999: 127–37; 2002: 3; 2006: 3–4, 6), Belletti (2001b: 493; 2004a: 4–7), Rizzi (2004: 4).

184

From Latin to Romance

system that is not subject to (macro)parametric variation.2 Hence to assume its absence in some varieties such as Latin is incompatible with the general view that ‘in the number and type of functional projections available to different languages [ . . . ] U[niversal]G[rammar] allows no variation at all’ (Cinque 1999: 127), as well as with current assumptions summed up in the Chomsky-Borer Conjecture (Baker 2008: 353) which identify the locus of crosslinguistic variation with the variable featural content and lexicalization of individual functional projections. Furthermore, such an approach would presumably lead us to question, if not refute outright, the consistent presence of (individual) functional structure(s) in a whole host of Romance contexts too, if we interpret the lack of overt functional categories as direct evidence for the lack of corresponding functional projections. For example, many dialects of the extreme south of Italy and varieties of the north-western Iberian Peninsula, the Canaries, and Latin America continue to express the present perfective through synthetic forms cognate with the Latin synthetic perfect (5a–c),3 in contrast to most Romance varieties that formally mark the present perfective through the innovative auxiliary + participle periphrasis (5d; also }2.2.1, }4.3.1.4). (5) a Sta sira quanta pisci pigghjiasti? (SCal.) this evening how.many fish you.took ‘How many fish have you caught this evening?’ b ¿Oíste lo que digo? (Leo.) you.heard the which I.say ‘Have you heard what I’m saying?’ c ¿Todavía no llegó tu padre? (Cmb.Sp.) still not arrived your father ‘Has your father not arrived yet?’ d Li bregands m’ an cremat l’ the brigands me= have burnt the ‘The brigands have burnt my house down!’

ostau! (Prv.) house

Are we to conclude from this that the functional projection evidenced by the lexicalization of the auxiliary an ‘they have’ in (5d) is absent from the grammars of the varieties in (5a–c)? Indeed, it is widely assumed that clauses containing periphrastic verb forms are necessarily richer in functional structure than clauses with

2 Giorgi and Pianesi (1996; 1997) propose a slightly different interpretation which allows some crosslinguistic microvariation in terms of ‘scattered’ vs ‘syncretic’ realization of a universally-given array of functional features in individual vs composite functional projections. 3 García de Diego (1946: 169), Zamora Vicente (1967: 434), Ambrosini (1969: 154–5), Alfonzetti (1998: 34), Lipski (1994: 173), Penny (2000: 159–60), Leal Cruz (2003: 132), Howe and Schwenter (2008). For an overview of the development and distribution of the two perfective paradigms across Romance, see Harris (1982) and Squartini and Bertinetto (2000).

A configurational approach

185

synthetic verb forms in order to accommodate additional grammatical elements such as auxiliaries. However, as Cinque (1999: 131) and Benincà and Munaro (2010: 5–6) observe, this is merely a superficial illusion, with the only difference between examples such (5a–c) vs (5d) a question of language-particular morphology: Provençal morphology makes available a particular morpheme (viz. an) to lexicalize the marked value of the functional head associated with present perfective (TAnterior in Cinque’s system), which is simply not available in the morphology of such varieties as southern Calabrian, Leonese, and Colombian Spanish which, nonetheless, may be able to license the same functional head through other mechanisms such as overt raising of the synthetic verb form to this position (}4.3.2). Mutatis mutandis, these same arguments carry over straightforwardly to Latin. In short, the only way to rescue the fundamental idea outlined in chapter 4 that the passage from Latin to Romance is characterized by the progressive emergence of functional structure is to make the ad hoc assumption that functional projections were indeed present in Latin, but in some sense dormant or inactive and hence inaccessible to the computational system (cf. Ledgeway 2007c: 363–5). Even on this view, however, we are still left without a formal explanation or understanding of how and why such projections should have become active and accessible in Romance. In any case we have already seen in the previous section (see also }5.4.3) how it is necessary to assume for Latin a fully active and accessible set of functional projections, whose left edge regularly serves to host fronted topical and focal elements. If such projections were not available in Latin, then the nature and identity of the position of focused adjectives such as summo/summam ‘highest’ in (4a,d) would remain puzzling, despite the fact that the positions they lexicalize are universally available across all categories instantiating an entirely regular and predictable structural pattern rather than an ad hoc option particular to these examples. 5.1.2 Early evidence for configurationality Even keeping our discussion at a very superficial level, there are very good reasons to take a fully configurational approach to Latin. For example, while the flexibility of classical Latin word order is, as we have observed on several occasions in the preceding chapters, widely recognized and reported in the literature, it is less frequently remarked that such flexibility typifies root clauses. The word order of embedded clauses, by contrast, is much less flexible (pace Salvi 2004: 43), generally following a very rigid SOV arrangement.4 Illustrative in this respect are the statistical findings of Charpin (1989: 518) who observes in his examination of classical prose that

4 Linde (1923: 154, 156–7), Marouzeau (1938: 49, 104–7), Ramsden (1963: 43–4, 114), Amacker (1989: 493), Bolkestein (1989: 23), Pinkster (1991: 69), Bauer (1995: 91; 2009a: 270), Oniga (2004: 99), Baños Baños and Cabrillana (2009: 696).

186

From Latin to Romance

[E]mbedded clauses behave like fixed structures in which, save very few variations, all elements are arranged according to an identical and easily identifiable order (Charpin 1977: 404). The distribution of constituents according to their function represents a fundamental characteristic of embedded clauses: with the exception of a few cases of extraposition, the clause is introduced by a conjunction or a relative; the verb occurs constantly in final position (70 to 90% of cases) and the subject in initial position. No matter what the textual sample, the sequence of elements is always tightly grouped together and cannot be broken up by the intrusion of any foreign elements.

This distinction between root and embedded clauses is not limited to classical Latin, but is readily observable in the ‘Latin of all periods, including that of very late antiquity’ (Adams 1976a: 93 n. 61; cf. also Adams 1992: 21–2), continuing even into early Romance.5 By way of illustration, consider the consistent final position of the embedded verb (in bold) in contrast to that of the main verb (underlined) in the following representative examples from archaic (6a; Clackson and Horrocks 2007: 29–30), classical (6b; Linde 1923: 154, 156–7), and late (6c; Adams 1976b: 136, 139; 1977: 69; Bauer 1995: 91) Latin: med mitat, nei ted endo cosmis (6) a iouesat deiuos qoi swears gods.ACC who.NOM me.ACC sends lest you.ACC in kind.NOM uirco sied (archaic Lat., CIL 12.4) maiden.NOM be.SBJV ‘The person who sends you to me prays to the gods, lest the girl be not kind towards you’ b et te exorabo profecto, ut mihi quoque et Catulo and you.ACC I.shall.beg truly that me.DAT also and Catulus.DAT tuae suauitatis aliquid impertias (classical Lat., Cic. De or. 2.16) your.GEN charm.GEN somewhat you.share.SBJV ‘and assuredly I shall implore you to spare a little of your amiability for me too and for Catulus’ c Et quoniam nobis ita erat iter, ut prius montem Dei and since us.DAT thus was route.NOM that first mount.ACC god.GEN ascenderemus (late Lat., Per. Aeth. 2.3) we.ascended.SBJV ‘And as our route was first to ascend the mount of God’ 5 Linde (1923: 156), Salvi (2004: 85–90), Bauer (2009a: 272). As observed by Bauer (2009a: 270), an identical pattern is found in Oscan where ‘in [ . . . ] texts after 200 BCE there are instances of VO (none before that time), but these instances are only attested in main clauses, not in subordinate ones’. Representative of Romance are the Strasbourg Oaths of 842, in which all 5 embedded verbs occur in clause-final position (namely, in quant Deus [ . . . ] dunat ‘as God . . . will grant’, cum om [ . . . ] dist ‘as one . . . must’, quid il [ . . . ] fazet, et [ . . . ] prindrai ‘that he . . . may do, and . . . I shall take’, qui [ . . . ] sit ‘who . . . be’).

A configurational approach

187

It is not at all obvious how the approach to Latin sentence structure developed in chapter 3 can readily account for this asymmetry between root and embedded clauses. If Latin phrase structure is genuinely claimed to be organized according to predominantly non-configurational principles, then the rigid and syntactically predictable word order of embedded clauses is totally unexpected. Certainly, it would represent an unprecedented, not to say theoretically questionable, move to maintain that Latin root clauses all but lacked any configurational structure whereas embedded clauses were entirely configurational. Even more problematic in this light is the implication of the hypothesis that language acquisition operates on the basis of data from main clauses alone (cf. Lightfoot’s Degree-0 Learnability), such that language change is predicted to originate in main clauses with linguistic innovations only potentially spreading to embedded clauses in subsequent stages of development.6 Consequently, subordinate clauses are thought to exhibit a more conservative behaviour,7 which, in the particular case of the development of Latin, can only be interpreted to mean, rather counterintuitively and in contradiction to what was argued in chapter 3, that the ostensible non-configurational organization of root clauses represents an innovation, rather than the original situation preserved in the conservative configurational word order patterns of embedded clauses. In any case, even the word order of main clauses is not entirely grammatically free, as non-configurational approaches would incorrectly lead us to expect. Even scholars such as Panhuis (1984: 156) who is at pains to highlight how ‘pragmatically, [word order] is not free, but organized in a communicative perspective’ is nonetheless obliged to recognize that ‘from a syntactic point of view, word order in Latin is [ . . . ] almost free’ (underlining A.L.). To be sure, Latin word order displays rather consistently fixed patterns of arrangement under pragmatically neutral readings,8 which can readily be taken to indicate the existence of configurational structure, witness the template in (7a) taken from Devine and Stephens (2006: 79) and some representative classical examples in (7b–d). (7) a S O IO/Obl Adjn Goal/Source non-ref. O Verb b Caesar omnem ex castris equitatum suis auxilio misit Caesar.NOM all.ACC out.of camp.ABL cavalry.ACC his.ABL.PL help.DAT sent (Lat., Caes. B.G. 4.37.2; S O IO Adjn V) ‘Caesar sent all the cavalry in the camp as a relief to his men’

6

Lightfoot (1991: ch. 3), Dixon (1994: 206), Harris and Campbell (1995: 27). Fleischman (1982: 112), Cabrillana (1999: 320 n. 9), Salvi (2004: 36–8). Panhuis (1982: 125), Salvi (2004: 43–5), Devine and Stephens (2006: 36–98), Hewson and Bubenik (2006: 2), Bauer (2009a: 243), Polo (2004: 383). 7 8

188

From Latin to Romance c Menapii legatos ad eum pacis petendae causa Menapians.NOM envoys.ACC to him.ACC peace.GEN seeking.GEN causa.ABL mittunt (Lat., Caes. B.G. 6.6.2; S O IO Goal) they.send ‘The Menapians send envoys to him for the sake of seeking peace’ d Germani una in parte confertis turmis in hostes Germans.NOM one.ABL in part.ABL packed.ABL.PL squadrons.ABL in enemy impetum fecerunt (Lat., Caes. B.G. 7.80.6; S Adjn Goal non-ref. O V) attack they.made ‘The Germans, with their cavalry squadrons massed together in one part of the battlefield, made an attack against the enemy’

We conclude therefore with Elerick (1992: 24) that ‘the term “free” has no utility in discussions of Latin word order’, inasmuch as ‘Latin word order is syntactically principled.’ Consequently, near–minimal pairs like (8a–b) with different linear arrangements of the same content words do not count as mere repetitions of each other, as was originally argued for classic non-configurational languages such as Warlpiri (Hale 1983: 5),9 but convey different information structures. For instance, the S O Goal V order in (8a) represents the neutral unmarked arrangement in which nothing is presupposed and the whole clause presents new information. By contrast, the O S Goal V order in (8b) constitutes a marked order in which the object constituent copias suas has been deliberately fronted, since it is contrastively focused (cf. also the postnominal position of the possessive suas (cf. }5.4.3)) with the object constituent equitum- of the following coordinated clause. (8) a Caesar suas copias in proximum collem subducit (Lat., Caes. Caesar.NOM his.ACC troops.ACC in next.ACC hill.ACC withdraws B.G.1.22.3) ‘Caesar leads off his forces to the next hill’ b copias suas Caesar in proximum collem subduxit troops.ACC his.ACC Caesar.NOM in next.ACC hill.ACC withdrew (equitatumque, qui sustineret hostium impetum, cavalry.ACC=and who.NOM sustained.SBJV enemies.GEN attack.ACC misit) (Lat., Caes. B.G. 1.24.1) he.sent ‘Caesar drew off his forces to the next hill(, and sent the cavalry to sustain the attack of the enemy)’

9 Following Hale’s early studies of Warlpiri, subsequent research showed that word order in Warlpiri is also pragmatically determined (Swartz 1987; 1988), a view ultimately recognized also in Hale (1992: 76). For an overview, see Austin (2001: }2).

A configurational approach

189

Indeed, despite traditional, often excessive, claims about the indiscriminate (grammatical) freedom of Latin word order, it is not difficult to find in the literature references to prevailing tendencies, if not hard-and-fast rules (for a classic case, see the list in Linde 1923: 178), regarding the linearization of particular sequences.10 It will suffice here to consider just a few relevant examples. Beginning with the adverb, Ricca (2010: 165) observes that ‘the position of the adverb as adjective modifier is remarkably rigid compared to the usual mobility of constituents in Latin’. For instance, unlike degree adverbs (e.g. TAM ‘so much’, UALDE ‘intensely’, NIMIS ‘excessively’, SATIS ‘quite, enough’, ADMODUM ‘completely’) which typically (though not exclusively) precede the constituent they modify (9a–b), standard handbooks widely report a fixed postmodifier order for focalizing adverbs such as (NE . . . ) QUIDEM ‘(not)

10 There are also some claims in the literature which do not stand up to close scrutiny (I thank J. N. Adams for critical discussion of the following data). For instance, on the basis of a classical corpus, Charpin (1989: 508–10) maintains that, with the exception of V-initial sentences or those containing a weak Wackernagel clitic, the direct object in classical Latin can only be separated from the verb by as much as one constituent, hence the apparent absence of such sequences as **DO XP XP V or **(S) V XP XP DO. In turn, this assumption allows him to conclude that in an example such as (i.a), despite the morphological ambiguity of the nominative and accusative plurals -es desinence, only ciuitates can be understood as the object of the sentence leaving parentes to be interpreted as the subject. It is not at all difficult, however, to find counterexamples to this generalization, witness (i.b) where the direct object Turnum is separated from the verb by a participial construction, a three-word prepositional phrase, and a two-word ablative phrase.

(i)

a raptarum parentes tum maxime sordida ueste lacrimisque abducted.GEN.FPL parents.NOM then precisely shabby.ABL clothing.ABL tears.ABL=and et querellis ciuitates concitabant (Lat., Liv. 1.10.1) and complaints.ABL states.ACC stirred ‘the parents of the adbducted maidens, in mourning garb and with tears and lamentations, were attempting to arouse their states to action’ b Turnum nouantem res pro manifesto parricidio Turnus.ACC renew.PRS.PTCP.ACC.SG thing.ACC for detected.ABL treason.ABL merita poena adfecissent (Lat., Liv. 1.52.1) deserved.ABL penalty.ABL they.treated ‘on account of his rebellious attempt, they justly punished Turnus for the treason in which he had just been taken’

Equally problematic is his claim (pp. 510–11) that nominal forms with formally identical desinences cannot be juxtaposed unless they belong to the same syntagm, thereby ruling out sequences such as MILITES (soldiers.NOM/ACC) + DUCES (leaders.NOM/ACC) NECANT (kill.3PL). Nonetheless, such cases (termed homoeoteleuton) are indeed attested, witness the final line in (ii) where every word ends in -(i)s but does not contain a single syntagm: (ii)

inferius uictae sidus Carthaginis arces lower.NOM.N vanquished.GEN.FSG star.NOM.N Carthage.GEN.FSG citadels.ACC.FPL et Libyam Aegyptique latus donataque rura and Libya.ACC.FSG Egypt.GEN= and flank.ACC granted.ACC.NPL=and lands.ACC Cyrenes lacrimis radicis Scorpios acris (Lat., Manil. 4.778–81) Cyrene.GEN tears.ABL root.GEN Scropio.ACC sharp.GEN ‘Scorpio, the sign beneath, (chooses) the towers of vanquished Carthage, Libya, the flank of Egypt, and the territory of the Cyrene, gifted with tears of a pungent root’

190

From Latin to Romance

even’, QUOQUE ‘also, too’, DENIQUE / DEMUM ‘just, precisely’, FERE / FERME / PAENE / PROPE ‘almost, nearly’, POTIUS ‘rather’, IPSE ‘self-, very’, and TANTUM ‘only’ (10a–b).11 (9) a [ . . . ] ut et minus late uagarentur et minus facile that and less widely they.wandered.SBJV and less easily finitimis bellum inferre possent (Lat., Caes. B.G. 1.2.4) neighbours.ABL war.ACC bring they.could.SBJV ‘(it resulted) that they could range less widely, and could less easily make war upon their neighbours’ b tam necessario tempore, tam propinquis hostibus ab iis non so necessary.ABL time.ABL so near.ABL.PL enemies.ABL by them.ABL not subleuetur (Lat., Caes. B.G. 1.16.6) he.assists.SBJV.DEP ‘he is not assisted by them on so urgent an occasion, when the enemy were so close at hand’ (10) a cum patri populus statuam posuisset, filio quoque when father.DAT people.NOM statue.ACC had.placed.SBJV son.DAT also dedit (Lat., Nep. Timot. 2.3) gave ‘the people, having erected a statue in honour of the father of Timotheus, gave one to the son also’ b Hic locus aequum fere spatium a castris this.NOM place.NOM equal.ACC almost space.ACC from camps.ABL Ariouisti et Caesaris aberat (Lat., Caes. B.G. 1.43.1) Ariovistus.GEN and Caesar.GEN was.distant ‘This spot was at nearly an equal distance from both Ariovistus’ and Caesar’s camps’ The positional asymmetry between degree and focalizing adverbs illustrated in (9)–(10) is clearly not free and finds no immediate explanation under the non-configurational approach. Rather, the distribution is highly suggestive of a structural distinction. For instance, whereas degree adverbs can be argued to lexicalize the associated left-peripheral modifier (viz. Spec(ifier)) position of the adjectival/adverbial category they modify (11a), we can hypothesize that focalizing adverbs head their own focus projection (FocP) whose associated leftperipheral modifier/Spec(ifier) position attracts the constituent over which they take scope (11b).

11

Gildersleeve and Lodge ([1895] 1997: 289, 431), Hale and Buck ([1903] 1994: 337), Ernout and Thomas ([1953] 1993: 162–3), Rosén (2009: 338), Spevak (2010: 20–2).

A configurational approach

191

(11) a [AdvP [Spec minus] [Adv' late]] less widely b [FocP [Spec filio] [Foc' quoque [DP filio]]] son.DAT also Another class of words which is well-known to display a strongly preferred, if not fixed, order regards so-called second-position Wackernagel elements (traditionally termed postpositives), whose syntactic placement cannot be adequately handled merely in phonological terms. Among these, we recall here a number of clausal connectors such as AUTEM ‘however’, UERO ‘indeed, but’, TAMEN ‘yet’, ENIM ‘for’, and 12 IGITUR ‘therefore’ (12a–b). (12) a a nullo uidebatur, ipse autem omnia uidebat by nobody.ABL he.was.seen self.NOM.MSG however all he.saw (Lat., Cic. Off. 3.38) ‘he was seen by nobody, yet he himself saw everything’ b Pisces [ . . . ] oua relinquunt, facile enim illa aqua sustinentur fishes.NOM eggs.ACC leave easily for that.ABL water.ABL are.supported (Lat. Cic. Nat.D. 2.129) ‘fish leave their eggs, for they are easily kept alive by the water’ Traditionally, weak pronouns have also been included in this group, although as Adams (1994a) has convincingly demonstrated, such pronouns do not so much seek out the clause-second position but, rather, are attracted to a focal host within a prosodically and semantico-syntactically self-contained portion of the clause termed a colon (Fraenkel 1932; 1933; 1964; 1965), a clausal subdomain which comes very close to the autonomous derivational domains identified as ‘phases’ in current minimalist syntax (see }5.4.3.1).13 Thus, although Latin weak pronouns do not necessarily occur in clause-second position, they do nonetheless occur in a fixed position (13a–b), namely in a right-adjacent position to a focal host (below indicated in small caps) which itself regularly occurs in colon-initial position (colon boundaries indicated below by a vertical line; cf. also Bauer 2009a: 294–9).

12

Hale and Buck ([1903] 1994: 168–9), Ernout and Thomas ([1953] 1993: 453–4), Janson (1979: 90–5), Pinkster (1990a: 164), Adams (1994a), Vincent (1998a: 420–42), Salvi (2004: 128–9, 136–52; 2011: 363–5), Rosén (2009: 407–8), Spevak (2010: 16–7). 13 Similar arguments apply to the position of the weak forms of the verb ESSE ‘be’ (Adams 1994a: 113–4, 123–4, 127, 129–30; 1994b); see further }5.4.2.4.

192

From Latin to Romance

(13) a j non PUBLICO me praesidio, j sed priuata diligentia not public.ABL myself.ACC protection.ABL but private.ABL diligence.ABL defendi (Lat., Cic. Cat. 1.11) I.defended ‘Not by PUBLIC protection, but by private diligence did I defend myself ’ me cupiditas umquam tenuit b j de triumpho autem j NULLA not.one.NOM me.ACC desire.NOM ever it.held about triumph.ABL yet (Lat., Cic. Att. 7.2.6) ‘but as to the triumph, I never felt any great desire for it’ Widely observable and recognized facts like these clearly reveal that Latin word order was not in any useful sense free. Instead it was constrained by hierarchically organized, configurational structures in which the linearly predictable placement of individual classes of items such as connectors and weak pronouns underlines the stable phrasestructural organization of the language, which is otherwise often superficially obscured by the frequent displacement (so-called scrambling) of individual lexemes and phrases under local and long-distance pragmatic fronting for topic and focus (cf. }5.4.3). In short, the steadfast positions of such elements as connectors and weak pronouns in (12)–(13) can be profitably exploited to identify and locate specific fixed positions within the clause and the functional categories associated therewith. A case in point is Salvi’s (2004) admirable analysis of Latin word order which relies on the placement of these elements as a key discriminant in detecting the rightmost and leftmost margins of the clausal left periphery (CP) and the sentential core (IP), respectively. Another example of fixed ordering concerns the development of the HABERE future periphrasis examined in }4.3.1.5, where it was noted following Adams (1991) how HABERE could occur in both pre- and post-infinitival positions. Rather than constituting a case of free variation, however, the evidence discussed by Adams reveals that while HABERE in post-infinitival position had not yet unambiguously grammaticalized as a future auxiliary, it was nonetheless this marked position alone, whose emphatic interpretation made it compatible only with the ‘obligation/necessity’ reading (Coleman 1971: 219), that was most readily associated with the future interpretation and which ultimately would give rise to the Romance future. Once again we see how Latin word order is not at all random, with conventionalized orders serving to convey specific grammatical information and contrasts that can only be read off surface linear order, itself ultimately the product of a fixed underlying configurational phrase structure. In this particular case, the two contrasting positions of HABERE can be taken to spell out two distinct positions within the extended functional structure of the sentential core (I-domain), with pre-infinitival HABERE lexicalizing the head of a higher epistemic modal projection of ‘possibility’ and post-infinitival HABERE ambiguously spelling out the head of a lower future time temporal projection or a deontic modal projection of ‘obligation’ in accordance with Cinque’s (1999: 78–83; 2004b; 2006) functional hierarchy (viz. ModPEpistemic > TPFuture > ModPObligation).

A configurational approach

193

However, by far the clearest evidence for configurational structure comes from a consideration of prepositions and complementizers. We have already noted how prepositional phrases (PPs) display a very rigid linear organization from earliest times such that ‘the order is always P NP and the two items are very rarely separated by intervening material’ (Vincent 1999: 1122).14 Thus, Adams (1971: 13) observes how rare examples of discontinuity between preposition and complement such as (14a) from Maecenas (inter . . . fraxinos) met with the disapproval of Quintilian (Inst. 9.4.28), who saw in such uses a ‘faulty structure’. And in a similar vein, Hewson and Bubenik (2006: 2–3) remark that the only observable restriction on the otherwise very liberal word order in the first two opening lines of Vergil’s Eclogue in (14b) is that the preposition sub and its complement tegmine occur in strict contiguity, there being no alternative order available in this case. This conclusion is further reinforced by Devine and Stephen’s (2006: 568) observation that, contrary to English, preposition stranding in wh-interrogatives is not licensed in Latin prose (14c): (14) a inter se sacra mouit aqua fraxinos (Lat., Maecenas in among self sacred.NOM moved water.NOM ash.trees.ACC Quint. Inst.9.4.20) ‘the sacred stream ran among the ash-grove’ b Tityre, tu patulae recubans sub tegmine fagi Tityrus.VOC you.NOM broad.GEN lying.NOM under cover.ABL beech.GEN siluestrem tenui Musam meditaris auena(Lat.,Virg.Ecl.1.1–2) wooded.ACC slender.ABL Muse.ACC ponder.DEP.2SG pipe.ABL ‘Tityrus, you, lying under the cover of a broad beech tree, compose the woodland Muse on slender pipe’ c

in quibus statuis ista tanta pecunia in which.ABL.PL statues.ABL this.NOM so.much.NOM money.NOM (**in) consumpta est? (Lat., Cic. Verr. 2.2.142) in spent is ‘on what statues was all that money spent?’

Similarly, an early fixed order has already been observed (cf. }4.4.1) for complementizers and their dependent clauses (CPs), inasmuch as complementizers absolutely never occur in clause-final position,15 witness (15a–b):

14 See also Gildersleeve and Lodge ([1895] 1997: 432), Hale and Buck ([1903] 1994: 337), Leumann and Hofmann (1928: 495), Marouzeau (1949: 42), Ernout and Thomas ([1953] 1993: 118), Rubio ([1972] 1982: 199–200), Lehmann (1985: 96), Bauer (1995: 131), Oniga (2004: 80, 94–5). Cf. however the discussion of anastrophe of prepositions in }5.3.1.1 and }5.4.3.3. 15 Gildersleeve and Lodge ([1895] 1997: 430), Hale and Buck ([1903] 1994: 336), Rubio ([1972] 1982: 199–200), Oniga (2004: 94–5).

194

From Latin to Romance

(15) a Optauit, ut in currum patris tolleretur (Lat., Cic. Off. 3.93) he.desired that in chariot.ACC father.GEN he.lifted.PASS ‘he desired to be lifted up into his father’s chariot’ b si spiritum ducit, uiuit (Lat., Cic. Inv. rhet. 1.86) if breath draws he.lives ‘if he is drawing his breath, he is alive’ Turning now to the verb and its arguments, we observed in }3.4 how among the classic tests for non-configurationality the absence of subject-object asymmetries is standardly assumed to constitute a significant diagnostic for the absence of a VP constituent and hence a flat sentential structure. Nonetheless, there is evidence for a VP constituent in Latin, in that a number of syntactic processes make reference to the lexical verb and its complement(s) to the exclusion of the subject. Here we shall consider just two such processes. The first concerns examples of VP-ellipsis such as (16a) in which an infinitival VP complement (underlined) of a modal predicate (in bold) such as UELLE ‘want’ and POSSE ‘can’, although realized in the first sentence, is absent, though understood, in the following sentence.16 These examples demonstrate how this specific rule of Latin ellipsis is sensitive to the VP constituent, inasmuch as it phonologically deletes just the verb and its complement(s), but not the sentential subject situated in a hierarchically superordinate position. Similar considerations apply to the example in (16b), in which the understood infinitival complement of the aspectual predicate soleo ‘I am wont’ explicitly realized in the first sentence as the complement of the modal potuisti ‘you could’ is anaphorically picked up by a proform ita ‘thus’ in the second sentence. In this particular case, the latter proform is more accurately described as a pro-VP in that it references the verb, its complement, and temporal adjunct to the exclusion of the subject. (16)

16

a Agorastocles: Eho tu, uin tu facinus facere hey you you.want=Q you deed.ACC do.INF lepidum et festiuom? – witty.ACC and merry.ACC Milphio: Uolo. – Ag.: Potesne mi auscultare? I.want you.can.Q me.DAT listen.to.INF Mil.: Possum (Lat., Pl. Poen. 308–9) I.can ‘Agorastcoles: How now, you; do you want to play a merry and a frolicsome prank? – Milphio: I do. – Ag.: Can you, then, give attention to me? – Mil.: I can.’

As pointed out to me by I. Roberts (p.c.), it is not at all obvious that infinitival complements to predicates such as UELLE ‘want’ are VP constituents, as predicted under recent cartographic accounts (Cinque 2004b; 2006), rather than TP or CP complements (cf. also Ledgeway forthc. c). Under either hypothesis, the tests in (16) still serve to show that it is necessary to recognize constituency in Latin.

A configurational approach

195

b Menaechmus: Non potuisti magis per tempus mi aduenire quam not you.could more for time.ACC me.DAT arrive.INF than aduenis. Peniculus: Ita ego soleo (Lat., Pl. Men. 139–40) you.arrive thus I am.wont ‘You couldn’t have come to me at a more opportune moment than you have. I’m in the habit of doing so’ A priori there are no reasons why such operations should make reference only to the verb and its complement(s) to the exclusion of the subject if the phrase structure of Latin were flat, since neither the subject nor the object are predicted to enjoy a privileged (namely, more local) relationship with the verb. However, the asymmetry in the behaviour of the external (subject) and internal (object) arguments of the verb witnessed in (16a–b) provides clear evidence that the verb and its complement are intimately related, entering into a more local relationship with one another than either of them does with the subject. In short, the privileged relationship of the verb and its complement is a consequence of their forming an immediate constituent (VP), which only in turn jointly combine together with the subject to form a larger constituent (vP-IP). Finally, we turn to some limitations on local hyperbaton (cf. }3.2.1.1, }5.4.3.3) which provide some subtle yet significant evidence for the presence of both configurationality and functional structure in Latin. Taking as our example the P(repositional) P(hrase) in (17a),17 it is frequent in such sequences for an A(djective)P(hrase) such as magna ‘big’ modifying the nominal head of the N(oun)P(hrase) complement to undergo focus-fronting,18 which in the present context can be understood as movement to the left-peripheral modifier (namely, Spec(ifier)) position of the PP (17b; Vincent 1999: 1128; Bauer 2009a: 288). Strikingly, though, the option of raising the nominal head of the NP complement to what appears to be the same position, a case of so-called anastrophe (17c), proves not only extremely rare and stylistically very marked (generally limited to verse; Marouzeau 1953: 67), but is also considered to be highly influenced by Greek and ultimately an ‘artificial affectation’ (Leumann and Hofmann 1928: 495).19 This consequently leads Bauer (1995: 132), for instance, to conclude that it does not represent a genuine feature of the language. Given, however, standard assumptions about phrase structure, the general absence from prose of strings such as (17c) follows straightforwardly from conditions on chain uniformity

17 Mutatis mutandis, identical arguments can be readily replicated for other categories such as the D(eterminer)P(hrase) and the A(djectival)P(hrase). 18 Gildersleeve and Lodge ([1895] 1997: 432), Marouzeau (1953: 68), Adams (1971: 12), Bauer (1995: 136; 2009a: 287, 291–2), Vincent (1999: 1126–9). 19 See also Draeger (1882: 92–4), Marouzeau (1949: 58-60), Ernout and Thomas ([1953] 1993: 120), Ostafin (1986: 120), Bauer (1995: 132, 136–7; 2009a: 287), Kessler (1995: 7), Penney (1999), Vincent (1999: 1127–8), Devine and Stephens (2006: 568–70, 572–3), Sekerina (1997: 195), Fortson (2008: 111–14).

196

From Latin to Romance

(Devine and Stephens 2006: 528–9), since it implies the illicit raising of a simple head (X) category cura to the left-peripheral modifier/Spec(ifier) position of PP, a maximal phrasal position (XP).20 Under a non-configurational approach, by contrast, the effective absence of this sequence from canonical (non-artificial) registers of Latin remains unexpected and somewhat baffling. (17) a [PP [P' cum [NP [N' cura [AP magna] ]]]] (Lat., Pl. Men. 897) with care.ABL big.ABL b [PP [Spec [AP magna]] [P' cum [NP [N' cura [AP magna] ]]]] (Lat., PL. Men. 895) big.ABL with care.ABL c ??[PP [Spec [N

CURA]]

care.ABL ‘with great care’

[P'

CUM

[NP [N' [AP

with

MAGNA] big.ABL

CURA]]]]

(Lat.)

5.2 Head parameter: traditional observations It has long been noted that basic ordering of head and dependency in Latin and Romance are diametrically opposed.21 For instance, Ernout and Thomas ([1953] 1993: 162) highlight how ‘[i]n general, the determiner tends to precede the determined’ in Latin, whereas ‘in French as in other Romance languages [ . . . ] the dependent word is placed after that which governs it’ (Weil ([1844] 1887: 54) according to a typological distinction from which many other basic properties are said to follow.22 Thus, to take three simple examples, we can see that, at least as a possibility (though for the various markedness patterns, see the discussion in }3.2.1 and below), Latin places the modifier (underlined) before the nominal (cf. 18a), verbal (cf. 19a), and clausal (cf. 20a) head (in bold), whereas in the corresponding Romance structures, here exemplified by Portuguese, the modifier typically follows: 20 As noted, however, the offending structures do occur, albeit under the observed stylistic restrictions, and their less than fully acceptable derivation and status therefore requires an explanation, a topic to which we return in }5.4.3.3.

(i) [PP [Spec damno [AP magno]] [P' cum [DP [Spec magno] [D' [NP [N' damno [AP magno] ]]]]]] damage.ABL with great.ABL (Lat., Pl. Asin 187) ‘at considerable cost’ 21 Grandgent (1907: 30), von Wartburg ([1934] 1971: 256), Rubio ([1972] 1982: 199–200), Harris (1978: 16), Tovar (1979), Renzi (1985: 131–7), Vincent (1988: 55–6; 62–3; 1997c: 166), Bauer (1995), Oniga (2004: 52), Clackson and Horrocks (2007: 280–1), Baños Baños and Cabrillana (2009: 683–5). Head and dependent are also variously termed in the literature, albeit not with necessarily identical values, modified/modifier, qualified/qualifier, governing/governed, déterminé/déterminant (or déterminatif), operand/operator, specified/specifier (for an overview, see Bauer 1995: 21–2). In his pioneering work, Weil ([1844] 1887: 59–70) talks of a ‘descending construction’ (head + dependent) and an ‘ascending construction’ (dependent + head). 22 Schmidt (1926), Greenberg (1963; 1966), Lehmann (1974), Harris (1978: 4–6), Bauer (1995: 13).

A configurational approach (18)

197

a mortis timor haut quemquam uexaret eorum (Lat., Lucr. 5.1189) death.GEN fear.NOM not.at.all anyone.ACC vexed.SBJV of.them b o medo da morte não atormentava nenhum deles (Pt.) the fear of.the death not tormented nobody of.them ‘the fear of death vexed not one of them’

(19) a Et hi mortem timent [ . . . ]? (Lat., Sen. Ep. 20.122.3) and these.NOM death.ACC fear b e eles temem a morte? (Pt.) and they fear the death ‘and do they fear death?’ (20)

a mori times (Lat., Sen. Ep. 9.18) die.DEP.INF you.fear b temes morrer (Pt.) he.you.fear die.INF ‘you fear dying’

This simple observation regarding the basic ordering of head and modifier, which classifies Latin essentially as head-final and Romance as head-first, captures the essence and more of the traditional synthesis-analysis distinction examined in chapter 2. In particular, the head parameter is generally taken to characterize not only syntactic structure, but also morphological structure,23 witness Latin head-last compounds like UEXILLIFER ‘flag-bearer’ which contrast with Romance head-first compounds like It. portabandiera lit. ‘bearer-flag’.24 In this way, we can bring under the same roof changes that took place in morphology and syntax, treating them as reflexes of the same overall development towards a consistent manifestation of the head parameter. Thus, in the transition from Latin to Romance the principal change evident in morphosyntax was not the replacement of synthetic with analytic forms, but the reversal of the order of head and modifier, namely the substitution of Postdeterminierung by Prädeterminierung (Baldinger 1968: 88). Consequently, in the same way that the complement comes to follow its verb (21a–b), so the lexical element in an inflected nominal or verbal form comes to follow its preposition (22a–b) or auxiliary (23a–b; Bauer 2009a: 247–8). (21)

a ancillam mittes (Lat., Mart. 11.23) girl.servant you.will.send ‘you will send (me) your maidservant’

23 von Wartburg ([1934] 1971: 256), Renzi (1985: 132), Bauer (1995: 4, 7, 24–5), Zamboni (2000: 123–4), Oniga (1998; 2004: }3.3). 24 Weil ([1844] 1887: 72), Bader (1962), Bauer (2011: 541–2).

198

From Latin to Romance b mandó a la criá (Ext.) he.sent ACC.MRK the maid ‘he sent for the maid’

(22)

a perpulerunt ut [ . . . ] Romam proficiscerentur (Lat., Liv. 3.50.11) they.pushed that Rome.ACC they.started.for.SBJV b ils poussèrent les soldats à partir pour Rome (Fr.) they pushed the soldiers to leav.INF for Rome ‘They urged the soldiers to set out for Rome’

(23)

a patrum uirtutes celebrantur (Lat., Cic. Rab. Post. 1.2) fathers.GEN virtues.NOM are.praised b le virtù dei padri vengono celebrate (It.) the virtues of.the fathers come celebrated ‘the virtues of their fathers are celebrated’

As well as subsuming much of the synthesis-analysis distinction, an approach in terms of the head parameter also provides further theory-internal evidence for the existence of functional structure in Latin. Following Bauer (1995: 35–9; 2006: 293), we observe that head-last (namely, left-branching) structures produce or align with autonomous morphologically complex words (e.g. Lat. LAUauerat ‘wash.PLPF.3SG’), while head-first (namely, right-branching) structures yield syntactic phrases in which the grammatical categories of the construction as a whole are realized by a lexicalized functional head (e.g. Sp. [IP [I' había [VP lavado]]] ‘he.had washed’). By this same line of reasoning, we are forced to assume then, given the nature of the head parameter, that the corresponding Latin construction contains the same functional head, albeit ordered head-finally (namely, [IP [I' [VP LAU-]AUERAT]]), thereby refuting the claims made in chapter 4 that Latin, unlike Romance, lacked functional projections. 5.2.1 Some Romance counterexamples? There are a number of apparent exceptions to the generalization of the head-first setting in Romance. For example, in }}3.2.2.1.–1 it was noted that alongside the unmarked postnominal position of the Romance adjective, just about the only position available in some varieties such as Sardinian and Asturian, a number of Romance varieties, with growing degrees of productivity the further one goes back in time,25 also license prenominal adjectives under particular circumstances (e.g. Fr. un ancien collègue ‘a former colleague’). These cases, rather than constituting a genuine problem for the head parameter, still allow us to maintain a head-first analysis even in these cases, if we assume that at an underlying level the adjective is 25

Huber (1933: 148), Alisova (1967), Ledgeway (2007a: 115–21), Vincent (2007: 61–4).

A configurational approach

199

always generated in postnominal position and that apparent prenominal adjectives constitute derived structures.26 Alternatively, others have maintained, both from a crosslinguistic perspective (Dryer 1988a), as well as on the evidence of Romancespecific data,27 that there is no compelling correlation between the order of the adjective and other word order patterns, in which case the Romance prenominal adjective would not constitute a genuine exception to the parameter setting. Another potential counterexample to the head-first setting includes, according to Renzi (1985: 135), Romance determiner + noun sequences (24a), in which he sees a continuation of the Latin head-last pattern within the N(oun)P(hrase) (24b). (24) a [NP [Spec i] [N' g jáur]] (Vgl.) ‘the hours’ b [NP [Spec illas] [N' horas]] (Lat., August. Conf. 9.8.17) ‘those hours’ However, as noted in }4.2.2, with the rise of configurationality and functional structure determiner + noun sequences should be considered D(eterminer)P(hrase) structures (Longobardi 1994; Giusti 2002), in which the head of the construction is the determiner which selects for an NP complement (identical considerations carry over to the Romanian enclitic definite article where its postnominal position is derived by N- or A-raising; cf. discussion of (32a–b) in chapter 4). On this view, Romance determiner + noun sequences represent canonical head-first structures (25a). Indeed, whenever determiners such as demonstratives function as modifiers rather than heads in Romance according to the archaic Latin pattern, a possibility preserved in Ibero- and Daco-Romance varieties (cf. discussion of (37a–d) in chapter 4), they consistently follow their associated noun on a par with adjectives (25b): (25)

a [DP [D' i [NP jáur]]] (Vgl.) ‘the hours’ b [DP [D' los [NP [N' [N ingleses] [AP aquellos] ]]]] (Sp.) those the English.PL ‘those English (people)’

Apparently more problematic for the head parameter, and indicative of a more general problem with the same, is the simple sentential negation of many GalloRomance varieties, Aragonese, and northern Catalan,28 where the negator does not precede its associated finite verb as in other Romance varieties (26a–b), but stands

See Vanelli (1980), Salvi (1985); for an alternative analysis, see the discussion in }3.2.2.1.1. Berruto (1998), Pountain (1998), Vincent (2007: 57–8 n. 3). Price (1962; 1986), Posner (1984; 1985a, b), Schwegler (1990), Zanuttini (1997: ch. 3), Parry (1997: 183–5), Veny (1982: 51). 26 27 28

200

From Latin to Romance

immediately after the finite verb (26c–e), so-called Stage III in Jespersen’s (1917) negation cycle (cf. }2.2.2) (26)

armes (Cat.) a no portarem not we.shall.carry arms ‘we shall not bear arms’ ll’ate (Nap.) b nu gghioca cu not he.plays with the.others ‘he does not play with the others’ c elle fume pas (coll. Fr.) she smokes not ‘she doesn’t smoke’ d mi capis miga (Eml.) I understand not ‘I don’t understand’ pas? (Lgd.) e O sabiatz it= you.knew not ‘Did you not know (it)?’

According to one view, canonical Romance negation structures like (26a–b) involve a head-first structure in which the negator constitutes the head that selects for a complement consisting of the verbal constituent, in generative terms a NegP that selects for a VP complement (27a). In this light, the structures in (26c–e) can be brought into line with those in (26a–b), if we assume that they are derived from raising of the verb to a position higher than that of the negator (27b; see Renzi 1985: 136–7; Zanuttini 1997), as suggested by the underlying position of the negator to the left of the verb in certain non-finite contexts (27c–d). (27)

a . . . [IP [NegP Neg [VP V . . . ]]] b elle [IP fume [NegP pas [VP fume ]]] (coll. Fr.) she smokes not ‘she doesn’t smoke’ c pour [IP [NegP pas [VP fumer]]] (coll. Fr.) for not smoke.INF ‘in order not to smoke’ d de [IP [NegP minga [VP credeg]]] (Mil.) of not believe.INF=therein ‘to not believe it’

A configurational approach

201

However, according to some analyses (see Rowlett 1998; Zanuttini 1997) postverbal negators like French pas represent, not the governing head of a NegP, but rather, its left-peripheral modifier/Spec(ifier) position. If so, then there is no obvious sense in which the negator can be formally considered the governing head of the construction.29 Moreover, this conclusion has a fundamental semantic, not to mention intuitive, appeal, in that negation is traditionally taken to be an operator that modifies the veracity of the verbal predicate (= the operand). Moreover, negators can be omitted without affecting the grammaticality of the remaining sentence, hence improbable candidates for head status. We are thus forced to conclude, assuming the verb to be the head of the construction, that Romance negation is, somewhat unexpectedly, invariably head-final (namely, Neg + V), but that the fixed linearization of head and modifier is often superficially disrupted in Gallo-Romance, Aragonese, and northern Catalan by the effects of verb raising in certain cases (namely, V + Neg + V; cf. also the discussion of Latin negation in }5.3.1.5).30 Apparently similar problems for the head parameter also result from a consideration of the development of word order. In }3.3.2 we identified in the passage from classical Latin (S)OV to modern Romance (S)VO an intermediate stage of Top/FocV (X), namely a V2 syntax. Interpreting these developments in terms of the head parameter, at the very most one can only talk of a change from OV to VO at the start and end points, whereas the intermediate V2 stage proves more difficult to capture. Of course, one could, as we did in }3.3.2, argue that underlyingly the V2 stage is part of the new VO order, inasmuch as the sentential core presents a fixed SVO order. However, this VO characterization fails to reflect the fact that in root clauses the finite verb is invariably fronted to the left periphery, where it is frequently 29 Of course, such analyses typically assume the presence of a null negator in the head of NegP licensing the negative adverb (e.g. Fr. pas) in its associated left-peripheral modifier/Spec(ifier) position. Under such an approach, orders such as Fr. pas + Vfinite can still be taken to instantiate the expected head + dependent order (viz. [NegP [Spec pas] [Neg’ Neg [VP Vfinite . . . ]]]). Examples like this highlight the difficulties in providing a consistent definition of the term head (pace Nichols 1986: 56–7) which, in some cases, is open to subjective interpretation or yields apparently counterintuitive results (Amacker 1989: 490; La Fauci 1997: 41–3). For example, in Bauer’s (1995) analysis, the head of a comparative like Lat. DULCIOR ‘sweeter’ is taken to be the suffix -IOR on account of the parallel with other inflectional forms, where the lexical stem is the modifier and the inflection is the head consistently surfacing to the left in the corresponding Romance structure (e.g. Lat. AFFIRMAUI ‘state.PRF.1SG’ > Cat. vaig afirmar ‘AUX.PRF.1SG state.INF’: ‘I stated’). However, in the case of the corresponding Romance comparative structure (e.g. It. più dolce lit. ‘more sweet’), conventional X-bar structural representations of the phrase would represent the degree adverb più ‘more’ as the left-peripheral modifier/Spec(ifier) of the A(djectival)P(hrase) headed by dolce ‘sweet’, hence not the head of the entire construction, unless the syntagm were to be reanalysed as a Degree Phrase. In other cases, the parameter yields apparently conflicting results. For instance, whereas the emergence of Romance Aux + V sequences conforms to the expected head-first/right-branching pattern, the inflectional structure of the auxiliary itself displays a head-final/left-branching pattern (La Fauci 1997: 42–3). 30 It has often been noted in the literature that negation fails to fit in with typological generalizations, with negators exhibiting an incredible degree of positional freedom (cf. Hawkins and Gilligan 1988; Cinque 1999: 120–6).

202

From Latin to Romance

preceded, among other constituents, by a fronted complement superficially yielding the older OV order. Apparently, then, what this intermediate V2 situation requires is a hybrid characterization of the head parameter, a setting, however, not envisaged by the theory. However, it must be recalled that object fronting under V2 is not grammatically driven (viz. a case of movement required to satisfy the formal linearization conditions of the system), but represents a case of pragmatically driven optional movement (viz. a case of A'(djunct)-movement) and therefore falls outside of the head parameter which, as a linearization principle of the computational system, continues to display a consistent VO setting in both root and embedded clauses.

5.3 Changing directions: Latin Even a most cursory examination of the Latin of any period, and even of the most formal and literary registers, reveals that the traditional view of Latin as a consistently head-final language (cf. }5.2) is completely untenable (Clackson and Horrocks 2007: 28–9). Already we saw in (2a–b) how in the classical language an adjectival head may either precede or follow its complement according to a pattern which, as we shall see, can be readily replicated for all categories (cf. Oniga 2004: 75). Indeed, as extensive typological research has revealed,31 very few languages exhibit a consistently harmonic linearization of head and dependent and in this respect Latin is no exception (Oniga 2004: 52, 102; Bauer 2009a: 244). Rather, the textual record highlights how even since its earliest attestations Latin already displays extensive evidence of an ongoing, yet well-advanced, progression from an inherited predominantly head-final order towards an innovative, increasingly head-initial order,32 yielding the harmonically inconsistent orders usually interpreted as evidence of so-called free word order. The head-final arrangement in Latin thus represents a conservative feature, inherited from a pre-historic stage of the language and ultimately Indo-European where headfinal orders were very much the norm,33 and are most robustly evidenced in the daughter languages the earlier the attested stage of their documentation.34 Deliberately oversimplifying, we can therefore identify the following four broad stages of development:35

31

See, for example, Hawkins (1983), Dryer (1992), Huang (1994). Adams (1976a), Porzio Gernia (1986: 6), Elerick (1994: 70–1), Oniga (2004: 102–3), Clackson and Horrocks (2007: 29, 31). 33 Lehmann (1974: 238; 1992: 237–47), Bichakjian (1987: 24, 96), Watkins (1998: 68), Clackson and Horrocks (2007: 28–9), Bauer (2009a: 251–2, 254), Magni (2009: 227). 34 Panhuis (1982: 110), Vincent (1988: 56), Bauer (1995: 213–14; 2009a: 250–1). 35 ‘H’ and ‘COMPL’ are intended here as cover terms for ‘head’ and ‘complement’ respectively, whereas the upper/lower case distinction indicates dominant/non-dominant order. 32

A configurational approach

203

(28) (PIE COMPL-H >) archaic Lat.: COMPL-H/h-compl > early Lat.: compl-h/ H-COMPL > classical Lat.: COMPL-H/H-COMPL > late Lat., early Romance: compl-h/H-COMPL The picture depicted in (28) underlines, in part, how The modern order was not abruptly substituted for the old. On the contrary, it is to be found in Latin, with generally increasing frequency, in inscriptions and popular writers, from the earliest texts down; it occurs sporadically also in literary authors, especially in Cicero. [ . . . ] Classic Latin may be said to represent an intermediate stage, while the revolution was in progress; there was a long struggle, and for centuries the ancient and the modern type were used side by side. By the fourth century the new order prevailed. (Grandgent 1907: 31)

Thus, although it is admittedly true that archaic Latin statistically displays more inherited head-final characteristics than classical and postclassical varieties, it must not be forgotten that Latin also inherited a number of structures (e.g. prepositions and right-branching relative clauses) which clearly had already undergone the shift to a head-initial order in a much earlier period (cf. }5.3.1.1), witness their precocious appearance across many of the earliest attestations of Indo-European.36 Traditionally, however, excessive attention has been focused on the position of the verb with respect to its object above all other things, such that OV order has notoriously acquired the status of an immediate and conspicuous indicator of high stylistic register, as famously encapsulated in Quintilian’s normative recommendation (Pinkster 1991: 70) that ‘it is far better to end a sentence with a verb if the composition so permits’ (Inst. 9.4.26). As a consequence, the unmistakable predominance of this stylizing OV order in literary registers (Watkins 1964: 1039; Adams 1976a: 92), undoubtedly artificially sustained as a literary convention under the strong normative pressures of Latinitas,37 has often been simplistically interpreted as the only and/or defining overt evidence for the headfinality of Latin to the detriment of all other evidence. Yet, as Adams (1976a) is at pains to demonstrate, even if we ignore linearization of V and O, for which the evidence of a universal OV arrangement is not so robust as is often simplistically assumed (cf. }5.3.2), the evidence for head-initiality in Latin is otherwise widespread and conspicuous in many other domains of the language which have traditionally been grossly overlooked. On the strength of this, Adams (1976a: 88) is able to conclude that ‘Latin is basically VO in type from the time of the earliest texts.’ Certainly, most of the typical correlating COMPL-H features (including such orders as relative-antecedent, genitive-noun, adjective-noun, noun-adposition) documented in the history of Latin already represent marked variants of the opposite H-COMPL correlating orders, ultimately suggesting that the shift from head-final to 36 37

Friedrich (1975: 39), Adams (1976a), Bauer (1995: 165, 168–9, 214; 2009a: 257, 272), Magni (2009: 227). Panhuis (1982: 99), Vincent (1988: 61), Clackson and Horrocks (2007: 280–1).

204

From Latin to Romance

head-initial, though often regarded as affecting the classical and late periods, had already occurred in large part before the time of early literary texts.38 It is for this reason that there is a non-linear representation of the shift from COMPL-H to H-COMPL in (28) above, with an apparent partial reversal in the predominant H-COMPL order in classical Latin following the early Latin period, since the classical language does not represent a chronologically linear variety of Latin as such but, rather, a highly stylized, literary variant of the language which deliberately harked back to the earliest formal attestations of the written language to incorporate antiquated features of the archaic period. Indeed, the facts to be reviewed below in this regard crucially highlight how ultimately the general linearization properties of a language cannot be exhaustively derived implicationally from the order of V and O (Bauer 2009a: 245–6), as assumed by the traditional typological approach.39 Rather, as we shall see (cf. 5.4.2), what appears to determine, or set the course for, the linearization of all elements within the clausal and nominal domains is the position of complementizers and adpositions, respectively, in line with the strong predictions of the Final-over-Final Constraint (FOFC).40 Thus, despite the order of V and O proving a less than reliable diagnostic for establishing the head directionality of Latin on account of the artificial preservation of a prestigious OV order as a prominent literary convention, other structures such as relatives, comparatives, complementizer and prepositional phrases, and the noun phrase show a very early and widely tolerated adherence to the head-initial pattern (Oniga 2004: 94; Bauer 2009a: 252–6). No doubt, head-initial linearization in these cases must have proved less conspicuous and stigmatized than in the case of VO sequences, readily passing under the stylistico-linguistic ‘radar’ and enjoying a high degree of acceptance. It is to these head-initial structures that we now turn. 5.3.1 Early head-initial structures The gradual reversal in the linearization of head and modifier witnessed in the passage from Latin (and ultimately Indo-European) to Romance did not affect all grammatical structures at the same time. For instance, it is generally acknowledged that this change is already observable in the nominal group in early Latin, but does not manifest itself in the verbal group until much later,41 highlighting once again how 38 Lehmann (1974: 238), Adams (1976a: 72), Panhuis (1982: 110–12), Clackson and Horrocks (2007: 28), Bauer (2009a: 256). Bauer (2009a: 254, 271) and Magni (2009: 246) maintain that such synchronic variation should not be interpreted in syntactic terms alone, but is often best understood in relation to contextbound stylistic and pragmatic factors that often characterize archaic and innovative usages (cf. }5.4.3.2.1). 39 Schmidt (1926), Greenberg (1963; 1966), Lehmann (1974; 1992: 239–40), Vennemann (1974), Harris (1978), Hawkins (1983). 40 See Biberauer, Holmberg, and Roberts (2008; 2009), Biberauer, Newton, and Sheehan (2009a, b), Biberauer, Sheehan, and Newton (2010), Biberauer and Sheehan (in press). 41 Adams (1976a), Bauer (1995: 11, 85, 90; 2009a: 268), Magni (2009: 226 n.3). A similar situation is reported for other branches of Italic by Konneker (1975: 367).

A configurational approach

205

the linearization of V and O cannot be considered the driving force behind the arrangement of other categories. Among the first structures to exhibit a move towards head-initiality are the adposition, the comparative, the relative clause, and the adjective (Bauer 1995: 168–9, 214; 2009a: 256–8), to which we can also add the complementizer. 5.3.1.1 Complementizers and adpositions As has already been highlighted in various places in the preceding chapters (cf. }3.4.1, }4.4.1.1), both complementizers and adpositions are the only categories to show a fixed head-initial order since our earliest texts.42 In the case of complementizers, we noted how in the history of Latin earlier paratactic, correlative, and participial constructions involving the simple juxtaposition of two clauses increasingly found themselves in competition with explicit hypotactic subordination structures with overt complementizers such as UT(I) ‘(so) that’ and, in particular, QUOD(/QUID) ‘because; that’ (> Gal./Sal. cu, Rom. che/que), and QU(I)A ‘because; that’ (> OIbR., SID ca, Ro. că; cf. }4.4.2.2).43 In such cases the complementizer precedes without exception its clausal complement (29a),44 as is also the case in Romance (29b), leaving us with no doubt that the complementizer phrase was head-initial from its earliest attestations.45 (29)

a legati Carteienses renuntiauerunt quod Pompeium deputies.NOM Carteian.NOM reported that Pompey.ACC in potestate haberent (Lat., B.Hisp. 36.1) in power.ABL they.had.SBJV ‘deputies arrived from Carteia reported that they had secured Pompey’ b pretind că şi-a mint¸it părint¸ii despre they.claim that self.DAT=she.has lied parents.the about fusese (Ro.) unde where she.had.been ‘they claim that she lied to her parents about where she had been’

42

Rubio ([1972] 1982: 199–200), Oniga (2004: 94), Hewson and Bubenik (2006: 1), Spevak (2010: 14–15). Palmer ([1954] 1990: 328), Haudry (1973), Bichakjian (1982), Calboli (1983b: 41–2), Lehmann (1984), Ramat (1994: 259–60, 264), Bauer (1995: 159–60; 2009a: 250, 258), Devine and Stephens (1999: 148). Indeed, examples of QUOD, and especially QUIA, following verba dicendi et sentiendi are extremely rare and sporadic until at least the end of the second century (Perrochat 1932: 141; Herman 1989: 145; Adams 2005: 195–7; Greco 2007: 12–26). 44 Note that cases like (i) below where two complementizers superficially occur in adjacent positions are not exceptions to this generalization, since it simply involves fronting of the topicalized si ‘if ’ clause to the left periphery (for discussion, see }4.4.1): 43

(i) Cibos uirides dare quod si tempus non patietur [ . . . ] utilissimum foods.ACC green.ACC.PL give.INF because if time.NOM not permits very.useful est (Lat., Pel. Ars vet. 152.2) is ‘If time is short, it is advisable to give them green fodder’ 45 Gildersleeve and Lodge ([1895] 1997: 430), Hale and Buck ([1903] 1994: 340). See, however, }}5.4.2.–1 for a discussion of the head-final analysis of the AcI construction.

206

From Latin to Romance

The case of the adposition is essentially analogous, although its original head-initial status remains controversial. What is clear, however, is that from the earliest texts the adpositional phrase is with very few exceptions head-initial,46 hence the traditional claim that Latin has prepositions. However, it is certainly true that Latin also displays, and with greater frequency in more formal and earlier texts,47 some marked examples of postposition when the complement is represented by a (relative/personal/ demonstrative) pronoun (30a–b).48 Quite exceptionally, the postpositional order had become firmly established as the norm with pronominal complements to the adposition CUM ‘with’ (30c), a pattern which survived into most early Romance varieties and is still found today in many varieties of Ibero-Romance (e.g. Pt. comigo ‘with me’, contigo ‘with you’, consigo ‘with him/her/them/you’, co(n)nosco ‘you’) and Italo-Romance (e.g. Cor. cu mecu/tecu/noscu/voscu ‘with me/you.SG/us/you.PL’). (30)

de agitur (Lat., Cic. Verr. 2.1.31) a quo which.ABL about it.be.at.stake.PASS ‘the point in question’ b Alpinas [ . . . ] niues et frigora Rheni / me Alpine.ACC.PL snows.ACC and cold.ACC.PL Rhine.GEN me.ABL sine sola uides (Lat., Virg. Ecl. 10.47–8) without along.NOM.FSG you.see ‘you gaze at Alpine snows and the frozen Rhine, without me, and alone’ c cum Caesare nobiscumque (Lat., Cic. Fam. 9.9.1) with Caesar.ABL us.ABL=with=and ‘with Caesar and with us’

Traditionally, such cases have been interpreted as (formulaic) relics of an earlier stage in which adpositions were left-branching in accordance with the general head-final tendency of (Proto-)Indo-European,49 although the switch to right-branching must have occurred quite early and, in any case, before the Italic period given the predominance of prepositions in both Umbrian and Oscan.50 Accordingly, traditional wisdom has it that Indo-European adpositions developed from original adverbial particles functioning as preverbs, which were subsequently reinterpreted

46 Gildersleeve and Lodge ([1895] 1997: 267, 432), Ernout and Thomas ([1953] 1993: 118–9), Adams (1976a: 88), Baldi (1979), Lehmann (1985: 96), Joseph (1991), Bauer (1995: 137; 2009a: 259), Vincent (1999), Hewson and Bubenik (2006: 3), Magni (2009: 229–33), Spevak (2010: 19–20). 47 Leumann and Hofmann (1928: 495), Marouzeau (1949: 42; 1953: 62, 67), Bauer (1995: 131–2, 146–7; 2009a: 291), Penney (1999: 266–7, 249–68). 48 Gildersleeve and Lodge ([1895] 1997: 267), Hale and Buck ([1903] 1994: 337), Marouzeau (1949: 44–57; 1953: 61–9), Ernout and Thomas ([1953] 1993: 119), Adams (1976a: 88), Bauer (1995: 132–7; 2009a: 260–7), Vincent (1999: 1125–6), Hewson and Bubenik (2006: 252), Magni (2009: 230). 49 Adams (1976a: 88), Bauer (1995: 137; 2009a: 252–3, 261). 50 Buck ([1933] 1979: 205–10), Bauer (1995: 130–1), Magni (200: 229).

A configurational approach

207

as postpositions since they frequently followed preverbal complements (viz. O + [preverb + V] ) [O + Postposition] + V).51 By contrast, Vincent (1999) argues that Latin (and, more generally, IndoEuropean) prepositional syntax was original. In addition to occurring in unstressed preverbal position, original preverbs could also bear focal stress and target the clauseinitial position from where they could precede the verb’s complement and, in time, be reinterpreted as governing the latter (viz. preverb + [O + V] ) [preverb + O] + V ) [preposition + O] + V). On this view, postpositional examples like those in (30a–c) are not exceptions or residues of an earlier arrangement but, rather, demonstrate that both the preverb and the pronoun could compete for the clause-initial stressed position with the preverb encliticizing to the pronoun whenever the latter was fronted under focal stress. Whatever the correct analysis, for our purposes it is sufficient to note that under both approaches adpositions represent, together with complementizers, the first categories in which a systematic head-initial pattern was to emerge. Below in }5.4.2 we shall demonstrate how the precocious changes witnessed in these two categories are responsible for triggering an overall change in the linearization of head and complement across all categories. 5.3.1.2 Comparatives In line with many other of the Indo-European languages, Latin presents two comparative constructions: the ablative and particle construction.52 The former, in which the standard of comparison originally appears in the ablative before its adjectival head, is considered to represent the older construction (Tarriño Ruiz 2009c: 650), witness its predominance in a number of older and/or set expressions such as negatives (31a), rhetorical questions implying a negative (31b), and proverbial expressions (31c). In the more recent particle construction, by contrast, 51 Joseph (1991), Bauer (1995: 130–7; 2009a: 259–61), Cuzzolin (1995), Vincent (1998a: 424 n. 11), Oniga (2004: 94), Hewson and Bubenik (2006). In a small number of other cases, postpositions were confined to adpositions of (recent) nominal origin such as CAUSA ‘cause.ABL’ > ‘for the purpose of’ (i.a) and GRATIA ‘favour.ABL’ > ‘for the sake of’ (i.b), whose postpositional status can be readily understood as the result of the original preposing of genitive nominal complements (cf. TIMOR BELLI ‘fear war.GEN’ ) BELLI timor ‘war. GEN fear’; cf. }5.3.1.4.2), although even in these cases, and especially in later texts, the complement is frequently and increasingly found following the adposition in accordance with the growing tendency not to prepose nominal complements (Ernout and Thomas [1953] 1993: 117–8; Bauer 2009a: 253; Magni 2009: 232).

(i)

a sororis causa (Lat., Pl. Trin. 686) sister.GEN cause.ABL ‘for the sake of your sister’ nuntii perferendi gratia (Lat., Caes. B.C. 2.7.3) b huius this.GEN news.GEN bring.GRD.GEN favour ‘for the sake of bringing this news’

52 Gildersleeve and Lodge ([1895] 1997: 188–9), Hale and Buck ([1903] 1994: 217–18), Bennett (1914: 292–7), Ernout and Thomas ([1953] 1993: 168–72), Adams (1976a: 83–6), Bauer (2009a: 253, 263), Magni (2009: 243–6). Besides these two constructions, Latin also boasts two further comparative structures involing AB ‘(away) from’ + ablative and DE ‘(down) from’ + ablative.

208

From Latin to Romance

the adjective precedes the standard of comparison introduced by the complementizer QUAM ‘than’ (31d). Again it is this latter head-initial construction which continues into Romance (31e). (31)

homine audacius (Lat., Pl. Men. 631) a nil hoc nothing this.ABL man.ABL braver ‘none is braver than this man’ alter audacior homo? (Lat., Pl. Amph. 153) b qui me who me.ABL other.NOM braver man.NOM ‘What other man is braver than me?’ dulci dulcior tu es (Lat., Pl. Truc. 614) c Oh melle oh honey.ABL sweet.ABL sweeter you are ‘O! Sweeter than honey you are to me’ d hominem callidiorem uidi neminem / quam Phormionem Phormio man.ACC cleverer.ACC I.saw nobody.ACC than (Lat., Ter. Phorm. 591–2) ‘I never saw a more cunning fellow than this Phormio’ e Peyres es plus savis que Guilhems (Occ.) Peyres is more clever than Guilhems ‘Peyres is cleverer than Guilhems’

Although the head-final ablative construction represents the original structure— and was to enjoy a new lease of life in later Latin (J.N. Adams p.c.)—the headinitial particle structure ‘is already well established in our earliest extant literature’ (Bennett 1914: 292). Indeed, the growing force of the innovative headinitial pattern is increasingly manifested in the ablative construction, even in early authors such as Plautus, in the placement of the standard of comparison following the adjective (32a; Adams 1976a: 83, 86). Similarly, Adams (1976a: 85), following Bennett (1914: 296–7), observes how in archaic Latin the adverbs PLUS/AMPLIUS ‘more’ and MINUS ‘less’ originally stood in apposition to a numerical phrase (32b), whereas in a subsequent development they began to be used with a numerical phrase placed in the ablative which, despite the archaicizing choice of case form, was regularly placed following the adverb in accordance with the more recent headinitial pattern (32c). (32)

a malleum sapientiorem uidi excusso manubrio hammer.ACC wiser.ACC I.saw shaken.ABL handle.ABL ‘the hammer should be wiser than the handle’

A configurational approach

209

b ego illam in aluo gesto plus annos decem (Lat., Pl. St. 160) I her in stomach.ABL carry more years.ACC ten ‘I have been carrying her in my stomach for more than ten years’ c plus triginta annis natus sum (Lat., Pl. Men. 446) more thirty years.ABL born I.am ‘I was born more than thirty years ago’ In sum, the evidence and distribution of the competing Latin comparative constructions force us to conclude with Adams (1976a: 85) that ‘[h]ere then is a further indication that by the time of Plautus the language had lost its OV characteristics’. 5.3.1.3 Relatives If Latin were a head-final language, then, typologically, we should a priori expect the construction of the relative clause to conform to that pattern with the relative preceding its antecedent (cf. however Hawkins 1983: 336, 344). Indeed, in archaic Latin this original pattern lingers on in the Proto-Indo-European correlative construction, particularly common in proverbs, directives, and legal texts such as the Leges XII Tabularum where it constitutes the predominant order (33a–c).53 (33)

a cui testimonium defuerit, is [ . . . ] obuagulatum ito wail.SUP will.go.FUT.IMP who.DAT witness.NOM lack.FUT.PFV.3SG he (Lat., Leg. XII Tab. 2) ‘He whose witness has failed to appear may summon him by loud calls’ arida erunt [ . . . ] ea omnia eximito b quae those.NOM dry.NOM will.be those.ACC all.ACC remove.FUT.IMP.2SG (Lat., Cato Agr. 44) ‘those which are dry, you have to trim them all’ uos cupide per hosce annos adpetistis atque uoluistis, ea c quae which.ACC you eagerly for these years sought and wanted it.NOM si temere repudiaritis (Lat., C. Sempronius Gracchus in Gell. 13.3) if rashly you.rejected.SBJV [ . . . ] ‘those things, which you eagerly sought and desired during that period, should you reject them rashly . . . ’

However, it is evident that this archaic pattern constituted a stylistically and pragmatically marked structure, which by early Latin (e.g. Plautus) had already long yielded to the head-initial structure (34), which now constituted the unmarked and by far most frequent construction of the relative clause.54 Once again, we see that from an early period Latin presents extensive evidence for the growing predominance of a head-initial organization. 53 Hale and Buck ([1903] 1994: 156–7), Kroll (1912: 8–13), Adams (1976a: 86–7; 1977: 71–2; 1994a: 146), Bauer (2009a: 258–9). 54 Hale and Buck ([1903] 1994: 336), Ernout and Thomas ([1953] 1993: 332–4), Bauer (1995: 159, 165; 2009a: 258–9).

210

From Latin to Romance

(34) ego quoque etiam, qui Iouis sum filius (Lat., Pl. Amph. 30) I too as.well who Jupiter.GEN I.am son.NOM ‘I too, as well, who am the son of Jupiter’ 5.3.1.4 Noun phrase

5.3.1.4.1

ADJECTIVES

Though as a typological diagnostic the position of the adjective with respect to the noun has been claimed on crosslinguistic grounds,55 as well as on the evidence of Latin and Romance,56 not to correlate robustly with other word order patterns, there are nonetheless some consistent Latin-internal patterns which incontrovertibly point towards an early shift from an original head-final AN order to the head-initial NA order (Adams 1976a: 89; Bauer 2009a: 263) that continues into Romance (cf. }3.2.2.1). Thus, although there is admittedly variation in the position of the Latin adjective,57 the available evidence would suggest that the two positions are not in free variation,58 with the NA order representing the unmarked position of the adjective in all periods. Indeed, this is by far the majority view in the literature.59 Accordingly, the basic order is NA, with objective adjectives—including relational (e.g. ROMANUS ‘Roman’), distinguishing (e.g. ALBUS ‘white’), and possessive (e.g. MEUS ‘my’) adjectives—typically following the noun from the time of the earliest texts (35a),60 while subjective adjectives—including chiefly emphatic adjectives of judgement (BONUS ‘good’) and size or quantity (MAGNUS ‘big’, MULTUS ‘much’)—are usually preposed (35b).61 (35)

55

a a senatu populi Romani amicus appellatus by senate.ABL people.GEN Roman.GEN friend.NOM called erat (Lat., Caes. B.G. 1.3.3) he.was ‘[who] had been called a friend by the senate of the Roman people’

Hawkins (1983), Dryer (1988a; 1992; 2005c), Huang (1994). Berruto (1998), Pountain (1998), Vincent (2007: 57–8 n. 3), Bauer (2009a: 245), Magni (2009: 234). 57 For an overview, see }3.2.1, and Giusti and Oniga (2007: 81–5) and Langslow (forthc.: }2). 58 Pace Feix (1934: 27), Vincent (1988: 54), Herman (2000: 83), Oniga (2004: 95), Crisma and Gianollo (2006: 73–4), Gianollo (2007: 68–9). 59 Hale and Buck ([1903] 1994: }624), Haida (1928: 11), Adams (1976a: 88; 1977: 71), Pinkster (1990a: 185), Coleman (1991: 326), Lehmann (1991: 223), Bauer (1995: 74; 2006: 298; 2009a: 265), Clackson and Horrocks (2007: 28), Magni (2009: 236–7), Tarriño Ruiz (2009a: 260–1), Spevak (2010: 225), Langslow (forthc.). Only Gildersleeve and Lodge ([1895] 1997: 430) claim the prenominal position to constitute the unmarked position in the classical language. In contrast to Adams (1977), Rizzi and Molinelli (1994: 117) find in their African Latin texts a predominance of AN over NA (15:4), with the 4 observed occurrences of NA resulting from emphatic postposing of a descriptive/subjective adjective. 60 For a more nuanced view of the position of possessives, see De Melo (2010). 61 Adams (1971: 12; 1976a: 80–1, 88–9; 1977: 71), Coleman (1991: 326), Bauer (1995: 72–3; 2006: 298), Baños Baños and Cabrillana (2009: 703). 56

A configurational approach

211

b magnam copiam in castris habebat (Lat., Caes. B.G. 1.16.5) great.ACC quantity.ACC in camps.ABL he.had ‘he had a great number in his camp’ In short, prenominal position proved stylistically marked and was reserved for adjectives with an emphatic reading, including those with a strong emotive value such as evaluative and affective adjectives and those indicating size. On this basis, Adams (1976a: 89–90) hypothesizes that with the passage from AN ) NA, which must have occurred long before Plautus as witnessed by adjectival distribution in archaic texts such as the Leges XII Tabularum (36a–b),62 it is natural to assume that objective adjectives moved to the new unmarked postnominal position, whereas subjective adjectives, with their principally emphatic readings, remained in the original prenominal position which had now become the marked position (cf. also Bauer 1995: 72–3).63 (36) a si aqua pluuia nocet (Lat., Leg. XII Tab. 7.7) if water.NOM rain.NOM harms ‘if runoff causes damage’ b Qui malum carmen incantassit (Lat., Leg. XII Tab. 8.2) who.NOM bad.ACC song.ACC had.incanted.SBJV ‘he who has incanted an evil song (= is guilty of insult)’ Thus, when objective adjectives occur in prenominal position (37a), this constitutes a marked option conveying an emphatic reading, typically contrastive focus.64 By the same token, when subjective adjectives convey a less than banal degree of emphasis, these are thrown into relief by being moved to the postnominal position (37b).65 (37)

a ad inferiorem partem fluminis Rheni (Lat., Caes. B.G. 1.1.6) to lower.ACC part.ACC river.GEN Rhine.GEN ‘to the lower (i.e. not higher) part of the Rhine’

62 Adams (1976a: 90), Bauer (1995: 66–7; 2009a: 265). On the comparative Italic evidence, see Buck ([1933] 1979: 224), Konneker (1975: 367–70), Bauer (1995: 66; 2009a: 263). 63 On the unmarked prenominal position of the Proto-Indo-European adjective, see Delbrück (1990: 98, 100), Lehmann (1974: 69), and Bauer (1995: 65–6); for an alternative view, see Hofmann and Szantyr (1965: 406). Somewhat implausibly, Bauer (2001a; 2009a: 265) claims that the retention of the AN pattern once the AN ) NA shift had taken place was made possible by the fact that such sequences were short and noncomplex constructions. 64 Marouzeau (1922: 17–32, 34–56), Adams (1976a: 89), Väänänen (1982: 260), Pinkster (1990a: 185), Bauer (1995: 68–9; 2009a: 263), Crisma and Gianollo (2006: 82), Gianollo (2006: 153; 2007: 68). Significantly, Langslow (forthc.) presents powerful evidence from classical prose to show that, contrary to what has often been claimed in the literature (Marouzeau 1922: 17–56; Bauer 1995: 70; 2009a: 263; Vincent 2007: 64–5), the prenominal position does not license per se figurative readings, which are instead regularly licensed in the unmarked postnominal position, but is specialized in marking contextual focus alone. 65 Marouzeau (1922: 54–6; 1953: 4), Adams (1971: 12), Bauer (1995: 70).

212

From Latin to Romance b C. Uolusenus, tribunus militum, uir et consilii C. Volusenus.NOM tribune.NOM soldiers.GEN man and judgement.GEN magni et uirtutis (Lat., Caes. B.G. 3.5.2) great.GEN and valour.GEN ‘C. Volusenus, a tribune of the soldiers, a man of great skill and valour’

Furthermore, Adams (1976a: 89) argues that the early shift from an original AN to an NA order is betrayed in a number of adjective-noun formulae of great antiquity (see also Spevak 2010: 228–9).66 Starting from the observation that in old Latin religious epithets were thrown into prominence by postposition, the original marked position (cf. }5.3.1.4.2), then early fixed collocations such as CAMPUS MARTIUS ‘field of.Mars’ (‘The Field of Mars’), UIRGO UESTALIS ‘virgin of.Vesta’ (‘The Vestal Virgin’), and COLLIS QUIRINALIS ‘hill of.Quirinus’ (‘The Quirinal Hill’) necessarily imply that there must have existed a corresponding unmarked AN order. Furthermore, he highlights (pp. 89–90) a fossilized example of the archaic unmarked AN order in the set phrase ERILIS FILIUS ‘master’s son’ (38a), which contrasts with other more spontaneous collocations of ERILIS ‘of the master’ where it usually assumes the modern unmarked NA order (38b): (38)

filio hanc fabricam dabo (Lat., Pl. Bacch. 366) a erili of.master.DAT son.DAT this.ACC trick.ACC I.will.give ‘I’ll tell my master’s son this contrivance’ b Di te seruassint semper, custos erilis (Lat., Pl. Asin. 655) gods.NOM you had.served.SBJV always guardian of.master ‘May the Deities ever preserve you, protector of your master’

To these examples, we may also add the Romance adverbial formation in -ment(e) (39a), which has its origins in the Latin periphrastic formation adjective + MENTE ‘mind.ABL’,67 attested, though not necessarily with a high degree of frequency (Karlsson 1981: 44), since early Latin (39b). The steadfast position of the prenominal adjective in this erstwhile periphrasis undoubtedly reflects a fossilization of the unmarked status of the prenominal adjectival position in the archaic period. (39)

a cobrava es preu que realment volia cobrar (Men.) he.received the price that really he.wanted receive.INF ‘he would receive the pay he really wanted’

66 By contrast, Baldi and Nuti (2010: 366–8) argue that the shift from AN to NA occurred later with the possessive adjective, inasmuch as the postnominal possessive still carried a marked interpretation in the archaic age and only became numerically predominant and unmarked in the early period. 67 Cf. }2.2.2; Karlsson (1981), Bauer (2001b; 2003; 2006), Ricca (2010: 181–5).

A configurational approach

213

b ille trauersa mente mihi hodie tradidit repagula he unexpected.ABL mind.ABL me.DAT today handed bars.ACC (Lat., Enn. Trag. 229) ‘he has today unexpectedly delivered to me the bars’68 In summary, the evidence of adjectival positions corroborates the patterns already observed above for other categories which uniformly point to the ongoing establishment of an unmarked head-initial order since earliest times.

5.3.1.4.2

GENITIVES

The placement of genitive constituents in the history of Latin is not so different from that of the adjective just witnessed (cf. Magni 2009: 241–2). In particular, genitives may occur both before and after their associated nominal head. Once again, however, the observed variation is not free but is constrained, among other things, by both chronological and pragmatic factors. In relation to the former, it is widely acknowledged that in the archaic period prenominal position was favoured (e.g. senatuos sententiad (S.C. Bacch. 17) ‘by decision of the senate (lit. senate.GEN decision.ABL)’),69 whereas by the late period postposition had came to predominate (e.g. passionem Domini (Per. Aeth. 43.3) ‘the passion of the Lord (lit. passion.ACC Lord.GEN)’) in line with the established Romance pattern (e.g. Glc. o noso pan de cada día ‘our daily bread (lit. the our bread of each day)’).70 Indeed, this conclusion is confirmed by a crude examination of the various statistical surveys reported in the literature,71 reported here in Table 5.1, which highlight a shift away from the near-equal distribution of pre- and postposition genitives in the classical language towards the predominance of postposition by the late period, with the exception of some artificially archaic religious (e.g. Braulius) and medieval texts. On a par with what was seen with the adjective, the unmarked prenominal position of the genitive in archaic Latin is clearly betrayed by its exceptional survival into early and classical Latin in a number of formulae and set expressions of considerable antiquity (e.g. senatus consultum (Liv. 2.48.10) ‘decree of the Senate (lit. senate.GEN decree.ACC)’, plebi scitum (Liv. 4.49.6) ‘plebiscite (lit. people. GEN decree.ACC)’), patronymic expressions (e.g. Alcumena [ . . . ] Electri filia

68 Cf. Rosén (1999: 57) for the translation ‘unexpectedly’ of trauersa mente, which minimally differs from the Loeb translation ‘with misguided mind’. 69 Bakkum (2009: 301–2) also concludes that the unmarked order in Faliscan is regularly GN, although he concedes, in view of the limits of the corpus, that the pronominal position might be due to pragmatic factors such as focus. 70 Gildersleeve and Lodge ([1895] 1997: 431), Ernout and Thomas ([1953] 1993: 162), Adams (1976a: 73–4; 1976b: 140; 1977: 70–1), Gianollo (2007: 76–7), Bauer (1995: 59–62; 2009a: 265–8), Magni (2009: 238), Baldi and Nuti (2010: 373). 71 Cf. also Gianollo (2005; 2007: 66–7), Baños Baños and Cabrillana (2009: 701), Mclachlan (forthc.: ch. 3).

214

From Latin to Romance

TABLE 5.1 Distribution of GN / NG across different Latin authors/texts Text / Author

GN / NG

S.C. Bacch. (Álvarez Pedrosa 1988)

41.7%:58.3%

Leges 2-c. B.C. (Álvarez Pedrosa 1988)

54.2%:45.8%

Var. Rust. (Mclachlan forthc.)

49.3%:50.7%

Cic. Att. 1 (Cabrillana 1993a)

42.5%:57.5%

Cic. Att. 1–3 (Mclachlan forthc.)

45.1%:54.9%

Cic. Cat. 1–2 (Mclachlan forthc.)

48.4%:51.6%

Cic. Mil. (Panchón 1986)

50.6%:49.4%

Cic. De or. (Lisón 2001)

40.2%:59.8%

Cic. Fam. 1.9 (Gianollo 2007)

51.9%:48.1%

Caes. B.G. 1 (Panchón 1986)

48.5%:51.5%

Caes. B.G. 1–3 (Mclachlan forthc.)

46.3%:53.7%

Sall. Cat. (Gutiérrez Ordóñez 1994)

42.5%:57.4%

Sall. Cat. (Mclachlan forthc.)

43.6%:56.4%

Sall. Iug. (Mclachlan forthc.)

42.1%:57.9%

Liv. 26–28 (Lisón 2001)

47.7%:52.3%

Petr. Cena Trim. (Hinojo 1985)

44.1%:55.9%

Petr. Cena Trim. (Gianollo 2007)

28.4%:71.6%

Per. Aeth. (Hinojo 1986)

7.5%:92.5%

Per. Aeth. (Gianollo 2005)

6.5%:93.5%

Vulg. Joel (García de la Fuente 1983) Vulg. Mat. (Gianollo 2007) Vulg. Mar. (Gianollo 2007) Braulius (8th-c. AD; García Sanchidrián 1994) Medieval Latin (Hinojo 2002)

0%:100% 3%:97% 3.6%:96.4% 45.8%:54.2% 48%:52%

(Pl. Amph. 99) ‘Alcmena [ . . . ] the daughter of Electrus (lit. Alcmena . . . Electrus.GEN daughter.NOM)’), and legal expressions (e.g. vitae necisque potestas (Sall. Iug. 14) ‘power of life and death (lit. life.GEN death.GEN=and power.NOM)’),72 whereas many of these same genitives shift to the postnominal position when employed in later expressions (e.g. acta senatus (Suet. Aug. 36) ‘minutes of the senate (lit. minutes.NOM senate.GEN)’). Within this category also fall a number of postpositions of nominal origin such CAUSA (cause.ABL) > ‘for the purpose of’ and GRATIA (favour.ABL) ‘for the 72

Bennett (1914: 51–2), Ernout and Thomas ([1953] 1993: 162), Adams (1976a: 74–5), Bauer (2009a: 266–7), Magni (2009: 240), Baldi and Nuti (2010: 370–2).

A configurational approach

215

sake of’, which were seen (cf. footnote 51) to take, at least originally, a preposed genitive complement in accordance with the unmarked GN order at the time of their coinage. As with prenominal adjectives (cf. }5.3.1.4.1), indirect evidence for the unmarked nature of prenominal genitives in archaic Latin is provided once again by the observation that religious epithets and titles, whether genitives or adjectives, were thrown into prominence in archaic Latin by postposition, giving rise to fossilized collocations such as pater familias (Cic. Cat. 4.12) ‘head of the household (lit. father.NOM family.GEN)’, tribunus plebis (Cic. Brut. 1.1) ‘tribune of the people (lit. tribune.NOM people.GEN)’, sacerdos Cereris (CIL I2.974) ‘priestess of Ceres, Goddess of agriculture (lit. priestess.NOM Ceres.GEN)’.73 By contrast, in those increasingly rare cases in which the prenominal position obtains in late Latin, by which time the synthetic genitive had also begun to find itself in competition from a postnominal prepositional periphrasis with DE ‘of’ in subliterate registers, the examples mostly involve invariable formulae and official phrases such as acquae ductum (A.Vales. 71) ‘acquaduct (lit. water.GEN line.ACC)’ and dei famulus (A.Vales. 88) ‘servant of God (lit. god.GEN servant.NOM)’.74 Aside from such examples, during this period the GN order was increasingly becoming a purely prestige order with little or no place in subliterary registers (Adams 1976a: 73–4), only marginally surviving as a deliberate stylistic affectation in the literary writings of a highly educated minority who continued to have some grasp of the artificial classical language (Adams 1976a: 82). Revealing in this respect are the results of a survey of the first 10 pages of Pompeius’ fifth–sixth-century commentary of Donato’s Ars gramatica (Keil 1868), which reveals 44 examples of dependent genitives, the vast majority of which were found to occur in postnominal position (viz. 32:12), e.g. ab octo partibus orationis ‘from eight parts of speech’ (96.9), officio nominis ‘role of the noun’ (97.4), motu linguae (97.7) ‘with the movement of the tongue’, nomina partium ‘the name of the parts [of speech]’ (98.4), pars vocis ‘part of the word’ (99.20), mugitus boum ‘the lowing of the cows’ (99.18) vs oris ratio ‘the reason for the voice’ (96.19), nominis iteratio ‘repetition of the noun’ (97.2), vocalium sonos ‘the sounds of the vowels’ (100.7). Classical Latin, by contrast, represents an intermediate position in this development in which the inconsistent statistical evidence of individual texts and/or authors has led to the traditional claim that it is not possible to identify a predominant order, with most scholars concluding that pre- and postposed genitives occur

73

Adams (1976a: 75–6), Bauer (1995: 53; 2009a: 265), Magni (2009: 238). Oscan and Umbrian present similar evidence for the unmarked and marked interpretation of pre- and postnominal genitives, respectively (Rosenkranz 1933; Adams 1976a: 73; Bauer 1995: 51, 53; Magni 2009: 238). 74 Adams (1976a: 73; 1976b: 140; 1977: 71), Gianollo (2007: 68), Bauer (2009a: 267), Baldi and Nuti (2010: 372).

216

From Latin to Romance

in equal number in the classical language.75 Rather than constituting free variation, however, genitive placement appears to follow precise pragmatic principles (pace Devine and Stephens 2006: 314), largely similar to those already identified above for the adjective in classical Latin (cf. }5.3.1.4.1). More specifically, prenominal position constitutes a stylistically marked option from as early as the time of Plautus (cf. Mclachlan forthc.: ch. 3, }1), typically obtaining when the genitive is placed under contrastive focus (40a) or is highly topical as typically obtains with pronominal genitives (40b).76 Indeed, the parallel with adjectival positions is further substantiated by the observation that if the genitive is qualified by a focused adjective, itself therefore placed in prenominal position, then both adjective and genitive are preposed (40c) in accordance with a tendency towards crosscategorial harmonic linearization (cf. }5.4.2.3). (40)

a sub Ueneris regno uapulo, non sub Iouis under Venus.GEN realm.ABL I.am.flogged not under Jupiter.GEN (Lat., Pl. Pseud. 15) ‘under the sway of Venus am I harassed, not under that of Jove’ legam (Lat., Pl. Pseud. 414) b horum sermonem these.GEN conversation.ACC I.will.pick.up ‘I will pick up their conversation’ ponderis saxa (Lat., Caes. B.G. 2.29.3) c magni great.GEN weight.GEN rocks ‘rocks of heavy weight’

However, recent work by Mclachlan (forthc.) convincingly demonstrates that pragmatic factors alone cannot explain all such variation in the classical language, and that genitive placement must ultimately be explained by a combination of conditioning factors. Thus, while she finds that pragmatically marked constituents, be these the dependent genitive or the head noun, bearing emphasis or contrast are more likely to be preposed—hence the order GN (41a) predominantly encodes a pragmatically marked interpretation of the dependent genitive, whereas the opposite NG order (41b) typically correlates with a marked reading of the nominal head—not all cases can be so straightforwardly accounted for in terms of information structure, witness (42a–b).

75

Adams (1976a: 77–8), Elerick (1994: 70), Bauer (1995: 55–9; 2009a: 266), Herman (2000: 83), Oniga (2004: 76), Crisma and Gianollo (2006: 73), Gianollo (2006: 152–3; 2007: 66–7), Baños Baños and Cabrillana (2009: 704–5), Baldi and Nuti (2010: 368–9), Spevak (2010: 265). 76 Adams (1976a: 78), de Jong (1983: 131–2, 135–4), Bauer (1995: 56–7, 64; 2009a: 266), Herman (2000: 83), Baldi and Nuti (2010: 371–2).

A configurational approach (41)

217

a Caesar obsidibus acceptis primis civitatis Caesar.NOM hostages.ABL received.ABL.PL foremost.ABL.PL state.GEN atque ipsius Galbae Regis duobus and self.GEN Galb.GEN king.GEN two.ABL filiis [ . . . ] Suessiones accepit (Lat., Caes. B.G. 2.13.1) sons.ABL Suessiones.ACC he.received ‘Upon receiving as hostages the leading men of the state and the two sons of king Galba himself [ . . . ] Caesar received the surrender of the Suessiones’ b me uero teste producto credo te ex acclamatione me.ABL in.truth witness.ABL produced.ABL I.believe you.ACC out.of shout.ABL Clodi aduocatorum audisse quae consurrectio Clodius.GEN supporters.GEN hear.INF.PFV which.NOM.FSG standing.up.NOM.FSG iudicum facta sit (Lat., Cic. Att. 1.16.4) jurors.GEN done.NOM.FSG is.SBJV ‘When I myself was called as a witness, you must have heard from the shouts of Clodius’ supporters how the jury rose in a body [what an uprising of the jurors occurred]’

publicae magnitudine belli (42) a Ad hoc mulieres, quibus rei to this.ACC women.NOM who.DAT.PL thing.GEN public.GEN greatness.ABL war.GEN timor insolitus incesserat, adflictare sese (Lat., Sall. Cat. 31.3) fear.NOM strange.NOM had.attacked distress.INF selves.ACC ‘The women also, seized by the terror of war, from which the vastness of the republic had shielded them, were greatly distressed’ b Certe, inquit Merula; nam ibi uidi greges magnos certainly said Merula.NOM for there I.saw flocks.ACC big.ACC.PL anserum, gallinarum, columbarum, gruum, pauonum, nec non geese.GEN chickens.GEN pigeons.GEN cranes.GEN peafowl.GEN nor not glirium, piscium, aprorum, ceterae uenationis (Var. Rust. 3.2.14) dormice.GEN fish.GEN boars.GEN other.GEN game.GEN ‘ “You are quite right,” said Merula; “for I have seen there large flocks of geese, chickens, pigeons, cranes, and peafowl, not to speak of numbers of dormice, fish, boars, and other game” ’ Unfortunately, here it is not possible to do justice to Mclachlan’s highly illuminating study which presents the first and much-needed in-depth investigation of genitive placement in classical Latin. Here it will suffice to point out the main conditioning factors identified by Mclachlan. In particular, she notes that in many cases genitive placement conflicts with pragmatic expectations, highlighting how linearization of head and genitive dependent may be determined, among other things, by the function of the genitive (e.g. subjective, objective, partitive, possessive) and the lexical

218

From Latin to Romance

category instantiated by the genitive (e.g. noun, pronoun, substantivized adjective, gerund). For instance, her findings highlight how, with the exception of Sallust’s Iug., in possessive function genitives are typically fronted (43a), whereas in partitive function they usually follow their head (43b). Similarly, a mixed picture emerges from a consideration of the different functions of the genitive dependent including, for example, the typically preposed and postposed positions, respectively, of subjective (43c) and objective (43d) genitives in most (though not all) texts. copias a nostris milia passuum quattuor (43) a Ariouisti Ariovistus.GEN troops.ACC by ours.ABL thousand steps.GEN four et uiginti abesse (Lat., Caes. B.G. 1.41.5) and twenty be.distant.INF ‘Ariovistus’ troops were twenty-four miles away from ours’ b nec dat tantum olei (Var. Rust. 1.55.1) nor it.gives so.much.ACC oil.GEN ‘and does not yield so much oil’ impetus tardatus est (Lat., Caes. B.G. 2.25.3) c paulum hostium a.little enemies.GEN attack.NOM hindered is ‘the enemy’s attack was delayed for a while’ d in suppliciis deorum magnifici [ . . . ] errant (Lat., Sall. Cat. 9.2) in prayers.ABL gods.GEN grand.NOM.PL they.were ‘they were lavish in their worshop of the gods’ In short, the results of Mclachlan’s study point to a highly nuanced and more complex series of distributional patterns than has traditionally been recognized in the classical texts, which show how in the transition from an original head-final to an innovative head-initial order the earlier arrangement acquires new, and increasingly pragmatically marked, functions and nuances (cf. also }5.4.3.2.1).

5.3.1.4.3

CONCLUDING REMARKS

Although our examination of the position of adjectives and genitives has revealed that in the classical language the placement of head and complement/modifier is not rigidly fixed (Oniga 2004: 75; Baños Baños and Cabrillana 2009: 701), the overwhelming evidence speaks in favour of an unmarked head-initial order, attested at least since early Latin and firmly established by the late period, contrasting with a marked head-final order that represents a residue of an archaic stage. Furthermore, unlike the clausal domain where, as we have seen (cf. also }5.3.2), the final position of the verb in the SOV arrangement is often, rightly or wrongly, considered to represent the unmarked linearization, many scholars have noted that in the nominal domain the unmarked and most frequent linearization of genitive subject, genitive object, and

A configurational approach

219

nominal is SGenNOGen (44a).77 Thus, although it is also possible to find both subjective and objective genitive complements following the nominal head, namely NSGenOGen (44b), it is much rarer to find them both preceding the noun, namely SGenOGenN (44c).78 (44)

a repentinam eius defensionem Gabini (Lat., Cic. Fam. 1.9.20) sudden.ACC his defence.ACC Gabinius.GEN ‘his sudden defence of Gabinius’ b memoria tua nostrae coniunctionis (Lat., Cic. Fam. 6.17) memory.NOM your.NOM our.GEN union.GEN ‘your remembrance of our friendship’ c C. Caesaris pecuniarum translatio (Lat., Cic. Off. 1.43) C. Caesar.GEN money.GEN.PL transfer.NOM ‘C. Caesar’s transfer of money’

Data like these and, in particular, observations regarding the relative frequency and rarity of SNO and SON orders, respectively, serve once again to highlight how headinitial SNO order, which ‘corresponds to a finite verb sentence-type SVO’ (Adams 1976a: 81), had already become firmly established as early as the classical period, eventually ousting SON entirely by the late period (Crisma and Gianollo 2006: 86–8; Gianollo 2007: 76–7). Once again it would appear that, while the stylized head-final (S)OV order was artificially preserved under the normative pressures of formal and/ or literary language, the corresponding (S)ON order in the nominal domain had long been surpassed by the rise of a predominant (S)NO order which, despite its headinitial arrangement, escaped the attention of grammarians and rhetoricians. 5.3.1.5 Other categories In addition to the cases just reviewed, there are also a number of other categories and structures which in the classical language, and even earlier, present evidence for a head-initial order. One of the most important of these is coordination.79 Latin boasts two series of copulative and disjunctive coordinators, one postpositive and enclitic (viz. -QUE ‘=and’, -UE ‘=or’) and the other prepositive (viz. ET, AC/ATQUE ‘and’, AUT, UEL ‘or’). According to Stassen (2000; 2001), typologically postpositive coordinators characterize head-final languages and prepositive coordinators 77 Adams (1976a: 81, 98), Gianollo (2005; 2006: 155–6; 2007: 73), Devine and Stephens (2006: 316), Giusti and Oniga (2007: 87). 78 Adams (1976a: 81), Gianollo (2007: 74). In her corpus of classical Latin authors, Gianollo (2005; 2007: 74) identifies the following statistical distributions: SGenNOGen = 18, OGenNSGen = 0, SGenOGenN = 10, OGenSGenN = 0, NSGenOGen = 6, and NOGenSGen = 3. For an analysis of these variations in terms of variable degrees of N(oun)-raising, namely S O N ) S N O N ) N S N O N, see Giusti and Oniga (2006; 2007). By contrast, we shall argue below that (early/classical) Latin lacks N-raising, in the same way that it lacks V(erb)-raising, with the variable position of the N(ominal) (and V(erbal)) head reflecting the raising or otherwise of the object (cf. }5.4.1, }5.4.3). 79 Ernout and Thomas ([1953] 1993: 440–50), Hofmann and Szantyr (1965: 469–84, 498–504), Pinkster (1969; 1990a, c), Magni (2009: 233–4), Torrego (2009a), Spevak (2010: 17–19).

From Latin to Romance

220

head-initial languages, which formally we can capture in terms of a coordination phrase (viz. &P) as in (45a).80 Here the coordinator is understood as the head of the phrase and the first and second conjuncts its left-peripheral modifier/ Spec(ifier) and complement, respectively (Kayne 1994). If the complement, the second conjunct, remains in situ, then a prepositive coordination structure obtains as in (45b). If, however, the complement is preposed (to an inner left-peripheral modifier/Spec(ifier) position), then a postpositive head-final construction like (45c) is derived.81 (45) a

[&P [Spec&P conjunct1] [&' coordinator [COMPL conjunct2] ]]

publica] ]] b meruimus . . . de [&P [Spec&P uobis] [&' et [COMPL re we.deserved from you.ABL and thing.ABL public.ABL (Lat., Pl. Amph. 39–40) ‘we deserved it [ . . . ] from you and from your state’ c [&P [Spec&P domi] [Spec&P duelli] [&' que [COMPL duelli] ]] male home.LOC war.LOC =and badly fecisti (Lat., Pl. Asin. 559) you.did ‘in peace and in warfare you have performed villainously’ Significantly, already in Latin the evidence for the postpositive coordinators inherited from Indo-European is rather meagre, with ‘the loss of postpositive connectors at a relatively early period’ (Torrego 2009a: 455; cf. also Torrego 2009b: 507–8). For example, the postpositive disjunctive coordinator -UE ‘or’, while still frequent in archaic legal texts such as the Leges XII Tabularum (e.g. morbus aeuitatue (1.4) lit. ‘illness age=or’), was already recessive by the time of Plautus (Torrego 2009a: 456–7) where it is chiefly found in fossilized formulae (e.g. plus minusue (Capt. 995) lit. ‘more less=or’) and generally replaced by prepositive AUT (e.g. iniuste aut grauiter (Capt. 308) lit. ‘unjustly or severely’), eventually falling out of the language as a productive coordinator by the first century BC (Hofmann and Szantyr 1965: 502–3; Janson 1979: 103). By contrast, postpostive -QUE persists longer,82 still constituting the principal copulative coordinator in early writers such as Cato (e.g. de domino bono bonoque aedificatore (Agr. 1) lit. ‘from owner.ABL good.ABL good.ABL=and builder.ABL’), though not Plautus where it had already been replaced by ET (e.g. uestem et nomina (Capt. 37)

80 It is also widely argued that the difference between the two series was also pragmatic; for an overview, see Magni (2009: 234). 81 Note that the possibilities are actually richer than this, in that the head alone of the second conjunct may raise, stranding adjectives and other modifiers in their in situ position (Hale and Buck ([1903] 1994: 165). In such cases, it may be argued that the raised head left-adjoins into the conjunction head. 82 Gildersleeve and Lodge ([1895] 1997: 300), Hofmann and Szartyr (1965: 473–5), Torrego (2009a: 457).

A configurational approach

221

lit. ‘garment.ACC and names.ACC’; Torrego 2009a: 458), and also in legal language (Penney 2005). The evidence briefly reviewed here from the distribution of post- and prepositive coordination replicates very closely the patterns seen above time and time again for other categories. In particular, we witness from earliest times an unmistakable shift away from an archaic head-final pattern (postpositive coordination) inherited from Indo-European towards a rapidly expanding innovative head-initial pattern (prepositive coordination) that survives into Romance (e.g. Ro. pâine şi brânză ‘bread and cheese’). Very briefly, similar evidence for early head-initial orders is also observable with polarity markers such as polar interrogative particles like postpositive -NE ‘is it the case that . . . ?’ (46a) and NUM ‘surely it’s not the case that . . . ?’ (46b), which since earliest times occur in clause-initial position (Brown, Joseph, and Wallace 2009: 502–6; Spevak 2010: 15) in accordance with a head-initial pattern.83 (46) a uoltisne eamus uisere? (Lat., Ter. Phorm. 102) you.want=Q we.go.SBJV visit.INF ‘do you want us to go and visit her?’ b num negare audes? (Lat., Cic. Catil. 1.8) Q deny.INF you.dare ‘you don’t dare deny it, do you?’ An analogous picture arises from a consideration of Latin negation. Putting aside a number of exceptions, typological investigations have revealed that, whether as a prefix or an independent word, SVO languages most typically display preverbal negation,84 whereas SOV languages commonly show postverbal negation (see also Dahl 1979; Payne 1985). Within this perspective, the preverbal position of Latin negation, whether as an independent word (47a) or as an incorporated prefix (47b), thus proves entirely consistent with a head-initial typology.85 To these examples, we may also add the case of the negative imperative formed by means of the negative modals NOLI/NOLITE (lit. ‘not.wish.IMP2SG/PL’) + infinitive,86 which equally show the head-initial order with the infinitive following the auxiliary (47c). (47) a etsi non ualuit (Lat., Nep. Milt. 3) although not had.force ‘although it did not take effect’ 83 Dryer (1992; 2005a, b), Biberauer, Newton, and Sheehan (2009a: 703–5; 2009b: 11-15), Biberauer and Sheehan (in press: }2.1), Biberauer, Sheehan, and Newton (2010). 84 Dryer (1988b; 1992), Morimoto (2001), Dryer and Haspelmath (2011: ch. 144). 85 However, Biberauer, Holmberg, and Roberts (2010: 81–7) argue that negation is a syncategorematic element which falls outside of the predictions of the FOFC. 86 Ernout and Thomas ([1953] 1993: 231–2), Palmer (2001: 106), Magni (2010: 265).

222

From Latin to Romance b Ita credo hoc illi nesciebant (Lat., Cic. Orat. 155) thus I.believe this.ACC those.ones.NOM not.they.knew ‘I thus believe that they did not know this’ c nolite existumare (Lat., Sall. Cat. 52.19) not.wish.2PL esteem.INF ‘do not go thinking’

We must also discuss here the placement of adverbs. Expectations in an OV language are that adverbs should precede the verb, whereas they should follow in a VO language. While Ricca (2010: 123–7) rightly urges caution against reading too much into adverb placement in view of the role of pragmatic factors in Latin word order, he acknowledges that in some cases such as (48a–b) ‘it would be quite a challenge to find any motivation for the [ . . . ] alternation between pre- and post-verbal position of the adverb, given the absolute identity of wording, writer, text type, general discourse organization, and local information structure’ (p. 126): (48) a L. Cossinio, amico et tribuli tuo, ualde familiariter L. Cossinius friend.ABL and tribesman.ABL your.ABL very intimately utor (Lat., Cic. Fam. 13.23.1) I.use ‘I am on very familiar terms with your friend and fellow-tribesman, L. Cossinius’ b C. Flauio, honesto et ornato equite Romano, utor C. Flavius, honest.ABL and distinguished.ABL knight.ABL Roman.ABL I.use ualde familiariter (Lat., Cic. Fam. 13.31.1) very ‘I am on very familiar terms with G. Flavius, an honourable and distinguished Roman knight’ Indeed, the frequency of postverbal placement is highlighted by Adams (1976a: 90–1) who, in his survey of prepositional expressions in adverbial function, finds that in early (e.g. Plautus) and late (e.g. Anonymus Valesianus) texts postverbal position predominates, markedly so in the latter case.87 By contrast, in archaic texts (e.g. Leges XII Tabularum) and in classical authors (e.g. Caesar) preverbal position prevails. This distribution allows him to conclude that the change from pre- to postverbal position most likely occurred in early Latin, but was artificially obscured in the classical literary language where there was an inexorable normative tendency to place the verb in final position. This view finds support in the observation that in 87 See also Linde (1923: 171), Adams (1976b: 139), Bauer (2009a: 270). These figures, however, fail to take into consideration the role and influence of pragmatic factors, which will have to be included in any future large-scale investigations of adverb(ial) placement in Latin, a major desideratum for future research.

A configurational approach

223

subliterary texts such as Terentianus postposition is clearly preferred, namely 47:16 (Adams 1977: 72). Finally, we conclude with a brief consideration of the order of the auxiliary ESSE ‘be’ + PPtP. We begin by noting that, although the preferred classical order is generally considered to be PPtP + auxiliary (49a),88 the opposite order is nonetheless equally found (49b). (49) a cum ab hora septima ad uesperum pugnatum although from hour.ABL seven.ABL to evening.ACC fought sit (Lat., Caes. B.G. 1.26.2) is.SBJV ‘although the battle lasted from noon until eventide’ b Diu cum esset pugnatum (Lat., Caes. B.G. 1.26.4) long.time when was.SBJV fought ‘after the battle had lasted some time’ Significantly, in her statistical survey Bauer (2006: 294) finds that head-initial order shows a steady increase in the history of Latin with a concomitant decline in the original head-final order. For instance, the head-final order is the only attested pattern in the archaic Leges XII Tabularum (50a) and the early S.C. de Bacchanalibus (50b), whereas in Cato the order already co-occurs alongside the innovative headinitial order, albeit at a ratio of 15 to 1. In later authors, the incidence of the head-final order steadily decreases, 10 to 1 in Plautus and 4 to 1 in Terence (50c–d; Muldowney 1937: 130; Marouzeau 1938: 24), and is even overtaken by the head-initial order in Gaius (Orinsky 1923: 91). Although clearly tentative, evidence like this, which needs to be extended to cover a greater textual and chronological sample, is indicative once again of a shift from an original head-final towards a head-initial positioning of the auxiliary. (50) a eo dies diffissus esto (Lat., Leg. XII Tab. 2.1) that.ABL day.ABL divided be.FUT.IMP ‘on that day the hearing shall be deferred’ esent (Lat., S.C. Bacch. 5) b ubei eorum u[e]r[b]a audita when their words.NOM heard.NOM.PL were.SBJV ‘when their words had been heard’ c paene oblitus sum relicuom dicere (Lat., Pl. Poen. 118) almost forgot I.am remainder.ACC say.INF ‘but I almost forgot to tell you the rest’

88

Möbitz (1924: 124, 126), Marouzeau (1938: 23–7), Bauer (1995: 104–5; 2006: 294), Salvi (2004: 112 n. 62), Devine and Stephens (2006: 180–98), Vincent (2007: 65), Bailey (2008: }2), Spevak (2010: 150–1, 155).

224

From Latin to Romance d est quod domi dicere paene fui oblitus (Lat., Pl. Pseud. 171) is who home.LOC say.INF almost I.was forgot ‘there is something which I almost forgot to say at home’

5.3.1.6 Summary The preceding evidence leads us to conclude that Latin can be considered to have reached an advanced stage in a transition from an original headfinal to a head-initial arrangement, whilst still recognizing that (classical) Latin displays a less than consistent setting in this respect.89 In this light, the head-initial sequences identified above for Latin can be best captured in terms of an ongoing unmarked order which, since at least early Latin, had for the most part begun to replace an archaic head-final order which, although attested throughout the history of Latin, especially in the more normative registers of the literary language, was now largely relegated to functioning as a marked, stylistic variant. Superficially, then, the word order of Latin appears to admit a certain degree of freedom, since the shift from a relatively loose head-final order to an increasingly rigid head-initial order is a long and slow process which only comes to completion in the late period (e.g. NG order) and, in some cases, as late as the early Romance period (e.g. AUX + (P)PtP). By contrast, the head parameter assumes a much more rigid interpretation by the time of Romance, where, among other things, dependent adjectives, genitives, and relatives consistently follow nominals (51a–b), nominals follow adpositions (51c), objects and adverbs follow verbs (51d–e), and lexical verbs follows auxiliaries (51f): (51)

a sa mákkina ruja de Juanne (Srd.) the car red of Juanne ‘Juanne’s red car’ b Guillermo quien construyó esta casa (Sp.) Guillermo who built this house ‘Guillermo who built this house’ mûrs (Frl.) c sui on.the walls d menèri las vacas a l’ abeurador (Occ.) I.led the cows to the trough ‘I led the cows to the drinking-trough’ e u 7v 7aŋ 7aniŋ (Cai.) it= rains slow slow.DIM ‘it is raining very lightly’ f quere chover (Glc.) it.wants rain.INF ‘it’s about to rain’

89

Grandgent (1907: 31), Bauer (1995: 4), Oniga (2004: 52).

A configurational approach

225

Overall, then, the conclusion to be drawn from the Latin evidence is that of a language displaying two tendencies which contrast an archaic pattern of head-last structures with an innovative pattern of head-first structures, in which ‘Classic Latin may be said to represent an intermediate stage, while the revolution was in progress’ (Grandgent 1907: 31). Indeed, on this point there is general consensus in the literature that IndoEuropean was predominantly head-last, while its grammatical structures in the transition to the daughter languages increasingly became head-first,90 a change largely completed by the early Latin period but whose effects in the classical period were in many cases masked by the deliberately archaizing patterns of the literary language. To this it must be added that the relevant change in directionality of the parameter did not occur at the same time in all areas of the grammar, but, rather, displays a staggered development proceeding at different rates in different morphosyntactic domains. To begin with, it is widely acknowledged that head-first structures emerge much earlier in syntactic than in morphological structures (Oniga 2004: 52, 103): in the area of derivational morphology, for example, Latin compounds, apart from a handful of rare (and usually late) exceptions (Adams 1976a: 91) such as FULCIPEDIA lit. ‘prop.feet’ (‘harlot’) and UERSIPELLIS lit. ‘change.skin’ (‘werewolf’), still strongly resist head-first formations, and the head-final patterns of Latin inflectional morphology are still relatively well preserved in Romance verbal and nominal formations (e.g. Srs. ils fretgs ‘the.PL fruits’, (literary) Pt. chamara ‘call.PLPF3SG’). As for syntactic structures, the first signs of the head-first setting have been shown to occur in complementizer and adposition phrases, then in the nominal group, and only spreading to the verbal group at a later date,91 with a number of head-last constructions surviving well into the Romance period (e.g. verb-final orders in French relatives until the seventeenth century) and, in some cases, even up until the present-day (e.g. Sp. synthetic future llorará ‘cry.FUT3SG’, though now rivalled by the head-first GO-future va a llorar lit. ‘he.goes to cry.INF’). It is to the verbal group that we now turn. 5.3.2 Verbal group It is generally assumed that the order of verb and complement in Indo-European was OV (for an overview, see Bauer 1995: 86–9), an order widely preserved in conservative and, in particular, legal texts such as the Leges XII Tabularum and the Senatus Consultum de Bacchanalibus where OV occurs to the total exclusion of VO (viz. 34 and 11 occurrences, respectively).92 By the later period, by contrast, VO had become established as the norm.93 For example, in sections (35–57) of the Anonymus 90

Lehmann (1974), Bichakjian (1987: 94; 1988), Bauer (1995: 213–4), Oniga (2004: 103). Vincent (1988: 63), Bauer (1995: 85, 89–90, 168–7), Buridant (2000: 747–51), Oniga (2004: 94). 92 Adams (1976a: 92), Panhuis (1982: 109), Álvarez Pedrosa (1988), Del Vecchio (1989: 551–2), Clackson and Horrocks (2007: 28), Bauer (2009a: 268). 93 Lehmann (1972b: 272; 1974: 238), Calboli (1983a: 129), Miller (1975: 33), Lightfoot (1979: 394–7), Hinojo (1985; 1986), Väänänen (1987: 106), Baños Baños and Cabrillana (2009: 682, 691–2), Bauer (1995: 91, 98–9; 2009a: 69–70); for a more nuanced interpretation of the late evidence, see Haida (1928), Koll (1965: 252–6), 91

226

From Latin to Romance

Valesianus VO outnumbers OV in main clauses by 22:2, and by 42:9 in chapters 1–6 of the Peregrinatio Aetheriae (Adams 1976a: 93). Superficially, at least, this view finds some support in Linde’s (1923) classic survey of the incidence of verb-final positions in some 16 authors/texts across 700 years of Latinity, which we reproduce below in Table 5.2: TABLE 5.2 Incidence of verb-final positions in Latin according to Linde (1923)94 Author / Text

Root (%)

Embedded (%)

Cato (Agr. 1–27)

70

86

Var. (Rustica 1–11)

33

44

Cic. (Inv. rhet. 1.1–22)

50

68

Cic. (Rep. 1.1–32)

35

61

Cic. (various Epistulae)

54

62

Cic. (Phil. 1)

52

70

Caes. (B.G. 2)

84

93

Sall. (Cat. 1–36)

75

87

Bellum Africum (1–35, 81–)

68

73

Liv. (33.30–45)

63

79

Sen. (Ep. 1, 2, 6–10, 15, 16, 24)

58

66

Petr.

51

67

Tac. (Germ. 1–37)

64

86

Plin. (Ep. 1.1,2,6,9,13; 2.1,2,6,11,12,14,15,17,20; 3.4,5)

50

68

Gai. (1.1–38; 4.160–87)

65

80

Apul. (Psyche 1–20)

58

62

Firm. (Err. prof. rel. 1–12)

56

64

V.Vit. (Hist. persecut. 1)

37

63

Per. Aeth. (1–20)

25

37

and Pinkster (1991: 79–80). Text type is also an important factor here, such that even in the late period legal documents and wills containing a high percentage of formulaic language artificially preserve a high incidence of V-final structures (Muller 1939: 53; Politzer 1958: 179–82; Cabrillana 1999: 320). Similarly, Cabrillana (1999: 231–2) explains the almost inverted distributions of OV and VO in the fourth-century Peregrinatio Aetheriae (37.06% vs 62.94%) and Mulomedicina Chironis (79.56% vs 20.44%) as a reflection, at least in part, of the different textual types involved, a fact confirmed by the similar high incidence of OV in other late technical treatises such as Appicius’ De re coquinaria and Palladius’ De agricultura (though the material of the latter is in large part taken verbatim from the first-century Columella). The incidence of OV is also particularly high in the fourth-/fifth-century Papyrus Bononiensis studied by Rizzi and Molinelli (199 119, 122), where the V-final position prevails at a ratio of 3:1 in both root and subordinate clauses. This distribution coincides with the authors’ general conclusion (p. 124) about the conservative nature of the language of the text. 94

To these data we may now also add the distribution of verb-final structures in root and embedded clauses in Ovid’s Met. (viz. 47.6%/51.9%; Amacker 1989: 493), in Livy (viz. 70.9%/75.7%; Amacker 1989: 493), and in the letters of Terentianus of the early second-century AD (viz. 26.7%/37.2%; Adams 1977: 73).

A configurational approach

227

The real problem, however, lies in the period between the archaic and late Latin periods: given what we have seen in }5.3.1 regarding the early emergence of a headinitial order in other phrases, we might quite legitimately expect to find a parallel shift within the verb phrase, especially on the strength of traditional assumptions which ultimately correlate the cross-categorial order of head and complement with the order of verb and complement (Greenberg 1963; 1966; Lehmann 1972b: 271–2). Yet, as we have frequently observed (cf. }3.3.1), the final position of the verb, and hence OV, is firmly established as the preferred order in the classical language,95 at least among certain authors (Pinkster 1991: 71–2; Cabrillana 1993b: 228), witness its prevalence in Caesar, Sallust, and the author of the Bellum Africum in Table 5.2. Furthermore, this view is confirmed by the behaviour of gapping in the classical language (cf. Ross 1970), which Panhuis (1980) and Elerick (1989a) have shown on the evidence of Caesar (B.G. 1–2) and Cicero’s speeches to overwhelming prefer the typical SOV pattern of backward-gapping (see also Spevak 2010: 97), namely SO + SOV (52a–b), with very few examples of the typical SVO pattern of forward-gapping, namely SVO + SO (52c). (52) a Neque certa subsidia collocari neque quid in quaqua nor certain.NOM.PL reserves.NOM place.INF.PASS nor what in any.ABL parte opus esset prouideri neque ab uno from one.ABL part.ABL necessary was.SBJV provide.INF.PASS nor omnia imperia administrari poterant (Lat., Caes. B.G. 2.22.1) all.NOM.PL orders.NOM direct.INF.PASS could.3PL ‘Neither could the adequate reserves be posted, nor the necessary means for every section be provided, nor all the commands be given by one person’ b non modo praetoris ius uerum etiam ciuis amiserat not only praetor.GEN right but even citizen.GEN he.had.lost (Lat., Cic. Cat. 3.15) ‘not only had he lost the rights of a praetor but also those of a citizen’ c Iam intelleges multo me uigilare acrius ad already you.will.understand much.ABL me.ACC watch.over.INF keener to salutem, quam te ad perniciem rei publicae (Lat., Cic. Cat. 1.8) health.ACC than you.ACC to ruin.ACC thing.GEN public.GEN ‘You will already understand that I look much more carefully after the interest of the state than you do after its destruction’

95 Kroll (1912), Linde (1923: 154), Möbitz (1924: 126), Perrochat (1926), Fankhänel (1938), Marouzeau (1938: 47–9), Watkins (1964: 1039), Adams (1976a: 92), Tovar (1979), Vincent (1988: 60), Elerick (1989a: 559), Bauer (1995: 89–90, 92), Magni (2009: 227 n. 6), Clackson and Horrocks (2007: 28).

228

From Latin to Romance

Indeed, this conclusion is further supported by the figures in Table 5.3 summarizing a selection of the some of the most widely reported statistics regarding OV and VO orders across a wide sample of archaic, early, classical, late, and medieval texts (see Pinkster 1991: 72; Baños Baños and Cabrillana 2009: 689). TABLE 5.3 Distribution of OV / VO across different Latin authors/texts96 Text / Author

OV / VO

S.C. Bacch. (Álvarez Pedrosa 1988)

100%:0% (100%:0%)

Leges 2-c. B.C. (Álvarez Pedrosa 1988)

96.2%:3.8% (96%:4%)

Pl. Capt. Adams (1976a: 94–5)

39:45 (43:15)

Pl. Amph. 1–400 (Adams 1976a: 95)

33:22 (20:7)

Pl. Aul. 1–325 (Adams 1976a: 95)

28:20

Pl. Asin. 1–380 (Adams 1976a: 95)

30:15

Pl. Mil. 1–500 (Adams 1976a: 95)

35:45

Ter. (Moreno Hernández 1989)

67:33

Cic. Cat. (Koll 1965: 246–7)

14:7

Cic. Leg. (Koll 1965: 246–7)

18:4

Cic. Att. 1 (Pinkster 1991: 72)

17:0

Cic. Att. 1 (Cabrillana 1993a)

81%:19%

Cic. S. Rosc. (sects 1–34) Adams (1976a: 94)

71:3

Cic. Deiot. (sects 1–34) Adams (1976a: 94)

60:10

Cic. philosophical writings (Bolkestein 1989: 14–15)

79%:21%

Cic. Mil. (Panchón 1986)

54:45 (71:28)

Caes. B.G. I 1–15 (Adams 1976a: 94)

75:3

Caes. B.G. 1–7 & B.C (Pinkster 1991: 72)

393:49

Caes. B.G. 1 (Panchón 1986)

73:26 (90:9)

Vitruvius 1.1–4 (Pinkster 1991: 72)

8:4

Ov. Met. (Amacker 1989) V+ 2 elements

166:125 (72:64)

Petr. Sat. 26–68 (Polo 2004: 378–9) O = NP

352:139 (237:57)

Petr. Sat. 26–68 (Polo 2004: 378–9) O = PP

123:27 (98:16)

Petr. Sat. (Pinkster 1991: 72)

52%:25%

Petr. Cena Trim. (Hinojo 1985)

57.6%:42.4% (82%:18%)

Celsius 1–6 (Pinkster 1991: 72)

66:11

Celsius 1–4 (Pinkster 1992: 522)

86.7%:13.26%

Where root and embedded clauses have been conflated, figures/percentages appear in italics; otherwise figures/percentages refer to root clauses with those for embedded clauses, where available, in brackets. 96

A configurational approach Pompey Inscriptions (Ramat 1984)

64.2%:35.8%

Liv. (Amacker 1989) V+ 2 elements

234:73 (149:35)

Terent. (Adams 1977: 68, 74–5)

9:23

Terent. (Pinkster 1991: 72)

4:14

Vetus, Ruth (Talavera 1981)

9.8%:90.2%

Per. Aeth. (Väänänen 1987: 106)

53:99

Per. Aeth. (Cabrillana 1999: 321) O = NP

37.06%:62.94%

Per. Aeth. 1–2 (Pinkster 1991: 72)

37:102

Per. Aeth. (Hinojo 1986)

38.5%:61.5%

Mul. Ch. (Cabrillana 1999: 321) O = NP

79.56%:20.44%

Anon. Val. II (Adams 1976b: 136)

41.3%:58.7%

229

Vulgata (100 sentences; Pinkster 1991: 72)

15:8

Braulius (8th-c. AD; García Sanchidrián 1994)

68.2%:31.8% (79.5%:20.5%)

Lib. reg. (10th-c. AD; Carrera 1983)

61.9%:38.1% (67.6%:32.4%)

Clearly the data in Table 5.3 are not immediately comparable in all respects: among the numerous differences and shortcomings, not all scholars distinguish, for example, between root and embedded clauses, finite and non-finite verbs, and nor do we necessarily know whether ‘O’ is taken to include complements other than canonical noun phrases such as pronouns, prepositional complements, and complement clauses (cf. also Pinkster 1991: 71; Baños Baños and Cabrillana 2009: 689). Moreover, in the case of the Pompeian inscriptions the strings involved are so short that arguably very little can be securely gleaned about genuine word order patterns (Adams 1977: 67). Bearing in mind such limitations, it is still possible to infer from the data in Table 5.3 that the exceptional steadfastness of the verb-final order in the classical language must represent, rather than an unexpected interruption in the otherwise well-established rise of head-initial structures (Linde 1923: 156), a prestige order artificially preserved under the strong normative pressures of the literary language,97 although long abandoned in lower non-literary registers and poetry.98 Indeed, this view finds strong, albeit indirect, support in four principal observations.

97

Adams (1976a: 93), Panhuis (1982: 145), Vincent (1988: 61), Vineis (1993: lif.), Herman (2000: 86), Oniga (2004: 99), Clackson and Horocks (2007: 28); cf. however Pinkster (1991: 73–4), Baños Baños and Cabrillana (2009: 689–90). 98 Väänänen (1982: 259–60), Herman (2000: 86), Polo (2004: 383), Bauer (2009a: 269–70), Danckaert (2010).

230

From Latin to Romance

The first concerns the high frequency of VO order in early Latin (Bauer 1995: 97; Clackson and Horrocks 2007: 28, 31), as revealingly evidenced in the plays of Plautus in Table 5.3. Even in the archaic period there is reason to believe that OV might not have been as rigid as conservative formulaic texts as the Leges XII Tabularum and the Senatus Consultum de Bacchalibus would seem to suggest.99 Indeed, it is telling to observe that in their examination of the only 6 extant inscriptions from before 400 BC containing verb phrases with 2 or more words, Clackson and Horrocks (2007: 29–30) highlight that out of a total of 11 verb phrases verb-final order is only attested in just 1 root clause and in both examples of embedded clauses.100 Evidence like this has even led some to hypothesize that OV order was never a genuine feature of the spoken language of any period,101 witness the high frequency of VO orders in early authors such as Varro, Plautus, and Terence.102 Rather, it would have constituted a highly stylized feature of specialized bureaucratic prose, especially legalese and solemn writing,103 which was subsequently artificially adopted in (post)classical Latin as a ‘distinguishing feature of written Latin and of good writing’ (Panhuis 1982: 116).104 Appealing though this may appear, this view seems rather unlikely (cf. also Vincent 1988: 61) in light of the fact that embedded clauses in all periods of Latinity and registers show an overwhelming preference for (S)OV (cf. Table 5.2).105 On the uncontroversial assumption that linguistic innovations and changes typically begin in root clauses, only potentially spreading to embedded clauses in subsequent stages of development, and for this reason generally assumed to reflect more conservative

Linde (1923: 169), Clackson and Horrocks (2007: 29), Adams (forthc. f: }1). Table 5.2 corroborates our frequent observation above that verb-final position, and indeed SOV order, is much more robustly preserved in embedded contexts, even when verb-final orders are otherwise greatly reduced in root clauses (Linde 1923: 154, 156–7; Marouzeau 1938: 49, 104; Ramsden 1963: 43–4, 114; Amacker 1989: 493; Bolkestein 1989: 23; Pinkster 1991: 69; Bauer 1995: 91; 2009a: 270; Oniga 2004: 99; Baños Baños and Cabrillana 2009: 696). 101 Meillet ([1928] 1977: 120), Del Vecchio (1989: 553), Clackson and Horrocks (2007: 29–30). 102 Pinkster (1990a: 169), Bauer (1995: 89; 2009a: 270); cf. though Moreno Hernández (1989: 528) and Baños Baños and Cabrillana (2009: 690). According to Adams (1976a: 97), the alternation between VO and OV in Plautus, roughly of equal frequency, reflects his characteristic constant register-switching. 103 Revealing in this respect is the high incidence of OV, not only in early legal texts such as S.C. Bacch. and Leges, but also in the late and medieval technical (e.g. Mul. Ch.) and religious (e.g. Vulgata, Braulius, and Libellus) texts reported in Table 5.3. 104 However, Panhuis (1982: 109) maintains that in legal texts such as the Leges XII Tabularum the verb functions as the rheme since it serves to focus the action ordered or prohibited by the law, which ultimately ‘confirms that this tendency is not mechanic, but sensitive to the communicative perspective’. By contrast, Del Vecchio (1989: 551–3) points out a number of cases in the same text where the final verb (such as the bland ESET ‘was.SBJV’) cannot possibly be given a rhematic reading, underlining, at least for cases like these, the futility of assuming a functional sentence perspective. 105 Linde (1923: 154, 156–7), Marouzeau (1938: 49, 104), Ramsden (1963: 43–4, 114), Amacker (1989: 493), Bolkestein (1989: 23), Pinkster (1991: 69), Bauer (1995: 91; 2009a: 270), Oniga (2004: 99), Baños Baños and Cabrillana (2009: 696). 99

100

A configurational approach

231

stages of the same language,106 it is highly likely that OV order, still robustly preserved in Latin embedded contexts, represents a once unmarked order of an archaic spoken variety only indirectly accessible to us. Indeed, this view finds support in the systematic SOV order (and other OV patterns) attested in Faliscan (Bakkum 2009: 299–300, 304–5), a variety which Bakkum takes to be an early form of Latin rather than a separate language. The second observation relates to the differential distribution of verb-final placement in authors such as Caesar, Sallust, Varro, and Cicero as highlighted in Table 5.2. Although contemporaries of each other, and hence we might legitimately presume speakers of the same language, there is a striking discrepancy in their recourse to the verb-final position in writing, 84 percent and 78 percent in Caesar and Sallust, respectively, but 33 percent in Varro and between 54–35 percent in Cicero (cf. Pinkster 1990a: 168). This incongruity can only be the product of a deliberate diamesic and diaphasic choice on the part of Caesar and Sallust to adopt in writing an artificially archaic, and what was surely by then a prestige, order, to the extent that their prose displays, not accidentally, a greater propensity towards verb-final order than that of Cato writing about a century and half earlier. The third observation concerns subliterary (though not necessarily late) texts. For example, in the semi-literate Latin passed down to us in the letters of Terentianus from the early second century AD, Adams (1976a: 94; 1977: 67–8) notes a significant preference for VO over OV (viz. 40:14), a striking finding given that in contemporary educated writers such as Suetonius and Tacitus OV continues to outnumber VO by a fair margin (e.g. 63:8 in Suetonius’ Cal. (1–16); cf. also 64 percent of root clauses with verb-final order in Tacitus reported in Table 5.2). Rather, such contrasts serve to highlight ‘the gulf which existed between learned written forms of the language and the speech of ordinary people’ (Adams 1976a: 94). The preference for VO orders in the everyday registers of the spoken language may also perhaps be indirectly reflected in the higher frequency of VO registered for the Cena Trimalchionis in Table 5.3, which, although undoubtedly an artificial concoction written by a man of learning (Adams 1976a: 93–4; 1977: 67), can be interpreted as Petronius’ hypercharacterized portrayal of colloquial Latin. Finally, our fourth observation, which provides an important indicator of the nonmarked nature of VO order, comes from a consideration of UNDERGOER subjects, namely the subjects of unaccusative and passive clauses. As is well known, unaccusative and passive subjects are both semantically and syntactically akin to the objects of transitive clauses and, as such, in many languages, including Romance, they pattern together in a number of significant ways (cf. }7.3). Crucially, one of these

106

}5.1.2.

Lightfoot (1991: ch. 3), Dixon (1994: 206), Harris and Campbell (1995: 27). Cf. also the discussion in

232

From Latin to Romance

similarities concerns word order, witness the unmarked postverbal position not just of the transitive object (53a), but also of the unaccusative (53b) and passive (53c) subjects in the following Catalan examples: (53) a un submarí va enfonsar la nau (Cat.) a submarine PRET.3SG sink.INF the ship ‘a submarine sank the ship’ b es va enfonsar la nau (Cat.) self= PRET.3SG sink.INF the ship ‘the ship sunk’ c va

ser enfonsada la Nau (Cat.) be.INF sunk the ship ‘the ship was sunk’

PRET.3SG

Significantly, an analogous behaviour has been noted for Latin, including the classical language, where unaccusative subjects (54a) and passive subjects (54b) are also frequently found in postverbal position:107,108 (54) a uenerat iam tertius dies (Lat., Petr. Sat. 26) had.come already third day ‘the third day had come already’ b relinquebatur una per Sequanos uia (Lat., Caes. B.G. 1.9.1) through Sequani road remained.PASS a ‘there remained one other line of route, through the borders of the Sequani’ Indeed, in the same way that objects frequently follow the verb in Plautus, Adams (1976a: 96) notes how passive subjects also favour postverbal position (cf. also Clackson and Horrocks 2007: 31), whereas active and transitive subjects typically precede the verb. For example, in his examination of root clauses in the Captivi he finds 19 occurrences of postverbal passive subjects in contrast to just 7 examples of

107 Linde (1923: 160), Adams (1976a: 95; 1976b: 122–3, 126–7), Pinkster (1991: 78; 1993a: 649; 1993b: 246), de Jong (1994: 92), Rosén (1998), Cabrillana (1999: 326), Spevak (2008b: 119), Bauer (2009a: 280–1). 108 de Jong (1994: 92) incorrectly claims that Cato’s De agri cultura is an exception, insofar as (unaccusative) presentatives do not display VS. However, his observation is based on examples such as (i.a–b), which involve embedded clauses that have been shown above (}5.1.2) to display a rigid S(O)V order.

(i)

a si herbae natae erunt (Lat., Cato Agr. 151.4) if weeds born will.be ‘if weeds spring up’ b si fistula erit (Lat., Cato Agr. 157.14) if fistula will.be ‘there is a fistula’

A configurational approach

233

preverbal passive subjects, whereas preverbal transitive subjects are massively preferred to postverbal subjects by about 41:6.109 Even in early authors where under strong normative pressures transitive objects precede the verb, passive subjects, by contrast, typically follow the verb. This is the case, for instance, in Terence’s Andria (Adams 1976a: 96): here postverbal passive subjects outnumber their preverbal equivalents by 20:7 (unlike postverbal transitive subjects which are far less numerous than their preverbal equivalents: 70:14), whereas non-pronominal transitive objects prefer the preverbal position by 26:11 in root clauses. Similar evidence comes from formal classical prose such Tacitus’ Annals (2.1–50; Adams 1976b: 125–6): although postverbal passive subjects are outnumbered by their preverbal equivalents (31:52), the difference between the two is not so great as that between preverbal and postverbal transitive subjects (90:14), a fact which highlights, albeit indirectly, the distinctly unmarked nature of postverbal UNDERGOERS in this period (be they subjects or objects). Unsurprisingly, an analogous pattern is found in late Latin texts (Adams 1976b: 122, 127) such as in the Anonymus Valesianus and the Vulgate (Mark 1–10), where once again overt subjects with passive verbs in root clauses predominantly occur in postverbal position (namely, 14:10; 37:29), in contrast to active transitive subjects which overwhelmingly prefer the preverbal position (51:12; 25:16). A similar picture emerges from the Peregrinatio, where both Adams (1976b: 124) and Cabrillana (1999: 326 n. 39) highlight a preponderance of postverbal subjects with passives: the former in his 25-page sample counts 54 examples of postverbal subjects in contrast to just 18 cases of canonical preverbal subjects, and the latter, focusing solely on the passive predicates LEGITUR/LEGUNTUR ‘is/are read’ and DICITUR/DICUNTUR ‘is/are said’, finds that in 88 percent of cases the subject is postverbal. Also telling is the observation that in Anthimus there are 21 examples of postverbal passive subjects, but not one example of a postverbal transitive subject (Adams 1976b: 125). Data like these serve to highlight how across different periods and registers of Latin there is a strong tendency for AGENT and UNDERGOER subjects to be structurally distinguished according to an emerging active-stative opposition (}7.3), which prototypically places the former in preverbal position and the latter in postverbal position. This observation highlights once again how, despite the strong normative pressures of the literary language which placed great importance and prestige by the OV order, UNDERGOER subjects, which correspond to the logical object, were much more readily tolerated in the postverbal position according to a pattern which presumably must have also characterized transitive objects in all other registers. Indeed, given the frequent crosslinguistic structural parallels between transitive 109 Adams notes a similar distribution of VPassive+S/S+VPassive in other Plautine comedies, namely Casina (12:6), Curculio (7:8), Epidicus (15/12), Menaechmi (14/12), Amphitruo 1–300 (7/3), Asinario 1–500 (9/0).

234

From Latin to Romance

objects and unaccusative/passive subjects, we can infer from the behaviour of Latin unaccusative/passive subjects that the postverbal position must have constituted the unmarked position of all UNDERGOER arguments, including objects, a pattern otherwise obscured in formal and literary registers where UNDERGOER objects (but not subjects) were typically fronted to the preverbal position in accordance with a wellestablished prestige order. 5.3.2.1 Auxiliary and dependent infinitive Finally, we consider very briefly the position of infinitival complements with respect to their governing verb on the basis of Adams’ (forthc. f) recent work on this topic. Beginning with a comparison of the position of dependent infinitives in a late first-century BC classical text, the Bellum Africum, and the fifth–sixth-century AD treatise of the African grammarian Pompeius, Adams finds that, whereas anteposition of the infinitive (like that of the nominal complement) predominates in the former (55a–b), in the latter postposition of the infinitive (like that of the nominal complement) massively predominates (56a–b). (55) a itaque castra cum mouere uellet (Lat., B.Afr. 6.1) thus=and camp.ACC when move.INF he.wanted. ‘and so as he was drawing off his men’ coeperunt (Lat., B.Afr. 6.1) b eius agmen extremum insequi his army.ACC outer.ACC pursue.INF they.began ‘they began to harass his rear’ (56) a sed si uolueris conponere (Lat., Pomp., Keil G.l. 5.180.33) but if you.will.have.wanted compose.inf ‘but if you want to compose’ b quo modo hoc potest fieri? (Lat., Pomp., Keil G.l. 5.180.34) become.INF which.ABL way.ABL this.NOM can ‘how can this be?’ Also interesting are the data he presents (}4) on the use of COEPI ‘I have begun’ + infinitive, which in addition to marking inceptive aspect, is also frequently employed as a periphrastic perfect (Petersmann 1977: 191–2). In the early and classical periods, Adams observes that preposing of the infinitive in conjunction with this aspectual predicate proves the norm, outnumbering postposition of the infinitve by 507:62 in his representative sample of nine early and classical authors. Yet, by the late period he finds that postposition of the infinitive comes to overwhelmingingly predominate by 158:46 in his sample of six representative late texts. He also documents (cf. }5) a similar distributional reversal in pre- and postposing of the infinitive in conjunction with OPORTET ‘it is necessary’ in the transition between the early/classical and later Latin periods.

A configurational approach

235

Finally, Adams (}6) also offers a more general overview of infinitival placement (excluding AcI) with all governing predicates across 27 different authors/texts, ranging from the early Plautus to the late Anthimus. Abstracting from the details of Adams’s nuanced interpretation of the differing distributions of the two positions in terms of text type, register, and period, it is possible to recognize in his detailed statistical overview unmistakable signs, even from a quite early period but especially in ‘many late texts without literary aspirations’, of a preference for pospositioning of the dependent infinitive in ordinary speech, in contrast to the more variable position of the infinitive in the literary language where, at least in certain authors and texts, it was often more frequently preposed than postposed. In conclusion, we witness on the basis of Adams’ (forthc. f) recent discussion of infinitival placement that there is once again considerable evidence to suggest a progressive replacement of an earlier head-final order with preposed dependent infinitive by an innovative head-first order with postposted dependent infinitive, which would subsequently generalize in Romance. 5.3.3 Conclusion The preceding discussion has served to show that, despite traditional claims based principally on the deliberately archaizing prestige registers of the classical literary language and, in particular on the arrangement of the verb and its complement, Latin cannot be readily described as a head-final language by any means. Rather, we have seen that since at least the early Latin period the arrangement of head and dependent already betrays an unmistakable shift from an original head-final order, still largely discernible in archaic Latin and the earliest attestations of the other Italic varieties, towards an increasingly unmarked head-initial order across a broad range of categories and constructions. Thus, where the original head-final order persists, this represents either a deliberate literary affectation, as in the case of the verb and its complement where OV constitutes the unmarked order in many classical texts, or a now marked order that serves to convey a pragmatically distinctive interpretation, as in the case of prenominal adjectives and genitives (cf. }5.4.3.2.1).

5.4 Changing directions: from Latin to Romance As the preceding discussion has to some extent already illustrated, the perceived nonconfigurationality of Latin can be broken down into two main ingredients, grammatically free word order resulting from an ongoing change in the directionality parameter, and pragmatically driven word order as a result of the unrestricted freedom of left-edge fronting to different functional projections. The emergence, apparently, of configurationality in Romance is therefore to be understood as the surface effect of the rigidification of the directionality parameter and a more

236

From Latin to Romance

restricted accessibility of the left-edge of functional structure. We now turn to exemplify these keys ideas in the following sections. 5.4.1 Head-last ) head-first: roll-up An important aspect of the so-called freedom of Latin word order and sentential organization concerns the linear arrangement of head and complement, an area of variation of a purely formal nature with no discernible effect on interpretation.110 Above we have presented extensive evidence to this effect to demonstrate how the oscillation between head-last and head-first structures in the history of Latin can be captured along two axes of variation, the first in terms of diachronic variation (headlast (archaic Latin) ) head-initial (early/late Latin)) and the second in terms of diaphasic (and no doubt diastratic and diamesic) variation (head-final (formal, literary) vs head-initial (subliterary, colloquial)). As hinted above, we propose to capture these facts in terms of a progressive reversal of the directionality parameter from a regular head-final setting towards a head-initial setting,111 with classical Latin displaying an ambivalent behaviour on account of its non-uniform characterization in relation to these two dimensions of variation (cf. 2a–b), namely, non-archaic (predominantly head-first), but formal and literary (predominantly head-final).112 Formally, we can translate this linear variation and ambivalence in terms of the application or otherwise of a structural operation known as ‘roll-up’. Following Kayne’s (1994) Linear Correspondence Axiom, we assume here the so-called Universal Base Hypothesis according to which all languages present the same underlying SVO word order, with identical arrangement of external (subject) and internal (object) arguments across all other categories, namely SXO (57a). This means that when the primary complement of a verb, for example, surfaces to the left of its head, as in OV languages, it must have moved leftwards across the verb from its base-generated complement position to a derived (inner) left-peripheral modifier/ Spec(ifier) position (viz. Compl ) Spec movement), a case of what we shall informally term ‘roll-up’ (57b).

110

That is not to say, however, as noted above (cf. also Bauer 2009a: 254, 271; Magni 2009: 246), that synchronic variation in the placement of head and complement cannot, with time, be reinterpreted in specific registers and styles as pragmatically motivated, as with the case of the marked focus reading of prenominal adjectives and genitives in classical Latin (cf. }5.3.1.4, }5.4.3.2.1). Nonetheless, in many cases the less than stable arrangement of head and dependent in (classical) Latin represents the output of a purely arbitrary principle of linearization without any descernible effect on semantico-pragmatic interpretation, a view convincingly supported by Adams’s (forthc. f: }7) and Mclachlan’s (forthc.) recent discussions of the placement of dependent infinitives and genitives, respectively, where different orders are shown not to necessarily correlate with any pragmatic motivation. 111 Adams (1976a), Porzio Gernia (1986: 6), Vincent (1988: 56), Elerick (1994: 70–1), Oniga (2004: 102–3), Clackson and Horrocks (2007: 29, 31), Bauer (2009a: 251–2, 254), Magni (2009: 227). 112 Cf. also Devine and Stephens’s (2006) distinction between ‘complement syntax’ and ‘specifier syntax’.

A configurational approach (57)

a

XP 3 Spec X' Subject 3 X Compl Object

b

237

XP 3 Spec X' Subject 3 Spec X' Object 3 X Compl Object

We exemplify both options from the verbal and nominal domains with the minimal pairs in (58)–(59), together with their structural representations in (60), where the (a) examples illustrate the head-first option with the complement in situ and the (b) examples the derived head-final option where the complement has undergone roll-up to the head’s left-peripheral Spec(ifier) position.113 (58) a instruit aciem (Lat., Liv. 33.15.19; complement in situ) draws.up.3SG battle.line.ACC b aciem instruit aciem (Lat., Liv. 7.37.7; roll-up movement of battle.line.ACC draws.up.3SG complement) ‘(he/the Samnite army) draws up the battle line’ nouarum (Lat., Cic. Att. 5.21.3; complement in situ) (59) a metum rerum fear.ACC things.GEN new.GEN ‘fear of revolution’ b urbanarum rerum metum urbanarum rerum (Lat., Cic. Att. 5.18.1; roll-up) urban.GEN things.GEN fear.ACC ‘because of concern of the situation in the city’ (60) a

V/NP 3 Spec V/N' 3 V/N DP instruit 4 metum aciem rerum n.

b

V/NP 3 Spec V/N' aciem 3 u. rerum V/N DP instruit 4 metum aciem u. rerum

In this light, the distinct grammatical organization of Romance and, in particular, its rigid head-complement order, now finds a straightforward explanation in terms of 113

For some further (near) minimal pairs illustrating the in situ and roll-up options across a range of head categories, see the examples (2b–c), (3b–c), (5a–b), (7a–h), and (8a–e) in chapter 3.

From Latin to Romance

238

the loss of roll-up from the grammar, as exemplified in the corresponding French structures in (61): rangèrent en ligne (Fr.) (61) a ils (**en ligne) se they in line selves= arranged in line b la the

(**des choses) crainte des choses (Fr.) of.the things fear of.the things

However, it must be remembered that the loss of roll-up from the grammar of both Latin and Romance occurs at different rates in accordance with the root vs embedded distinction (cf. also Roberts (1997) for word order change in English). As already observed (}5.1.2), in all periods of Latinity and registers, and even up until early Romance (Bauer 1995: 107–8; Salvi 2011: 366–8), the original head-final (S)OV order was retained much more robustly and for much longer in embedded clauses in accordance with the widely observed conservative linguistic behaviour of such clauses. 5.4.2 Rise of head-initiality Having identified the loss of roll-up as the structural correlate of the observed rise of head-initiality in the passage from Latin to Romance, it now remains to provide some explanation for this change. The solution, we propose, is to be sought in the Final-over-Final Constraint (FOFC),114 a structural condition intended to capture a robust empirical generalization regarding a notable asymmetry in the possible disharmonic word orders attested crosslinguistically. Exemplifying with English lexemes and the Infl auxiliary head (must) and its VP complement (eat, pasta), the FOFC states that whereas a head-initial category may select equally for a headinitial (62a) or a head-final (62b) category, giving rise to harmonic and disharmonic linearizations, respectively, a head-final category may only select for a harmonically head-final complement (62c), but not for a disharmonically head-initial complement (62d). (62) a

c IP IP IP **IP d b 3 3 3 3 Infl VP Infl VP VP Infl VP Infl must 3 must 3 3 must 3 must V NP NP V NP V V NP eat pasta pasta eat pasta eat eat pasta

Significantly, the structural possibilities in (62) do not hold only of synchronic stages of individual languages, but equally apply to the various historical stages of individual 114

Biberauer, Holmberg, and Roberts (2008; 2009), Biberauer, Newton, and Sheehan (2009a, b), Sheehan (2009), Biberauer, Sheehan, and Newton (2010), Biberauer and Sheehan (in press).

A configurational approach

239

languages and thus can be profitably used to make robust predictions about the direction of language change. In particular, given the impossibility of disharmonic orders in which a head-final category dominates a head-initial phrase (cf. 63d), this predicts that the change from head-final (archaic Latin) to head-initial (early/late Latin, Romance) must proceed top-down, starting with the C(omplementizer)P(hrase) in the sentential domain and the P(repositional)P(hrase) in the nominal domain, before possibly percolating down into other categories (namely, CP > IP > VP, and PP > DP > NP). If the change were to proceed bottom-up (namely, VP > IP > CP and NP > DP > PP) or to start at some intermediate category (e.g. IP and DP), then it would invariably violate FOFC along the way, producing an illicit structure such as (62d) with a head-initial phrase dominated by a head-final category. By way of illustration, consider the possible structural scenarios outlined in (63a–d) and (64a–d): (63) a b c d

[[John [pasta eat V] must T] if C] (harmonic head-final: cf. 63c) [C if [ John [pasta eat V] must T]] (disharmonic: cf. 63b: Comp changes) [C if [T John must [pasta eat V]]] (disharmonic; cf. 63a–b: Aux changes) [C if [T John must [V eat pasta]]] (harmonic head-initial; cf. 63a: V changes)

(64) a [[John [pasta eat V] must T] if C] (harmonic head-final; cf. 63c) b *[[T John must [pasta eatV]] ifC] (starts at T ) disharmonic FOFC violation; cf. 63d) c *[[T John must [V eat pasta]] ifC] (starts at T ) disharmonic FOFC violation; cf. 63d) d *[[John [V eat pasta] mustT] ifC] (starts at V ) disharmonic FOFC violation; cf. 63d) (63) illustrates the predicted top-down development of a hypothetical change from head-finality to head-initiality in the clausal domain, beginning at the level of CP with the complementizer if and with every subsequent step in the change conforming to FOFC. (64), by contrast, illustrates what happens if the change starts lower down, at the level either of the auxiliary (62b,c) or the verb phrase (62d) while the complementizer remains head-final, producing in each case a FOFC violation. Returning now to the rise of head-initiality in the history of Latin and, ultimately, in the passage to Romance, FOFC predicts then that the relevant change in head directionality must have proceeded top-down. Indeed, it is not by chance that we have already observed (cf. }3.4.1, }4.4.1, }5.3.1.1) that the first robustly head-initial structures to emerge in Latin involve the C(omplementizer)P(hrase) and the P(repositional)P(hrase).115 In relation to the former, we saw that Latin presents two This conclusion comes very close to the claim in }3.4.1 that the passage from Latin to Romance can be characterized in terms of a gradual top-down development of configurationality from CP and PP. 115

240

From Latin to Romance

rival patterns of complementation, the accusative and infinitive (AcI) construction, an archaic pattern inherited from Indo-European in which there is no overt subordinator,116 and an innovative pattern, albeit attested since the archaic Latin period, in which the complement clause is introduced by an overt complementizer such as UT ‘so that’ and QUOD/QUIA ‘that’. As the older inherited construction, we claim that the AcI construction correspondingly instantiates an earlier head-final construction introduced by a null head-final C(omplementizer), as illustrated in (65a),117 which was increasingly rivalled by the innovative head-initial complementation structure introduced by an overt head-initial complementizer UT or QUOD/QUIA (65b).118 (65) a [[tacitum te dicere V-IP] Ø CP] credo (Lat., Mart. 6.5.3) silent.ACC you.ACC say.INF I.believe ‘I fancy you say to yourself ’ b scis enim [CP quod [V-IP epulum dedi]] (Lat., Petr. Sat. 71.9) you.know for that feast.ACC I.gave ‘for you remember that I gave a public banquet once’ By the same token, we see a parallel development in the nominal domain, where original head-final adpositional structures (postpositions) have been predominantly replaced by head-initial adpositional structures (prepositions), witness examples such as (66a) where the postpositional use of CUM ‘with’ and similar (monosyllabic) prepositions is chiefly restricted to pronominal complements.119 Taking this idea further, we can also naturally extend it to bare case-marked (oblique) nominals, here analysed as silent postpositional phrases headed by a head-final null adposition (66b; cf. also Emonds 1985: 224), which increasingly came to be rivalled by innovative headinitial prepositional structures (66c). (66) a te [PP cum [Caesare]] [[nobis]cum PP] que coniungeres =and you.joined.SBJV yourself with Caesar.ABL us.ABL=with (Lat., Cic. Fam. 9.9.1) ‘you should join with Caesar and with us’

116 Palmer ([1954] 1990: 328), Haudry (1973), Bichakjian (1982), Calboli (1983b: 41–2), Lehmann (1984), Ramat (1994: 259–60, 264), Bauer (1995: 159–60; 2009a: 250, 258), Devine and Stephens (1999: 148). 117 Although adopting a head-initial structure, Cecchetto and Oniga (2002) and Oniga (2004: 132–6) also assume that the AcI construction involves a null complementizer, similar in many respects to the English overt prepositional complementizer for (e.g. For them to pay would be absurd; I prefer for them to pay). 118 Perrochat (1932), Adams (1976b: 94; 1977: 61; forthc. a: }3.5), Bolkestein (1989), Calboli (1989a), Herman (1989), Cuzzolin (1994a, b), Bauer (1995: 165), Karlsen (2001), Ferraresi and Goldbach (2003), Greco (2007). 119 By contrast, Vincent (1999) maintains that prepositional structures were original to Latin and IndoEuropean more generally (cf. }5.3.1.1). For our purposes, it is sufficient that adpositional phrases are among the first categories to show head-initiality, a view compatible with either approach.

A configurational approach

241

b Pompeius [ . . . ] proficiscitur [[Canusium] Ø PP] (Lat., Caes. B.C. 1.24.1) Pompey.NOM sets.out Canusium.ACC ‘Pompey . . . sets out for Canusium’ c miles [PP ad [Capuam]] profectus sum (Lat., Cic. Sen. 10) solider.NOM to Capua.ACC set.out I.am ‘I set out as a soldier for Capua’ We thus witness a parallel development in the sentential and nominal domains where an archaic head-final structure headed by a null complementizer (AcI) or null adposition (bare case-marked nominal) is progressively rivalled, and eventually replaced entirely, by an innovative head-initial structure headed by an overt complementizer and preposition, respectively.120 Now, whereas traditional scholarship has nothing of great explanatory interest to say about the whys and wherefores of the loss of the AcI and its replacement by the finite complementizer construction in Romance (Cuzzolin 1994a; Oniga 2004: 136), and simplistically tends to see the loss of the Latin morphological case system as a consequence of a tendency towards greater analyticity triggered by processes of morphophonological weakening and erosion,121 under the current approach both developments can be brought under the same roof and given a principled explanation. In particular, we now have a highly natural unified account of the eventual demise of the Latin AcI and of morphological case in Romance and their replacement with the finite complementizer and prepositional constructions, since the former both involve head-final structures which, as we know, did not survive into Romance following the loss of roll-up, whereas the latter both instantiate the emergent head-initial pattern (cf. also Herman 1989: 147–8).122 In 120

According to the definition of FOFC in Biberauer, Holmberg, and Roberts (2008; 2009) and Biberauer, Newton, and Sheehan (2009a), if the verbal (phase) head induces roll-up (yielding OV order), then all the non-distinct categories it embeds within its complement domain are predicted to harmonize with it and display the same roll-up inducing property, ultimately yielding a consistent pattern of headfinality. On this view, the failure of finite complement clauses and prepositional complements, which are both unmistakably head-initial, to harmonize in accordance with the head-finality of the selecting verb is a consequence of their categorial distinctness: while the verb is clearly verbal [+V], finite clausal and prepositional complements are argued to be nominal (or in any case [V]); indeed the majority of Latin conjunctions (e.g. UT(I), QUOD, QUIA, QUONIAM, QUAM, CUM) are of nominal origin and ultimately go back to an interrogative-indefinite relative stem *kw- (Ernout and Thomas [1953] 1993: 293–4; Palmer ([1954] 1990: 328, 332–40; Sihler 1995: 395–401). In a similar fashion, many prepositions, themselves originally preverbs (cf. }5.3.1.1), have their origins in grammaticalized nominal forms (Silher 1995: 438–441) such as PIE *H2enti ‘face.LOC’ > ANTE ‘before’, PIE *op- ‘back’ > OB ‘near (OLat.); in front of ’, CAUSA ‘cause.ABL’ > ‘for the sake of ’. By the same token, there is no implication, however, on the basis of this analysis, that the null headfinal complementizer and adposition in the AcI and bare case-marked nominals are necessarily [+V]. 121 Grandgent (1907: 42–8), Muller and Taylor (1932: 65), Löfstedt (1959: 126), Väänänen (1966: 115–19; 1982: 195–7), Lakoff (1972: 189), Adams (1976b: 49), Bauer (1995: 137–9), Molinelli (1998: 147); cf. also }2.2.3. 122 It is well known that crosslinguistically VO languages do not have clause-final complementizers (Hawkins 1990: 256–7; Dryer 1992: 102; Biberauer, Holmberg, and Roberts 2008: 102). It is significant to note in this respect that Romanian, the only language to retain a robust binary case opposition (cf. }7.3.2.2), principally marked in the determiner system, is also the only Romance variety to display an enclitic definite article (e.g. câinele ‘dog=the.NOM/ACC’ vs câinelui ‘dog=the.GEN/DAT’; cf. }4.2.2.2). Historically, these two

242

From Latin to Romance

short, the loss of the AcI and of morphological case in Romance now follows without further stipulation as a concomitant of the more general shift from head-finality to head-initiality, here interpreted as the eventual loss of roll-up. It is therefore our claim here that the passage from Latin to Romance is characterized by a principle of cross-categorial harmonization, such that once head-initiality becomes established in the topmost CP and PP layers, it is then free to percolate down harmonically to the phrases that these in turn embed.123 Significant evidence for this view can be gleaned from the differential placement of the older head-final structures (AcI and bare case-marked nominals) and their more recent head-initial equivalents (finite complementizer and preposition constructions). 5.4.2.1 Position of complement clauses Ramat (1994: 259) observes that ‘[p]ostverbal complementation has been an innovation of several IE dialects’, which explains why ‘Latin had a reduced postverbal complementation but a rich system of nonfinite verbal forms to express clause combining and particularly subordination’ (p. 261). In this respect, it has been repeatedly noted in the literature that ‘embedded clauses introduced by quod and ut [ . . . ] tend to be placed after the governing verb whereas infinitival complements with Subject in the Accusative (AcI) tend to be preverbal’ (Baños Baños and Cabrillana 2009: 698), witness the positional contrast in (67).124 (67)

suisque finibus atque oppidis uti iussit their.ABL=and territories.ABL and towns.ABL enjoy.PASS.INF he.ordered et maleficio et finitimis imperauit ut ab iniuria and neighbours.ABL he.commanded that from injury.ABL and mischief.ABL se suosque prohiberent (Lat., Caes. B.G. 2.28.3) self.ACC theirs.ACC.=and they.restrained.SBJV ‘he ordered them to enjoy their own territories and towns, and commanded their neighbours that they should restrain themselves and their dependents from offering injury or outrage [to them]’

facts might be considered as interrelated, with the exceptional retention of case in the forms of the article ultimately interpreted as a consequence of the original head-final structure (i.a) in which the Romanian enclitic article arose, though since reanalysed as a head-initial structure (i.b; cf. also the retention of a binary case opposition in the pronominal article of some Franco-Provençal and Ræto-Romance varieties discussed in Salvi 2011: 320–1) as outlined in }4.2.2.3. (i)

a [DP [Spec

CANEM] [D ILLUM [NP CANEM]]] '

dog

(Lat.; roll-up)

that

b [DP [D' câinele [NP câine]]] (Ro.; N(oun)-raising) dog=the 123 In various joint works (cf. footnote 40), Biberauer, Holmberg, Roberts, and Sheehan capture the fact that the linearization properties of phase heads such as complementizers and prepositions percolate down to all categorically non-distinct phrases they embed in terms of the Phase Impenetrability Condition (PIC). 124 See also Calboli (1978: 224, 252–3; 1983b: 51). Bolkestein (1989: 14–15), Herman (1989: 137–40; 2000: 90), Ramat (1994: 259), Pinkster (1996: 225), Karlsen (2001: 376–7), Devine and Stephens (2006: 124–5, 560), Greco (2007: 26–7, 373), Baños Baños (2009b: 540), Bauer (2009a: 249).

A configurational approach

243

In a similar fashion, Devine and Stephens (2006: 124) find in their corpus of classical Latin that ‘[i]nfinitival complements of iussit [‘he ordered’; A.L] are regularly preverbal’ while ‘[f]inite indirect commands with ut, on the other hand, are mostly postverbal’. Statistical data to this effect are also presented in Bolkestein (1989: 14–15), who reports the following figures for her sample of classical prose: TABLE 5.4 Placement of AcI and Finite complements in classical prose (Bolkestein 1989) Preverbal (%) AcI QUOD-clause AcI UT-clause AcI UT-clause

Cicero’s philosophical writings 32 15.7 DICERE and IMPERARE in Livy 84 6.5

Postverbal (%) 68 84.3 16 93.5

Causative FACERE and EFFICERE in Cicero, Caesar, and Livy 89.7 10.3 23

77

With the exception of Cicero, the AcI consistently appears in preverbal position and the corresponding finite construction with QUOD or UT in postverbal position. While the distribution in Cicero would seem to weaken this generalization somewhat, it is significant to note that the percentage of preverbal examples of the AcI is more than double that of preverbal QUOD-clauses. Comparable are the data reported in Spevak (2010: 157–62) for her corpus of three classical authors (Caesar, Cicero, and Sallust), inasmuch as she finds with verbs of thinking (e.g. ARBITRARI ‘to judge’, EXSTIMARE ‘to consider’, PUTARE ‘to think’) that the AcI predominantly precedes its governing verb (53.3 percent), whereas with verbs of saying (e.g. DICERE ‘to say’, NEGARE ‘to deny’, NUNTIARE ‘to announce’) it shows a very slight preference for postposition (50.5 percent). However, she also notes a number of examples in both cases where the AcI is wrapped around the governing verb (namely, 28.5 percent and 21.4 percent respectively). If we interpret these latter examples as partial roll-up of the AcI complement, as argued in }5.4.3.3 below, then these cases can be assimilated to the canonical preverbal examples, yielding the following AcI+V/V+AcI individual and global distributions: (verbs of thinking) 81.8%:18.2%, (verbs of saying) 49.5%:50.5% > (total) 65.65%:34.35%. A similar picture is found in the later period (cf. Adams 1976a: 94; forthc. a: }3.5). For example, in his survey of four Christian authors/texts from different regions, periods, and sociocultural levels (namely Saint Cyprian, Lucifer of Cagliari,

244

From Latin to Romance

Peregrinatio, and Salvian of Marseille), Herman (1989: 137) highlights the preponderance of the preverbal position of the AcI in contrast to finite QUOD-/QUIA-complements which, save a handful of exceptions, systematically occur in postverbal position: TABLE 5.5. Placement of AcI and QUOD-/QUIA-clauses in Christian authors (Herman 1989) S. Cyprian AcI

QUOD, QUIA

Lucifier of C. AcI

QUOD, QUIA

Peregrinatio AcI

QUOD, QUIA

Salvien of M. AcI

QUOD, QUIA

Postverbal

55

98

44

95

56

100

40

100

Preverbal

45

2

56

5

44

—

60

—

As Table 5.5 illustrates, once again it is not only the predominant preverbal position of the AcI that is remarkable in these texts, but also the (near-)absence of preverbal examples of QUOD-/QUIA-complements which highlights the exceptional nature of the former.125 Indeed, the extremely rare examples of preverbal finite complement clauses in just two of these authors are all considered by Herman (1989: 138–9) to be highly marked structures, involving a case of focus fronting. This leads him to conclude (pp. 139–40) that, following verba sentiendi and dicendi, there was effectively only a choice between the AcI and the finite construction when the complement clause followed the verb, inasmuch as the AcI was the only available option in preverbal position. An analogous distribution is identified by Rizzi and Molinelli (1994: 122) for the fourth-/fifth-century Papyrus Bononiensis, where the AcI occurs predominantly in preverbal position (10:2) and all 7 occurrences of finite QUOD-/UTclauses are invariably postverbal. The same distribution continues even in the medieval period where, judging by Karlsen’s (2001: 376–7) survey of St Bridget of Sweden’s Revelaciones (Books I–VII), the AcI is once again found both in pre- and postverbal position in contrast to finite QUOD-clauses, which are restricted to postverbal position. Even more compelling in this respect are the results of Greco’s (2007: 76–7, 95–6, 161–2, 251–2, 320–1, 372–5) analysis of the four tenth–twelfth-century texts Chronicon Salernitanum, Chronicon of St Benedict of Sant’Andrea del Soratte, Chronica Monasterii Casinensis, and Chronicon Vulturnense, in which he documents (cf. Table 5.6) the predominant preverbal position of the AcI (57.6 percent of cases), in contrast to finite complements which, save few exceptions, systematically occur in postverbal position (93.2 percent 125 While it is true that preposed (adverbial) QUOD/QUIA-clauses generally receive a causal interpretation (cf. QUOD/QUIA ‘because’, as well as the more neutral ‘that’), a functional argument based on potential ambiguity does not explain why complement clauses headed by QUOD/QUIA cannot a priori be preposed. What we see here is a structural effect.

A configurational approach

245

of cases). These facts lead Greco (2007: 373) to conclude, in accordance with what we have already seen above for earlier texts, that ‘while infinitival clauses can occur both before and after their governing predicate, finite complements are practically never found preceding their governing predicate’. TABLE 5.6 Placement of AcI, (Greco 2007)

QUOD-/QUIA-,

and

AcI

UT-clauses

in 10th–12th-c. authors

QUOD, QUIA

UT

preverbal

postverbal

preverbal

postverbal

preverbal

postverbal

27

15

0

19

5

44

Chron. St Benedict

18

22

0

13

0

26

Chron. Mon. Casinensis

66

39

1

15

5

44

Chron. Vulturnense

22

22

0

8

2

8

Chron. Salernitanum

By contrast, Adams’s (2005) investigation of infinitival and finite complementation in non-literary texts and in the speeches of the freedmen in Petronius reveals a departure from the picture above. In particular, he identifies in the Vindolanda letters, the letters of Claudius Terentianus, two letters of Rustius Barbarus, and letters from Bu Njem a strong preference for postposition of the AcI (namely, 12:59), a preference which is still in evidence even if we interpret the 10 examples in which the AcI is wrapped around the governing verb as cases of partial roll-up, hence instances of preposing (namely, 22:59). Significantly, however, Adams (2005: 201) also notes that out of 59 cases of postposition, 36 of these involve an accusative subject pronoun attached to the governing verb, an observation which he hypothesizes might point to a structural change in the syntax of the AcI with reinterpretation of the pronoun as the direct object of the verb.126 In short, then, what superficially pass for the AcI in these texts might actually instantiate a quite distinct construction (or, at the very least, a construction undergoing transformation) with a syntax increasingly divergent from that of the classical construction, which in turn could explain the rise in the postposition of the infinitival clause with respect to other more canonical texts. Additional evidence to this effect discussed by Adams (2005: 201–2) which is suggestive of a progressive breakdown in the classical AcI and a growing

126 Similar facts are noted by Adams (2005: 204–5) for the speech of the freedmen in Petronius where, once again, postposition of the AcI prevails over anteposition by a considerable margin (6:39), including if we count cases of the governing verb placed within the AcI as instances of anteposition deriving from partial roll-up (namely, 28:39). The stereotyped nature of the construction is further revealed by the fact that postposition of the AcI typically triggers a mechanical placement of the accusative subject immediately following the main verb.

246

From Latin to Romance

inability at lower social and educational levels to successfully manipulate the construction to full effect includes the anacoluthic coordination of the AcI with indicative clauses, the avoidance of the AcI in favour of simple direct speech, the near mechanical placement of the AcI following the governing verb with cliticization of the accusative subject to the latter, and the tendency to avoid the AcI in all but those contexts in which it is used to report short utterances free of subordination and other forms of clausal complexity.127 In short, these tendencies are indicative of a classical construction in decline in the relevant registers, whose syntax displays severe restrictions in those increasingly rare cases in which the non-literary authors feel sufficiently confident to employ it (cf. also Greco 2007: 21–2). Indeed, revealing in this respect is Adams’s (2005: 201) observation that in the letters of Claudius Terentianus two of the three examples of anteposition of the AcI are found in the greeting formula bene ualere te opto ‘I pray that you are in good health’ (lit. ‘well be. in.good.health.INF you.ACC I.wish’; cf. also Adams 1977: 62), which he concludes ‘is unlikely to have been constructed by Terentianus himself but was probably fixed and conventional in the circles in which he moved’. In short, anteposition would appear to be strongly associated with the original, classical uses of the AcI, whereas postposition predominates in those more genuine uses of a dependent infinitival in which the syntax of the construction, as we have observed, has changed quite radically. The evidence of these non-literary authors highlighted by Adams (2005) does not therefore so much contradict our conclusions about placement of the AcI in classical and more literary authors and texts as highlight how, with the progressive shift from head-finality to head-initiality, no doubt more evident in lower social and educational milieux, the original head-final AcI construction was kept alive for a time by radical reanalysis of its syntax, including its mechanical postverbal placement in accordance with the emerging head-initial order.128 To sum up, the entire body of data presented above highlight a differential distribution of both complementation types and, in particular, how ‘it is the SOV word order with infinitive sentence (AcI) which was originally placed before the verb of the main sentence’ (Calboli 1983b: 52), while finite complement clauses were always postverbal (Pinkster 1996: 255). This is not a trivial finding, inasmuch as it provides direct support for the analysis above of the AcI as an archaic head-final structure headed by a null complementizer. More specifically, we have seen that the head-initial finite complementizer construction is only compatible with the postverbal in situ position, which follows from FOFC which predicts harmonic head-initial orders like (68a), but rules out disharmonic orders like (68b) produced by roll-up in which a head-initial phrase

127

Cf. also Adams (1977: 61–4, 72), Calboli (1983b: 126–8), Bourgain (2005: 94), Greco (2007: 25–6, 31). It must be recalled that even the postverbal placement of the AcI is entirely in accord with the FOFCbased analysis in the text, which predicts that finite complements can occur exclusively in postverbal position, whereas AcI complements may occur in both pre- and postverbal positions. 128

A configurational approach

247

(here the UT-/QUOD-clause) is dominated by a head-final category. Now, if the AcI were also a head-initial phrase, then we should expect it to pattern on par with the finite complement construction in (68). Instead we have seen that the distribution of AcI is less constrained and occurs not only in situ in postverbal position but also, and more frequently, in a derived preverbal position under roll-up. Again this distribution follows from FOFC, which predicts both the less frequent disharmonic (69a) and the more frequent harmonic (69b) orders, and at the same time provides independent proof of the head-final nature of the AcI.129 (68) a [VP V [CP UT/QUOD [IP SNOM OACC Vfinite]] ] (finite complement clause) b **[VP [CP UT/QUOD [IP SNOM OACC Vfinite]] V [CP UT/quod . . . ] ] (finite complement clause) (69) a [VP V [CP [IP SACC OACC VINF] ] ] (AcI) b [VP [CP [IP SACC OACC VINF] ] V [CP . . .  ] ] (AcI)

129 FOFC further predicts that a constituent headed by a clause-final complementizer such as the AcI should be harmonically head-final across the whole infinitival clause, yielding the order S IO O V Aux. Although clearly a topic for future research, a preliminary examination of available examples in the literature would seem to support this prediction, witness the following representative S O V Aux (i.a), IO SPASS V (i.b), and S V Aux (i.c) orders.

(i)

a aio [te, Aeacida, Romanos uincere posse] (Lat., Cic. Diu. 2.116) I.say you.ACC Aeacida Romans.ACC defeat.INF be.able.INF ‘I predict that you, son of Aeacus, can defeat the Romans’ b Galli [ . . . ] [Italiae bellum inferri] audiebant (Lat., Liv. 21.24.2) Gauls.NOM Italy.DAT war.ACC wage.PASS.INF they.heard ‘The Gauls were told that it was against Italy that war was being waged’ c sperat [se absolutum iri] (Lat., Cic. Sull. 21) he.hopes self.ACC acquitted.FUT.PASS.PTCP go.INF ‘he hopes that he shall be acquitted’

As remarked in Charpin (1989: 519) and Devine and Stephens (2006: 560), deviations from this pattern appear amenable to an explanation in terms of pragmatic-fronting (cf. }5.4.3), witness the topic- (ii.a) and focus-fronting of the object in (ii.b) and the focus-fronting the gerundival verb in (ii.c). (ii)

a

REM [te rem ualde bene gessisse] rumor erat (Lat., Cic. Fam. 1.8.7) thing.ACC you.ACC very well manage.PRF.INF rumour was

‘there is a report that you have won a great victory’ b

GALLIAE [sese totius Galliae potiri posse] sperant (Lat., possess.DEP.INF be.abl.INF they.hope all.GEN Gaul.GEN self.ACC Caes. B.G. 1.3.8) ‘they hope to be able to master the whole of Gaul’

TOTIUS

ADOPTANDUM [se a Traiano adoptandum c a Sura conperit from Sura.ABL he.discovered adopt.GRD.ACC self.ACC by Trajan.ABL

esse (Lat., Hist. Aug. Hadr. 3.10) be.INF ‘he learned from Sura that he was to be adopted by Trajan’

From Latin to Romance

248

These facts also find further confirmation in the distribution of complement clauses when selected by an embedded predicate. As already highlighted, embedded clauses in Latin are typically SOV, namely verb-final, an observation which, in accordance with FOFC, correctly predicts that the AcI is the only option in such contexts (70a; Bolkestein 1989: 28; Karlsen 2001: 374–5, 379).130 Consequently, if a finite complement is selected, then the usual embedded verb-final order is necessarily substituted by a verb-initial or verb-medial order (70b) or, alternatively, the finite complement is extraposed and is referenced within the sentential core by a proleptic expletive pronominal ILLUD ‘that’ (70c; Salvi 2004: 59–60). suos fines (70) a uel ui coacturos ut per or force.abl PTCP.FUT.ACT.3PL.ACC so.that through their.ACC territories.ACC eos ire paterentur (Lat., Caes. B.G. 1.6.3) them.ACC go.INF they.permitted.SBJV ‘or compel them by force to allow them to pass through their territories’ b Labienus, ut erat ei praeceptum a Caesare ne proelium Labienus as it.was to.him instructed by Caesar.ABL lest battle.ACC committeret (Lat., Caes. B.G. 1.22.3) he.initiated.SBJV ‘Labienus, as he had been ordered by Caesar not to come to an engagement’ c et qui illud exploratum habeat, quieturos milites, si and who that.N.ACC established he.had rest.ACT.FUT.PTCP.PL soldiers.PL if retineantur (Lat., Liv. 40.36.3) diutius in prouincia longer in province.ABL retain.SBJV.PASS.3PL ‘and who had also definitely ascertained whether the soldiers would take it quietly if they were retained in the province’ Finally, it is significant to note that the head-final analysis above of the Latin AcI is directly supported by the empirical findings of recent typological research. For instance, in his broad crosslinguistic sample of OV languages Dryer (2009: 200) finds that, while complementizers may either be clause-final, as in Japanese, or clause-initial, as in Supyire, this difference is, in turn, robustly correlated with the position of the complement clause, as summarized in (71a–b): (71) a

OV languages in which the complement clause precedes the verb normally have clause-final complementizers rather then clause-initial complementizers.

130

Revealing in this respect is Karlsen’s (2001: 375) observation that verbs which typically select for a in root clauses tend to prefer the AcI when they themselves occur in embedded contexts. For instance, in the Revelaciones QUOD-clauses outnumber examples of the AcI when selected by root DICERE ‘to say’ (129:28), but not when the latter occurs itself in an embedded context (16:48). QUOD-clause

A configurational approach b

249

OV languages in which the complement clause follows the verb normally have clause-initial complementizers rather then clause-final complementizers.

Thus, the relevant generalization is that OV languages will have final complementizers if the complement clause precedes the verb and initial complementizers if the complement clause follows the verb, a distribution which is perfectly mirrored by the evidence of the Latin AcI and UT-/QUOD-clauses, respectively. In addition, Dryer’s study (p. 204) further highlights how complementizers are by far more frequently overt in VO languages than in OV languages, a finding once again entirely in line with our proposed head-final and head-initial analysis of the Latin AcI and finite complementation which respectively involve covert and overt complementizers. 5.4.2.2 Position of nominal complements It has long been noted, since at least Linde (1923: 170–1, 175), that variation in the preverbal and postverbal position of nominal complements is not random, but displays a sensitivity to the nature of the nominal complement (cf. also Möbitz 1924: 124). Putting aside constituents of a ‘heavy’ nature, appositions, and lists (cf. Linde 1923: 175), Linde observes that nominals expressing the goal of an action (72a) or an adverbial function (72b), in both cases typically realized as P(repositional)P(hrase)s, are placed more frequently in postverbal position than bare case-marked nominals (72c): (72) a ligna conficit ad fornacem (Lat., Cato Agr. 16) wood.ACC he.cuts to kiln.ACC ‘he cuts the wood for the kiln’ b auxilio cohorti uenit cum legionibus duabus (Lat., Caes. B.C. 3.51.1) help.DAT cohort.GEN he.came with legions.ABL two.ABL ‘he came to the aid of the cohort with two legions’ c nouitas tamen omnium conuertit oculos (Lat., Petr. Sat. 35.1) novelty.NOM however all.GEN turned. eyes.ACC ‘still the oddity of the thing drew the eyes of all’ Similar observations also abound in many recent analyses of Latin word order,131 which equally highlight the greater propensity of PPs to occur in postverbal position than bare case-marked nominals, which only begin to surface in postverbal position with any regularity in later and/or subliterary texts (Bauer 2009a: 271). In short, we witness since early times the emergence of a significant distributional difference between bare case-marked nominals and prepositional nominals, with the former typically surfacing in preverbal position under roll-up and the latter in their in situ

131

Bauer (1995: 97–8; 2009a: 269–71), Salvi (2004: 59–60), Polo (2005: 397–8), Devine and Stephens (2006: 119–23), Danckaert (2010).

From Latin to Romance

250

postverbal position. This distribution is anything but accidental, and cannot be attributed to pragmatic factors given the possible ‘range of pragmatic values found with these postverbal prepositional phrases’, but, rather, suggests that ‘it is the syntactic presence of the prepositional head that is the main licensing factor for postverbal position’ (Devine and Stephens 2006: 122). Indeed, the distribution evidenced here mirrors with remarkable similarity that of the AcI and finite complementation examined above and provides, we suggest, compelling evidence for interpreting these facts in terms of FOFC: a prepositional nominal complement may, as a head-initial category, enter into a harmonic head-initial order (73a), but not a disharmonic order in which it is dominated by a head-final category (73b),132 whereas a head-final bare case-marked nominal—interpreted here as a concealed adpositional phrase with null head—may occur either in the more frequent harmonic head-final order (74b) or the rarer disharmonic order (74a). (73) a [VP V [PP p [DP O]]] (prepositional complement) b **[VP [PP p [DP O]] V [PP p . . . ] ] (prepositional complement) (74) a [VP V [PP [DP O] P] ] (bare case-marked complement) P] ] (bare case-marked complement) b [VP [PP [DP O] P] V [PP . . . . By contrast, traditional attempts to account for this positional difference between prepositional and bare case-marked nominals prove less than satisfactory. In particular, there is a tendency to treat postverbal complements, whether nominal or clausal, as in some sense as ‘heavy’ or ‘long’ (often in relation to Behaghel’s Law; cf. Bauer 2009a: }3.3.2) and therefore as having undergone rightward extraposition.133 While this analysis might prove appropriate in those cases where the extraposed constituent (including complement clauses, quotations, appositions, and relatives) is in construction with a proleptic proform or antecedent in preverbal position within 132 This implies, as with the case of preverbal finite complements, that when prepositional complements occur in preverbal position this is a result of their having undergone (locally bound) pragmatic fronting (cf. Devine and Stephens 2006: 120), a conclusion corroborated by Bauer’s (2009a: 271) claim that postverbal prepositional complements represent the unmarked order. It is also worth noting, in this respect, that Latin (i.a) differs from languages like German (i.b) in which prepositional complements systematically surface in preverbal position. As is implicit in the text above, this difference suggests a dimension of variation in the application of FOFC across languages in terms of whether it holds across phasal boundaries (verbal and prepositional in this case) as in Latin or not as in German (see further Sheehan 2008).

(i)

a [VP concurrent [PP ad Aristium]] (Lat., Caes. B.G. 7.43.1) they.rush.together to AristiusACC ‘they run together to Aristius’ b Johann ist [VP [PP nach Berlin] gefahren] (Ger.) Johann is to Berlin gone ‘Johan has gone to Berlin’ 133

Salvi (2004: 59–60), Polo (2005: 397–8), Bauer (2009a: 269–70).

A configurational approach

251

the sentential core (cf. 70c), its explanatory adequacy proves highly dubious in other cases such as prepositional phrases which never co-occur with a preverbal proleptic element. First, ‘heaviness’ or ‘length’ is, at best, an intuitive term but, in any case, not an objective, formally definable linguistic concept that can be unambiguously applied to individual constituents without some subjective judgement. In particular, it is not at all obvious in what sense prepositional complements, but not bare case-marked nominals, invariably qualify as ‘heavy’ or ‘long’, when both categories are equally capable of showing various degrees of structural complexity and concomitant phonological heaviness. Indeed, considerations like these lead Devine and Stephens (2006: 121) to conclude that ‘[t]he fact that a locative argument is a heavy or complex phrase [ . . . ] may also contribute to the choice of postverbal position [ . . . ] [a]lthough one suspects that the presence of a prepositional head is the main licensing factor for postverbal position.’ Second, as syntactic objects, complements of the verb, whether nominal (DP), adpositional (PP), or clausal (CP), should be treated uniformly by the ‘narrow’ syntax, which is blind to their phonological properties and any subjective concept of supposed ‘heaviness’, whatever that turns out to be. In short, an autonomous syntax deals just with constituents (e.g. DPs, PPs, CPs), but is ultimately unable to interpret these as ‘heavy’/‘light’ or ‘long’/‘short’. Third, if following Salvi (2004: 59–60, 85–6, 97–8, 202, 204) ‘heaviness’ is to be interpreted in structural terms as a reflex of the absence of overt case-marking and checking, then the rise in postverbal nominal complements in later/subliterary texts remains highly problematic. According to this view, the change from OV to VO is considered a consequence of the erosion of the case system, such that nominal complements are no longer required to raise to the preverbal position (> OV) to check morphological case, but can now remain in situ in postverbal position (< VO).134 Yet, as observed in }2.2.3, the relevant changes in sentential word order producing (S)VO preceded the loss of the case system,135 although its erosion should be expected to have taken place before or, at the very least, at the same time as the change from OV to VO, but certainly not after it.136 134

Along the lines of Magni’s (2000) analysis of Pompeian Latin, Polo (2006: 230) claims that (S)VO arises first with postverbal bare case-marked objects which fail to denote prototypical patients specified positively for the features [animate], [human], and [definite], noting in particular that the chief nonprototypical patient feature trigger for VO is definiteness. However, this generalization is weakened by the observation that, even when marked [+definite], objects may occur in preverbal position in her sample. By the same token, she also identifies a number of cases of postverbal complements where the three features prove entirely irrelevant. Finally, it significant to note that, contrary to Polo’s hypothesis, features such as animacy and definiteness have frequently been invoked in the literature as the trigger for object shift, yielding the preverbal position (Torrego 1998; Ledgeway 2000: ch. 2). 135 Adams (1976a: 98; forthc. a: }3.1), Sasse (1977), Bichakjian (1987: 89), Bauer (1995: 7–8), La Fauci (1997: 41), Magni (2009: 247). 136 Inflectional parameters are notoriously unreliable indicators of syntactic processes, witness the difficulties of generative attempts such as the Rich Agreement Hypothesis (Bobaljik 2002) and the Mirror

From Latin to Romance

252

Finally, if postverbal PPs, together with all other supposed ‘heavy’ constituents, were extraposed, bearing the relevant unreconstructable intonational pattern of extraposed constituents, then they would prove completely irrelevant for patterns of unmarked word order and would not be interpreted as SVOPP sequences by speakers but, rather, as marked S(O)V, OPP sequences. Yet, as we have already observed (cf. footnote 132 above), the postverbal position represents the unmarked, neutral position of PPs, rather than a pragmatically marked derived position. To conclude, we have seen that the FOFC-based account proposed here not only explains the directionality of the changes in Latin and ultimately Romance word order, but it also provides us with a highly natural and principled explanation of the differential placement of prepositional and bare case-marked complements, parallel to that of finite complements and the AcI, which does not rely on such subjective notions as ‘heaviness’. 5.4.2.3 Other patterns of harmonization Besides the placement of clausal and adpositional complements, FOFC also makes important predictions about the possible harmonic and disharmonic linearizations of other categories, which are fully substantiated once again by the Latin evidence and provide the key to understanding the word order patterns that would subsequently become established in Romance.137 Of particular note in this respect is Elerick’s (1994) insightful examination of linearization patterns across a series of categories in classical Latin where, contrary to the robustly attested harmonic (H + [H-Compl], [Compl-H] + H) and disharmonic (H + [Compl-H]) linearizations, the FOFC-violating disharmonic order ([H-Compl] + H) is barely attested. In particular, Elerick considers three case studies (cf. Table 5.7): (i) noun + gerundive and object; (ii) preposition and postposition (viz. CAUSA ‘for the purpose of ’; GRATIA ‘for the sake of ’) + gerundive and object; and (iii) noun + genitive complement and modifying adjective: TABLE 5.7 Some linearization patterns in classical Latin Head-initial Harmonic

Head-initial Disharmonic

Caesar

N + [VGer O]: 14

Caesar

P + [VGer O]: 36

Cicero

N + [GA]: 33

33.5

Total (%) 29.5

N + [O VGer]: 19

Head-final Harmonic

Head-final Disharmonic

[O VGer] + N: 19

[VGer O] + N: 1

P + [O VGer]: 44

[O VGer] + P: 21

[VGer O] + P: 5

N + [AG]: 31

[AG] + N: 44

[GA] + N: 14

29.9

7.1

Principle (Baker 1985; 1988: 13) to relate overt morphological richness to the availability of syntactic operations such as verb movement and pro-drop phenomena (though now see Koeneman and Zeijlstra forthc.). 137

Cf. also the discussion of pronominal adjectives qualifying preposed genitives in }5.3.1.4.2.

A configurational approach

253

As is immediately obvious, the incidence of the FOFC-violating linearization given in the final column of Table 5.7 is significantly lower across all three structures than the other three competing orders, accounting for just 7.1 percent of cases overall. This skew in the statistics is so strong and consistent across all three structures that it cannot be readily dismissed as an accident of the sample but, rather, must reflect the ill-formedness of a head-initial structure dominated by a head-final category. Indeed, if we are to take FOFC seriously, then the small number of apparent cases of a headinitial phrase dominated by a head-final category reported in Elerick’s sample must be explained otherwise. Unfortunately, Elerick provides very few examples in his study, and illustrates the FOFC-violating orders with just two examples, which are reported in (75a–b). consilium (75) a neque aliter Carnutes [interficiendi Tasgeti] not=and otherwise Carnutes.ACC kill.GER.GEN Tasgetius.GEN plan.ACC fuisse capturos (Lat., Caes. B.G. 5.29.2) take.FUT.ACT.PTCP.ACC.PL be.PRF.INF ‘as the Carnutes would not otherwise have taken the measure of slaying Tasgetius’ b equitatumque omnibus locis [iniciendi timoris] cavalry.ACC=and all.ABL.PL places.ABL cause.GER.GEN fear.GEN causa ostentare coeperunt (Lat., Caes. B.G. 7.55.9) for.the.purpose.of display.INF they.began ‘they began to display the cavalry on all sides to strike terror into the Romans’ Although future research is required to verify all such cases, including across a broader range of authors and texts, FOFC predicts that the pre-head position of the [VGEROGEN] complement in (75a–b) must be a result of pragmatically driven word order, and not the result of roll-up to the Spec(ifier) position of the selecting nominal or adpositional head. In particular, we take the order witnessed in (75a–b) to be a case of locally bound focus-fronting of the head-initial VP constituent from its underlying post-head position (cf. }5.4.3). The data in Table 5.7 also prove significant in that they highlight how the principle of harmonic linearization may provide the stimulus for change, ultimately overriding other factors and driving change. By way of example, one only need consider the behaviour of gerundival VPs in Caesar reported in Table 5.7. Now, in }5.3.2 it was noted how verb-final order is a hallmark of Caesar’s prose with as much as 84 percent of all main-clause verbs occurring in final position (cf. Table 5.2), a characteristic that led Linde (1923: 154) to label him as a ‘fanatic for verb-final position’. Indeed, the incidence of OV in conjunction with gerundives in Caesar reported in Table 5.7 unquestionably continues to represent the dominant order, accounting for 64.8

254

From Latin to Romance

percent of all cases. However, at the same time we see that the frequency of the otherwise non-dominant VO order exceptionally increases when dominated by a head-initial category (viz. preposition), coming to account for as much as 31.4 percent of cases, an observation which appears to indicate that ‘the harmonic linearization seems to occasion tolerance for the non-dominant VO order’ (Elerick 1994: 68). Indeed, the effects of harmonic linearization also seem to be at play in other areas examined by Elerick (1994). One of these concerns the linearization of nominal head and genitive complement: as observed in }5.3.1.4.2, in classical prose NG and GN essentially occur in similar proportion, a distribution which is broadly reflected in Elerick’s survey of book 6 of Caesar’s de Bello Gallico (namely, 66:57). If, however, the nominal head is itself embedded within a prepositional phrase, then the distribution of the two orders now differs considerably, producing a striking skew in the number of examples of the harmonic P + [NG] and disharmonic P + [GN] orders (namely, 42:14), which can only be understood as the effect of a principle of harmonic linearization (cf. Elerick’s principle of Harmonic Phenotypic Linearization).138 In short, we claim that the presence of a superordinate head-initial category like the preposition in this and similar cases acts as a stimulus for harmonic linearization (classical Latin: P + [NG] / p + [gn]), such that once the harmonic order becomes established (late Latin / Romance: P + [NG]), it can then spread in the speech of subsequent generations beyond the original conditioning environment and generalize to simple cases of linearization of nominal and genitive complement, namely classical Latin: NG / GN > late Latin: NG / gn > Romance: NG (Elerick 1994: 71). Indeed, this is precisely the scenario which Elerick claims can be readily observed from a comparison of the distribution of the (P +) NG/GN orders seen above in Caesar with those in Livy (36.1–19) reported in Table 5.8: TABLE 5.8 (P+) N/G and (P+) G/N linearizations in Caesar (B.G. 6) and Livy (36.1–19) NG

GN

P + NG

P + GN

Caesar (B.G. 6)

66

Livy (36.1–19)

74

57

42

14

33

48

12

Unlike in Caesar, in Livy the head-initial harmonic NG order is no longer broadly codominant with the disharmonic GN order, but now, almost two generations later, its frequency is more than double that of GN. This illustrates how the skew between the two orders produced by the introduction of a preposition in Caesar (42:14) in

138

For an apparent counterexample to this tendency with genitives dependent on the unmodified nominal CUPIDITAS ‘desire’, see the discussion in Devine and Stephens (2006: 337).

A configurational approach

255

accordance with the principle of harmonic linearization has, in the intervening time, become increasingly conventionalized, such that the NG order now represents the dominant order in Livy, even when it occurs outside of the original conditioning environment.139 5.4.2.4 Clausal word order Finally, we turn to the development of clausal word order. Judging by the robust conservation of SOV in embedded clauses and the evidence for the archaic pattern of head-finality in the AcI examined in }5.4.2.1, we can infer from this that proto-Italic must have been strictly head-final in both root and embedded clauses. However, by the time of our earliest records root clauses must have developed a head-initial CP, as substantiated, among other things, by a number of leftward verb-fronting operations to the vacant C(omplementizer) position within the left periphery (}4.4.1). Now, in line with the predictions of FOFC (cf. examples (63)–(64) above), the establishment of a head-initial CP necessarily constitutes the first step in the clausal domain for a change from head-finality to head-initiality, but does not inevitably imply that all the lower categories it dominates (viz. IP > VP) must undergo this change. In theory, this means that Latin root clauses can be associated with at least two different structural representations, a disharmonic order in which the locus of sentential inflection IP (hosting such categories as auxiliaries) remains head-final (76a), and an innovative harmonic order in which IP has also become head-initial (76b):140 (76) a [CP C [IP [VP OV] Aux]] b [CP C [IP Aux [VP OV]]] However, in most cases the Latin evidence is such that it is impossible to tell which of the two structures in (76) a particular root clause instantiates since, apart from auxiliary ESSE ‘be’ employed to form the perfective of the middle voice (namely, passive and (semi-)deponent verbs; cf. }4.3.1.4, }7.2), Latin does not have any other auxiliaries to overtly mark the position of the Infl head (cf. }3.2, }4.3). Consequently, in most clauses where the verb assumes a synthetic form (namely, in the imperfectum of the active and middle voices and in the perfectum of the active voice), a surface structure such as (77a) proves a priori compatible with both the underlying disharmonic (77b) and harmonic (77c) representations (‘p(arameter)-ambiguous’ in the terminology of Clark and Roberts 1993):

139 Although Mclachlan (forthc.) also observes that the harmonic NG order increases when the NP is governed by a preposition in three (Varro, Caesar, and Sallust) of the four authors in her classical corpus, it is surprising to note that in Cicero, by contrast, she finds an increase in the disharmonic GN order in the same context. 140 Of course, if IP is head-initial, then the VP it embeds may also shift to a head-initial linearization in accordance with FOFC. For simplicity, we ignore this possibility in what follows.

From Latin to Romance

256

(77) a forum saepietur (Lat., Cic. Phil. 1.25) forum.NOM will.surround.PASS.3SG ‘the forum will be surrounded’ b [CP C [IP [VP forum saepietur] Aux]] c [CP C [IP Aux [VP forum saepietur]]] It is only in the perfective paradigms of the middle voice through recourse to auxiliary ESSE ‘be’ that any overt evidence surfaces for the position of Infl and the headedness of IP. As is well documented (cf. }5.3.1.5),141 the unmarked classical order is that exemplified in (78a) with the head-final linearization participle + Aux, instantiating the disharmonic order in (76a). This, however, is not the full story since it is well known, especially following the detailed work of Adams (1994a: 113–14, 123–4, 127, 129–30. 1994b), that, like weak pronouns, the various forms of ESSE ‘be’ behave phonologically like enclitics and, as such, are very frequently detached from their associated participles and attracted to a focal host in colon-initial position much earlier in the clause (Fraenkel 1932; 1933; 1964; 1965). In most, cases—though by all means not all—the colon in question corresponds to the top-most layer of the clause, namely CP (cf. }5.4.3.1), such that forms of ESSE occur in clause-initial position immediately following a left-peripheral fronted focal host, as illustrated in (78b) where auxiliary EST ‘is’ stands enclitic to the focal demonstrative IS ‘that’. Indeed, the high frequency of this pattern is bolstered by the fact that one of the recurrent focal hosts involved is the ubiquitous verbal negator NON ‘not’, itself an Infl-related category, such that ‘a periphrastic verb form (e.g. factum est [lit. done is]) if negated regularly shows est attached to the negative, which entails either of the orders factum non est [done not=is] or non est factum [not=is done] (rather than non factum est [not done is])’ (Adams 1994a: 127). (78) a Cur armatorum corona senatus saeptus est why armed.GEN.PL crown.ABL senate.NOM surrounded.NOM is (Lat., Cic. Phil. 2.112) ‘Why has the senate been surrounded with a belt of armed men?’ b

est hodie locus saeptus (Lat., Cic. Diu. 2.85) that.NOM =is today place.NOM guarded.NOM ‘THAT site is guarded to this day’ IS

It is easy to see how a frequent surface string like (78b), originally the product of an independent phonological linearization rule, could be readily reanalysed by speakers as a reflex of a particular syntactic choice, namely the head-initiality of IP, implying

141

Möbitz (1924: 124, 126), Marouzeau (1938: 23–7), Bauer (1995: 104–5; 2006: 294), Salvi (2004: 112 n. 62), Devine and Stephens (2006: 180–98), Vincent (2007: 65), Bailey (2008: }2).

A configurational approach

257

the postulation of the underlying harmonic representation in (76b) for such strings. Once such a structural reanalysis becomes available, then it can also be assumed, by speakers, to underlie ambiguous strings with a synthetic verb form like (77a), in which the position and headedness of IP is not overtly reflected. Now, although the high frequency and regularity of the erstwhile phonological fronting of auxiliary ESSE arguably provides, with time, the syntactic cue and necessary impetus for speakers to increasingly reanalyse IP as being head-initial (and, in turn, VP), as sketched in (79a), this nonetheless produces an asymmetry in the system. As repeatedly noted throughout this chapter, embedded clauses in the whole of Latinity (with strong residues well into early Romance) show a very marked conservative tendency towards rigidly maintaining an archaicizing SOV order. Consequently, in embedded clauses ESSE is considerably less susceptible to undergo fronting, but generally prefers to maintain its clause-final position, as illustrated in (79b). (79)

a Root clauses: (C) (InflESSE) (O)VO (InflESSE) OV (InflESSE) b Embedded clauses: QUOD/UT

The contrast in (79a–b) highlights the (increasingly) asymmetric distribution of the Infl-auxiliary ESSE in root and embedded clauses, which surfaces in clause-final and ever more frequently also in clause-initial position (following a left-peripheral focus) in root clauses, but is typically confined to the clause-final position in embedded contexts. This asymmetry is ripe for reanalysis, in that it provides speakers with the necessary surface cues to reinterpret the pattern in (79a–b) as the output of a VerbSecond (V2) syntax, which would subsequently characterize the late Latin and early Romance periods (}3.3.2, }4.4.1). In particular, the root vs embedded distributional asymmetry illustrated in (79a–b) provides speakers with a robust alternation, in which ESSE is seen to undergo fronting to a clause-initial position immediately right-adjacent to a left-peripheral focus in root contexts,142 but, significantly, not in embedded contexts where the C(omplementizer) position is lexicalized by a complementizer (e.g. UT ‘so that’, QUOD ‘because; that’). From this speakers were able to deduce that ESSE could be fronted if the C(omplementizer) position was not lexicalized, as in root clauses (80a), but not if that position was already filled by a complementizer, as happens in embedded clauses (80b). The ultimate consequence of this is that the original asymmetric distribution of ESSE in (79a–b), the erstwhile output of a phonological fronting rule, is now reinterpreted as the output of an

142 The account of ESSE ‘be’ fronting developed here provides a principled and natural explanation for Salvi’s (2004: 55–6, 94–8, 101–7; 2011: 356–63) observation outlined in }4.4.1 regarding the rise of generalized verb fronting (V2) in conjunction with fronted focused constituents, previously in complementary distribution, since ESSE ‘be’ precisely seeks out a focal host to which to attach.

258

From Latin to Romance

asymmetric V2 syntax which fronts the verb to the vacant C(omplementizer) position when available.143 (80)

a Root clauses: [CP ESSE [[(O)VO VP] b Embedded clauses: [CP UT/QUOD [[OV VP]

ESSE IP]] ESSE IP]]

5.4.3 Pragmatic variation: left-edge fronting Traditional arguments in favour of the non-configurational status of Latin have often pointed to the existence of pragmatically driven word order and discontinuous structures (cf. }3.2.1.1, }3.3.1.1), both of which were considered by Hale (1983) among the most robust criteria in establishing non-configurationality (}3.4). Rather than treating them as two separate and ultimately unrelated phenomena, with discontinuity interpreted as a purely structural concomitant of the absence of any hierarchical phrasal organization, we interpret phrasal discontinuity as a reflex of pragmatically driven word order that, as part of a focusing/topicalizing device, targets individual pieces of structure fronting them to various positions within the left-edge of individual (e.g. nominal, prepositional, verbal, and clausal) syntactic domains distributed across the clause where they receive a pragmatically salient interpretation. Consequently, we shall not only discuss here cases of hyperbaton proper in which ‘words that are or seem to be syntactically connected [ . . . ] occur some distance apart, separated by other words that are in grammatical terms less closely connected’ (Powell 2010: 163), as with illustriorem . . . locum in (81a), but also cases like (81b) where the adjective clarissimae and noun uictoriae ‘are separated, but [ . . . ] the noun phrase is not discontinuous, as everything inside it clearly belongs to it’ (p. 169; cf. also Foubert 2007).144 Thus, whether discontinuity outside of the nominal group obtains or not, in both cases we are dealing with a (one-word) phrase (e.g. illustriorem, clarissimae) which has uniformly been fronted to the left-edge of a superordinate phrase containing it such as the verb phrase in (81a) and the noun phrase in (81b). (81) a illustriorem obtinebat locum (Lat., Cic. Brut. 238) more.illustrious.ACC he.obtained place.ACC ‘he held a more illustrious place’ b clarissimae testis uictoriae (Lat., Cic. Off. 1.75) very.distinguished.GEN witness.NOM victory.GEN ‘witness of his glorious victory’

143 We witness in this development the first signs of an important change in the grammar of Latin in the transition to Romance which produces a number of head (viz. verb and noun) raising operations. 144 In all following examples, we indicate with italics the head and tail of those constituents whose linear adjacency has been disrupted.

A configurational approach

259

As these examples illustrate, the intended interpretations of the head and tail of such discontinuous strings produced by left-edge fronting is ensured by Latin’s rich case and agreement system (what was termed in chapter 3, following Bresnan, ‘lexocentricity’), which, as a nominal category, entails that edge-fronting ‘typically affects noun phrases, prepositional phrases [ . . . ], and verb phrases to the extent that they include a noun phrase’ (Bauer 2009a: 286).145 In what follows we do not intend to provide an exhaustive treatment of edgefronting (for which see the excellent recent treatments in Devine and Stephens (2006: ch. 6) and Powell (2010)), but limit ourselves to showing how the key patterns of pragmatically driven word order witnessed in the omnipresent left-edge movements of (classical) Latin interact with, and further substantiate, the strictly configurational structure of Latin, including operations such as roll-up, as well as provide strong empirical evidence for the presence of functional structure. It should also be recalled (though for a fuller discussion, cf. }3.2.1.1) that we assume here that, apart from some specific stylistic formulae, the discontinuities and structural reorganizations produced by left-edge fronting are not necessarily ‘either a formal rhetorical feature or a colloquial feature’ (Powell: 2010: 184), but, rather, simply represent the way in which discourse information is conveyed in Latin (Kessler 1995: 1; Spevak 2009). More specifically, we follow Fraenkel (1968: 76) who highlights how such patterns arise whenever ‘the speaker is required to place strong emphasis on a particular element, sometimes in contrast to something else’, an observation which undoubtedly aligns left-edge movement with a pragmatic device and, in particular, with the licensing of focused and, to a lesser extent, topicalized readings, as widely recognized in the literature.146 This view is further independently supported by the widespread recognition that the words typically involved in edge-fronting belong to the typical subclasses of focus and topic modifiers such as interrogatives, relatives, demonstratives, quantifiers, possessives, adjectives of quantity and size, comparatives and superlatives, restrictive adjectives, and degree and measure adverbs,147 the same classes of ‘preferential words’ (Powell 2010: 174) that Adams independently identifies as focal hosts for clitic pronouns (1994a: 122–30) and clitic forms of ESSE ‘be’ (1994b: 19–24). 5.4.3.1 Identifying the left-edge: cola and left peripheries If the pragmatic dimension of Latin word order is to be reduced to left-edge fronting, then we need to establish the proper identity and location of the structural phrases whose left-edge is targeted 145 It is also widely acknowledged (Herman 1985: 348; de Jong 1986: 295; Bolkestein 2001: 253; Bauer 2009a: 291) that hyperbaton (what we are calling more generally edge-fronting) typically affects argument noun phrases, and only much more rarely, if at all, noun phrases in adjunct function. 146 Hale and Buck ([1903] 1994), Marouzeau (1922: 216), de Jong (1986: 295), Kessler (1995: 10), Bolkestein (2001), Pinkster (2005b), Devine and Stephens (2006: 542–6), Clackson and Horrocks (2007: 227), Foubert (2007: 57, 60), Spevak (2009), Powell (2010: 171–2, 174–5, 176). 147 Bolkestein (2001 252), Devine and Stephens (2006: 542–4), Powell (2010: 174–5).

260

From Latin to Romance

in this process. The starting point lies in a well-established distinction of traditional philological scholarship, namely the colon. In a series of seminal, though largely neglected, studies on the metrical properties of the sentence, Fraenkel (1932; 1933; 1964; 1965) demonstrates on the evidence of such phenomena as the distribution of enjambment, the position of Wackernagel elements, and the placement of vocatives that the Latin sentence can be broken down into a series of rhythmical clausulae of differing sizes called cola (for an overview, see Laughton 1970; Habinek 1985b: 5–8) which, in turn, correspond to different autonomous pieces of grammatical structure.148 Already we have seen, following Adams (1994a, b), how the recognition of such cola proves essential in understanding the distribution of non-emphatic pronouns (}5.1.2) and the verb ESSE ‘be’ (}5.4.2.4) which, as Wackernagel elements, are regularly attracted to a focal host situated in colon-initial position. These discrete subdivisions of the sentence identified by Fraenkel include any expanded nouns (regardless of case), appositives, prepositional phrases, ablative absolutes, adverbial clauses, participial and infinitival clauses (including with accusative subject), embedded clauses, and verb phrases (see Habinek 1985b: 127; Bauer 2009a: 294–9). Despite the apparent array of different categories implied by the traditional terminology of this classification, it is possible to reduce the various categories of cola identified by Fraenkel to just four fundamental underlying types that encompass all relevant examples. In particular, Fraenkel’s classification of cola, which identifies metrically and semantico-syntactically autonomous units of structure, mirrors with surprising accuracy the autonomous derivational subdomains termed phases in current minimalist syntax (Chomsky 2001). Within the sentential domain we can therefore recognize two major sentential subdivisions (82a), one at the level of the C(omplementizer)P(hrase), hence including adverbial clauses, participial and infinitival clauses, embedded clauses, ablative absolutes, and the other at the level of the verb phrase (henceforth vP), including the lexical V(erb)P(hrase) and all its arguments and associated satellites. Consequently, the left edge of the colon which has traditionally been identified with the area of the clause hosting topicalized and focused elements can now be equated with the left periphery of the CP, already examined in detail in }4.4.2, and the left periphery of the vP, the details of which have been recently discussed in a number of works (cf. Belletti 2004b; 2005a; Poletto 2006; 2007; 2010: 71–2, 73–4). Thus, exemplifying from old Tuscan, we can identify within the clausal structure two phases (cola), a higher one related to the CP superordinate functional structure and a lower one related to the subordinate vP functional structure, whose left edges offer competing topic and focus positions, as illustrated by the higher and lower positions of the focused quantifier tutto ‘everything’ in the near minimal pair (82b–c): 148

For an elaboration of Fraenkel’s theory with additional evidence from the ancient practices of punctuation and a further distinction between rhythmical and rhetorical cola, see Habinek (1985b).

A configurational approach

261

(82) a [CP _____ [IP Aux [vP _____ [VP V]]]] b [CP quando TUTTO [IP ebbe [vP _____ [VP dato tutto]]]] (OTusc., Nov. 16) when all he.had given c [CP ch’ _____ [IP avea [vP TUTTO [VP donato tutto]]]] that he.had all given (OTusc., Nov. 18) ‘[ . . . ] when/that he had given everything’ In a similar vein, within the nominal domain (83a) we can recognize two major phasal subdivisions corresponding to the P(repositional)P(hrase) and the D(eterminer)P(hrase), whose left peripheries can host topicalized and focused constituents. Once again, we exemplify this from old Tuscan: in (83b) the left periphery of the PP headed by dipo’ hosts the modifying focused measure phrase anni cinquanta (cf. Andreose 2010: 621–4), and in (83c) the left periphery of the DP hosts the focused adjectival phrase gentile which, in raising from within the NP, has stranded its associated degree modifier molto and crossed the nominal head aspetto (cf. Giusti 2010b: 396).149 (83) a [PP _____ P [DP _____ [NP N]]] b [PP ANNI CINQUANTA dipo’ [DP _____ la [NP seconda guerra years fifty after the second war anni cinquanta]]] (OTusc., Giamboni Orosio 4.24) ‘fifty years after the war’ c [PP _____ di [DP GENTILE of gentle (OTusc., Dante VN 8.1) ‘of very gentle appearance’

[NP aspetto molto gentile]]] aspect very

It is only by appealing to these four phases (cola) within the clausal (CP, vP) and nominal (PP, DP) domains and their associated left peripheries, to which pragmatically salient constituents of various types can be fronted under topic and focus, that the pragmatically driven dimension of Latin word order can be properly understood. Revealing in this respect is the frequent observation that hyperbaton, and hence leftedge fronting, may mark the confines of individual phrases (Adams 1971: 10), ultimately serving to frame and define the colon (Habinek 1985a) or, in our terms, 149 Cf. the base position of the measure moifier phrase within the DP in (i.a) and that of the adjective piacevole within the NP in (i.b) (both examples taken from Dante’s Vita nuova):

(i)

a a[PP _____ appresso [DP_____ la [NP morte after the death

di questa donna alquanti die]]] (9.1) of this lady some days

‘some days after this lady’s death’ b [PP _____ di of

[DP _____

‘of most pleasant appearance’

[NP aspetto molto piacevole]]] (5.1) aspect very pleasant

262

From Latin to Romance

spell out the confines of the phase. Thus, following Habinek, Kessler (1995: 10) maintains that in a string such as (84) fronting of the adjectival modifier tacitum from its associated noun uulnus serves to mark the left and right boundaries of the colon, ensuring that tacitum uiuit sub pectore uulnus is cohesively read as a single phrase (in our terms a CP or vP phase), thereby extending the phonological phrase from outside the original NP (tacitum uulnus). (84) interea et tacitum uiuit sub pectore uulnus (Lat., Virg. Aen. 4.67) meanwhile and silent.ACC lives under breast.ABL wound.ACC ‘and meanwhile a silent wound is alive in her breast’ In what follows, we shall outline how an appropriate recognition of the role of the left periphery in these four phasal domains (cola) holds the key to a proper understanding of the structural and pragmatic organization of, as well as the restrictions on, Latin prose word order. 5.4.3.2 Modifier fronting A fundamental distinction in the analysis of edgemovement, and hence also hyperbaton, is that illustrated in (85a–c) and (86a–b): (85) a agit maximas gratias (Lat., Cic. Att. 1.20.7) he.does greatest.ACC.PL thanks.ACC b maximas gratias agit (Lat., Cic. Ad Att. 3.8.4) greatest.ACC.PL thanks.ACC. he.does ‘she expresses her warmest gratitude’ c maximas agit gratias (Lat., Cic. Ad Att. 3.5.1) greatest.ACC.PL she.does thanks.ACC ‘s/he expresses her warmest gratitude’ (86) a gratias agere maximas (Lat., Cic. Att. 10.4.11) thanks.ACC do.INF greatest.ACC.PL ‘that he thanked him warmly’ b lacunam intus magnam facito (Lat., Cato Agr. 38.1) hole.ACC inside big.ACC do.IMP.FUT.2SG ‘make a large pit inside’ Irrespective of the position of the verb, in the examples in (85a–c) it is the attributive adjective maximas which has been displaced from its underlying, unmarked postnominal position and fronted under focus to the left periphery of a relevant phase ([N + Adj] ) Adj ( . . . ) [N + Adj]), whereas in (86a–b) it is the nominals gratias and lacuna which have undergone fronting to the left periphery of

A configurational approach

263

a relevant containing phase (namely, [N + Adj] ) N ( . . . ) [N + Adj]).150 We can thus distinguish between edge-fronting of a modifier and edge-fronting of a governing nominal category, henceforth referred to as modifier fronting and nominal fronting, respectively.151 In practice, however, the distributional frequency of the two constructions is not equal, with edge-fronting predominantly targeting modifiers, whether adjectives as in (85a–c) or, for example, genitive or ablative nominal modifiers (see below), and only more rarely, if at all, the governing nominal category.152 Taking sentences (85a–c) then as our representative examples of modifier fronting, we begin by noting that the difference between (85a) and (85b) is not one of verb raising as, for example, claimed in Devine and Stephens (2006), but simply represents the optionality of roll-up of the complement to the preverbal Spec(ifier) position in classical Latin observed in }5.4.1, with the verb remaining in situ in all cases (namely, (S)VO vs (S)OVO). Now in all three examples we claim that, in accordance with its scalar interpretation, the superlative adjective maximas is extracted from its base postnominal position (87a) and raised to (a focus Spec(ifier) position within) the left periphery of the DP phase containing the NP (87b), a case of local fronting that spells out part of the nominal functional structure not otherwise lexicalized in Latin.153 This both predicts and explains the widely observed reading of fronted modifiers like the adjective in cases such as (85a–c), where the modifier receives a (contrastively/ informationally) focused reading and the nominal tail usually a topical or, at any rate, non-focal reading.154 150 Clearly, if the whole D(eterminer)P(hrase) undergoes roll-up to the preverbal position (i), then the nominal head will not be interpreted as having undergone edge-fronting to a higher phase outside of the containing DP, unless other elements are present in the clause such as intus in (86b) which can intervene between the nominal head and its modifier.

(i)

gratias maximas egit (Lat., Cic. Fam. 13.54.1) thanks.ACC greatest.ACC.PL he.did ‘he expressed his warmest gratitude’

151 These two patterns have been variously termed in the literature ‘backward’ and ‘forward’ discontinuity (Pinkster 2005a) and ‘premodifier’ and ‘postmodifier’ hyperbaton (Devine and Stephens 2006). 152 Kessler (1995: 7), Bolkestein (2001: 252), Pinkster (2005b: 250), Devine and Stephens (2006: 572), Bauer (2009a: 289–90, 292). 153 de Jong (1986: 295), Bolkestein (2001: 252), Pinkster (2005b: 251), Devine and Stephens (2006: 544–7), Foubert (2007: 60), Powell (2010: 171–2). In }5.4.2 we argued that bare case-marked nominals should be interpreted as concealed P(ostpositional)P(hrase)s headed by a null adposition (namely, [PP [Spec DP] [P’ P DP]]). For expositional simplicity, in what follows we shall simply represent them as simple D(eterminer)P(hrases)s since, in any case, they always escape the confines of the PP by raising to the latter’s left-periphery under roll-up. 154 As pointed out by Devine and Stephens (2006: 546), revealing in this respect are examples such as (i), where the stranded nominal recitatores ‘readers’ clearly represents old, assumed information, as illustrated by its appearance in the immediately preceding discourse (viz. recitatoribus), whereas the fronted modifier tres ‘three’ is clearly focused in that it contrasts with the preceding cardinal duobus ‘two’.

From Latin to Romance

264

As they stand, (85a–b) might in theory not differ at all in their pragmatic interpretation, but the structure of (85b), together with that of (85c), is superficially ambiguous since once the adjective undergoes local fronting to the left periphery of the containing DP phase, which itself has been raised to the left of the verb under roll-up, it is free, in turn, to use this left-peripheral position as an escape hatch from which to raise outside of the DP phase to undergo further, non-local fronting operations to more remote and pragmatically more salient left-peripheral positions within the clausal superstructure at the level of the vP (87c) or CP phase (87d). This movement out of the DP phase and higher containing phases follows from well-formedness conditions on the locality of movement, as currently formalized in the Phase Impenetrability Condition (PIC) which requires movement out of the containing phase (colon) to transit through the left-edge (viz. head or Spec(ifier) positions) of the left periphery, as represented schematically in (88).155 (87) a [DP [Spec _______] [NP gratias [MAXIMAS] ]] b [DP [Spec MAXIMAS] [NP gratias [MAXIMAS] ]]

(i)

excitauit recitatores (Lat., Cic. Clu. 140–4) duobus recitatoribus [ . . . ] TRES ipse three self.NOM he.called.forth readers.ACC two.ABL readers.ABL ‘two readers [ . . . ] he himself called upon three readers’

155 Contrary to the expectations of the PIC outlined above, Ross (1967) and Devine and Stephens (2006: 559–60) suggest that the application of hyperbaton, and consequently edge-movement, is limited to the confines of the immediately containing finite/tensed clause, but can operate across non-finite clauses (i.a–b; cf. also example (91) below and the discussion in Charpin (1989: 519)). However, as Kessler (1995: 7–10) demonstrates, finite clauses are not necessarily islands for edge-movement, witness examples such as (i.c) where the focused genitive modifier illorum is extracted from the embedded quam-clause, and (i.d)— although not a case of hyperbaton—where ablative complement quibus omnibus has been fronted across two embedded clauses.

(i)

a De quibus TRES uideo [tres sententias ferri] (Lat., Cic. De Amic. 56) of which.ABL.PL three I.see opinions.ACC hold.INF.PASS ‘about these things I see that three opinions are held’ b

EXIGUAM dixit [fortunam exiguam interuenire] (Lat., Cic. De fin. 1.63) little.ACC he.said fortune.ACC interfer.INF ‘he said that fortune interferes with the wise man only a little’

c illorum uides [quam niteat oratio illorum] (Lat., Cic. Fin. 4.3.5) those.GEN you.see how shines.SBJV speech.NOM ‘you see how elegant their speech is’ d quibus omnibus dici uix potest [quantum intersit [quem which.ABL.PL all.ABL say.INF.PASS scarcely it.can how.much differs.SBJV which.ACC ad modum utatur quibus omnibus orator]] (Lat., Cic. De or. 17.55)) to way.ACC uses.PASS orator.NOM ‘the way in which the orator uses these [gestures] makes a difference which can scarcely be described’

A configurational approach

265

c [vP [Spec MAXIMAS] (agit) [DP [Spec MAXIMAS] [NP gratias [MAXIMAS] ]] (agit) ] d [CP [Spec MAXIMAS] [vP [Spec MAXIMAS] (agit) [DP [Spec MAXIMAS] [NP gratias [MAXIMAS] ]] (agit) ]] (88)

CP 2 Spec C' 2 C IP 2 Subj I' 2 Infl vP 2 Spec v' 2 v VP 2 V PP 2 Spec P' 2 P DP 2 Spec D' 2 D NP 2 N XP Modifier

Unlike in (85b–c), in other cases it is possible to identify non-local fronting if other elements are present in the structure.156 For instance, a modifier fronted to the lower vP left periphery can be identified as such if other elements intervene between it and the lexical verb (89a–b) and, especially, if the fronted modifier is preceded by a preverbal subject (89c–d), presumably located in the Spec(ifier) position of the I(nflectional)P(hrase) projection. 156 Obviously, if roll-up fails to obtain, then edge-fronting of the modifier to a higher vP-related (i.a) or CP-related (i.b) left periphery is immediately revealed by the resulting discontinuity between the preverbal modifier and its associated postverbal nominal tail:

(i)

a Latrocinia NULLAM habent robberies.NOM none.ACC have b

‘robberies involve no disgrace’ MEO tu epsitulam dedisti my.DAT you letter.ACC you.gave ‘To my slave you gave a letter?’

NULLAM

MEO

infamiam (Lat., Caes. B.G. 6.23.6) disgrace.ACC

servo? (Lat., Pl. Pseud. 1203) servant.DAT

From Latin to Romance

266

[IP Subj [vP [Spec MODIFIER] [ XP . . . MODIFIER-N [VP V]]]] (89) a qui SUMMAM inter eos habet SUMMAM auctoritatem who.NOM highest.ACC among them has authority.ACC (Lat., Caes. B.G 6.13.8) ‘who has the highest authority among them’ civitati Haeduae TANTAM dignitatis tribuebat b quod TANTAM because so.much.ACC state.DAT Aeduan.DAT esteem.GEN he.granted (Lat., Caes. B.G. 5.7.1) ‘He attached so much importance to the Aeduan state’ c Uercingetorix MINORIBUS Caesarem MINORIBUS itineribus subsequitur marches.ABL follows.PASS Vercingetorix.NOM lesser.ABL.PL Caesar.ACC (Lat., Caes. B.G. 7.16.1) ‘Vercingetorix follows Caesar with fairly short marches’ e propinquitas castrorum CELEREM superatis ex fuga proximity.NOM camps.GEN quick.ACC surpassed.ABL.PL out.of flight.ABL CELEREM receptum dabat (Lat., Caes. B.C. 1.82.5) withdrawal.ACC gave ‘the proximity of the camps gave the defeated a quick refuge from flight’ Similarly, a modifier fronted to the higher CP left periphery is often made visible by its position to the left of a preverbal subject (90a–c), a (lower) complementizer (90d), or a postpositive clause connector (90e). [CP [Spec MODIFIER] Comp [IP Subj [v-VP MODIFIER-N V]]] (90) a AGRARIAM Ti. Gracchus AGRARIAM legem ferebat (Lat., Cic. Sest. 103) agrarian.ACC T. Gracchus.NOM law.ACC proposed ‘Ti. Gracchus proposed an agrarian law’ MAGNA Pompeius b cui to.whom great.ACC.PL Pompey.NOM (Lat., Caes. B.C. 3.4.5) ‘to who Pompey gave great awards’

MAGNA

praemia tribuit awards.ACC granted

c

MAXIMAS tibi omnes MAXIMAS gratias agimus (Lat., Cic. Marc. 33) greatest.ACC you.DAT all.NOM thanks.ACC we.do ‘we all express the greatest thanks to you’

d

HANC

cum habeat HANC praecipuam this.ACC since he.has.SBJV particular.ACC (Lat., Cic. Brut. 261) ‘since he has this particular merit’

laudem praise.ACC

A configurational approach

267

e MULTORUM enim capita MULTORUM ciuium uiderat citizens.GEN he.had.seen many.GEN.PL for heads.ACC (Lat., Cic. Deiot. 12.33) ‘for he had seen the heads of many citizens exposed’ There are even cases where both higher and lower left peripheries can be simultaneously lexicalized by raised focused modifiers, cases of what we may term ‘double focus’ (cf. Devine and Stephens 2006: 547–8; Powell 2010: 177–8, 181 n. 39). For example, in (91) we see that the quantifier nullam and restrictive adjective grauem have been extracted together from their associated nominal tail causam and raised, via the lower vP periphery, to the CP left periphery of the containing AcI clause, from where the quantifier alone is finally raised to what appears to be the left periphery of the matrix vP:157 (91)

quia [vP [QP NULLAM] uideo [CP [QP NULLAM [AP GRAUEM]] [vP [NULLAM [GRAUEM]] serious.ACC because none.ACC I.see subesse [DP [NULLAM [GRAUEM]] causam [GRAUEM]] ]]] (Lat., Cic. Att. 1.10.2) cause.ACC exist.INF ‘because I cannot see that there is any serious motivating reason’

Furthermore, given what we have already established about the fine structure of the left periphery of the CP in }4.4.2, where topics were seen to precede foci, it is possible to have multiple frontings to the same phasal left periphery with the expected Topic + FOCUS order: [DemP haec [NP calamitas [DP [PRIORUM] (92) a [DP [TopP haec] [FocP PRIORUM] this.NOM previous.GEN.PL disaster.NOM [NP consulum [AP priorum]]]]]] (Lat., Liv. 3.31.6) consuls.GEN ‘this disaster that befell the previous consuls’ b [CP cum [TopP ipse] [FocP LITTERAM] [IP-vP Socrates though self.NOM letter.ACC Socrates.NOM ipse nullam LITTERAM reliquisset]] left.SBJV none.ACC (Lat., Cic. De or. 3.60) ‘although Socrates himself had not left a single letter’ 157 The initial step of the structural representation in (91) is greatly simplified for expositional convenience. In particular, we must assume that the restrictive adjective grauem is first fronted to the DP left periphery (i.a) and that the quantifier nullam is subsequently fronted to the left-peripheral Spec(ifier) position of the fronted adjective (i.b), which then, in turn, raise together out of the DP to target the left periphery of higher phasal domains (i.c) before the quantifier floats off to a left-peripheral position within the matrix clause (i.d):

(i)

a b c d

. . . [DP [AP GRAUEM] [QP nullam [NP causam [AP grauem]]]] . . . [DP [AP [QP NULLAM] GRAUEM] [QP nullam [NP causam [AP grauem]]]] . . . uideo [CP [AP [QP NULLAM] GRAUEM] subesse [DP [AP [QP NULLAM] GRAUEM] . . . ]] quia . . . [vP [QP NULLAM] uideo [CP [AP [QP NULLAM] GRAUEM] subesse . . . ]]

268

From Latin to Romance c [CP [TopP hic] [FocP OPTIMUS] [IP-vP hic illis temporibus est best.NOM those.ABL times.ABL is this.one.NOM patronus optimus habitus]] (Lat.,Cic. Brut. 106) lawyer.NOM considered ‘in those days he was considered the best lawyer’ [IP-vP in terris NULLA d [CP [TopP animorum] [FocP NULLA] souls.GEN none.NOM in lands.ABL origo animorum inveniri potest]] (Lat., Cic. Tusc. 1.27) find.INF.PASS can.3SG origin.NOM ‘for the souls of men no origin can be found in the earth’

Turning finally to prepositional phrases, these too frequently display nominal modifiers which migrate under focus to a left-peripheral Spec(ifier) position within the containing DP phase, a case of local edge-fronting: (93) a in [DP [Spec OSTIARII ILLIUS] [NP cella in door.keeper.GEN that.one.GEN storeroom ostiarii illius]] (Lat., Petr. Sat. 37.8) ‘in the storeroom of that door-keeper of that man’ b in [DP [Spec PATRIS] [NP memoria patris]] (Lat., Cic. Att. 4.16.6) memory.ABL in father.GEN ‘in their memories of his father’ ueteribus populi c pro [DP [Spec UETERIBUS] [NP Heluetiorum iniuriis for old.ABL.PL Helvetii.GEN injustices.ABL people.GEN Romani]] (Lat., Caes. B.G. 1.30.2) roman.GEN.SG ‘for the old wrongs done by the Helvetii to the Roman people’ Of course, once raised to the DP left periphery, the modifier is then free and able to escape the DP phase entirely and raise further to the left periphery of the superordinate PP containing the DP (94a–c), and even beyond to higher clausal left peripheries (94d–e). Though in origin this non-local edge-fronting to the prepositional phase must no doubt have carried a strong focused or topicalized reading, it has been widely reported in the literature how this practice became somewhat ‘routine’ and conventionalized in classical prose, ultimately functioning as an indicator of good style rather than just as a pragmatic device.158

158

Gildersleeve and Lodge ([1895] 1997: 432), Vincent (1999: 1128), Bauer (2009a: 287–8), Powell (2010: 173).

A configurational approach

269

(94) a [PP [Spec MAGNO] cum [DP [Spec MAGNO] [NP dolore magno]]] great.ABL with sorrow.ABL (Lat., Cic. Phil. 1.12.31) ‘with great grief ’ b [PP [Spec TRISTI] cum [DP [Spec TRISTI] sad.ABL with ‘with a sad heart’

NP

corde tristi]]] (Lat., Enn. Ann. 482) heart.ABL

c [PP [Spec DEORUM] in [DP [Spec DEORUM] [NP numero deorum]]] number.ABL gods.GEN in (Lat., Cic. Nat.D. 1.118) ‘numbered among the gods’ d [vP [Spec TRIBUS] existat [PP [Spec TRIBUS] ex [DP [Spec TRIBUS] out.of three.ABL arises.SBJV [QP tribus [NP rebus]]]]] (Lat., Cic. Off. 2.65) things.ABL ‘it originates from three things’ e [vP [Spec NULLA] deseruit [PP [Spec NULLA] in [DP [Spec NULLA] in none.ABL he.left [QP nulla [NP re]]]]] (Lat., Nep. Han. 23.8.4) thing.ABL ‘he never deserted him’

5.4.3.2.1

REANALYSIS: ROLL-UP

) EDGE-FRONTING

Finally, given the textual nature of our evidence, we must note that in many cases it is not possible to tell whether a given string is to be interpreted as a case of grammatically induced raising with no pragmatic import, namely roll-up (95a, 96a), or as a case of pragmatically induced raising under focalization/topicalization, namely edgefronting (95b, 96b), since the surface landing position of the modifier, be it the leftperipheral Spec(ifier) position of the NP (roll-up) or of the DP (edge-movement), is invariably to the left of the nominal head.159

159 Clearly no ambiguity arises when other disambiguating elements occur within the nominal group such as (i.a–b), where the prepositional adjunct ex urbe and the object complement mulierum must have been fronted under roll-up to the Spec(ifier) of the immediately containing N(oun)P(hrase) and not fronted under focus to the left periphery of the D(eterminer)P(hrase), since they are preceded by elements (demonstrative hosce, focused adjective nocturna) situated above the NP:

(i)

a [DP [DemP hosce [NP [Spec ex urbe] sicarios ex urbe]]] (Lat., Cic. S.Rosc. 74) these.ACC out.of city.ABL killers.ACC ‘these city assassins’ b [DP [Spec NOCTURNA] [NP [Spec mulierum] sacrificia mulierum nocturna]] (Lat., Cic. Leg. 2.21) nocturnal.NOM.PL wives.GEN sacrifices.NOM ‘the nocturnal sacrifices of the women’

270 (95)

From Latin to Romance a [DP [NP [Spec familiaris funeris]

excusatio familiaris funeris]]

b [DP [Spec familiaris funeris] [NP excusatio familiaris funeris]] domestic.GEN funeral.GEN excuse.NOM ‘the pretext of a domestic calamity’ (Lat., Cic. Rab. Perd. 8) (96)

a [DP [NP [Spec uinaria]

cella uinaria]]

b [DP [Spec uinaria] [NP cella uinaria]] of.wine.ABL cellar.ABL ‘wine cellar’ (Lat., Col. Rust. 12.18.4) Now, we have already noted in }5.3.1.4 how in an archaic period, only partially accessible to us, adjectival and nominal modifiers must have stood in prenominal position under unmarked pragmatic readings, but had already largely shifted to the postnominal position by the time of most of our earliest records, with the prenominal position surviving typically as the position that licenses marked focused and topicalized readings (cf. also Langslow forthc.). Now, these facts find an immediate explanation in terms of our conclusions about roll-up and edge-movement above. With the progressive loss of roll-up as an option in the history of the Latin and in the passage to Romance, we can immediately understand how the erstwhile roll-up operation which, as a purely grammatical principle of linearization devoid of any pragmatic import, originally placed complements and modifiers to the left of their associated head (cf. 95a, 96a), was progressively reanalysed, albeit with some degree of ambiguity, through time as a marked pragmatic order, no longer derived through roll-up, but increasingly through edge-movement to the left periphery of the relevant containing phase (95b, 96b). Indeed, this view finds strong independent support in Devine and Stephens’s (2006: chs 4, 5) survey of the position of nominal arguments and modifiers within the nominal group in classical prose. On the whole, they find that the derived pre-head position tends to correlate with marked pragmatic readings (topic and focus), with very few cases of pragmatically unmarked preposing. A similar picture was noted above in }5.4.2.1 in relation to the increasingly marked partial anteposition (so-called ‘splitting’) of the AcI in non-literary texts and authors, which Adams (2005: 197–8, 204–5) argues to correlate quite systematically with a marked (topic/focus) interpretation of the anteposed material. Both of these findings are highly indicative of an ongoing, but apparently rapidly increasing, reanalysis of obsolescent p(arameter)-ambiguous (Clark and Roberts 1993) roll-up orders as pragmatically driven edge-movements, in generative terms a shift from A(rgument)movement (triggered by an EPP feature) to A'(djunct)-movement (triggered by an edge feature). 5.4.3.3 Nominal fronting As already noted, nominal fronting proves to be a much more marked and considerably less frequent (indeed even entirely absent from some

A configurational approach

271

authors and texts) variant of edge-movement. Consequently, any structural analysis should be able explain the more marked nature of nominal fronting, as well as the pragmatic interpretations it conveys. We begin then by considering (97a), a classic example of nominal fronting where the nominal legiones has raised to the left of the verb, stranding its associated adjective nouas in the postverbal position. Structures like this, however, appear to involve problematic derivations which violate the principle of categorial uniformity required of the head and tail positions occupied by the fronted nominal (the so-called Chain Uniformity Condition). In particular, in the representation in (97b) the nominal legiones constitutes in its base position just part of the overall NP, namely its head (viz. N), but apparently raises to the leftperipheral Spec(ifier) position of the immediately containing D(eterminer)P(hrase) and from there to the Spec(ifier) position of the lower vP left periphery, although both of these latter positions are full phrasal (namely, XP) positions. (97) a legiones conscripsit NOUAS (Lat. Cic. Phil. 11.27) legions.ACC he.enlisted new.ACC ‘he enlisted new legions’ b

vP 2 Spec v' legiones 2 v VP 2 V DP conscripsit 2 Spec D' legiones 2 D NP 2 N AP legiones nouas

However, the structural ill-formedness of examples like (97a) is only apparent, in that there is another available analysis for such structures which is, in fact, immediately corroborated by their pragmatic interpretation. As noted in Devine and Stephens (2006: 531–40) and Powell (2010: 172), in nominal fronting the stranded modifier normally receives the same focused reading that obtains under modifier fronting, whereas the extracted nominal usually conveys predictable information, hence typically interpreted as topical and, at most, sometimes a weak focus. In short, the pragmatic interpretation of nominal fronting examples like (97a) appears to be identical to that of modifier fronting examples such as (98), the only difference being one of syntax and, more specifically, word order.

272

From Latin to Romance

(98) ut NOUAS scriberet [DP [Spec NOUAS] [NP legiones [nouas]]] that new.ACC.PL he.enrolled.SBJV legions.ACC (Lat., Liv. 29.13.1) ‘that he should enroll fresh legions’ To account for the pragmatic interpretation of nominal fronting structures such as (97a), we argue that these too involve the intermediate step of local edge-fronting of the modifier to the Spec(ifier) position of the DP left periphery, where it receives the observed focused reading (99a). If nothing else happens, then we derive a case of local modifier fronting along the lines of examples such as (85a), repeated here as (99b). [NP legiones nouas]] (99) a conscripsit [DP [Spec NOUAS] he.enlisted new.ACC.PL legions.ACC b agit [DP [Spec MAXIMAS] [NP gratias maximas]] (Lat., Cic. Att. 1.20.7) thanks.ACC he.does greatest.ACC.PL However, we know that in classical Latin there still existed the increasingly obsolescent and marked option of ‘rolling-up’ the complement to a Spec(ifier) position to the left of the verb (or other selecting head), as seen in (85b). In a similar vein, we argue that nominal fronting examples such as (97a) also involve the additional step of roll-up, but not full roll-up of the entire DP category to the Spec(ifier) position of the governing verb, but just partial roll-up of a subpart of the available DP structure (cf. also Polo 2004: 420–1 n. 43; Devine and Stephens 2006: 552). In particular, once the modifier has been locally fronted to the left periphery of the DP phase, then either the full DP can undergo roll-up, as in (100a) with local modifier fronting of nouas, or a subpart thereof such as the NP can undergo partial roll-up giving rise to what we have been calling nominal fronting (100b). This, in turn, provides a highly natural explanation of the observed pragmatically neutral interpretation of the raised nominal, which has not migrated to a phasal left edge under pragmatic fronting, but has undergone a purely grammatically driven case of movement required to satisfy the formal linearization conditions of the system. (100) a [DP [Spec

NOUAS]

[NP legiones [nouas]]] scribendas new.ACC.PL legions.ACC write.GER.ACC.PL [DP [Spec nouas] [NP legiones [nouas]]] (Lat., Liv. 6.27.9) ‘fresh legions should be enrolled’

A configurational approach

b

273

vP Spec v' legiones nouas v VP V DP conscripsit Spec D' NOUAS

D

NP

N AP [legiones [nouas]] It turns out then that what we have been calling nominal fronting is actually just an underlying case of modifier fronting, the application of which is superficially concealed by the effects of partial roll-up. The marked nature and relative infrequency of nominal fronting therefore follows straightforwardly as a general consequence of the already widely observed growing obsolescence and concomitant markedness of roll-up.160 This is a pleasing result as it means that pragmatically driven edge-fronting can now be reduced to a single operation, with the apparent superficial variation between modifier and nominal fronting (and their associated degrees of markedness) falling out in a principled manner from the already independently established variation in roll-up. Crucially, as the derivation in (100b) reveals, what is extracted under nominal fronting is not just the nominal head, but a full phrasal category containing at least the latter. Indeed, this analysis of nominal fronting as a surface reflex of partial roll-up predicts that any maximal phrasal category below the DP containing the fronted focused modifier, and not just the nominal head, can be

160 Furthermore, the less-than-perfect status of nominal fronting, which it will be recalled is often entirely absent from some authors and texts, can also be ascribed to the fact that the derivation in (100b) involves a questionable step. In particular, following fronting of the A(djectival)P(hrase) nouas to the leftedge Spec(ifier) position of DP, subsequent remnant movement of the NP [NP legiones [nouas]] to the leftperipheral Spec(ifier) position of lower vP left periphery creates a configuration, in which the ‘trace/copy’ of the moved AP nouas in its base position is carried along with the head legiones under remnant NP movement and is thus forced to cross its own (now lower) surface spell-out position in the left-peripheral Spec(ifier) position of the DP (for further discussion of the problems involved with such derivations, see Agbayani and Golston 2010: 149–54).

274

From Latin to Romance

rolled-up including Q(uantifier)P(hrase)s (101a), Dem(onstrative)P(hrase)s (101b), and NPs (101c).161 (101) a [QP omnes [NP cogitationes [suas]]] terminaret [DP [Spec SUAS] all.ACC.PL thoughts.ACC it.confined.SBJV its.ACC.PL [QP omnes [NP cogitationes [suas]]]] (Lat., Cic. Arch 29) ‘it confined all its thoughts’ b [DemP earum [NP templum [dearum]]] inflammauit [DP [Spec DEARUM] temple.ACC he.burned goddesses.GEN those.GEN [DemP earum [NP templum [dearum]]]] (Lat., Cic. Orat. 211) ‘he set fire to the temple of those goddesses’ c [NP mentionem [Sullae]] facere [DP [Spec SULLAE] [QP nullam mention.ACC do.INF Sulla.GEN none.ACC [NP mentionem [Sullae]]]] (Lat., Cic. Sull. 37) ‘to make no mention of Sulla’ This approach to edge-fronting incorporating the notion of partial roll-up thus allows us to understand a wide variety of complex structures and recognize in them an underlying regularity which is often superficially disrupted and frequently interpreted as evidence for the non-configurational structure of Latin (cf. chapter 3). Illustrative in this respect are examples such as (102). (102) remque commouisti noua disputatione dignam (Lat., Cic. Brut. 297) thing.ACC=and you.caused new.ABL discussion.ABL worthy.ACC ‘you have raised a matter worthy of fresh discussion’ In this structure the object complement rem is modified by the adjective dignam, which itself takes an ablative nominal complement noua disputatione, itself the product of local modifier fronting (103a). From this underlying structure, the ablative complement is first ‘rolled-up’ to the Spec(ifier) position of the governing A(djectival)P(hrase) (103b). In turn, the whole AP is fronted to the left edge of the containing DP under focus (103c), after which partial roll-up of the remaining NP constituent now containing just the head rem in its phonological representation takes place (103d).

161 The fact that the subpart of the DP that undergoes partial roll-up does not, and indeed cannot, transit through the left edge of the DP phase (already occupied by the focused modifier) to reach its surface position to the left of the governing verbal or nominal head does not violate general principles on the locality of movement, as embodied in the Phase Impenetrability Condition (PIC), since this is not a case of subextraction proper out of the DP (cf. Left Branch Condition) but, rather, a case of argument raising (albeit a subpart thereof) to satisfy the linearization principle.

A configurational approach

275

(103) a . . . rem [AP dignam [DP [Spec NOUA] disputatione [noua]]] b . . . rem [AP [Spec NOUA DISPUTATIONE] dignam [DP [Spec NOUA] disputatione [noua]]] c . . . [DP [Spec NOUA DISPUTATIONE DIGNAM] [NP rem [AP [Spec NOUA DISPUTATIONE] dignam . . . d [vP [Spec rem . . . ] commouisti [DP [Spec NOUA DISPUTATIONE DIGNAM] [NP rem . . . ]]] Finally, we turn to prepositional phrases where, as previously observed (}3.4.1, }5.1.2), nominal fronting proves even more marked and rare and is chiefly limited to verse, witness the following examples: (104) a [PP [Spec numine [dominae]] sub [DP [Spec DOMINAE] deity.ABL under mistress.GEN [NP numine [dominae]]]] (Lat., Ov. Met. 15.546) ‘under the deity of my mistress’ b [PP [Spec damno [magno]] cum [DP [Spec damage.ABL with great.ABL (Lat., Pl. Asin. 187) ‘at considerable cost’

MAGNO] [NP

damno [magno]]]]

c [PP [Spec aciemque [mediam]] per [DP [Spec MEDIAM] through middle.ACC line.ACC=and [NP aciem [mediam]]]] (Lat., Liv. 9.43.15) ‘and through the middle of the line’ d [PP [Spec litus harenosum [Libyae]] ad [DP [Spec LIBYAE] to Libya.GEN shore.ACC sandy.ACC [NP litus harenosum [Libyae]]]] (Lat., Virg. Aen. 4.257) ‘to the sandy shore of Libya’ Structurally, the proposed analysis is identical to that of the previous cases of nominal fronting examined. For example, in (104a) the genitive complement dominae of the nominal head numine first undergoes focus-fronting to the left-periphery of the containing D(eterminer)P(hrase), following which the remnant NP now containing just the nominal head numine in its phonological representation undergoes partial roll-up to the left-peripheral Spec(ifier) position of the P(repositional)P(hrase). Additional support that partial roll-up in (104a–c) involves raising of a phrasal category (viz. NP) is provided by examples such as (104d) in which, following modifier fronting of the modifier Libyae to the DP left periphery, roll-up raises not

276

From Latin to Romance

just the nominal head litus, but the entire NP containing the modifying A(djectival)P(hrase) harenosum.162 Despite the structural similarities with the other cases of nominal fronting, prepositional structures like those in (104a–d) prove most rare and are extremely marked stylistically, even in verse. This fact, we suggest, can be independently explained once again by appealing to roll-up. Whereas we have seen that roll-up producing headfinal structures represents an obsolescent feature of the language which, under the deliberately conservative normative pressures of the literary language, was artificially maintained (albeit alongside of innovative head-initial structures) in the classical literary language with most categories (}5.4.1), this option already represented a markedly archaic and non-productive pattern in conjunction with complementizers and adpositions since the earliest texts (cf. }5.3.1.1). In this light, the unmistakable rarity and highly marked nature of structures such as (104a–d) now follows automatically, not from any concept of structural ill-formedness, but from the fact that they involve (partial) roll-up with a category, namely an adposition, which, save very few exceptions, had largely become consistently head-initial.163 In short, structures such 162 Partial roll-up is also needed in certain cases of modifier fronting such as (i), where in order for the quantifiers compluribus and aliis to undergo focus edge-fronting to the left periphery of the P(repositional) P(hrase) stranding the lower nominal head causis, the NP containing the latter must first be fronted to the left edge of the D(eterminer)P(hrase). Only then can the remnant Q(uantifier)P(hrase) containing compluribus and aliis undergo partial roll-up to the prepositional left periphery.

(i)

[causis]]] de [DP [Spec causis] [QP compluribus [aliis [PP [QP compluribus [aliis several.ABL other.ABL.PL of reasons.ABL [NP causis]]]]] (Lat., Caes. B.G. 5.54.5) ‘for several other reasons’

163 It goes without saying that the analysis of edge-fronting outlined here equally applies to all other categories. For example, in (i.a) the adverbial degree modifier aeque has been extracted from the leftperipheral Spec(ifier) position of its associated adjective iucunda and raised to the left periphery of the DP containing the NP headed by uita. Whereas in most cases it appears possible to explain derived orders such as (i.b–c) in terms of (partial) roll-up to the left-peripheral Spec(ifier) position of the governing lexical category, here the A(djectival)P(hrase), in other cases (i.d) it appears necessary to assume a further functional layer (e.g. aP on a par with vP) whose left periphery hosts a raised focused constituent before partial roll-up of the remaining constituent takes place.

(i)

a [DP [Spec

AEQUE] [NP

equally

uita [AP [aeque] iucunda]]] (Lat., Cic. Diu. 4.30) life.NOM pleasant.NOM

‘an equally pleasant life’ b [AP [Spec

MEA]

carior [DP [Spec MEA] [NP uita [mea]]]] (Lat., Cic. Sest. 45) my.ABL dearer.NOM life.NOM

‘dearer than my own life’ c [AP [Spec

OMNIS] expers [DP [Spec OMNIS] [NP curae]]] (Lat., Liv. 1 Praef. 5) all.GEN free.NOM concern.GEN ‘free of all concern’

d [DP [NP [Spec similes [imperitorum]] philosophos [aP [Spec IMPERITORUM] similar.NOM.PL philosophers.NOM uneducated.GEN.PL [AP similes [imperitorum]]]]] (Lat., Cic. Nat.D. 2.45) ‘philosophers similar to the uneducated’

A configurational approach

277

as (104a–d) instantiate rare cases of (partial) postpositional phrases, and hence their obvious marked status. 5.4.3.4 From Latin to Romance As noted by Väänänen (1982: 259), hyperbaton, and hence edge-fronting, represents ‘one of the most distinctive features of Latin with regard to Romance’, a feature which, although surviving residually in some of the earliest Romance literary texts (cf. Poletto forthc.), already shows signs of an inexorable decline in subliterary and late Latin.164 It is tempting to think of the decline of edge-fronting in the passage from Latin to Romance as a direct concomitant of the decline of the case system, especially since pragmatically driven word orders and discontinuity in Latin appear to be licensed by the availability of a rich case and agreement system (Bauer 2009a: 286). Yet, as convincingly argued by Herman (1985: 355), the case and agreement system of Latin was, apart from perhaps some confusion between the accusative and ablative singular in some declensions, essentially still very much intact at the time of the first signs of the decline in edge-fronting, forcing us to recognize that ‘the loss of structural disjunction was not a consequence of the loss of the case system’. There must therefore be some other explanation for the loss of hyperbaton and edge-fronting in the passage from Latin to Romance, a formal account of which suggests itself in terms of the preceding analysis of edge-movement. In particular, we have seen that edge-movement targets a piece of nominal substructure and fronts it under focus or topic to at least the left periphery of the immediately containing DP, from where it can then escape the confines of the containing DP phase (colon) and raise further to higher and pragmatically more salient focus and topic positions in the left edge of PP, vP, and CP. However, in the absence of raising to the DP left periphery, all other potential higher edge-movements are blocked, an observation which highlights how the essential difference between Latin and Romance crucially lies at the level of the DP. Indeed, as already observed in }4.4.2 and }5.4.3.1, Romance presents extensive evidence for left-peripheral topic and focus positions at the level of CP and vP, suggesting that what prevents these from being targeted by nominal substructures must be the result of some barrier effect of the Romance DP which blocks subextraction out of the NP, a restriction traditionally known as the Left Branch Condition (Ross 1967; cf. }3.2). Consequently, alongside the postnominal base position (105a), Latin also freely allows extraction of an adjectival modifier out of the NP to the DP left periphery (106a), or from there to a higher left periphery (107a, 108a), whereas Romance allows neither of these latter two options (106b, 107b, 108b) inasmuch as the modifier cannot vacate the containing DP phase (colon). As a consequence, in Romance the only legitimate way for such modifiers to receive a

164

Adams (1971: 10–11; 1976b: 140–1), Herman (1985: 351–3; 2000: 82), Bauer (2009a: 288–92).

278

From Latin to Romance

pragmatically marked reading and to reach the lower vP or higher CP left periphery is for the whole DP containing them to undergo edge-fronting (109a–b). (105)

a inter [DP [NP aues [albas]]] uetuit consistere among birds.ACC white.ACC.PL he.forbade stay.INF coruum (Lat., Ov. Met. 2.631–2) raven.ACC b entre [DP [D' las [NP aves [blancas]]]] vedó al among the birds white he.forbade to.the cuervo asentarse (Sp.) raven settle.INF=self ‘he stopped the raven from living among the white birds’

(106)

a [DP [Spec cumque with= medullis]]]] [ . . . ] marrow.ABL.PL

albis] [D'  [NP ossa [[cum albis] and white.ABL.PL bones.ACC condebat in aluum (Lat., Ov. Met. 14.208–9) he.buried in belly.ACC

b sepultaba [DP (**[Spec blancas]) [D' los [NP huesos [con médulas he.buried white.FPL the bones with marrow.FPL [blancas] ]]]] (Sp.) white.FPL ‘he filled his belly with bones full of white marrow’ (107)

a [albas] flexit [DP [Spec albas] [D' [NP aues birds.ACC.PL whites.ACC.PL he.turned albas]]] illuc (Lat., Ov. Met. 10.720) thither b (**[blancas]) giró [DP (**[Spec blancas]) [D' las white he.turned the él (Sp.) [NP aves [blancas]]]] hacia birds white towards him ‘he turned the white birds towards him’

(108) a [PP [Spec albis] in [DP [Spec albis] [D'  [NP turribus towers.ABL white.ABL.PL in [albis] ]]]] (Lat., Ov. Met. 4.48) b [PP (**[Spec blancas]) en [DP ([Spec blancas]) [D' las [NP torres white in the towers [blancas] ]]]] (Sp.) white ‘among the white dovecotes’

A configurational approach (109)

279

a **[blanca] escogieron [DP [Spec blanca] [D' la [NP limusina [blanca] ]]] (Sp.) white they.chose the limousine b [DP la limusina BLANCA] escogieron [DP la limusina [blanca] ] (Sp.) the limousine white they.chose ‘they chose the WHITE limousine’

The Latin vs Romance contrast in (106)–(108) is not arbitrary but reveals a strict complementary distribution between fronting to the DP left periphery and the presence of articles (cf. Bošković 2005a,b; 2008). The pattern suggests that the accessibility or otherwise of the DP left periphery hinges directly on properties of the D(eterminer) position which, if inactive or deficient as in Latin which lacks articles (represented by  in (106)–(108) above), freely allows fronting to its associated left-peripheral Spec(ifier) position (110a), but blocks any such fronting in languages with articles like Romance where the D(eterminer) position is very much active (110b). (110) a . . . [DP [Spec Top/Foc] [D'  [NP]]] (Latin) b . . . [DP [Spec ] [D' ART [NP]]] (Romance) Indeed, this robust empirical generalization finds further support in Poletto’s (forthc.: ch. 3) analysis of early Tuscan (cf. also Poletto 2010: 74–5), where she observes that in those now very rare cases in which ‘an XP is preposed in front of the N, the N never has a definite determiner’, witness some of her examples in (111a–c):165 (111) a Facestilo tu per dare [DP [Spec di me] [NP esemplo [di me]]] you.did=it you for give.INF of me example alle genti? (OTusc., Giamboni LVV. 2.1) to.the peoples ‘Did you do it to make an example of me to the people?’ b [DP [Spec Di dolor] [NP madre antica [di dolor]]] (OTusc., Dante VN 42.2) Of sorrow mother ancient ‘old mother of misery’ c gli altri c’ han [DP [Spec d’ amor] [NP neente [d’amor] ]] the others that have of love nothing (OTusc., Davanzati Rime 11.14) ‘the others who love not’

165

For some rare counterexamples to this generalization, see, however, Giorgi (2010: 277–8).

280

From Latin to Romance

As these old Tuscan data highlight,166 there is a complementary development in the gradual loss of edge-fronting and the progressive emergence of the article (}4.2.2), which allows us to firmly situate the cause of this change with properties of the D(eterminer) head of the DP phase. Formally, there are a number of ways of implementing this empirical generalization (for an overview, see Bošković (2005a, b; 2008) and Bošković and Gajewski (forthc.)). For example, we have already seen (}3.2) how one approach is to assume that Latin entirely lacks the DP functional structure, although this fails to explain the position targeted by Latin modifiers under local edge-fronting. Another idea is to assimilate nominal edge-fronting structures to clausal wh-extraction structures (Corver 1992): in the same way that realization of the complementizer that in (112a) is said to block the tail of the extracted wh-phrase who from connecting with its higher surface and intermediate positions (a so-called that-trace effect), the ill-formedness of (112b) can be understood in terms of the intervening effect of the definite article the, which, on a par with that in (112a), prevents the tail position of the adjective red from being properly bound by its higher intermediate and surface positions. Such an analysis, however, incorrectly predicts that extraction via the left periphery of all phases, and not just DPs, should prove ill-formed when the head of that phase is also simultaneously filled. Yet, we have already seen (cf. 90d–e) that extraction via the Spec(ifier) position of the CP left periphery is indeed possible when the head position is also filled by an overt complementizer. (112) a [Who] do you think [CP [Spec who] [C' (**that) [IP [who] spoke?]]] [NP [red] book]]] b **[Red] he bought [DP [Spec red] [D' the Finally, the observed Latin vs Romance contrast is also reminiscent of the traditional parametric variation observed in subjacency effects across languages (Rizzi 1982), where it is noted, for example, that, besides NP, I(nflectional)P(hrase) is a bounding node (namely, a barrier to movement) in a language like English, but not in Italian where C(complementizer)P(hrase) is a bounding node. This difference therefore accounts for the fact that movement of the relative, which can cross at most only one bounding node, is illicit in the English example (113a) where it crosses two bounding nodes (I can . . . , and you think . . . ), but well-formed in the Italian (113b) where it crosses only one bounding node (che cosa . . . ).

166

Data like these can be readily replicated from all other early Romance varieties, witness the opening line from the ninth-century Strasbourg Oaths in (i), where the genitive Deo and adjective Christian have been fronted to the left of their governing nominal heads: (i) Pro Deo amur et pro Christian poblo (OFr.) for God love and for Christian people ‘For the love of God and for the Christian people’

A configurational approach

281

(113) a **George’s new idea [CP [of which] [IP I can imagine [CP [what] [IP you think of which]]]], will soon become known to everybody b la nuova idea di Giorgio [CP [Spec di cui] [IP immagino the new idea of Giorgio of which I.imagine di cui ]]]] diverrà [CP [Spec che cosa] [IP pensi what thing you.think will.become presto di pubblico dominio (It.) soon of public domain Along similar lines, we might hypothesize that whereas in Romance both DP and NP are bounding nodes, in Latin only NP counts as a bounding node and therefore fails to inhibit extraction out of DP.167 At any rate, and however we ultimately decide to express it formally, for our purposes it is sufficient to note that the observed blocking effect, absent in Latin, is related to the presence/absence of articles in the two languages. In short, the defective nature of the Latin D(eterminer) position, reflected in its failure to lexicalize this functional head, neutralizes the potential barrierhood (or phasehood) of DP, thereby licensing subextraction out of the NP and edge-fronting. In Romance, by contrast, this same D(eterminer) head is fully active, as formally manifested by its ability to license articles, and is therefore able to endow DP with the observed barrierhood.168

5.5 Conclusion In contrast to chapters 3 and 4, which considered an approach to the changes in word order and sentence structure from Latin to Romance in terms of the progressive rise of configurationality and concomitant functional structure, this chapter has attempted to demonstrate how the same empirical generalizations can be captured within an approach that assumes the presence of both configurational and functional structure already in Latin. In essence, the perceived non-configurationality of Latin has been broken down into two main ingredients: (i) grammatically free word order resulting from an ongoing change in the head directionality parameter (ultimately interpreted as the progressive loss of roll-up), which a priori allows dependents/ complements to occur on either side of their head (114a–b); and (ii) pragmatically driven word order, often producing discontinuous structures, resulting from the greater accessibility of topic- and focus-fronting to positions situated in the left

167

The illicit nature of this sort of movement observed in Romance has in many recent approaches been ascribed to the fact that it is ‘too’ local/short (cf. Ishii 1999; Grohmann 2000; 2003; Abels 2003; Bošković 2005a: 16–17). 168 Cf. a similar distinction between the transitive/unergative and unaccusative/passive verb phrase (vP) in terms of the non-defective vs defective status of the v head (Chomsky 2001: 12).

282

From Latin to Romance

edge of individual (phasal) functional projections (115a). Interpreted in this manner, the apparent emergence of configurationality in Romance is to be understood as the surface effect of the rigidification of the directionality parameter (114c) and of the restricted accessibility of edge-fronting to left-peripheral positions within the functional structure (115b). (114)

a [NP []

dux [militum]] (Lat., Liv. 10.40.2) leader.NOM soldiers.GEN

b [NP [militum] dux [militum]] (Lat., Liv. Per. 25.) soldiers.GEN leader.NOM c le (**des soldats) commandant des soldats (Fr.) the of.the soldiers leader of.the soldiers ‘the general’ (115)

celeris subsidii]] a [DP [Spec celeris] [NP spe hope.ABL assistance.GEN swift.GEN confirmata (Lat., Caes. B.C. 3.69.2) encouraged b assurée par [DP [Spec (**prompt)] l’ [NP espoir d’ assured by swift the hope of un prompt secours]] (Fr.) a swift assistance ‘encouraged by the hope of speedy assistance’

If, as argued in chapter 3, Latin syntactic structures were predominantly of an exocentric, ‘flat’ design in which all elements can, in principle, occur in all positions, then there would be no way in which the strict generalizations of the head parameter and the regular surface positions targeted by pragmatically salient (focused and topicalized) elements, defined in both linear and hierarchical terms, can be formally stated. In contrast, the approach adopted here crucially recognizes an unmistakable regularity in the recurrent structural patterns of Latin, assuming the design of Latin syntax, like that of all languages, to be fully configurational in that it makes available fixed positions for heads and their complements—with the latter oscillating between an immediately post-head base position (114a) and an immediately pre-head derived position (114b)—and the full range of functional projections that were assumed to be missing in chapter 4. In a language like Latin, such functional projections are not typically spelt out overtly by functional categories such as determiners and auxiliaries lexicalizing the relevant head positions, as happens in Romance (cf. the French determiner l’ ‘the’ in (115b)), but are made visible instead through their left-peripheral

A configurational approach

283

modifier/Spec(ifier) positions which frequently serve to host pragmatically salient (focused and topicalized) elements (115a). It is these formal changes in the directionality parameter and the differential role of functional structure in the two varieties which, we argue, conspire to give the superficial impression of differences in configurationality and functional structure in Latin and Romance. Above we have described in great detail these grammatical and pragmatic aspects of Latin word order, showing how the apparent free variation in the head parameter witnessed in (114a–b) is in large part to be understood in terms of diachronic and diaphasic variation, and how the apparent freedom of Latin word order driven by pragmatic factors witnessed in discontinuous structures like (115a) follows a well-defined set of structural principles, ultimately ensuing from the defective nature of the Latin D(eterminer) position which surfaces in the failure of DP to act as a barrier to the extraction of elements from within the NP. We thus see that whereas Latin typically spells out left-peripheral Spec(ifier) positions through roll-up and left edge-fronting, Romance typically spells out the corresponding head positions, including through (nominal/verbal) head-movement operations entirely absent from Latin (cf. }3.2.2.1.1, }4.2.2.2, }4.3.2), giving rise to an important typological difference in the syntax of Latin and Romance which privilege an XP-type (phrasal) and X-type (head) syntax, respectively. As we shall see in the following chapter, this difference finds a direct parallel in the typological distinction between dependentmarking and head-marking languages, which Latin and Romance can arguably be considered to exemplify.

6 Head-marking and dependentmarking 6.1 Introduction Following the seminal work of Nichols (1986), a valuable and insightful typological distinction in grammatical systems is that between head-marking and dependentmarking.1 In dependent-marking constructions, the relation between head and dependent is marked directly on the dependent itself, as in the Latin example (1a), where the accusative case marking borne by the noun portas immediately identifies its syntactic relation as that of direct object. In head-marking constructions, by contrast, the relation between head and dependent is marked on the head, as in the corresponding colloquial Spanish example (1b), where the verbal head abrió bears a feminine plural accusative clitic las that spells out the direct object relation of the nominal las puertas with matching number and gender features. In this light Romance clitics, often viewed as the spell-out of the verb’s Case feature (see Borer 1984), receive a highly natural interpretation, if they can now be understood as the overt realization of such a feature on an independent D-head adjoined to the verb.2 (1)

a portas gates.ACC

aperuit (Lat., Caes. B.C. 3.11.4) he.opened

b las abrió them.FSG= he.opened ‘he opened the gates’

las puertas (Sp.) the.FPL gates.F

Contrasting examples like these highlight an important typological shift in the marking of grammatical relations in the transition from Latin to Romance, involving a gradual, but still generally incomplete, move away from dependent-marking towards head-marking (Vincent 1997c: 164; Cennamo 2001a: 16). This shift finds

1

See also van Valin (1987), Vincent (1993; 1997c: 164), Bresnan (2001: 111–13), Mereu (2004: 63–72). In what follows, morphologically marked heads appear in bold and morphologically marked dependents are underlined. If unmarked, they appear in normal type. 2

Head-marking and dependent-marking

285

further support in Nichols’s (1986: 104) observation regarding the distributional skew in word order patterns: whereas predominantly head-marking languages display the greatest freedom, allowing, for example, both verb-initial and verb-final orders, languages that show an overwhelming tendency towards dependent-marking prove much more restricted with ‘neutralization of word-order types to a single possibility, SOV’ (cf. also van Valin 1987: 393). Indeed, we have already seen in the previous chapters how in the transition from Latin to Romance word order progressively developed from (S)OV to (S)VO, a fact which now independently corroborates the rise of head-marking at the expense of dependent-marking. Furthermore, this typological distinction is entirely compatible with, and provides additional support for, the (albeit opposing) theories regarding configurationality and the rise of functional categories developed in chapters 3–4 and chapter 5. In chapter 3 the transition from Latin to Romance was characterized in terms of the rise of configurationality, a distinction which Nordlinger (1998) and, in turn, Bresnan (2001: 113–4) integrate with the head-/dependent-marking distinction to yield four basic language types: head-marking, configurational/non-configurational languages (e.g. Navajo/Mohawk), and dependent-marking configurational/non-configurational languages (e.g. Icelandic/Dyirbal). Within this typology Latin is then similar to Dyirbal, marking within a predominantly non-configurational syntax grammatical dependencies lexocentrically through the distinct inflectional forms assumed by its dependents, whereas in the configurational syntax of Romance grammatical dependencies, as in Navajo, are marked partly endocentrically through fixed, hierarchical structure and partly through head-marking. By the same token, the head-/ dependent-marking distinction provides us with an immediate and complementary account for the discussion in chapter 4 of the emergence in Romance of the functional heads D, I(nfl), and C (and their associated projections) which, with the shift from dependent-marking to head-marking, make available the means to directly encode grammatical information relating to nominal, verbal, and clausal dependents. By contrast, in chapter 5 we saw how Latin displays, especially in its more archaic and literary varieties, the phenomenon of roll-up with raising of dependents to the left of their associated head, and that under the right pragmatic circumstances it exhibits extensive edge-fronting of dependents to left-peripheral positions of individual functional projections. In the passage to Romance, by contrast, both roll-up and liberal edge-fronting were lost and in their place emerged head-marking strategies in the form of head movement operations, in particular N(oun)-raising (}3.2.2.1.1, }4.2.2.3, }4.2.3.1) and V(erb)-raising (}3.3.2, }4.3.2), and the appearance of functional categories such determiners, auxiliaries, and complementizers. In short, we see a move away from a Latin dependent-marking syntax in which various dependents are raised to lexical and functional left-peripheral modifier/Spec(ifier) positions towards a Romance head-marking syntax in which functional head positions are increasingly spelt out under direct lexicalization or movement.

286

From Latin to Romance

6.2 Variation in marking As with the synthesis-analysis distinction discussed in chapter 2, it is not appropriate to talk of head-marking and dependent-marking languages (Nichols 1986: 73); rather, this distinction must be predicated of particular constructions (van Valin 1987: 375), although it cannot be denied that, when considered from this perspective, individual languages show an overwhelming tendency to consistently employ one type over the other (Nichols 1986: 66; Vincent 1993: 141). For example, Latin is dependent-marking, not only at the level of the verb phrase (cf. 1a), but also at the level of the nominal/prepositional phrase, witness in (2a) the agreement on the dependent attributive adjective ueteribus, the genitive case on the possessor Ariouisti, and the ablative case marking on the nominal head and attributive adjective ueteribus copiis determined by the governing prepositional head cum. Similarly, at the level of the clause (2b) non-finite complements are (indirectly) marked as such by the accusative case borne by the infinitival subject (cf. uos) and by the infinitival marking -RE (cf. audire), itself an original locative case ending, of the embedded verb form, whereas finite complements are marked by the presence of a complementizer (cf. ut) and, in many cases, also by subjunctive verb forms (cf. sedeant; Nichols 1986: 151). (2)

a cum [ueteribus copiis with old.ABL.PL troops.ABL ‘with Ariovistus’ old troops’

[Ariouisti]] Ariovistus.GEN

(Lat., Caes. B.G. 1.37.3)

iubet uos imperator histricus, /bonoque b audire listen.INF orders you leader.NOM of.actors.NOM good.ABL.SG=and ut animo sedeant in subselliis (Lat., Pl. Poen. 4–5) that heart.ABL they.remain.SBJV in benches.ABL ‘It is the head-manager who bids you listen, that with a good grace they may be seated on the benches’ Thus, although ‘Latin is a prototypical example of a D[ependent-]M[arking] language’ (Vincent 1997c: 164), it does nonetheless exhibit some head-marking strategies. An obvious example is the referencing of nominative subjects through person and number agreement on the finite verb, as illustrated in the 3SG/PL opposition in (3a–b; cf. Lehmann 1985: 85), as well as in the indexing of number and gender through participial agreement in the perfective paradigms of the middle voice (4a–b; cf. }7.3.1.2). These exceptions are not exclusive to Latin, but reflect a crosslinguistic preference in otherwise consistently dependent-marking languages for head-marking at the level of the verb phrase (Nichols 1986: 77, 104).

Head-marking and dependent-marking (3)

287

a hic uester hostis uestram rem publicam oppugnat this.NOM your.NOM enemy.NOM your.ACC thing.ACC public.ACC attacks (Lat., Cic. Phil. 4.14) ‘but this enemy of yours is attacking your state’ b hostes [ . . . ] castra oppugnant (Lat., Caes. B.G. 5.40.3) enemies.NOM camp.ACC.PL attack.3PL ‘the enemy attack the camp’

(4)

oppugnata est (Lat., Liv. 5.4.11) a Decem quondam annos urbs ten once years.ACC city.NOM.F attacked.NOM.FSG is ‘a city was once besieged for ten years’ b castra consulis oppugnata [ . . . ] camp.NOM.NPL consul.GEN attacked.NOM.NPL ‘the consul’s camp was attacked’

sunt (Lat., Liv. 35.11.2) are

Similarly, although modern Romance varieties increasingly show a strong tendency towards head-marking, as we shall see below, there are still a number of core dependent-marking strategies in evidence,3 including, for example, subordination where sentential complements, be they finite (5a) or non-finite (5b), are not marked on the governing predicate but directly on the complement through an overt complementizer (non-finite: reflexes of DE, AD; finite: reflexes of QUOD/QUID, QUIA, SI). However, as Vincent (1993: 150) perceptively notes, the apparent retention of dependent-marking in the case of Romance subordination is not necessarily exceptional, but reflects the resolution of two competing forces between clauseinternal and clause-external syntax inherent in subordination. On the one hand, if we treat the sentential complement as an argument, then we should expect it be marked through head-marking on the governing predicate in line with the favoured head-marking strategy of clausal (namely, VP) relations (Nichols 1986: 75). If, on the other hand, we recognize that the sentential complement is an embedded clause, then we should expect it to show dependent-marking in accordance with the favoured crosslinguistic pattern at the sentence level (Nichols 1986: 76), albeit itself marked on the head of the sentence, namely Comp(lementizer). (5) a dicenu [chi saria a più bell’ ora di u ghjornu] (Cor.) they.say that it.would.be the more fine hour of the day ‘they say it is the finest part of the day’

3 Nichols (1986: 72) terms languages that exhibit some head-marked and some dependent-marked patterns ‘split-marking languages’, a characteristic which once again apparently places Romance with Bantu (cf. De Cat and Demuth 2008).

288

From Latin to Romance b vi tuccarà [à aspittà to.you= it.will.touch to wait.INF ‘you’ll have to wait a long time’

longu long

tempu](Cor.) time

Another important Romance example of dependent-marking, this time at the level of the VP, concerns the differential marking of specific, animate objects with reflexes of the preposition AD ‘to’ in southern Italian dialects (6a) and Ibero-Romance (6b) or PER ‘through > on’ in Daco-Romance (6c), the so-called prepositional accusative (cf. }7.4.2).4 (6)

a vulem’ accid’ a ffratete (Abr.) we.want kill.INF ACC.MRK brother=you ‘we want to kill your brother’ b els va emmantonar them= AUX.PFV.PST.3SG cover.with.shawl.INF ‘he covered the two of them with a shawl’ c le-am întâlnit them.f=we.have met ‘we met the girls’

pe ACC.MRK

a ACC.MRK

tots all

dos (Ibz.) two

fete (Ro.) girls

Paradoxically, however, as illustrated in the Ibizan Catalan and Romanian examples, this dependent-marking strategy frequently calls for and licenses a corresponding head-marking construction, namely clitic-doubling.5 Indeed, these latter examples illustrate how dependent-marking and head-marking do not exhaust the possibilities available to languages for marking syntactic relations (Nichols 1986: 65), which may, as in the Ibizan and Romanian examples, display double-marking with reciprocal morphological coding on head and complement, a pattern also observable in (3a–b) and (4a–b) where, in addition to person, number and/or gender indexing of the subject on the verbal head, the subject dependents themselves are also marked for nominative case. Another frequent example of double-marking in Romance concerns possessive structures like those in (7a–b), where the identity of the possessor is marked both on the determiner possessor (son, sua), the head of the D(eterminer)P(hrase), as well as on an adnominal P(repositional)P(hrase) dependent on the head noun (à lui, dele). In the former case the determiner possessive

4 See, among others, Meyer-Lübke (1899: }50), Kalepky (1913), Reichenkron (1951), Meier (1948), Rohlfs (1969: }}632, 639; 1971), Diaconescu (1970), Martín Zorraquino (1976), Villar (1983), Nocentini (1985), Pensado (1985; 1995), Stimm (1986), Green (1988a: 106), Bossong (1991), Zamboni (1992), Jones (1993: 65–8; 1995), Trumper (1996: 354–5), Sornicola (1997; 2011: 35–42), Vincent (1997b: 209), Torrego (1998; 1999), Ledgeway (2000: 20–1; 2009a: 831–42), Fiorentino (2003). 5 Rohlfs (1968: }468), Jaeggli (1981), Suñer (1988), Demonte (1995), Schmitt (1998), Torrego (1998), Ledgeway (2000: 37–8), Alboiu and Motapanyane (2000b: 11–12), Zagona (2002: 68–9).

Head-marking and dependent-marking

289

indexes the person feature of the possessor, as well as the gender of the possessed, while the adnominal PP indexes both the person, number, and gender features of the possessor. (7)

a elle a fait envoyer un sms sur son mobile à she has made send.INF an sms on POSS.3MSG mobile.M to ‘she had the message sent on his mobile’ b chegou em sua casa dele (Pt.) he.arrived in POSS.3FSG house of.him ‘hei arrived at hisj house’

lui (Fr.) him

The final possibility, particularly common in languages with little or no inflection, consists in the complete absence of any marking whatsoever. A striking case in point comes from examples of null object anaphora (namely, object pro-drop) in both Latin and, more limitedly, in Romance (cf. }3.4), where in contrast to the case of subject-referencing (cf. 3a–b), the verbal head shows no overt marking whatsoever for the object complement whose presence is licensed and recovered through its pragmatic salience alone. (8)

a nouistine hominemi? i you.know=Q man.ACC (him) ‘do you know the man? I do’

noui (Lat., Pl. Bacch. 837) I.know

em áreas que eram controladas pelos b Elesi ficarão they will.stay in areas that were controlled by.the guerrilheiros do Hezbollah. Os enviados especiais guerrillas of.the Hezbollah the envoys special.PL Marcos Uchôa e Sérgio Gilz acompanharam i (BPt.) Marcos Uchôa and Sérgio Gilz accompanied (them) ‘They will stay in areas that were controlled by the Hezbollah guerrillas. The special envoys Marcos Uchôa and Sérgio Gilz accompanied them’

6.3 Romance functional categories ) Romance head-marking As already observed above, the head-/dependent-marking distinction offers an immediate and complementary account for our previous discussion (cf. chapters 4–5) of the emergence of the functional head categories Det(erminer), Aux-Infl(ection), and Comp(lementizer) in Romance which, with the shift from dependent-marking to head-marking, provide the means to directly encode grammatical information relating to nominal, verbal, and clausal dependents. In what follows, we shall review some of the Romance evidence for the emergent headmarking patterns in relation to these three functional heads.

290

From Latin to Romance

6.3.1 Head-marking on D Beginning with the D(eterminer) head, the widely noted erosion of the Latin morphological case system (}2.2.3, }5.4.2) can now be construed as a reflex of the move away from dependent-marking. In part, and especially in the spoken varieties, the identification of argument functions has to varying degrees been taken over by the Dsystem (Vincent 1997c: 164), where accusative, and especially dative, nominals are referenced by doubling clitic pronouns affixed to the verb:6 (9) a no l’i havia conegut not you= I.had recognized ‘I had not recognized you’

[a ACC.MRK

b [copiilor]i nu lei children.the.DAT not them.DAT= ‘the children don’t lack anything’

vostè]i (Cat.) you

lipseşte lacks

nimic (Ro.) nothing

c lesi tengo que regalar [a los niños] la bicicleta nueva (Sp.) new them.DAT I.have that give.INF to the children the bike ‘I have to give the children a new bike’ d ccii ’aj dugnu him.DAT= it.F.ACC= I.give ‘I’ll give my brother the key’

[’a the

chiave]i key

[a to

fratimma]i (Cos.) brother=my

Moreover, as we have seen (}3.2.2.1), in a number of Romance varieties, especially Gallo-Romance and Brazilian Portuguese,7 the D(eterminer) position often emerges as the principal, and in many cases the sole, locus of grammatical information relating to its nominal dependent. For example, in (10a–e) the number and gender features of the NPs are exclusively spelt out in the accompanying D head: (10) a [DP

ce / ces this.M these ‘this/these bucket/s’

b [DP

sas the.FSG ‘the sister/s’

c [DP

/ l« the.FPL

quella / quelle those.F that.F ‘that/those lesson/s’

d [DP

[NP seau/x]] (Fr.; seau(x) [so]) bucket.M [NP sr"]] (Tor.) sister.F [NP lezione]] (Tsc.) lesson.F

sas [NP especialitat her.FPL speciality.F ‘her renowned specialities’

6 7

renomada]] (Auv.) renowned.F

Lipski (1994: 82–9), Gierling (1997), Alboiu and Motapanyane (2000b: 11–13), D’Introno (2001: 100–1). Thomas (1969: 49), Campos and Rodrigues (1992: 129), Azevedo (2005: 226).

Head-marking and dependent-marking

e [DP

as [NP primeira first.F the.FPL ‘the first rainfalls’

291

chuva]] (BPt.) rain.F

Another significant, although often overlooked, innovative head-marking strategy involves the replacement of semantically based case distinctions with the use of distinct prepositions. Latin prepositional phrases are clearly dependent-marking, in that the nominal complement bears the case determined by its governing prepositional head (11a–b). Given the loss of case in Romance (with the notable exception of Romanian discussed below), one might quite legitimately conclude that Romance prepositional phrases lack any marking whatsoever (11a'–b'). (11)

a propter on.account.of

clamorem cry.ACC

tuom (Lat., Pl. Truc. 291) your.ACC

a' par ta criaillerie (Fr.) by your shouting ‘by reason of your clamour’ b sine without

doloribus (Lat., Pl. Truc. 807) pains.ABL

b' sans douleurs (Fr.) without pains ‘without pain’ While this conclusion might appear valid for the Romance examples above, in other cases Romance appears to operate a case of head-marking through lexical choice of the prepositional head. For example, in Latin the distinction between position (inessive, adessive) and motion (allative) in conjunction with the same preposition (e.g. SUB ‘under, up to’, SUPER ‘above, on top of’, IN ‘in, (in)to’) was often marked by the ablative/ accusative alternation on the nominal dependent (12a–b), whereas in Romance the same distinction is marked by the choice of prepositional head (12a'–b'): (12) a [ . . . ]

ne lest

ille that.one.NOM

in in

uilla villa.ABL

a' [ . . . ]

che non rimanesse nella that not he.remain.SBVJ in.the ‘that he should not stay at his villa’

b deuertit he.turns.aside

in in

uillam villa.ACC

resideret (Lat., Pl. Mil. 51) remained.SBJV villa (It.) villa

Pompei (Lat., Pl. Mil. 54) Pompeius.GEN

b' si gira verso la villa di self= he.turns towards the villa of ‘he turns aside to the villa of Pompeius’

Pompeio (It.) Pompeius

292

From Latin to Romance

Finally, one further context in which there is strong evidence for head-marking in Romance comes from a consideration of Romanian case-marking. With the exception of feminine nouns and adjectives which overtly mark a binary nominative/ accusative vs genitive/dative case opposition in the singular (13a–b), albeit with almost complete syncretism between the genitive/dative singular form and all plural case forms (13b'), case is otherwise regularly marked in Romanian on the accompanying determiner head (14a–d): (13)

(14)

a casă house.F.NOM/ACC

frumoasă (Ro.) pretty.FSG.NOM/ACC

b case house.F.GEN/DAT

frumoase (Ro.) pretty.F.GEN/DAT

b' houses.F

pretty.FPL

a oraşul frumos / un oraş frumos (Ro.) a.M.NOM/ACC city.M pretty.MSG city.M=the.MSG.NOM/ACC pretty.MSG b oraşului frumos / unui oraş frumos (Ro.) city.M=the.MSG.GEN/DAT pretty.MSG a.M.GEN/DAT city.M pretty.MSG c oraşele frumoase / (nişte) oraşe frumoase (Ro.) cities.F=the.FPL.NOM/ACC pretty.FPL some.NOM/ACC cities.F pretty.FPL d oraşelor frumoase / unor oraşe frumoase (Ro.) cities.F=the.PL.GEN/DAT pretty.FPL some.PL.GEN/DAT cities.F pretty.FPL

6.3.2 Head-marking on Infl Turning now to the verbal I(nfl) head, this too emerges as a key marker of a number of head-dependent relations. For example, in Gallo- and Ræto-Romance varieties, as well as Tuscan, the object clitic referencing system observed above in (9a–d) carries over to the subject relation,8 where between the fifteenth and sixteenth centuries there emerges a series of subject clitics derived from weakened nominative subject pronouns (Poletto 1995).9 These erstwhile dependents migrated towards a verbal head (Nichols 1986: 86) and are now affixed to the auxiliary or lexical verb under I(nfl), where they spell-out (15a–b) and/or match (15c–d) the features of the subject in its canonical preverbal subject position in the Spec(ifier) of IP:10 8

Nichols (1986: 89), Saeed (1996), Vincent (1997: 164), Luraghi (2006: 249). For a detailed reconstruction and analysis of the order of emergence of subject clitics in Occitan based on dialect variation, see Oliviéri (2009). 10 Kayne (1975), Renzi and Vanelli (1983), Vanelli, Renzi, and Benincà (1985), Vanelli (1987), Roberge (1990), Polettto (1993; 1995; 2000), Cardinaletti (1997; 2004), Cardinaletti and Repetti (2010). For arguments that modern Romanian has developed a series of (doubling) postverbal subject clitics in the so-called double subject construction (i), see Cornilescu (2000). 9

Head-marking and dependent-marking (15) a [IP [Spec i] [I'

ai SCL.FSG

293

vegne]] (Gen.; a < ILLA ‘she’) comes

‘she comes’ b [IP [Spec i] [I'

ii SCL.3MPL

dizen]] (Bol.; i < ILLI ‘they’) say.3PL

‘they say’ c [IP [Spec les enfants]i [I' ilsi veulent [VP tout SCL.3PL want all the children ‘the children want to eat everything’ d [IP [Spec

Cec]i [I' ui Ciccio SCL.MSG ‘Ciccio is working’

bouffer]]] (coll. Fr.) eat.INF

travaja]] (Pie.; u < ILLU ‘he’) works

Another significant example of the rise of head-marking to reference the subject relation concerns the inflected non-finite forms of the verb found in several IberoRomance varieties and old Neapolitan, where the infinitive (Table 6.1) and, to a lesser extent in accordance with dialectal variation, the gerund, and present and past participles (Table 6.2) display (in some persons) overt marking for person and number on the verbal head (cf. }2.2.1).11 TABLE 6.1 Romance inflected infinitives (cant- ‘sing’) Infinitive

1SG

2SG

3SG

1PL

2PL

3PL

Glc.

cantar+

-

-es

-

-mos

-des

-en

OLeo.

cantar+

-

-es

-

-mos

-des

-en

ONap.

cantar(e)+

-

-

-

-mo

-vo

-no

(E)Pt.

cantar+

-

-es

-

-mos

-des

-em

Srd.

cantare+

-po

-s

-t

-mus

-dzis

-n

Thus, although traditionally considered non-finite parts of speech which fail to license overt subjects, in the following representative examples we see that the infinitive (16a–d), the gerund (17a–c), and the participles (18a–b) variously index through suffixal inflectional marking on the verbal head the person and number features of the (overt/covert) dependent subject: (i)

Ion vine el mai Ion comes=he more ‘Ion is arriving later’ 11

târziu (Ro.) late

Maurer (1968), Carballo Calero (1981), Loporcaro (1986), Raposo (1987), Jones (1992; 1993: 78–82), Longa (1994), Vincent (1996; 1998b), Ledgeway (1998: 41–6; 2000: 109–14; 2007c; 2009a: 585–90), Mensching (2000), Lobo (2001), Pires (2002), Sitaridou (2002).

294

From Latin to Romance

TABLE 6.2 Romance inflected gerunds and participles (cant- ‘sing’) Gerund Glc.

Pt. (Póvoa de Atalaia)

Present / Past Participles O.Nap.

cantando+

cantand(o)+

cantanno+

cantante+

cantato+

1sg

-

-

-

-

-

2sg

-

-s

-

-

-

3sg

-

-

-

-

-

1pl

-mos

-mos

-mo

—

—

2pl

-des

-eis

-vo (-ve)

—

—

3pl

-

-em

-no (-ne)

-no

-no

(16)

a é doado supoñeren as cousas (Glc.) it.is easy suppose.INF.3PL the things ‘it is easy for them to assume things’ viram as ruínas (Pt.) b depois de eles chegarem after of they arrive.INF.3PL they.saw the ruins ‘after they arrived, they saw the ruins’ c èy hora de nne levaremo da it.is time of us= raise.INF.1PL from ‘it is time for us to get up’

lo the

liecto (ONap., LDT 63.2–3) bed

d devo accabbare custu travallu prima de ghiraret su mere (Srd.) I.must finish.INF this work before of return.INF.3SG the boss ‘I must finish this job before the boss returns’ (17) a cantandodes así, gañaréde-lo premio (Glc.) sing.GER.2PL thus you.will.win=the prize ‘singing like that, you will win the prize’ b não saíndomos de casa, morrâmos à fome (Pt., Ervedosa do Douro) not leave.GER.1PL from house we.die at.the hunger ‘if we don’t leave the house, we’ll starve to death’ c Tirandono certi piscatori la reta dal mare, pull.GER.3PL certain fishermen the net from.the sea senterono essere pisante[ . . . ] credendonose haverno believe.GER.3PL=selves have.INF.3PL they.felt be.INF heavy incappati multi pesci (ONap., Brancati VFE 133.23–5) caught many fish ‘some fishermen drawing in the net, felt it to be heavy [ . . . ] believing themselves to have caught many fish’

Head-marking and dependent-marking

295

(18) a E datonosse insembla salute como convenne, and given.3PL=selves together greeting as it.was.necessary (ONap., LDT 120.4–5) ‘and after having greeted one another as was customary’ b in questa / cità di Napoli erano duo mariti in this city of Naples were two husbands e mugliere timentino Dio (ONap., CDP 17v.17–19) and wives fear.PCTP.3PL God ‘in this city of Naples there were two husbands and wives in fear of God’ I(nfl) is also the head, at least historically, that encodes through the HABERE/ESSE perfective auxiliary alternation a core distinction between the core participants of the sentence (Vincent 1982). More specifically, as we saw in detail above (}4.3.1.4), historically in the compound perfective forms of the verb AGENT/EXPERIENCER subjects align with the auxiliary HABERE (19a), whereas UNDERGOER subjects align with ESSE (19b). (19)

a i tozi i ga lavorà the boys SCL.3MPL have.3 worked ‘the boys worked until nine o’clock’ b i fiori i the flowers SCL.3MPL ‘the flowers have died’

ze be.3

fin until

ae at.the

nove (Pad.) nine

morti (Pad.) died.MPL

In a related development, a variant of I(nfl) dedicated to encoding and licensing the object features of the verb, generally referred to as light v in the generative literature (cf. discussion in }5.4.3.1), also spells out via participle agreement (Loporcaro 1998) the same distinction between AGENT/EXPERIENCER participants (no/default agreement; 20a),12 and UNDERGOER participants (with agreement; 20b). (20)

a elas auran perdut (Occ.) they.F will.have lost.MSG ‘they will have lost’ b sun arribats (Occ.) they.are arrived.MPL ‘they have arrived’

12 In Aromanian (Kramer 1989: 430), Megleno-Romanian (Dahmen 1989: 441), and western Romanian varieties (Urit¸escu 2007), it is the feminine singular form of the participle that serves as the default form (i).

(i)

am mâncată (ARo.) I.have eaten.FSG ‘I(M/F) have eaten’

Note furthermore that in cases like (20a–b) we clearly have an example of double-marking, where two heads in the construction, the inflectional head Infl (auxiliary) and the verbal head v (participle), reference and index grammatical information about the subject relation.

296

From Latin to Romance

Yet a further development is found in a number of varieties spoken in centralsouthern Italy, Piedmont (province of Alessandria), and northern Catalonia (cf. }7.3.1.1.2),13 where the two perfective auxiliaries are now distributed according to grammatical person (21–22), with ESSE tending to occur in the 1/2 persons and HABERE in the 3 persons.14 Synchronically, the I(nfl) head in these varieties can then be said to license through auxiliary selection a system of person-marking, namely, discourse participants vs non-discourse participants. (21) a so /si scritto (EAbr) I.am /you.are written ‘I/you have written’ b a scritto (EAbr) have.3 written ‘he has/they have written’ (22) a son vist (Cpc.) I.am seen ‘I have seen’ b ha/han vist (Cpc.) have.3SG/PL seen ‘he has/they have seen’ Other candidates for the reinforcement of the head-marking pattern in Romance, which limitations of space unfortunately do not allow us to explore in detail here, include the loss and replacement (beginning in Latin; cf. Baldi and Nuti 2010: 251–322) of ESSE with HABERE (and with TENERE in Ibero-Romance and upper southern Italy) in the possessive construction. In the former construction the possessor was marked by a dative dependent (23a), whereas in the latter the

Rohlfs (1969: }730), Tuttle (1986), Kayne (1993), Veny (1982: 51), Ledgeway (2000: 192–5, 204–5), Bentley and Eyþórsson (2001), Cennamo (2001b), Manzini and Savoia (2005 II: ch. 5), Loporcaro (2007), D’Alessandro (2010: 31–2), D’Alessandro and Ledgeway (2010b), D’Alessandro and Roberts (2010), Legendre (2010). 14 Auxiliary ‘be’ is also reported to occur with transitives, though in all six grammatical persons, in a number of Romanian dialects (i.a), possibly due to Bulgarian/Macedonian influence (M. Maiden p.c.), although the participle agreement here with the subject (i.b) suggests that the participle be treated as a verbal adjective (Dahmen 1989: 441; Atanasov 2002: 245; Avram and Hill 2007), witness the resultative English rendering below. 13

sam mănkát (MRo.) I.am eaten.MSG b sam mănkátă (MRo.) I.am eaten.FSG ‘I am done eating’

(i) a

Head-marking and dependent-marking

297

pronominal possessor is primarily marked through the verb inflection of the copular head (23b–d). (23)

forte fuisset a si ei if him.DAT by.chance it.had.been ‘if he by chance had had a fever’

febris (Lat., Pl. Mil. 719) fever

b si febrim non habebit (Lat., Cato Agr. 157.9) if fever.ACC not he.will.have ‘if he doesn’t have a fever’ c dacă if

are he.has

febră (Ro.) fever

d se tiver if has.FUT.SBJV.3SG ‘If he has a fever’

febre (Pt.) fever

Another striking pan-Romance example concerns the emergence of the Romance causative construction (}2.2.2). Although clearly analytic inasmuch as it involves two separate lexemes (viz. reflex of FACERE/LAXARE ‘make/let’ + infinitive), it mimics the syntax of a morphological causative (Comrie 1985) with incorporation of the lexical infinitive into FACERE or LAXARE to yield a single verbal complex in which causation is marked on the FACERE/LAXARE head (24a–b).15 This head-marking construction replaces earlier dependent-marking constructions like EFFICERE ‘to cause’ + dependent UT clause (24c; cf. Zamboni 2000: 120–1). (24)

a l’ ‘the

eau water

bout (Fr.) boils’

bouillir] b je [IP fais I make boil.INF ‘I boil the water’

l’ the

eau (Fr.) water

audiatis (Lat., Cic. Phil. 2.10) c efficiam ut [ . . . ] attente I.will.cause that carefully you.hear.SBJV ‘I’ll have you listen to me attentively’ Our final example involves the marking of transitivity in Romanian 2SG imperatives. Unlike other Romance varieties, Romanian displays in a large number of non-first conjugation verbs a transitive/intransitive distinction in the inflectional alternation -e/-i on the verbal head,16 referencing the presence (25a) and absence (25b) of a dependent in the clause, respectively:

15 16

Kayne (1975), Zubizarreta (1985), Burzio (1986), Alsina (1992; 1996: ch. 6), Guasti (1993; 1996). Lombard (1974: 250–1), Alboiu and Motapanyane (2000b: 31), Pîrvulescu and Roberge (2000).

298 (25)

From Latin to Romance a arde toate documentele! (Ro.) burn.IMP2SG.TR all.FPL documents.FPL ‘burn all the documents!’ b arzi! (Ro.) burn.IMP2SG.INTR ‘burn!’

6.3.3 Head-marking on C Turning now to the C(omplementizer) head, here too there is clear evidence for the rise of head-marking at the expense of dependent-marking. Already we have seen in relation to the discussion in (5a–b) how subordination in Romance is preferably marked on the head of the embedded clause through an overt complementizer, whereas in Latin subordination was not marked so robustly on the C(omplementizer) head, but more readily on the complement/dependent of the latter through infinitival verb morphology and accusative case-marking of the embedded subject. Another significant example involving subordination involves Romance dual complementizer systems (}4.4.2.2). More specifically, in embedded contexts in southern Italian dialects and Romanian the indicative/subjunctive distinction on the verb is only present, if it all, in the 3rd person (Salentino, Romanian) or has been entirely lost (remaining southern Italian dialects).17 However, the C head which selects the IP-VP dependent containing the verb now marks this same distinction through a complementizer alternation (Vincent 1993: 151–2): Cal. ca or chi (IND) vs mi, mu, or ma (SBJV); Ro. că (IND) vs să (SBJV); Sal. ca (IND) vs cu (SBJV).18 By way of example, consider the following minimal pair from the Calabrian dialect of Melicucco, where the declarative vs jussive reading of the complement of diri ‘to say’ is marked by the differing lexicalization of the C head and not on the verbal dependent: (26)

a ti dicu ca staju jendu (Mel.) you= I.tell that I.am going ‘I tell you that I’m leaving’ b ti rissi ’u vai rá (Mel.) you= I.told that you.go there ‘I told you that you should leave’

A number of Romance varieties also show a tendency to mark the interrogative force of the IP-VP dependent, not on the verb itself, as happens, for example, in Occitan

17 Rohlfs (1968: 301–2; 1969: 192 n. 3), Lombard (1974: 31), Mancarella (1998: 99–117), Loporcaro (1997: 345–6), Damonte (2006a), Bertocci and Damonte (2007), Ledgeway (2009a: 522–32). 18 However, for evidence that the dialects of Salento preserve marking of the IP-VP dependent through reanalysis of original consonantal lengthening on the embedded verb, see Ledgeway (forthc. c).

Head-marking and dependent-marking

299

and popular French through suffixation of the particle -ti to the verb (27a–b), but through lexicalization of the C(omplementizer) head (28a–c; cf. also }4.4.2.3), witness the use of est-ce que ["s(ə)kə] (< erstwhile cleft: ‘is-this that . . . ’) in standard French, a (< AUT ‘or’) in Sardinian,19 o(cche) ‘or (what)’ in Tuscan (Garzonio 2004), che/chi/ce ‘that/what’ in central-southern Italian dialects (Rohlfs 1969: 157–9; Manzini and Savoia 2005 I: }3.8.2; Cruschina 2008: ch. 5; Ledgeway 2009a: 811–13), and que/qué ‘that/what’ in Ibero-Romance and Occitan (Rigau and Prieto 2007): va-ti? (pop. Fr.) (27) a ça that goes=INT ‘how are things?’ caldrà-ti b me me= it.will.be.necessary=INT ‘will I have to take up arms’ (28)

prene take.INF

las the

armas? (Occ.) arms

a est-ce que

tu rentres? (Fr.) you go.home ‘are you going home?’

INT

b a

s’ at comporada sa domo self= he.has bought the house ‘has he bought himself a house in Nuoro?’

INT

in in

Nùgoro (Srd.) Nuoro

c Que

plou? (Cat.) it.rains ‘Is it raining?’ Q

6.3.4 Extreme head-marking: the case of Ripano Although we have identified above some significant patterns of extension of headmarking across Romance, albeit alongside many inherited dependent-marking strategies, the Ripano dialect of central Italy spoken in the south-eastern Marchigiano village of Ripatransone represents an extreme case within the Romance family, having extended the domain of head-marking, often in conjunction with dependent-marking in a double-marking strategy, to almost all categories of all areas of the grammar. The result, as we shall see, is a language which, from a typological perspective, stands out among the Romance languages for a distinctive cross-referencing system that relies on extensive head-marking.20

19

Jones (1993: 244–7), Manzini and Savoia (2005 I: 610–1), Mensching and Remberger (2010a, b). See further Egidi (1965), Parrino (1967), Lüdtke (1974; 1976), Mancini (1993), Harder (1998), Ledgeway (2004c; 2009c), Jones (2008), D’Alessandro (2011). Also of interest is Loporcaro and Vigolo ([2002–3] 2005). 20

From Latin to Romance

300

6.3.4.1 Verbal domain

6.3.4.1.1

SUBJECT-VERB AGREEMENT

As is typical within Romance, the Ripano system of nominal inflectional agreement exemplified in Table 6.3 marks syncretically both number and gender, both on the head (determiner) and on any associated dependents (noun, adjectives): TABLE 6.3 Ripano nominal agreement system Determiner

Nouns

Adjectives

quist- ‘this’

fijj- ‘offspring’

v(i)ecchi- ‘old’

masculine

feminine

masculine

feminine

masculine

feminine

Singular

quist-u

quest-e

fijj-u

fijj-e

viecchi-u

vecchi-e

Plural

quist-i

quest-a

fijj-i

fijj-a

viecchi-i

vecchi-a

Now, in most Romance varieties the finite verb follows a distinct verbal agreement paradigm, as in Italian where, for example, the finite verbal head indexes at the very least number (singular vs plural; cf. present subjunctive), but in most cases also person (1 vs 2 (or 1/2) vs 3; cf. present indicative, imperfective subjunctive), witness Table 6.4: TABLE 6.4 Italian finite verbal agreement system (aprire ‘to open’) Present Indicative Singular

Plural

1

apr-o

apr-iamo

2

apr-i

apr-ite

3

apr-e

apr-ono

Present Subjunctive

Imperfect Subjunctive

Singular

Plural apr-iamo

apri-ssi

apri-ssimo

apr-a

apr-iate

apri-see

apri-ste

apr-ano

Singular

Plural

apri-ssero

In Ripano, by contrast, the finite verb grafts the nominal agreement pattern for number and gender seen in Table 6.3 onto the traditional Romance verbal agreement system which, we have just seen, privileges marking for person. The result, as revealed in Table 6.5, is a system in which the finite verbal head privileges number and gender marking over person which, in the singular at least, is entirely neutralized.21 Indeed, this is quite a surprising result since Ripano is a null-subject language 21 In line with other central (and upper southern) dialects (cf. }4.2.2.2.2), Ripano nominal agreement also presents a distinct neuter agreement inflection (in -ə) for a subclass of non-count nouns which is also grafted onto the verbal agreement system, witness the masculine, feminine, and neuter contrasts exemplified in (i.a–c).

(i)

a

lu frəki cresciu (Rip.) the.MSG boy.M grows.MSG ‘the boy is growing’

Head-marking and dependent-marking

301

and traditional accounts of the (Romance) null subject parameter standardly link the licensing of null subjects precisely to the availability of rich person agreement on the verb. TABLE 6.5 Ripano finite verbal agreement system (magnà ‘to eat’) Present Indicative Masculine

Feminine

Singular

magn-u

magn-e

1PL

magn-em-i

magn-em-a

2PL

magn-et-i

magn-et-a

3PL

magn-i

magn-a

Far from being limited to finite verbs, such a rich and complex system of subject head-marking has come to permeate even non-finite verb forms such as past participles (29a–b), gerunds (30a–c), and (rhizotonic) infinitives (31a–d). (29) a so saputu I.am known.MSG ‘I found out’

/ sapute (Rip.) known.FSG

b lu frəki / le frəkine the girl.F the boy.M ‘the boy / the girl has slept’ (30)

è is

dd ərmitu / slept.MSG

a chissà chə vva fəcennu (Rip.) who.knows what he.goes doing.MSG ‘who knows what he’s up to!’ b mamme stieve cucənenne (Rip.) was cooking.FSG mum.F ‘mum was cooking’ c li frəkì stievi currenni (Rip.) the boys.M were running.MPL ‘the boys were running’

b

le frəkine cresce (Rip.) grows.FSG the.FSG girl.F ‘the girl is growing’

c

lə grà crescə (Rip.) the.N wheat.N grows.N ‘the wheat is growing’

dd ərmite (Rip.) slept.FSG

302 (31)

From Latin to Romance a nəm bozze not I(F)can ‘I can’t drink’

beve (Rip.) drink.INF.FSG

b sai skrivu? (Rip.) you(M.SG).know write.INF.MSG ‘do you know how to write?’ c tu sa leggu you(MSG) know read.INF.MSG ‘can you read Italian?’

l’ the

italiano? (Rip.) Italian(-language)

d səpeta scriva? (Rip.) you(FPL).know write.inf.FPL ‘do you know how to write?’ e lu bbova nən vò the bull.M not wants ‘the bull won’t drink’

6.3.4.1.2

bbevu (Rip.) drink.INF.MSG

SUBJECT-/OBJECT-VERB AGREEMENT

The observed patterns of subject head-marking on the verb are further augmented by a distinct series of objective agreement inflections that surface on the verbal head. In particular, we need to recognize two relevant morphosyntactic agreement patterns, a matching and a non-matching paradigm. The matching agreement paradigm is illustrated in Table 6.6, and invariably surfaces in reflexive constructions like (32a–e) where the number and gender features of the subject and the object always match exhaustively,22 as well as in those transitive clauses where subject and object dependents happen to share the same number and gender features (33a–d). TABLE 6.6 Ripano matching agreement SubjectÆ + ObjectÆ Masculine

(32)

22

Feminine

Singular

-u

-e

Plural

-i

-a

a mə myself.M=

rəlavu (Rip.) I(MSG).wash.MSG

Note that the respective person features of the subject and object dependents are not relevant to this agreement relation.

Head-marking and dependent-marking b mə myself.F=

rəlave (Rip.) I(FSG)wash.FSG

c və yourselves.M=

rəlavati (Rip.) you(MPL).wash.MPL

d və yourselves.F=

rəlavata (Rip.) you(FPL).wash.FPL

e issu / esse sə he / she self= ‘s/he gets dressed’ (33)

303

/

vestu dresses.MSG

veste dresses.FSG

a facettu nu ggira (Rip.) I(M).made.MSG a.MSG trip.MSG ‘I(M) undertook a trip’ b mette l’ acque (Rip.) she.puts.FSG the water.FSG ‘she pours the water’ c mazzi they(M).kill.MPL d oggia today

li the.MPL

magnema we(F).eat.FPL

cheppù (Rip.) capons.M

lə the.FPL

patata (Rip.) potatoes.FPL

We thus see in these examples that the surface realization of the final vocalic inflection (-u (MSG), -e (FSG), -i (MPL), and -a (FPL)) signalled on the verbal head represents the resolution of an exhaustive matching agreement relation between the subject and object dependents. Clearly, in many other transitive clauses the gender and number features of subject and object dependents will fail to match and in that case the nonmatching agreement paradigm in Table 6.7 comes into force, as illustrated in (34a–g):23 TABLE 6.7 Ripano non-matching agreement SubjectÆ + Object Masculine Singular

-ə

-ə

Plural

-i

-ə

(34) a magnə le eat.AGR? the.FSG ‘he eats polenta’ 23

Feminine

pəlende (Rip.) polenta.F

In the following examples, we indicate non-matching agreement with the gloss ‘AGR?’.

304

From Latin to Romance b li frəkì magni the.MPL boys.M eat.MPL ‘the boys eat the wallnuts’ c so magnatə I(M).am eaten.AGR? ‘I’ve eaten the pears’

lə the.FPL

lə the.FPL

nucia (Rip.) wallnuts.F

pera (Rip.) pears.F

d semə magnatə we(F).are eaten.AGR? ‘we ate the sandwich’

lu the.MSG

panì (Rip.) sandwich.M

e babbu e ddittə dad.M is said.AGR? ‘dad told the truth’

le the.FSG

vərità (Rip.) truth.F

f mamme e kkottə li mum.F is cooked.AGR? the.MPL ‘mum cooked the maccaroni’ g dovremmi magnà we(M).should.MPL eat.INF ‘should we eat the cake?’

le the.FSG

makkarù (Rip.) maccaroni.M torte? (Rip.) cake.F

With the exception of 3MPL subjects (34b,g), on the surface structures like these would seem to suggest that, whenever there is a mismatch in features between subject and object dependents, agreement fails and the inflection on the verbal head surfaces as a default vowel (-ə). This is the view espoused, for instance, in Mancini (1993: 120), who claims that in such examples ‘agreement with the subject fails to obtain and is replaced by a neutral, undifferentiated inflection (-ə/-a)’.24 However, the final schwa cannot be interpreted tout court as a case of failure of agreement on the verbal head, since the verbal head is clearly sensitive to the number and gender features of the subject and object dependents, with final -ə representing the resolution of non-matching agreement. Thus, although final schwa does not unambiguously signal the number and gender features of the verb’s subject and object dependents, it does nonetheless represent a case of, admittedly impoverished, agreement, variously spelling out the gender, and to a lesser extent, the number features of the subject, albeit with some degree of ambiguity. By way of example, consider (35a) where, given the MSG features of the object lu panì, the presence of the non-matching agreement signalled by the final schwa on the verbal head necessarily implies that the subject must be feminine, either singular (viz. I, you, she) or 3PL (viz. they), but cannot possibly be MSG or 3MPL which would have given rise to (matching) head-marking desinences -u and -i, respectively. Similarly, given the presence of a FSG object le 24 The distinction here between -ə and -a is one of position, with the latter occurring in sentence-final and pre-pausal position.

Head-marking and dependent-marking

305

patate in (35b), final schwa on the verbal head necessarily entails that the subject may be MSG (viz. I, you, he) or 3FPL (viz. they), but cannot possibly be 3MPL or FSG which would have licensed (matching) head-marking inflections, namely -i and -e, respectively. As can be readily verified, similar exclusions and inferences about the identity of the subject apply to the remaining examples in (35c–d). The exception is the 3MPL (35e) which always licenses distinctive -i agreement, whether matching obtains or not.25 (35) a magnə lu panì (Rip.; subject 6¼ (3)MSG, MPL) eat.AGR? the.MSG sandwich.M ‘I(F)/you(FSG)/she/they(F) eat(s) the sandwich’ le patate (Rip.; subject 6¼ MPL, (3)FSG) b magnə eat.AGR? the.FSG potato.F ‘I(M)/you(MSG)/he/they(F) eat(s) the potato’ li panì (Rip.; subject 6¼ MPL) c magnə the.MPL sandwich.M eat.AGR? ‘I(M/F)/you(MSG/FSG)/(s)he/they(F) eat(s) the sandwich’ d magnə lə patata (Rip.; subject 6¼ MPL, FPL) eat.AGR? the.FPL potatoes.F ‘I(M/F)/you(MSG/FSG)/(s)he eat(s) the potatoes’ e magni lu panì / le patate / li the.FSG potato.F the.MPL eat.PRS.3MPL the.MSG sandwich.M panì / lə patata (Rip.; subject = MPL) sandwich.M the.FPL potatoes.F ‘they(M) eat the sandwich / potato / sandwiches / potatoes’

6.3.4.1.3

OBJECT-VERB AGREEMENT

An important aspect of Ripano head-marking within the verbal domain concerns the agreement relations determined by clitic pronouns. Above we saw that when the number and gender features of the subject and the full object complement do not coincide, non-matching agreement is the norm (cf. 34–35). If, however, the object complement surfaces as a clitic pronoun, then it overrides the subject as potential 25 Jones (2008) finds that there is a tendency among younger speakers to move to partly Italian patterns of subject agreement, albeit with gender marking, under non-matching agreement:

(i)

ija vuoju le macchəne (Rip.) I(M) want.MSG the.FSG car.F ‘I want the car’ b parə chə e magnatu li it.seems.N that he.is eaten.MSG the.MPL ‘he apparently ate the snails’ a

chiuova (Rip.) snails.M

306

From Latin to Romance

agreement controller and its gender and number features alone are spelt out on the verbal head, as schematicized in Table 6.8 and exemplified in (36a–h): TABLE 6.8 Ripano object clitic agreement SubjectÆ/ + ObjectÆ

(36)

Masculine

Feminine

Singular

-ə

-ə

Plural

-i

-ə

a esse nnə mmə she not me.M= ‘she hugs me(m.)’

bbracciu embrace.AGR.MSG

b tə cunosciu you.MSG= know.AGR.MSG ‘I(F) know you(MSG)’ c ia le vede I(M) her= see.AGR.FSG ‘I(M) see her’

(**bbracce) (Rip.) embrace.AGR.FSG

(**cunosce) (Rip.) know.AGR.FSG (**vedu) (Rip.) see.AGR.MSG

d lə ggiovəne, tu le vide (**vidu) the.FSG youth.F you(M) her= see.AGR.FSG see.AGR.MSG ‘do you(MSG) see the girl often?’ e li pagna li so the.MPL clothes.M them.M= I(F).am ‘the washing, I(F) have hung it out’ f esse l’ è vvistu she him= is seen.MSG ‘she has seen him’

spasi hung.out.MPL

spessa? (Rip.) often (**spasa) (Rip.) hung.out.FPL

(**vviste) (Rip.) seen.FSG

g stə kuosa lə so saputa (**saputu) da lu curata (Rip.) these things.F them.F I(M).am known.FPL known.MSG from the priest ‘these things, I(M) learnt them from the priest’ h le frəkine, le si viste (**vistu) prima? (Rip.) the.FSG girl.F her= you(MSG).are seen.FSG seen.MSG before ‘The girl, have you(MSG) seen her before?’ Whether the verb is finite (38a–d) or non-finite (38e–h), the agreement features of the verbal head are invariably controlled by the object clitic and never by the subject. From a typological perspective, this is a significant finding since it goes against

Head-marking and dependent-marking

307

general crosslinguistic predictions about preferred agreement controllers, which lead us to expect the finite verb to display agreement with at least the highest potential controller in the hierarchy of argument relations (namely, subject > direct object > indirect object > oblique; Croft 1990: 101–11). Given, however, the observed nominal nature of the agreement pattern superimposed on the Ripano finite verbal paradigm, the precedence of object clitic agreement over that of subject agreement is quite understandable: in the same way that the Romance past participle, also characterized by a nominal gender and number agreement paradigm, variously displays agreement with a nominal associated at some level of representation with the verb’s object position (Loporcaro 1998), so does the Ripano finite verb which exhibits the same nominal agreement system. Consequently, the head-marking pattern of the Ripano finite verb observed in (36a–h) is comparable to that of the active past participle in a variety such as Italian, where, on par with (34)–(35), the participle fails to agree with a full nominal complement (37a), but does agree with the corresponding object clitic (37b). (37)

a ho steso / **stesi hung.out.MPL I.have hung.out.MSG ‘I’ve hung out the washing’ b li ho them.MPL I.have ‘I’ve hung it out’

stesi hung.out.MPL

i the.MPL

panni (It.) clothes.m

/ **steso (It.) hung.out.MSG

Indeed, this interpretation of the facts finds further support in the observation that, alongside object-controlled agreement like (36a–h), finite verbs in simplex tenses (38a–c), but significantly not non-finite participles in compound tenses (38d), may also display the expected non-matching agreement paradigm when the complement is represented by a clitic pronoun. From this perspective, head-marking of agreement features on the Ripano finite verb represents a hybrid system, which marries together nominal and verbal agreement systems as mirrored in the object-driven and the nonmatching types in (36a–d) and (38a–c), respectively. (38)

a esse nnə mmə she not me.M= ‘she hugs me(m.)’

bbracciu embrace.AGR.MSG

b ia le vede I(M) her= see.AGR.FSG ‘I(F) know you(MSG)’

(bbraccə) (Rip.) embrace.AGR?

(vedə) (Rip.) see.AGR?

c issu li cucini he them.M= cook.AGR.MPL ‘he’s cooking them at once’

(cucinə) cook.AGR?

subbəta (Rip.) at.once

From Latin to Romance

308

d li pagna li so the.MPL clothes.M them.M= I(F).am ‘the washing, I(F) have hung it out’

spasi hung.out.MPL

(**spasə) (Rip.) hung.out.AGR?

Head-marking of complements represented by clitics is not just limited to those bearing the direct object relation, but is also extended to the indirect object relation (Mancini 1993: 124–5). For instance, in (39a) both the subject esse and the full nominal direct object lə verità are feminine singular, which might lead us to expect the verb to be head-marked for FSG matching agreement in -e (viz. diche). However, the clause also contains an indirect object represented by 3MSG clitic iə and this, in conjunction with the FSG subject and object, licenses a non-matching agreement relation, yielding the verb form dichə. A parallel explanation applies to (39b), where the expected headmarking of the MSG matching agreement relation between subject and direct object on the finite verb (viz. diengu) is overridden by the presence of an indirect object clitic tə indexing a FSG referent. As a consequence, the verbal head bears the resolution of a non-matching agreement relation (viz. diengə).26 (39)

a esse iə dichə sembra she to=him say.AGR? always ‘she always tells him the truth’

lə the.FSG

b sə tə diengə if you(F)= I(M).give.AGR? ‘If I thump you’

chezzotta (Rip.) thump.M

’n a.M

verità (Rip.) truth.F

6.3.4.2 Contagious head-marking Far from being limited to the verb, the rich and complex head-marking system of Ripano agreement outlined above for the verbal domain has permeated other areas of the grammar, coming to contaminate many categories of the language which in Romance do not otherwise display inflection. As Mancini (1993: 129) puts it, ‘it is as if in Ripano all invariable parts of speech were forced to decline’. Thus alongside gerunds and infinitives already seen in (30)–(31), another notable category which is regularly head-marked to index its dependents is the preposition, thereby aligning Ripano with languages like Welsh in this respect. 26 Although we have seen that direct object clitics typically control participle agreement even when we might expect to find non-matching agreement (i.a), the antipassive construction in (i.b) demonstrates that, even if the direct object surfaces as a clitic, it will fail to control agreement if the construction contains a reflexive dative clitic coreferent with the subject.

(i)

le patate, lə son the.FSG potato.F it.FSG= I(F).am ‘I(F) have eaten all the potato’ b le patate, mə lə the.FSG potato.F me(M)= it.FSG= ‘I(M) have eaten all the potato’ a

magnate eaten.FSG

tutta (Rip.) all.FSG

son I(M).am

magnatu eaten.MSG

tutta (Rip.) all.FSG

Head-marking and dependent-marking

309

For example, in the contrasting pairs in (40a–c), we see that the prepositional head is marked by the same set of matching final inflections to index the gender and number of its complement: (40)

a dopu lu ddi after.MSG the.MSG day.M ‘following the day / night’

/ dope after.FSG

b sottu lu tavulì under.MSG the.MSG coffee.table.MSG ‘under the coffee table / the chair’ c vəcinu lu mara close.MSG the.MSG sea.M ‘close to the sea / house’

le the.FSG

notte (Rip.) night.F

/ sotte under.FSG

/ vəcine close.FSG

le the.FSG

le the.FSG

sedie (Rip.) chair.F

case (Rip.) house.F

In a similar fashion, we find that the C-head hosting complementizers (41a) and whrelatives and interrogatives (41b) is also regularly inflected to mark the gender and number features of its embedded verbal subject. Within the nominal domain, we see, for example, that all cardinal quantifiers are head-marked for the gender and number features of their associated nominal complement (42a–b), and predicative complements of nominal origin regularly bear head-marking (43a–c) to index the number and gender features of the subject dependent (presumably the subject of the underlying small clause headed by the predicative nominal). (41)

a quannu / quanne passu / passe when.MSG when.FSG pass.MSG pass.FSG ‘whenever I/you/(s)he pass(es) by there’ b commu te siendu? how you.SG= feel.2FSG ‘How do you feel?’

(42)

/ /

comme how.FSG

lloka (Rip.) there

te you.SG=

siende? (Rip.) feel.2FSG

a ci stie centi frəkì (Rip.) there= are hundred.MPL boy.MPL ‘there are a hundred children’ b ija vedu sola vinda I see only.FPL twenty.FPL ‘I can only see twenty chairs’

sedia (Rip.) chairs.F

(43) a c’ajju famu / setu / c’ajje fame / sete (Rip.) I.have.MSG hunger.MSG thirst.MSG I.have.FSG hunger.FSG thirst.FSG ‘I am hungry/thirsty’ b c’aveti you.have.MPL

peuri fear.MPL

/ c’aveta you.have.FPL

peura (Rip.) fear.FPL

310

From Latin to Romance c li viecchi a bbesogni the.MPL old.MPL have.3 need.MPL ‘old people need to be taken care of’

d’ of

essa be.INF

ssistiti (Rip.) helped.MPL

6.4 Conclusion Clearly, further research is necessary to determine the extent of the typological move from dependent-marking to head-marking in the transition from Latin to Romance, but the discussion above highlights some of the advantages that can be gained from adopting the head-/dependent-marking distinction in understanding a number of the fundamental changes in the morphosyntax of Latin and Romance. In particular, we have noted how this typological distinction can be profitably married together with the results of the previous chapters. In particular, the claim regarding the emergence of functional structure in the passage from Latin to Romance developed in chapter 4 provides a principled explanation for the gradual rise of head-marking in Romance, since it is precisely the availability of these functional head positions in the grammar which enables the Romance languages to overtly spell out and index the formal properties of their associated dependents. By the same token, the gradual demise of a Latin XP/Specifier-syntax, manifested in the movement operations of phrasal dependents to argument (roll-up) and left-peripheral positions (edge-movement), in favour of a Romance X/Head-syntax, manifested in the gradual rise of head-movement operations and the direct lexicalization of different functional head positions, developed in chapter 5 provides us with the necessary analytic tools to interpret the gradual shift away from dependent-marking (Specifier-syntax) to head-marking (Head-syntax). Finally, it is worth noting that one of the consequences of either an extensive system of dependent- or head-marking is the emergence of a relatively free word order, albeit pragmatically constrained (Vincent 1997c: 164–5). Indeed, we have already seen this to be the case in Latin, where rich case and agreement inflections on nominal dependents allows them, in principle, to occur in all possible positions within the clause. In a similar fashion, a rich system of head-marking in Romance should a priori lead us to expect a similar degree of freedom in word order. It is not then by chance that some linguists, notably Tesnière (1959: 175) and Harris (1978; 1988: 236), have argued that the rise in clitic doubling structures in the less conservative spoken varieties allow the nominals they reference to occur in all possible positions (Lehmann 1985: 97–8; Nichols 1986: 107), thereby placing them on a par with polysynthetic languages (Jelinek 1984; Baker 1995). This is illustrated in the colloquial French examples (44a–f), where commas are purely conventional (and hence not necessarily indicative of intonational breaks) and the presence of the nominative (je 1SG), accusative (le 3MSG), and dative (lui 3SG) clitics on the verbal head (ai donné ‘I.have given’) unambiguously identify all the verb’s dependents (moi

Head-marking and dependent-marking

311

‘I’, le livre ‘the.MSG book.MSG’, à Pierre ‘to Pierre’). In view of evidence like this, Harris has even argued, though not without some controversy, that such freedom has led to VSO (cf. 44a, b) emerging as the unmarked order in colloquial French. (44)

a je le lui ai donné (V), moi (S), le livre (DO), à Pierre (IO) b je le lui ai donné, moi (S), à Pierre (IO), le livre (DO) c je le lui ai donné (V), le livre (DO), moi (S), à Pierre (IO) d je le lui ai donné (V), le livre (DO), à Pierre (IO), moi (S) e je le lui ai donné (V), à Pierre (IO), moi (S), le livre (DO) f je le lui ai donné (V), à Pierre (IO), le livre (DO), moi (S) ‘I gave Pierre the book’

If such analyses are correct for some of the more colloquial varieties of Romance,27 then we should expect them to show the typical clustering of syntactic behaviours posited for typically head-marking languages (Nichols 1986: 114; Saeed 1996: 41), including (i) coding of the predicate argument structure through clitics on the verb; (ii) zero anaphora; (iii) free-floating NPs underspecified for case; (iv) a ‘flat’ structure devoid of a VP category; (v) free nominal order; and (vi) the importance of discourse notions (focus, topic) in determining word order. Indeed, all these properties appear to be covered by those varieties such as colloquial French in (44) which have grammaticalized clitic doubling structures (cf. also examples (9) and (15) above): in these varieties all grammatical relations are saturated and realized by the pronominal clitic heads on the verb, leaving their coreferent full nominals to be interpreted as adjuncts outside of the VP and hence not subject to grammatical ordering, but serialized in accordance with discourse factors.28 The result is a syntax which comes very close to the so-called non-configurational syntax of Latin examined in chapter 3.

27 As I. Roberts (p.c.) points out, this analysis still faces a number of difficulties including the fact that: (a) some enclisis may exist in imperatives even in basolects; (b) it still remains to be shown that arguments cannot appear in argument positions; and (c) it is not clear that there exist French varieties in which clitics always appear, with arguments of all kinds, including negative and non-referentially quantified nominals and wh-phrases. 28 This, of course, comes very close to the various versions of the pronominal argument hypothesis (cf. }3.2.1.2) advanced to explain so-called non-configurational languages (Jelinek 1984; Speas 1990; Hale 1994; Baker 2001; 2003).

7 The rise and fall of alignments 7.1 Introduction In this final chapter we discuss a number of core changes in the transition from Latin to Romance, many of which have already been discussed in detail above, in relation to an ongoing and as of yet unresolved conflict between two competing alignments in the marking of arguments (La Fauci 1988: 48; Zamboni 1998: 128). Although not necessarily providing direct support for our preceding formal approaches to the historical morphosyntax of Romance, much of what will be discussed below is entirely compatible with the conclusions of these approaches, especially those relating to the rise of configurationality and functional structure, whilst demonstrating how the core developments in the history of Romance morphosyntax can be integrated into a highly original and insightful theory of a centuries-old typological conflict between a nominative/accusative and an active/ stative syntactic orientation.1 In particular, it will be demonstrated that, despite both the start and the end points of our discussion, classical Latin and modern Romance, displaying a predominantly nominative/accusative orientation in their morphosyntactic systems, this does not represent a case of uninterrupted continuity but, rather, masks an intermediate stage, only in part inferable from documented sources, of an active/stative orientation (Zamboni 1998: 130). Before looking at the details of these developments, we must first establish some basic concepts and distinctions. Following a widely accepted typological distinction (Dixon 1994: 6–8; see also Comrie 1989: 110–16), we can distinguish three core sentential participants labelled A and O (1a), the subject and object, respectively, of a transitive construction, and S (1b–c), the subject of an intransitive construction: (1) a Guillermo (A) ‘Guillermo

1

perdió lost

las the

llaves (O) (Sp.) keys’

Central to our discussion are the seminal works of La Fauci (1988; 1991; 1997; 1998; 2011a) and Zamboni (1998; 2000).

The rise and fall of alignments b Guillermo (S) ‘Guillermo

313

perdió (Sp.) lost’

c me se perdieron me= self= lost.3PL ‘my keys got lost’

las the

llaves (S) (Sp.) keys

In a number of cases both Latin and Romance make a further distinction between two types of intransitive S(ubject), namely between: (i) an S with an agentive interpretation (1b) and hence, to all intents and purposes, identical to A(gent), bar the presence of an O(bject); and (ii) an S with an UNDERGOER interpretation (1c) and hence, to all intents and purposes, identical to O(bject), bar the presence of an A(gent). The former we may call SA and the latter SO. To varying degrees, languages make available the means to encode these three core participants through nominal marking systems (case, adpositions), verb marking systems (agreement, auxiliaries, and voice distinctions), and through sentential word order (La Fauci 1988: 54). Together these three mechanisms of argument marking variously place the three nuclear sentential participants into one of the following three typological organizations (La Fauci 1997: 12): i) A is formally distinguished from O and, in turn, shares the same formal marking as SA/O ii) O is formally distinguished from A, and, in turn, shares the same formal marking as SA/O iii) A is formally distinguished from O, but the formal marking of S is split between A (= SA) and O (= SO). The arrangement described in (i) is traditionally termed a nominative/ accusative alignment, while the arrangement described in (ii) yields an ergative/ absolutive alignment. The third and final alignment represents a compromise between the two preceding alignments, in that S is formally aligned in part with A and in part with O, as illustrated in Table 7.1. TABLE 7.1 Typological alignments of A, S, and O Nominative/Accusative

Active/Stative

Ergative/Absolutive

A

A

A

S

SA

S

SO O

O

O

314

From Latin to Romance

It is doubtful, however, that the full grammatical apparatus of any language can be consistently described in terms of just one of these three alignments (La Fauci 1988: 31–2), although it is often possible to associate particular languages with one predominant orientation. For example, below we shall see that classical Latin combines a nominative/accusative orientation in the nominal system with a partially active/stative orientation in certain areas of the verb system (La Fauci 1997: 17–19). In the later Latin period and continuing into the early Romance period, this active/stative orientation expands further into the verb system spreading even to the nominal system (and, in some cases, perhaps even to be considered an ergative/absolutive orientation), whereas at the level of the sentence there emerges a new nominative/accusative orientation in word order. However, in many cases and in some, especially southern Romance varieties, these new or expanded active/stative alignments were short-lived, coming to be replaced eventually by new nominative/accusative alignments. Significantly, the complex vicissitudes in the morphosyntactic alignments of the late Latin and Romance nominal, verbal, and sentential systems to be discussed below point to a new (though see Cremona 1970, and now also Green 2006) areal classification of the Romània in terms of a northern-southern continuum (Zamboni 1998: 128; 2000: 86, 104–5), rather than the traditional western-eastern (von Wartburg 1950) or central-peripheral (Bartoli 1925; 1929; 1933) divisions based on predominantly phonological and lexical criteria, respectively. This northern-southern continuum contrasts a northern Romània, coinciding with the historical areas of Gallia transalpina (northern Gaul: langue d’oïl, southern Gaul: langue d’oc), Gallia cisalpina (northern Italian dialects), and Rætia (Ræto-Romance varieties) on the one hand, and a southern Romània made up of the (historical) areas of central-southern Italy (central-southern Italian dialects), Sardinia, Iberia (Galician, Portuguese, Spanish, Catalan), Istria/Illyria (Dalmatian), and Dacia (Romanian) on the other. Putting aside details for the moment, these northern/southern areas can be very broadly, though not exhaustively or without exception, distinguished, respectively, in terms of the following structural oppositions: (i) prolonged retention/early loss of V2 syntax; (ii) marking of A/S (subject clitics, generalized preverbal position) vs marking of O (prepositional accusative, object clitic doubling); (iii) prolonged retention/early loss of binary (or ternary) case system; (iv) HABERE/ESSE auxiliary alternation vs generalized auxiliary (ether HABERE or ESSE depending on variety and/or syntactic context); (v) retention vs loss of participial agreement; (vi) loss vs retention (and reinforcement) of preterite (Zamboni 2000: 87).

7.2 Classical Latin The nominal system of Classical Latin can unequivocally be described in terms of a nominative/accusative alignment (Zamboni 2000: 103). By way of illustration, consider the three sentences in (2a–c):

The rise and fall of alignments (2)

a is [ . . . ] capite uelato uictimam he.NOM head.ABL covered.ABL victim.ACC ‘with veiled head he slays a victim’

315

caedet (Lat., Liv. 10.7.10) slays.FUT

minutus laborat (Lat., Petr. Sat. 44) b populus people.NOM small.NOM Suffers ‘and so poor folk suffer’ numerus hostium c magnus great.NOM number.NOM enemies.GEN ‘a great number of the enemy fell’

cadebat (Lat., Caes. B.G. 5.34.2) fell

Whether the Latin grammatical subject corresponds to the A of a transitive predicate (2a), the SA of an (intransitive) unergative predicate (2b), or the SO of an (intransitive) unaccusative predicate (2c), it invariably surfaces in the nominative (indicated by the final inflection -s borne by the three subjects in the examples above). By contrast, the grammatical O(bject) of a transitive verb surfaces in the distinct accusative form (marked in (2a) above by final inflectional -m on uictimam). It follows then that the nominal system of Latin formally contrasts A and S(A/O) (marked nominative) with O (marked accusative) to yield a canonical nominative/ accusative orientation which proves totally insensitive to the semantic characterization (AGENT vs UNDERGOER) of the subject. By contrast, the verb system of classical Latin is less consistent in its morphosyntactic orientation. As the examples in (2a–c) already clearly illustrate, in the active voice the verb system in the tenses of the imperfectum also operates according to a nominative/accusative alignment, in that the finite verb invariably agrees in person and number with the nominative subject (witness the final 3SG inflection -t in all three examples above), and not with the accusative object. However, classical Latin also has a middle voice (VOX MEDIA), which formally brings together intransitive UNDERGOER subjects variously drawn from the passive (3a) and deponent (3b) paradigms,2 which in the imperfectum align with verb forms distinctively marked by the middle formant -R (and variants): (3)

a Qui damnatus erit uirgis ad necem caedetur (Lat., Cic. Verr. 2.3.69) who.NOM damned will.be rods.ABL to death.ACC will.beat.PASS ‘Whoever is condemned, shall be beaten to death with rods’ b N. Fabius Massiliae N. Fabius.NOM Marseilles.LOC ‘N. Fabius died at Marseilles’

2

moritur (Lat., Liv. 42.4.1) dies.PASS

Deponents are those verbs which are active in meaning, but passive in form such as PROFICISCOR ‘I depart’ (cf. AMOR ‘I am loved’), PROFECTUS SUM ‘I (have) departed’ (lit. departed I.am; cf. AMATUS SUM ‘I have been loved’ (lit. loved I.am)).

316

From Latin to Romance

As the active-passive contrast between (2a) and (3a) reveals, the surface passive subject in the latter is underlyingly an O, hence its SO status. Analogously, the overwhelming majority of deponents are unaccusative predicates, whose surface subject is analysed in many current formal frameworks as a derived subject moved from or related to the verb’s complement position, hence its UNDERGOER interpretation and SO status. In the paradigms of the perfectum, the middle is further marked with respect to the active: whereas the latter employs a synthetic verb construction (cf. the use of the perfective reduplicative cecid- and -ui- formatives in (4a–b)), the former makes recourse to a periphrastic conjugation consisting of auxiliary ESSE ‘be’ and the perfect passive participle, the latter agreeing in gender and number with nominative subject (5a–b): (4) a hic est [ . . . ] qui legiones nostras cecidit (Lat., Liv. 22.6.3) this.one.NOM is who.NOM legions.ACC our.ACC.PL slew ‘here is the man who slew our legions’ b ille [ . . . ] si laborauit in nomen (Lat., Sen. Ep. 7.6) he.NOM if he.worked in name.ACC ‘if he has worked for reputation’ (5)

a uirgis caesi tribuni ab legato sunt (Lat., Liv. 29.18.13) rods.ABL beaten.MPL tribunes.NOM by lieutenant.ABL are ‘the tribunes were beaten with rods by the legate’ b Stoicus, qui Athenis nuper est mortuus (Lat., Cic. Off. 2.86) Stoic.NOM who.NOM Athens.LOC recently is died.MSG ‘a Stoic philosopher who recently died in Athens’

To conclude, the classical Latin verb system combines a formal distinction between a nominative/accusative alignment in the paradigms of the imperfectum of the active and middle voices, where in both cases the finite verb displays person and number agreement with its associated nominative subject (be it A, SA, or SO), and an active/ stative alignment in the paradigms of the perfectum (Zamboni 2000: 103), where A and SA are marked by a synthetic paradigm and SO alone is marked by a periphrastic paradigm with number and gender features referenced in the agreement of the verbal participle (La Fauci 1997: 20).

7.3 Late Latin and conservative Romance: active/stative syntax In the passage from classical Latin to Romance there is initially a notable decline in the nominative/accusative orientation of the nominal and verbal systems, paralleled by a corresponding expansion in the range of the active/stative alignment in the verbal and nominal domains. In the verb system, reflexes of an expanding active/stative alignment can be observed in the genesis of the perfective auxiliary

The rise and fall of alignments

317

construction and concomitant patterns of participle agreement, which, in turn, provide the necessary impetus for comparable realignments within the nominal system surfacing in the early Romance reduced case systems based on a binary opposition and, ultimately, in the gradual generalization of the accusative as the universal, unmarked case form. At the level of the sentence too, the effects of an active/stative alignment are clearly observable in emergent word order patterns. 7.3.1 Verbal group 7.3.1.1 Perfective auxiliary constructions

7.3.1.1.1

LATIN BACKGROUND

As discussed in detail in }4.3.1.4 above, traditionally the emergence of the perfective auxiliary + PtP periphrases is retraced, in the case of the HABERE ‘have’ construction, to the grammaticalization of an original resultative aspectual periphrasis, and, in the case of the ESSE ‘be’ construction, to an extension of the perfective paradigm of the Latin middle voice.3 The result, as we have seen, is a split in the perfective forms of the verb between a periphrasis with auxiliary HABERE for transitive/unergative predicates with A/SA subjects on the one hand, and a periphrasis with auxiliary ESSE for unaccusative and passive predicates with SO subjects on the other. Consequently, the verb system of late Latin and, in turn, (early) Romance develops an active/stative orientation (also often termed ‘split intransitivity’) in which intransitive S(ubjects) are formally distinguished according to their SA–SO characterization (La Fauci 1988: 51–2). Although accepting the core developments of the HABERE/ESSE periphrasis outlined in }4.3.1.4 above, it must be noted that this analysis takes the genesis of the HABERE periphrasis as essentially unrelated to the already established ESSE periphrasis, situating its origins wholly in the fortuitous reanalysis of an erstwhile resultative periphrasis. An empirically and theoretically more convincing alternative analysis is, however, to view the rise of the HABERE periphrasis as an integrated change in the verb system within a more general active/stative realignment already underway in classical Latin (La Fauci 1988: 46–50; 1997: 26; 2006: }}3–4; 2011b: 103–14). Recall that above we observed in relation to (5a–b) that in the tenses of the perfectum classical Latin already displayed a specialized perfective auxiliary periphrasis (ESSE ‘be’ + PPtP) for a subset of intransitive predicates (passives and deponents), whose S(ubjects) at some underlying level of representation are also O(bjects), namely SO. In the perfectum, therefore, the Latin verb tends to contrast SO (marked by a periphrastic formation: ESSE + PPtP) and A/SA (marked by a synthetic formation such as the -S- and -UIperfective formatives). The rise of HABERE + PtP in the late Latin/early Romance period can then be seen as an analogical response to the ESSE + PPtP periphrastic 3

Vincent (1982), Salvi (1982; 1987), Tuttle (1986), Zamboni (2000: 127–8).

318

From Latin to Romance

conjugation, reinforcing through the auxiliary alternation of otherwise parallel structures the original active/middle distinction of classical Latin to ultimately produce an active/stative alignment following the extension of ESSE to all constructions with SO subjects (namely, unaccusatives). On this view, the emergence of HABERE to mark all A/SA subjects in the perfective paradigms does not involve the fortuitous grammaticalization of HABERE in an original resultative periphrasis, but simply represents the extension of an already existing active/middle distinction in the Latin possessive construction, a case of conservative innovation (Zamboni 2000: 87). In this regard, La Fauci (1997: 24) notes that classical Latin had two possessive constructions (cf. }6.3.2) one involving ESSE (6a) and the other HABERE (6b; cf. Baldi and Nuti 2010: 251–322). (6)

a nulla tibi linguast (Lat., Pl. Stich. 260) none.NOM.FSG you.DAT tongue.NOM.FSG=is ‘have you got no tongue?’ b si decem habeas linguas (Lat., Pl. Bacch. 128) if ten you.have.SBJV tongues.ACC ‘even if you had ten tongues’

Although ESSE is not usually labelled a possessive, a term traditionally reserved for HABERE, (6a) demonstrates that it can assume possessive function. This leads La Fauci to conclude that the possessive value of ESSE and HABERE, just like the perfective value they assume in the so-called auxiliary periphrasis, is a global property of the constructions in which they appear, and not an inherent property of the predicates themselves. Returning then to the two possessive constructions in (6a–b), that formed with ESSE should be considered a middle construction, in that the surface nominative subject nulla . . . lingua (the possessed) is semantically to be identified with the UNDERGOER role, hence a subject of the SO type, whereas the LOCATIVE argument (the possessor) surfaces as a dative tibi. The HABERE possessive, by contrast, exemplifies an active construction in which the LOCATIVE and UNDERGOER arguments now surface, respectively, as the nominative subject (marked in the person agreement of the 2SG finite verb habeas) and accusative object (linguas) of a transitive construction (namely, as A and O). It follows that the HABERE/ESSE alternation we find in the late Latin perfective construction is simply a direct extension of this original active/ middle alignment of the possessive construction from nominal to verbal predicates. Indeed, this is the view also recently advanced in Adams (forthc. d: }2.2.1), who documents the extension of HABERE with participles expressing mental and other types of acquisition alongside the earlier ESSE periphrasis.4 On this point he notes, for example, that the relationship between the two predicates in the participial

4

Thielmann (1885b: 379–80, 515–16), Allen (1964), Pinkster (1987: 197, 219 n. 7).

The rise and fall of alignments

319

construction (e.g. EST MIHI EMPTUS/HABEO EMPTUM lit. ‘is me.DAT bought/I.have bought’) is identical to that found in the nominal possessive construction (EST MIHI LIBER/LIBRUM HABEO lit. ‘is me.DAT book.NOM/book.ACC I.have’). In support of this analysis, Adams provides a series of near minimal pairs involving both verbs such as (7)–(8),5 which highlight the intrusion of the active verb of possession HABERE into a construction which, for a long time, had been the sole reserve of the inactive ESSE + dative construction. (7)

a tanti sunt mi emptae? (Lat., Var. Rust. 2.2.5) me.DAT bought.FPL such.ABL are ‘have I bought them at such a price?’ b eum autem emptum habebat cum socio it.M.ACC but bought.ACC.MSG he.had with partner.ABL Cn. Acerronio (Lat., Cic. Tull. 16) Cn. Acerronius.ABL ‘but he had had a partner in the purchase, Cnaeus Acerronius’

(8)

a Metello iam antea experimentis cognitum Metellus.DAT already before experience.ABL.PL known.ACC.MSG erat genus Numidarum infidum (Lat., Sall. Iug. 46.3) it.was race.ACC Numidians.GEN treacherous.ACC ‘Metellus had already learned from experience that the Numidians were a treacherous race’ b haberem a Furnio nostro tua penitus I.had.SBJV from Furnius.ABL our.ABL your.NPL deeply consilia cognita (Lat., Cic. Fam. 10.12.1) views.ACC.N known.ACC.NPL ‘I had a thorough acquaintance with your views from our friend Furnius’

7.3.1.1.2

ROMANCE

Deliberately generalizing somewhat,6 the original active/stative alignment of the two perfective auxiliaries already evident in the Latin possessive constructions proves most resilient in the northern Romània:

5 In addition, Adams highlights how the expression of agency with the gerundive produced another environment in which the two possessive patterns alternated, providing further impetus for the pattern of HABERE/ESSE alternation with the participle. More specifically, alongside the classical construction with ESSE ‘be’ + dative (MIHI DICENDUM EST ‘I have to say’ (lit. ‘me.DAT say.GER it.is’)), from the early imperial period there emerges a rival periphrasis with HABERE ‘have’ (HABEO DICENDUM lit. ‘I.have. say.GER’). 6 For an exhaustive account of auxiliary distribution across (Italo-)Romance, see Loporcaro (2007).

320

From Latin to Romance

(9) a tu as pleuré / tu étais monté (Fr.) you have cried you were ascended ‘you have been crying / you had gone up’ b avèm susat / sem arribats we.have sweated we.are arrived ‘we sweated / we arrived at the market’ c a

i’ eu drimì / SCL I have slept ‘I have slept / we have left’

a SCL

d ha dormito / è he.has slept he.is ‘he has slept / he has fainted’

suma we.are

al at.the

mercat (Lgd.) market

partìi (Cai.) left

svenuto (It.) fainted

In the modern varieties of the southern Romània, by contrast, there has been a tendency towards the gradual loss of the original active/stative alignment, with the generalization of HABERE in many varieties (cf. }7.4.1.1). Of course, there are many exceptions to these broad areal generalizations, including the retention of the active/ stative HABERE/ESSE split in some parts of the southern Romània, including Alguerès and Balearic Catalan varieties (10a–b), though not the Pitiuses (Moll 1997: 134; 2006: 290; Veny 1982: 68, 81), Aragonese (10c; Nagore Laín 2001: 86–7), Sardinian (10d; Jones 1993: 130–7), and some southern Italian dialects such as Cosentino (10e; Lombardi 1997). (10) a he caminat / vengut sés? (Alg.) I.have walked come you.are ‘I have walked / Have you arrived?’ b Què has trobat? / som tornada de Barcelona (Bal.) what you.have found I.am returned from Barcelona ‘What have you found? / I have returned from Barcelona’ c ¿Has trobau as claus? / l’ augua you.have found the keys the water ‘Have you found the keys? / the water had boiled’ d at faeddatu / est he.has spoken he.is ‘he has spoken/returned’ e amu chiangiutu we.have cried ‘we have cried/fallen’

/

yera was

bullida (Ara.) boiled

ghiratu (Lul.) returned

simu we.are

caduti (Cos.) fallen

However, this simple approach to the active/stative split in perfective auxiliation, which only considers non-reflexive unaccusatives, is not exhaustive in that it fails to

The rise and fall of alignments

321

recognize the full extent of the rich variation found in the traditional domain of stative syntax in relation to distinct classes of reflexive predicate. Following Loporcaro (2001; 2007), the degree of conservation of the original active/stative split in the distribution of the perfective auxiliaries across Romance can be summarized by way of the six patterns illustrated in Table 7.2.

TABLE 7.2 Active/stative split in Romance perfective auxiliaries Stative Unacc.

Reflexives

Active Trans./Unerg.

inherent direct trans. indirect unerg. indirect trans. It.

E

H

Cos.

E

H

ORmc.

E

H

OTsc.

E

H

Lec.

E

H

Sp.

H

Thus, we see that a language like Italian represents a conservative variety in which HABERE is employed just with active syntax (namely, with A and SA subjects), while ESSE is reserved for stative syntax (namely with SO subjects), including all reflexives (11a–d). (11) a si erano vergognati (It.; inherent reflexive) selves= they.were ashamed ‘they had felt ashamed’ b si erano salutati (It.; direct transitive reflexive) selves= they.were greeted ‘they had corrected themselves’ c si erano telefonati (It.; indirect unergative reflexive) selves= they.were telephones ‘they had rung one another’

322

From Latin to Romance d si erano lavati le selves= they.were washed the ‘they had washed their hands’

mani (It.; indirect transitive reflexive) hands

In this respect, a variety such as Cosentino differs only minimally from Italian, in that it aligns indirect transitive reflexives with transitives and unergatives as part of the active alignment (12a). By contrast, in a variety such as old Romanesco (Formentin 2002: 236–7), the division between the original active/stative split is further weakened with the extension of HABERE into the domain of indirect unergative reflexives (12b), and still further in the early Tuscan of Dante, where stative syntax had shrunk to just unaccusatives and inherent reflexives following the expansion of HABERE to direct transitive reflexives (12c; La Fauci 2004; 2011c). A radically more reduced stative syntax is found in such varieties as Vallader (Ganzoni 1983), where auxiliary ESSE is restricted to unaccusatives (12d), which in Spanish (12e) have long succumbed to the generalization of auxiliary HABERE, thereby eliminating entirely the active/stative opposition from the system of perfective auxiliation. (12)

a s’ avìanu lavatu i selves= they.had washed the ‘they had washed their hands’

manu (Cos.; indirect transitive reflexive) hands

b secu(n)do ch(e) se ào lassato (ORmc.; indirect unergative reflexive) according that self= he.has left ‘in accordance with what he has left for himself (in his will)’ c la donna che [ . . . ] ci s’ hae the woman that us= self= has transitive reflexive) ‘the woman that showed herself to us’ d ella s’ ha she self= has ‘she got washed’

mostrata (OTsc., VN 38.3; direct showed

lavada (Vld., inherent reflexive) washed

e habían muerto / se they.had died selves= ‘they had died/repented’

habían they.had

arrepentido (Sp.) repented

Despite these variations in the distribution of the perfective auxiliaries, the original correlation between ESSE and SO is still preserved in all modern varieties, including Spanish (13b), in the retention of ESSE as the passive auxiliary.7

7

The Apulian dialect of Altamura (Loporcaro 1988) is an apparent partial exception, in that the passive in this dialect can be constructed with both ESSE and HABERE:

The rise and fall of alignments (13)

323

a lo Palais dei Papas fogùet bastit a l’ Edat mejana (Prv.) the palace of.the popes Was built at the age middle ‘The Palais des Papes was built in the Middle Ages’ b a lo largo de los años fueron publicados 381 to the length of the years were published 381 ‘381 issues were published over the years’

números (Sp.) issues

However, even the binary variations in active/stative syntax revealed in Table 7.2 do not exhaust all attested possibilities in Romance, with a number of Romance varieties displaying what Loporcaro (2007) perceptively describes as a case of triple auxiliation. In these varieties, broadly all found in the southern Romània, both auxiliaries continue to alternate, but not just according to the original active/stative split, but also according to grammatical person. Thus alongside pure person split systems in which, according to one pattern, 1/2nd persons typically align with ESSE and 3rd persons with HABERE with all verbs (14a–d; cf. also }6.3.2),8 we also find varieties in which such a split is combined with a traditional active/stative split, such that the choice of auxiliary is in part determined by the argument structure of the verb and in part by grammatical person. For instance, in the conservative variety of Altamurano described in Loporcaro (1988: 278–80, 290), HABERE/ESSE occur in free variation in the 1/2 persons with all verb classes (15a–b), whereas in the 3rd person there is an active/ stative split in certain verb classes with 3SG (15c), but not 3PL (15d), licensing ESSE with unaccusatives, and 3PL (15e), but not 3SG (15f ), licensing HABERE with transitives and unergatives.

(i)

a

MaRɪ Maria

b

l’ the

fo was

/

avi had

rrʊbbwɛtə robbed

lɪ the

krɛip (Alt.) goats

‘Mari’s goats were stolen’ emmə men

fuornə were

/ avernə had

pajɛit (Alt.) paid

‘the workers were paid’ Rohlfs (1969: }730), Tuttle (1986), Kayne (1993), Veny (1982: 51), Ledgeway (2000: 192–5, 204–5), Bentley and Eyþórsson (2001), Cennamo (2001b), Manzini and Savoia (2005 II: ch. 5), Bentley (2006: 56–64), Loporcaro (2007), D’Alessandro (2010: 31–2), D’Alessandro and Ledgeway (2010b), D’Alessandro and Roberts (2010), Legendre (2010). As alluded to in the text, the widely reported 1/2 (ESSE) vs 3 (HABERE) person split, though common, is not the only one found in (Italo-)Romance, with other dialects showing many different distributions, including in the dialect of Pompei, for example, the generalization of HABERE in all persons except the 3SG where only ESSE is available (Cennamo 2001b: 444). In other dialects, by contrast, we witness free variation of the two auxiliaries in all persons, as happens with reflexive predicates in Venetian (Lepschy 1984), or in specific persons, as happens in Mendrisiotto where ESSE occurs in all six person-number combinations with reflexives, though freely alternating with HABERE in the 3SG/PL (Lurà 1987: 169). Data like these lead Loporcaro (2007: 196) to conclude that ‘the distribution of aux[iliary] E/H across verb persons in mixed systems is not predictable, overall, on the basis of an alleged person ergativity split’. 8

324

From Latin to Romance

(14) a So / ha bist / vingut (Olo.) I.am he.has seen come ‘I have/he has seen/come’ b e’

süma vistla / we.are seen=her ‘we saw her/ he has seen her’ SCL

al SCL

a has

vistla (Chi.) seen=her

c so / si /a fatecate / ite (Acf.) I.am you.are have.3 worked gone ‘I/you/(s)he have/has worked/gone mːanʤɛit (Alt.) eaten 



(15)

a aɟɟə manʤɛit / I.have eaten ‘I have eaten’

sɔ I.am

ba mwert / you.have died ‘you have died’

sə you.are

ce /**a rːʊmwɛsə he.is he.has remained ‘he remained alone’ d awon:ə rʊmwɛsə they.have remained ‘they remained’

/

e nan awonːə /**dzɔ not they.have they.are ‘they’ve never written’ f nan a / e not he.has he.is ‘he’s never written’

mːwert (Alt.) died

sʊul (Alt.) alone sɔ they.are skrɪtːə written skrɪtːə written

rːʊmwɛsə (Alt.) remained mɛi (Alt.) ever mɛi (Alt.) ever

Variations like these highlight the complex interaction between an active/stative alignment overlaid by a person split, yielding at least five different triple auxiliation systems, as outlined in Table 7.3 from Loporcaro (2011b), where the distribution of HABERE/ESSE varies in accordance with three different clause types (unaccusative vs transitive/unergative vs reflexives). In particular, we witness in these systems a regular active/stative split between HABERE employed with transitives/unergatives and ESSE employed with unaccusatives, just as in standard Italian and many Occitan varieties, whereas reflexives variously exhibit variation in the distribution of the two auxiliaries in accordance with a structured person split or in free variation in all or specific persons of the paradigm. For example, in the Marchigiano dialect of Maceratese (Paciaroni 2009) reflexives generally display ESSE on a par with non-reflexive unaccusatives (16a),

The rise and fall of alignments

325

TABLE 7.3 Romance triple auxiliation systems Stative Unacc.

Active

Reflexives

Trans./Unerg.

inherent direct trans. indirect unerg. indirect trans. Mac.

E

Agn. A

E

Agn. B

E

Csv.

E

Trn.

E

E/H

H

E/H

H E/H

H

E/H

H E/H

H

whereas indirect transitive reflexives show free variation of both auxiliaries in all persons (16b). (16)

a essa s ɛ rviʃtita (Mac.) she self= is dressed ‘she got dressed’ b essa s ɛ rlaata she self= is washed ‘she washed her hands’

/ essa she

s self=

a has

rlaato washed

le the

ma (Mac.) hands

A slightly different pattern is found in the conservative and innovative Molisan varieties of Agnonese, henceforth labelled as AgnA and AgnB, respectively. Here reflexives standardly display ESSE (17a), with the exception of indirect unergatives (AgnA) which, together with indirect transitives (AgnB), display free variation of HABERE/ESSE (17b–c). (17)

a essa s e she self= is ‘she got up’

rrəttsiəta (Agn.A/B) raised

b kella femməna s e kkuʧənieta / s a kuʧənietə essa seula (Agn.A/B) that woman self= is cooked self= has cooked she alone ‘that woman cooked for herself’ c sɔrma s a missə / sister=my self= has put ‘my sister put her hat on’

e is

mmissə put

ru the

kappiəllə (Agn.B) hat

326

From Latin to Romance

Even greater variation is found in the northern Calabrian dialect of Castrovillari where, in contrast to dyadic reflexives which select HABERE (18a) on a par with transitives/unergatives, monadic reflexives display free variation of HABERE/ESSE (18b–c). (18) a lɔrʊ s annu allʊrdatʊ they selves= have dirtied ‘they got their shirt dirty’ b Maríja s ɛ / Maria self= is ‘Maria got dressed’

a has

c rɔsa s ɛ / a Rosa self= is has ‘Rosa answered herself ’

a the

kammɪsa (Csv.) shirt

vvɪstʊta (Csv.) dressed rrɪspʊsa replied

sʊla (Csv.) alone

The final type is represented by Trentino dialects like Valsuganotto (Loporcaro and Vigolo 1995: 98), where all reflexives display a classic person-related alternation with ESSE in 1/2 persons (19a–b) and HABERE in the 3rd persons (19c). (19) a me son spaurá / zbará / lavá (le myself= I.am frightened shot washed the ‘I got scared/shot myself/washed myself/my hands’

man) (Vsg.) hands

b te si spaurá / zbará / lavá (le yourself= you.are frightened shot washed the ‘you got scared/shot yourself/washed yourself/your hands’ c el s a spaurá / zbará / lavá (le he self= has frightened shot washed the ‘he got scared/shot himself/washed himself/his hands’

man) (Vsg.) hands

man) (Vsg.) hands

7.3.1.2 Participle agreement Together with the emergence of an active/stative split in the auxiliary alternation of the perfective paradigms, there emerges a parallel alignment in the novel system of participle agreement. In particular, in the ESSE periphrasis the participle variously agrees in gender and number with the UNDERGOER (20a), while in the HABERE periphrasis the participle agrees with the direct object (20b) but not with the AGENT subject (20c), as witnessed by the following examples from Majorcan Catalan: (20)

a ma mareta s’ és aufegada (Maj.) my mother.DIM self= is suffocated.FSG ‘my poor mother has suffocated’

The rise and fall of alignments b una altra cosa m’ an other thing.F me= ‘another thing she told me’

ha has

c vui fer lo que ha I.want do.INF the what has ‘I want to do what she has done’

dita told.FSG fet done.MSG

327

aquella (Maj.) that.one.FSG (**feta) done.FSG

aquesta this.one.FSG

The agreement paradigm evidenced in (20a–c) is that of a classic active/stative alignment, in which agreement marks all types of O (including those which subsequently surface as S), but never A (including SA). Now, while there does seem to be a strong correlation with the retention of the active/stative alignment in the auxiliary system and a corresponding retention of the same alignment in participle agreement, witness the evidence of Balearic Catalan (excluding the Pitiuses; Villangómez i Llobet 1978: 122), French, many Italo-Romance varieties (including Italian), and Occitan, there are some genuine counterexamples (Loporcaro 1998: 8–12; for standard Catalan, see }7.4.1.1). Particularly revealing in this respect are many of the dialects of central and southern Italy, where the original active/stative auxiliary alternation has been replaced by a nominative/accusative person-marking system (ESSE: 1/2 vs HABERE: 3; cf. }7.4.1.1), but participle agreement still operates according to the original active/stative alignment. For example, in the central Italian dialect of Cori spoken in Lazio, whether HABERE (3PL) or ESSE (all other persons) is selected, the participle only agrees with O (21a) and SO (21b), but never with A (21c) or SA (21d): (21)

a so cote le I.am picked.FPL the.FPL ‘I’ve picked the plums’

prunca (Crs.) plums.F

b issi s’ èo vergognati (Crs.) they.M selves= have shamed.MPL ‘they’ve shamed themselves’ c Maria è rutto jo Maria is broken.MSG the.MSG ‘Maria’s broken the piggy-bank’ d issi èo magnato (Crs.) they.M have eaten.MSG ‘they have eaten’

dindarolo (Crs.) piggy.bank.M

328

From Latin to Romance

7.3.2 Nominal group 7.3.2.1 Extended and generalized accusative It has long been observed that,9 apart from some very early examples, Latin texts from the southern provinces of the Empire (Italy, Iberian Peninsula, Africa, and the Balkans) from the end of the second century AD, and with rapidly increasing frequency in subsequent centuries (Löfstedt 1933: 329– 34; Norberg 1944: 21–32), frequently show an ‘extended’ use of the accusative in place of the nominative to mark the subject of finite clauses. This use of the ‘extended’ accusative (or ‘restricted’ nominative) does not involve, however, a random substitution of the nominative with the accusative but, rather, is structurally determined.10 In particular, it is found in unaccusative syntax where the appearance of the accusative appears to reflect the underlying semantic case of the UNDERGOER subject, thus typically surfacing in middle constructions with deponents (22a), anticausatives (22b), passives (22c), impersonal passives (22d), and existentials (22e), as well as in active syntax in conjunction with unaccusatives (22f ) and, in particular, the verb ESSE ‘be’ (22g; cf. Adams forthc. e }6.4). (22)

a nascitur ei genuorum contractionem aut claudicationem is.born.PASS him.DAT knees.GEN contraction.ACC.F or limp.ACC.F (Lat., Mul. Ch. 516) ‘his knees are developing a contraction or a limp’ b multos languores sanantur in ipsis locis many.ACC.MPL weaknesses.ACC.MPL heal.PASS.3PL in same places (Lat., Ant. Plac. Itin. 165.16) ‘many weaknesses are healed in these places’ c ipsas portas sames.ACC.FPL gates.ACC.FPL ‘the(se) gates are opened’

aperiuntur (Lat., Itin. Hier.. 11.1) open.PASS.3PL

d et sic fit orationem pro and thus is.done prayer.ACC.FSG for ‘and thus the prayer is made for everyone’

omnibus (Lat., Per. Aeth. 25.3) all.ABL.PL

e habebat de ciuitate forsitan mille quingentos passus had.3SG from city.ABL.F perhaps thousand five.hundred.ACC.MPL steps.ACC.M (Lat., Per. Aeth. 23.2) ‘it was perhaps 1500 paces from the city’

9

Löfstedt (1933: 329–34), Norberg (1944: 21–32), Gerola (1949–50), Bastardas Parera (1953: 16–20), Westerbergh (1956: 235–6), Herman (1966; 1987b: 102; 1995: 72–5), Durante (1981: 41), Pensado (1986), Väänänen (1982: 203–4), La Fauci (1988: 54–5), Zamboni (1998: 131–2), Pieroni (1999), Cennamo (2001a; 2009), Rovai (2005), Ledgeway (2011: 459–61), Adams (forthc. e: }6). 10 La Fauci (1988: 54–5), Zamboni (1998: 132), Cennamo (2001a: 4–5, 10–11).

The rise and fall of alignments

329

f ut sanguinem exeat copiosum (Lat., Mul. Ch. 618) so.that blood.ACC.M exit.SBJV copious.ACC.MSG ‘so that plentiful amounts of blood may run out’ g si sine uulnere erit, totam curationem if without wound.ABL he.will.be all.ACC.FSG healing.ACC.FSG haec est (Lat., Mul. Ch. 526) this.NOM.FSG is ‘if he is unwounded, this is all part of the healing process’ This ‘extended’ use of the accusative which increases greatly in frequency in later Latin texts can arguably be construed as a gradual active/stative alignment of the nominal system, whereby non-active subjects come increasingly to be formally codified on a par with transitive objects,11 parallel to the active/stative split that emerged in the compound perfective forms of the verb system. In the same way that the latter comes to formally distinguish A/SA from SO/O in auxiliary selection and participle agreement (namely, HABERE, -Agr vs ESSE, +Agr), the nominal system too comes to mirror this distinction in the southern provinces with nominative reserved for A/SA subjects and accusative for O(bjects) and, increasingly through time, also for SO subjects. However, as Adams (forthc. e: }6.3) is at pains to stress, not all purported cases of the extended accusative reported in the literature can be regarded as genuine examples of the accusative or of a genuine morphosyntactic development in the marking of non-active subjects. Rather, caution needs to be exercised in assessing all apparent cases,12 making due allowance for such factors as textual corruption (particularly evident in the case of the Mulomedicina Chironis which is often cited in this regard), morphological conflations, and anacoluthon. Particularly dubious in this respect are supposed cases of singular accusative subjects in -AM and -UM (1/2 declensions), where the addition and omission of final -M is notoriously unsystematic and cannot be consistently interpreted as reliable evidence for the accusative status of the subject (cf. }2.2.3). Illustrative in this respect are examples like (23), where the apparent accusative marking of the subject nominal causam is contradicted by its accompanying demonstrative haec, which occurs in an unambiguously nominative form. This rightly leads Adams to interpret causam as a scribal error, since Gregory was a learnèd writer and it would be ‘stretching credibility to suggest that Gregory wrote haec and then causam, given that there is no more frequent scribal error than the omission or false addition of final -m’. La Fauci (1988: 54–5), Cennamo (2001a: 4, 7), Adams (forthc. e: }6). For example, Herman’s (1995) assertions about a ‘neutralization’ of accusative and nominative in the second declension in the Visigothic slate tablets is based on a misrepresentation of all five examples she cites (J. Adams, p.c). 11 12

330 (23)

From Latin to Romance eice, eice haec a te, o sacerdos, throw.IMP.2SG throw.IMP.2SG these.ACC.NPL from you o priest.NOM ne faciat scandalum haec causam (Lat., Greg. Tur. 5.18) lest do.SBJV.3SG scandal.ACC this.NOM.F cause.ACC.F ‘Get rid of these things, bishop, get rid of them for fear that the matter may cause a scandal’

Equally unreliable are apparent cases of plural accusative subjects in -AS and, to a lesser extent, -OS (1/2 declensions) which, as Adams (forthc. e: }6.6) reminds us, had long served in different parts of the Empire, especially in low registers and inscriptions, as alternative and/or rival nominative plural forms alongside the prescriptively more correct forms in -AE and -I, respectively.13 Indeed, the forms, especially feminine plural -AS, are attested with such a high frequency—from perhaps as early as the Republic but, in any case, regularly from the first century AD and even becoming standard in Merovingian Latin (Vielliard 1927: 109)—that they are better considered as alternative non-standard forms of an (Indo-European >) Italic nominative plural morpheme. Consequently, many purported examples of the extended accusative are more convincingly explained in morphological, rather than syntactic, terms. Allowing for such factors, the residue of genuine cases of accusative subjects are not so numerous that it is genuinely possible to talk of a systematic replacement of the nominative by the accusative in late Latin unaccusative syntax. Rather, what we witness, especially in lower register pre-medieval texts, is the gradual encroachment of the accusative on the nominative in the formal marking of non-active (A/SA) subjects. Instructive in this respect are the following salutary observations on the topic by Adams (forthc. e: }6.4): [I]t would be misleading to imply that subject accusatives are common in late Latin or that there is an attested stage of the language in which they can be seen to be established, if still outnumbered. Examples are thinly scattered, and, worse, they tend to be in corrupt texts of which the manuscript or manuscripts cannot be relied on. [ . . . ] There is no receding of the nominative at all, merely the occasional use of an accusative where a nominative might have been expected, and such examples may be subject to other explanations (conflations, anacoluthon, false addition of final -m [ . . . ]).

Apart from some important exceptions to be discussed below (}7.3.2.2), by the time our written records begin the Romance texts of this southern area no longer show such an intransitive split in the nominal system, inasmuch as accusative had generalized to all nominals, A, SA/O, and O alike. Indeed, even in late Latin texts it has been claimed that there is early evidence of this development 13

Norberg (1944: 27 n. 2), Gerola (1949-50: 207–19), Väänänen (1966: 84), Rovai (2005: 71–6), Adams (2007: 674–5 n. 79).

The rise and fall of alignments

331

(Cennamo 2001a: 8–9), with the extension in some cases of the accusative even to active, dynamic intransitive SA subjects (24a) and, more rarely, to transitive A subjects (24b). (24)

a ipsos filios sedeant (Lat., CSV 42.16.1047) same.ACC.PL sons.ACC sat.3PL ‘the(se) children were seated’ b Iulia Crescensa cui filios et Iulia.NOM Crescensa.NOM who.DAT sons.NOM/ACC and nepotes obitum fecerunt (Lat., ILCV 3052 B) grandchildren.ACC tomb.ACC made.3PL ‘Iulia Crescensa for whom her sons and grandchildren a tomb’

However, Adams (forthc. e: }7) notes that extension of the accusative to active subjects in such cases is only weakly attested and, even then, often ambiguous. Indeed, this is the case in the examples above where the relevant forms (ipsos, filios) involve the ending -OS, which we noted above, can be taken to represent the morphological extension of a genuinely nominative, non-standard inflection probably by analogy with the feminine plural -AS ending. Moreover, this extension of the accusative is not limited to the verbal domain examined above, but apparently also involves the nominal domain, where the nominative may be replaced by the accusative in contexts such as lists, recipes, citation forms, appositions, exclamations, and commands (25a–c).14 It must be acknowledged however, following Adams (forthc. e: }5), that there had long been random fluctuation between the prescriptively unjustified use of the nominative and the accusative in these contexts, in many cases often quite simply explicable by the fact that writers were not paying due attention to the context. Nonetheless, Adams (}5.4) does concede that random variation of this type in, for example, lists might be indicative of a gradual destabilization of the traditional case functions conveyed by the traditional nominative/accusative formal opposition. (25) a Puteolos, Antium, Tegeanum, Pompeios: Pozzuoli.ACC.MPL Antium.NOM/ACC.N Tegeanum.NOM/ACC.N Pompei.ACC.MPL hae sunt uerae coloniae these are real.NOM.PL colonies.NOM (Lat., CIL IV.3525) ‘Pozzuoli, Anzio and Tegeanum are real colonies’

14

Gerola (1949–50), Herman (1966), Väänänen (1982: 203–4), Pensado (1986), La Fauci (1997: 34), Zamboni (1998: 131–2).

332

From Latin to Romance b Et “aquam foras, uinum intro” and water.ACC outside wine.NOM/ACC inside ‘he exclaimed “Out water, in wine!”’ c potionem ad eos, qui sanguinem drink.ACC to those who blood.ACC ‘a drink for those who pass blood’

clamauit (Lat., Petr. Sat. 52) he.called meient (Lat., Mul. Ch. 822) urinate.3PL

Significantly, it is generally claimed from a crosslinguistic perspective (Villar 1983: 31; Zamboni 1998: 131) that the case form used in so-called a-syntactic functions such as those in (25a–c) above represents the unmarked case. In short, the late Latin evidence might be taken to point to a gradual generalization of the accusative in verbal and nominal contexts as the unmarked case, a development already complete in the earliest texts of most of the southern Romània, but not in those of the northern Romània where a residual nominative-accusative/ oblique distinction continued until the fourteenth century (Gallo-Romance) or even the seventeenth/eighteenth century in the case of Ræto-Romance (namely, Surselvan and Vallader).15 Traditionally, the generalization of the accusative in Romance is simplistically claimed to be a consequence of the greater frequency of the accusative with respect to the nominative.16 However, in a language assumed to operate on a nominative-accusative alignment, the near-exclusive survival of the accusative stands out as an unexpected development, since the accusative represents the marked case in such an alignment; as Penny (2002: 119) succinctly puts it, ‘the traditional argument, while morphologically adequate, is syntactically inadequate.’ Rather, the extension and gradual generalization of the accusative in late Latin and Romance provides further proof for the loss of the original nominative/accusative orientation in the nominal system in favour of an active/stative or, better, ergative/ absolutive orientation (recall the possible extension of the accusative to active, dynamic intransitive subjects), in which it is precisely the accusative (= absolutive/ stative), and not the nominative (= ergative/active) which represents the unmarked case (La Fauci 1997: 56–58; Zamboni 1998: 131). Incidentally, this observation calls into question the traditional hypothesis (cf. }2.2.3) that the loss of the case system was due (in large part) to phonetic erosion.17 While perceptual saliency will undoubtedly have some impact on the survival of the case system, the evidence reviewed here suggests that the reduction and loss of the case system is fundamentally due to a typological realignment in the nominal system from an original nominative/accusative 15

In masculine adjectives in these Ræto-Romance varieties the original case distinction has been reinterpreted as a predicative (< nominative) vs attributive (< accusative/oblique) distinction (Haiman 1988: 366–7, 381–4; Haiman and Benincà 1992: 141–52). 16 Price (1971: 96–7), Harris (1978: 46), De Dardel and Wüest (1993: 52). 17 Lakoff (1972: 189), Vennemann (1975), Harris (1978: 8), Bauer (1995: 5).

The rise and fall of alignments

333

orientation towards an active/stative (or ergative/absolutive) orientation (La Fauci 1997: 41; Zamboni 1998). 7.3.2.2 Early Romance binary case systems Superficially, the Romance languages fall into two groups, those that from the earliest texts present a binary case distinction (Gallo-Romance, Ræto-Romance: nominative vs accusative/oblique; Daco-Romance: nominative/accusative vs dative/genitive) and those which have lost all case distinctions following the generalization of the accusative (southern Romance). De Dardel and Wüest (1993) interpret this distribution of case across Romance in terms of two areally and chronologically conditioned phases of simplification. Accordingly, in the first phase the case system of Latin would have been radically reduced to zero in all emergent Romance varieties on account of presumed creolization processes, in turn followed by a second phase of reconstruction which produced a ternary (nominative vs accusative vs oblique) case system. Consequently, the vast majority of southern Romance varieties with apparently no documentary evidence are assumed to be more archaic and belong to the first phase, whereas languages with (residual) case distinctions (Gallo-Romance, Ræto-Romance, Daco-Romance) belong to the second phase. The problems for this theory are numerous (for an overview, see Zamboni 1998: 129–30), suffice it to note here two points. First, it is inconceivable that uneducated, illiterate speakers (the overwhelming majority of the population) would be able to reconstruct, even partially, the case system of a (literary) language which had long ceased to be spoken. Second, there is considerable residual evidence in southern varieties to suggest the existence of a binary (or ternary) case system here too. Rather, all the available evidence from the northern and southern Romània points to a binary case system,18 whose original nominative-accusative orientation, as witnessed, for example, by early Gallo-Romance where all subjects, be they A, SA, or SO, are marked nominative in contrast to accusative-marked O(bjects), gradually develops into an active/stative orientation.19 Revealing in this respect is the observation that virtually all modern Romance residues of the nominative, including in southern Romance varieties where evidence has traditionally been deemed lacking, 18 This binary case system (Gallo-/Ræto-Romance: nominative vs accusative/oblique; Daco-Romance: nominative/accusative vs dative/genitive) is most likely derived from a preceding ternary nominativeaccusative-oblique system (La Fauci 1997: 37–53; Zamboni 1998: 137–42; 2000: 93, 110–15), evidence of which is still preserved in some pronominal systems (Salvi 2011: 322–4). In particular, this ternary system combined a subsystem of animate nouns operating on a nominative/accusative alignment incorporating a definite ‘marked nominative’ with three cases (nominative vs accusative vs oblique) and a subsystem of inanimate nouns operating on an active/stative alignment with two cases (nominative/accusative vs oblique). It is this first subsystem, following the early neutralization of -U/-O and -I/-E producing the neutralization of accusative and oblique, which surfaces in Gallo- and Ræto-Romance varieties, whereas the second subsystem underlies the case system of modern Daco-Romance. 19 Schsler (1984), van Reenen and Schsler (2000), Salvi (2011: 319–22), Smith (2005; 2011: 281–9), Sornicola (2011: 18–32).

334

From Latin to Romance

involve animate nouns:20 AMITA > ‘father’s sister’ > (O)Fr. (t)ante ‘aunt’ (cf. OFr. OBL antain); *kompanjo ‘companion’ > Cat. company, It. compagno; *kómpatre ‘godfather’ > OSp. cuémpadre; DEUS ‘God’ > OCat./Pt. Deus, RæR. Dius, Sp. Dios; DOM(I)NUS ‘master’ > Cat. en ‘the’ (used before male proper names; cf. }4.2.2.2), Sp. don ‘honorific title’ (cf. DOMINUM (ACC) > dueño ‘owner, landlord’); *draks ‘large serpent > dragon’ (cf. DRACONEM (ACC)) > Cat. drac, It. drago, Ro. drac ‘devil’; FRATE(R) ‘brother’ (cf. FRATREM (ACC)) > SID/Ro. frate, It. frate ‘monk’;21 HOMO ‘man(kind)’ > It. uomo, Cat./Occ. hom/ òm ‘one’ (indefinite subject; cf. HOMINEM (ACC) > Cat./Occ. home/òme ‘man(kind)’), Ro. om ‘person’; HOSPES ‘host; guest’ > Ro. oaspe ‘guest’ (alongside of HOSPITEM (ACC) > oaspete); JESUS > Fr./Pt./Sp. Jésus/Jesus/Jesús; LATRO ‘thief’ (cf. LATRONEM (ACC)) > Cat. lladre, OGen. layro, It. ladro, ONap. latro; MAGISTE(R) ‘master’ (cf. MAGISTRUM (ACC)) > Nap. masto; MAIOR (cf. MAIOREM (ACC)) > Srd. mère ‘boss’; MILES ‘soldier’ (cf. MILITEM (ACC)) > Ro. mire ‘bridegroom’; MULIE(R) ‘lady’ > It. moglie ‘wife’ (cf. MULIEREM (ACC) > Cat. & Pt./Sp. mulher/mujer); NEMO ‘nobody’ + NOM -S > Log. nemos: NEPOS ‘grandchild; nephew’ (cf. NEPOTEM (ACC)) > SLig./Pie. nevu, OVen. nievo, Rov. nevo (cf. PL navaudi); PATE(R) ‘father’ (cf. PATREM (ACC)) > Nap. pate; PAUO ‘peacock’ > Sp. pavo ‘turkey’ (cf. PAUONEM (ACC) > Fr./It. paon/pavone ‘peacock’); PRE(S)BYTE(R) ‘priest’ (cf. PRESBYTEREM (ACC)) > Cal. prìevite, OCat./OPt./OSp. prestre, Fr. prêtre, It. prete, OLmb. prèvido, Ro. preot; *pùllitr ‘colt’ > Sp. potro; RES ‘thing’ (cf. REM (ACC) > Fr. rien ‘nothing’; Pie. reŋ ‘(no)thing > not’) > Srd. arrèze ‘reptile; worm’; RE(X) ‘king’ (cf. REGEM (ACC)) > (O)Fr. roi(s), It. re; SARTO(R) ‘tailor’ (cf. SARTOREM (ACC)) > Cat./Sp. sastre, It. sarto; *sénior ‘elder’ > Cal./Occ ségnu/sénher ‘master!’ (VOC; cf. SENIOREM (ACC) > Cal./Occ. signòre/ senhór ‘mister, sir’), Fr. sire ‘sire’; SERPE(NS) ‘snake’ (cf. SERPENTEM (ACC)) > Nap./OIt. serpe; SORO(R) ‘sister’ (cf. SOROREM (ACC)) > Fr. soeur, Cal. suoru, Ro. soră, Srd. sòrre, It. suora ‘nun’. Also to be included here are the agentive suffixes -ATO(R) (e.g. ADUOCATOR > OVto. avogádro ‘lawyer’, CURATOR ‘guardian’ > Cal./Sic./Tar. curátulu ‘head shepherd’, PASTOR ‘shepherd’ > OBel. pástro ‘pastor’, *teksator ‘weaver’ > Trn. tesádro, TRADITOR ‘traitor’ > Fr. traître, PICTOR ‘painter’ > peintre), and -ONE + NOM -S (e.g. FILONES ‘spinner’ > Egd. filunz, Lmb. filonz, Egd. tissunz ‘weaver’, Mtv. crivlonz ‘jigger’), -a (F)/- (M) (with OBL -AN- (F)/-ON- (M); e.g. Lvl. muta/mut ‘girl/boy’ vs mutans/mutons ‘girls/boys’), many male first names in -s (e.g. Cat./Fr./Pt. & Sp. Carles/Charles/ Carlos, Fr. Jacques, Georges, Louis, Fr./Pt./Sp. Jésus/Jesus/Jesús, (Pt./)Sp. Marcos,

20 The bibliography here is too vast to cite in its entirety. Some recommended sources include: Rohlfs (1968: }344), Spence (1971), Harris (1978: 49), Tekavčić (1980: 36–7), La Fauci (1988: 55–6; 1991: 149), Stanovaïa (1993), Seidl (1995), Formentin (1998: 285–6), Zamboni (1998: 133–7, 139; 2000: 95–6, 108–12), Rovai (2005: 64), Smith (2011), Adams (forthc. e: }2). 21 The nominative origin of forms derived from the lexemes FRAT- ‘brother’, MAT- ‘mother’, and PAT‘father’ is not conclusive in all cases (Rohlfs 1968: 6; Adams forthc. e: }3.3). Alongside a possible nominative etymological base (e.g. FRATE(R) > It. frate or FRAT(E)R > *fratre > Fr. frère), an accusative etymology is also possible, not to say preferable in some cases (e.g. FRATRE(M) > OFr. fradre > Fr. frère).

The rise and fall of alignments

335

Pablos, Pilatos; cfr. also Fr. Eve, Berte (OFr. ACC Evain, Bertain)), and in ItaloRomance in (-ES >) -i (e.g. Cal. Biasi ‘Biagio’ (< BIASES), It. Gi(ov)anni (< IOHANNES)). For a detailed discussion of the retention of the nominative in Italo-Romance masculine palatal plurals in -ci/-gi (e.g. It. amici ‘friends’, greci ‘Greeks’, medici ‘doctors’, monaci ‘monks’, nemici ‘enemies’, porci ‘pigs’), see Maiden (2000). Within a typological perspective, the otherwise exceptional retention of a marked nominative in conjunction with animates clearly points to a previous active/stative alignment in the late Latin/early Romance nominal system,22 in which animate nouns, on account of their high dynamicity and definiteness, were typically encoded as A and SA and consequently often fossilized in their nominative (= active) form in Romance, whereas inanimates, on account of their low dynamicity and indefiniteness, were typically encoded as SO and O and therefore usually fossilized in the accusative (= stative) form in Romance (Zamboni 2000: 114). Exceptionally, in a small number of cases, especially in Gallo-Romance, both the active and stative forms of the same animate noun have survived (Harris 1978: 49), giving rise to lexically and/ or syntactically differentiated doublets (for detailed discussion, see Smith 2011), including *kompanio/*kompanionem > Fr. copain/compagnon ‘(boy)friend/companion’, Frk. *wrakkjo/*wrakkjone ‘vagabond’ > Fr. gars/garçon ‘lad/boy’, HOMO/ HOMINEM ‘man(kind)’ > Cat. hom/home, Fr. on/chomme, Occ. òm/òme ‘one (indefinite NOM only)/man(kind)’, PASTOR/PASTOREM ‘(shep)herd’ > Fr. pâtre/pasteur ‘shepherd/ pastor’, SE(N)IOR/SENIOREM ‘elder’ > Fr. sire/seigneur ‘sire/lord’. 7.3.3 Sentence: word order Above (cf. }3.3, }5.3) we reviewed abundant evidence to demonstrate that, following original Latin SOV and transient late Latin/early Romance XV(X) orders, the Romance languages have broadly converged towards an unmarked SVO word order. With the exception of some modern Gallo-Romance varieties, typified by French (see }7.4.2), this SVO order masks in most modern varieties an active/stative alignment, where S and O are to be understood more broadly as A/SA and O/SO, respectively. This explains why in the unmarked case (answering the question: What happened?) transitive (26a) and unergative (26b) subjects occur preverbally, whereas unaccusative subjects (26c) occur in a postverbal position corresponding to that occupied by the complement in transitive constructions:23 22

Villar (1983: 3–40), La Fauci (1991: 149; 1997: 56–8), Zamboni (1998: 131). For further detailed discussion, see Bentley (2006: 364–8). In literary/high registers of modern French, a relic of this active/stative alignment is still observable in the postverbal position of indefinite SO subjects in construction with a preverbal 3MSG expletive subject il (i.a), unavailable to indefinite transitive or unergative subjects (i.b). 23

(i)

a

il est arrivé une arrived a.F it.M is ‘there arrived a baker’

boulangère (Fr.) baker.F

336 (26)

From Latin to Romance a il treno bloccò la stazione / un desconegut portà la bandera (It./Cat.) the train blocked the station a stranger carried the flag ‘the train blocked the station/a stranger carried the flag’ b il treno fischiava / un desconegut passejava en el the train whistled a stranger walked in the ‘the train was whistling/a stranger was walking in the flat’ c arrivò il treno / arribà arrived the train arrived ‘the train/a stranger arrived’

un a

pis (It./Cat.) flat

desconegut (It./Cat.) stranger

Once again, evidence like this, with early attestations in (late) Latin (Cennamo 2001a: 15–16) as already discussed in }5.3.2, points to an active/stative orientation at the level of the sentence to parallel at every level the identical orientations examined above in the verbal domain (auxiliary selection, participle agreement) and nominal domain (restricted nominative for A/SA and extended accusative for OS). 7.3.4 Other patterns Finally, we briefly review how the active/stative alignment exemplified above in relation to the nominal and verbal systems and the sentence finds further support in a number of other phenomena which systematically distinguish between A/SA and O/SO. Within the nominal system there is some evidence to suggest that in the relative pronoun systems of many early Italo-Romance varieties an original nominative/accusative distinction QUI ‘who’ vs QUEM/QUOD ‘whom/which’ > qui/chi vs que/che was realigned according to an active/stative orientation.24 In these varieties, headed subject relatives are typically marked by qui when the antecedent is high in the animacy hierarchy, typically human and dynamic (27a), but marked by que when the antecedent is low in the animacy hierarchy, typically non-human and stative (27b). However, even with human antecedents, the original accusative que form is encountered when the verb denotes a non-controlled event or state with a non-agentive subject. This situation typically obtains with unaccusatives (27c), but, under the appropriate circumstances, can also involve unergatives and transitives as in the Neapolitan and Piedmontese

b

**il a chanté une boulangère it.M has sung a.F baker.F ‘there has sung a baker (a song)’

(une a

chanson) (Fr.) song

24

Formentin (1996), Parry (2007), Ledgeway (2009a: 963–7).

The rise and fall of alignments

337

examples (27d–e), where the acts of urinating and carrying an unborn child, respectively, can readily be construed as involving a non-controlled event with a nonagentive subject. (27)

a a quela santa inperarixe chi de lo mundo è guiarixe to that holy empress who of the world is guide (OGen., Anon. gen 98.31–2) ‘to that holy empress who leads the world’ b Aquesta cità que avea num this city which had name ‘this city which was called Jericho’

Iherico (OPie., Serm. sub. 246.12–13) Jericho

c quilli ke sono andai (OLmb., PSPDI 28.30-31) those who are gone ‘those who have gone’ d chillo che piscia raro (ONap., Bagni 382) that.one who pisses rarely ‘he who urinates rarely’ e som quella che lo portay (OPie., SCSG 33.18) I.am that.one who him= I.carried ‘I am the one who carried him (in my womb)’ In short, the qui/que alternation in the relative system of these early Italo-Romance varieties serves, not so much to mark a subject/object distinction, as is arguably the case with French qui/que, but, rather, to distinguish A/SA from O/SO (and even A from O/S according to an ergative/absolutive alignment). Within the verb system, there are further reflexes of the active/stative alignment in, for example, the distribution of the absolute participle. In this construction, the accompanying DP must function as the direct object of a transitive participle (28a–b) or the subject of an unaccusative (29a–b), but not the subject of an unergative (30a–b) or or transitive (31a–b) participle, thereby aligning once again O with SO and A with SA.25 (28)

a La lettre écrite, Jean sabla le champagne (Fr.) the.FSG letter.F written.FSG Jean sanded the champagne ‘Having written the letter, Jean opened the champagne’ b Lo castèl pres, la resisténcia de las Corbièras tombèt (Lgd.) the.MSG castle.M taken.MSG the resistance of the Corbières fell ‘After having taken the castle, the Corbières’ defence fell’

25 Though for a more nuanced interpretation of the evidence, see Loporcaro ([2003] 2004) and Bentley (2006: 323–61).

338

From Latin to Romance

(29) a Ses parents partis, Jean s’ est mis à organiser la fête (Fr.) his parents.M left.MPL Jean self= is put to organise.INF the party ‘His parents having left, Jean began to organize the party’ b Los amics tornats, soi partit lo the.MPL friends.M returned.MPL I.am left him= ‘Once my friends came back, I went looking for him’ (30) a **Les amis causé(s), ils se the.PL friends.M chatted.M(PL) they selves= ‘The friends having chatted, they kissed’

sont are

quèrre (Lgd.) search.INF embrassés (Fr.) kissed

b **Los malausses dormit(s), la miá maire se n’ es anada (Lgd.) the.MPL sick.MPL slept.M(PL) the my mother self= therefrom= is gone ‘The patients having slept, my mum left’ (31)

a **Les collègues préparé(s) le dîner, ils the colleagues.M prepared.M(PL) the dinner they sont mis à manger (Fr.) are put to eat.INF ‘The colleagues having prepared dinner, they began to eat’

se selves=

b **Los vesins traversat(s) lo vilatge, an plan manjat (Lgd.) the.MPL neighbours.M crossed.M(PL) the village they.have much eaten ‘After the neighbours walked across the village, they ate a hearty meal’ An identical split in the verb system can be seen in the central Marchigiano dialect of Sassoferratese studied by Peverini (2008). In addition to the widely attested active/stative alignment in the compound perfective paradigms where HABERE aligns with A/SA and ESSE with SO, simplex lexical verbs also display an identical alignment in the 3rd person. Typically, in the 3rd person lexical verbs do not mark a formal number distinction between singular and plural, where a form corresponding to the singular in most other Romance varieties marks both numbers (32a–b'). However, when the subject is left-dislocated, as is clearly shown when it precedes another topic or a focused phrase, full agreement in the 3PL becomes possible (though not obligatory) with unaccusatives (32c), but not with unergatives or transitives (32d):26 (32)

26

SIBI

a Marco magna siempre Marco eat.3SG always ‘Marco always eats peas’

i the

biseje (Sas.) peas

For a further example of the active/stative alignment in late Latin pleonastic reflexives involving a SE/ alternation, see Cennamo (1999).

The rise and fall of alignments a' Marco e Checco magna / **magnano siempre i Marco and Checco eat.3SG / eat.3PL always the ‘Marco and Checco always eat peas’

339

biseje (Sas.) peas

b arvenia Marco (Sas.) returned.3SG Marco ‘Marco was coming back’ b' arvenia / **arveniano Marco returned.3SG / returned.3PL Marco ‘Marco and Checco were coming back’

e and

Checco (Sas.) Checco

c Marco e Checco DA LA SCOLA arvenia / arveniano (Sas.) Marco and Checco from the school returned.3SG / returned.3PL ‘Marco and Checco were coming back from school’ d Marco e Checco dae nonnesi magna Marco and Checco at grandparents=their eat.3SG / **magnano siempre (i biseje) (Sas.) always the peas / eat.3PL ‘Marco and Checco, at their grandparents’ house, always eat (peas)’ Finally, at the level of the sentence the active/stative orientation surfaces in the distribution of INDE-cliticization (33; cf. Burzio 1986),27 bare plural NPs (34; Alsina 1996: 105–8; Zagona 2002: 21–2), and the agreement of adjectives in adverbial function (35; Ledgeway 2000: 273–4; 2003b: 117–19; forthc. d; Maturi 2002: 165). In all three cases, the licensing of the phenomenon in question is exclusively linked to the complement position, thus grouping together O and SO and contrasting these with A and SA. Consequently, only transitive objects (cf. (a) examples) and unaccusative subjects (cf. (b) examples), but not unergative/transitive subjects (cf. (c) examples), can be cliticized by INDE ‘of it/them’ (33), surface as bare plural NPs (34), and license adverbial agreement (35): (33)

he vist tres [en] (Cat.) a He vist tres [NP nois] ) n’ I.have seen three boys of.them= I.have seen three ‘I’ve seen three boys’ ) ‘I’ve seen three of them’ han vingut tres [en] (Cat.) b han vingut tres [NP nois] ) n’ have come three boys of.them= have come three ‘three boys came’ ) ‘three of them came’

27

For an alternative view, see Bentley (2006: ch. 6).

340

From Latin to Romance c tres [NP nois] menjan (tomàquets) ) **tres [en] en three boys eat (tomatoes) three of.them= menjan (tomàquets) (Cat.) eat tomatoes ‘Three boys are eating (tomatoes)’ ) ‘three of them are eating (tomatoes)’

(34)

a han visto animales (Sp.) they.have seen animals ‘they have seen animals’ b han muerto animales (Sp.) have died animals ‘animals have died’ c **animales han comido (la hierba) (Sp.) animals have eaten the grass ‘animals have been eating (the grass)’

(35)

a ci avia cunzatu her.DAT= I.had prepared ‘I had styled her hair well’

bbuoni good.MPL

i the.PL

capiddri (Cos.) hair.MPL

b si eranu priperati bbuoni (Cos.) selves= they.were prepared.MPL good.MPL ‘They had prepared themselves well’ c **Maria ha studiatu bbona (ssi Maria has studied good.FSG these ‘Maria has been studying (these facts) well’

fatti) (Cos.) facts

7.4 Innovative Romance: nominative/accusative syntax In the preceding discussion we have seen how in the transition from Latin to Romance there arose a number of new active/stative orientations replacing earlier nominative/accusative orientations. These, in turn, were countered in many cases by the emergence of new nominative/accusative alignments which, although clearly morphosyntactic innovations, recreate the conservative typological orientations of Latin, highlighting an ongoing, yet unresolved, conflict between opposing and incoherent argument-marking orientations (La Fauci 1988: 59–60; Zamboni 1998: 130). Although many of these new nominative/accusative orientations are often reported to affect predominantly the southern Romània (Zamboni 2000: 128), ostensibly on account of the subsequent changes in the systems of auxiliary selection and participle agreement, they are, as we shall presently see, just as prevalent in the northern Romània.

The rise and fall of alignments

341

7.4.1 Verbal group 7.4.1.1 Perfective auxiliary constructions Beginning with the verb system, the most notable development involves a tendency towards the gradual loss of the original active/stative alignment in the perfective auxiliary system in favour of generalized HABERE (/TENERE) ‘have’ in Ibero-Romance (36a–d), numerous southern Italian varieties (37a–c), and Dalmatian (38):28 (36) a hubiéremos bebío / we.would.have drunk ‘we would have drunk/come’

venío (Ast.) come

b ja havia fumat / arribat (Cat.) already I.had smoked arrived ‘I had already smoked/arrived’ c tinhas chorado / estado you.had cried been ‘you had cried/been ill’

doente (Pt.) ill

d he comido / vuelto (Sp.) I.have eaten returned ‘I have eaten/returned’ (37)

a dʤuwann ‘Giovanni

a has

kkɔtt cooked

b tu e manʤetə you have eaten ‘you ate/came’

a the

mənɛstrə soup

/ marí Maria

a mmɔrt (Tbs.) has died’

/ vənotə (Irs.) come

c avìanu manciatu / nisciutu (Sic.) they.had eaten gone.out ‘they had eaten/gone out’ (38) i ju insegnut / venájt (Vgl.) they have taught come ‘they have taught/come’ Of course, there are many exceptions to this broad areal generalization including, in the northern Romània, the extension of HABERE to (most) unaccusative predicates in Jèrriais, Acadian, and Canadian French (39a),29 and Venetian (39b; Marcato and Ursini 1998: 251–4). 28 Benzing (1931), Huber (1933: 221), England (1982), La Fauci (1992), Formentin (2001), Ledgeway (2003a), Stolova (2006), Loporcaro (2007: 178–80). 29 Haden (1973: 431), Canale, Mougeon, and Belanger (1978), Sankoff and Thibault (1980), Jones (2001: 109–10.).

342

From Latin to Romance

(39) a il a mny (Aca.) he has come ‘he has come’ b el ga invecià he has aged ‘he has aged a lot’

tanto (Ven.) so.much

Here the perfective auxiliary therefore no longer encodes the A/SA vs SO split of previous stages, but simply marks the agreement features of the nominative subject, be it A, SA, or SO. To conclude, we must finally consider three further dimensions of variation which may replace or overlay the original active/stative distributional splits in perfective auxiliation examined in }7.3.1.1.2. The first concerns tense, and is found in a number of Campanian varieties (Como 1999: }3.2; Ledgeway 2000: 201–6; 2009a: 624–6) such as Procidano (Parascandola 1976: 108) and San Leucio del Sannio (Iannace 1983: }119). In the former there operates a canonical (reflexive) transitive/unergative vs (reflexive) unaccusative split with the selection of HABERE with the former (40a–b) and ESSE with the latter (40c–d). (40) a hó mangéto I.have eaten ‘I had a bite’

nu a

muórzo (Prd.) bite

b t’ hé fètto yourself= you.have done ‘Did you take a bath?’

lu the

vègno? (Prd.) bath

c sóngo jut’ acco sòcrema (Prd.) I.am gone at mother-in-law=my ‘I went to my mother-in-law’s’ d me sóngo myself= I.am ‘I awoke’

scetéto (Prd.) awaken

In the dialect of San Leucio del Sannio, by contrast, HABERE has generalized in the present perfect to all verb classes (except the passive) and all grammatical persons, witness the following representative examples: (41) a eggio fatto tutto I.have done all ‘I’ve done all I could’

quello that

ch’ which

b iti venuto priesto? (S.Leu) you.have come early ‘did you arrive early?’

eggio I.have

pututo (S.Leu.) been.able

The rise and fall of alignments c nun ino not they.have ‘it wasn’t them’

state been

343

loro (S.Leu.) they

Outside of the present perfect, however, the distributional patterns observed for Procidano and the dialect of San Leucio del Sannio are both replaced by the generalization of ESSE with all verb classes and in all grammatical persons: (42) a jé fóvo ritto ca te fuve puósto cu Nannina I was said that yourself= you.were placed with Nannina ca fóve fètto sciàrra cu Satóre ca se fóve ’mbarchéto (Prd.) that was done quarrel with Satore that self= was embarked ‘I had said that you had got engaged with Nannina who had quarrelled with Salvatore who had joined the navy’ b cómme nen lu fusse as not him= I.were ‘as if I hadn’t seen him’

visto (Prd.) seen

c me fóv’ addunéto ca me myself= I.was realized that me ‘I had realized that he was tricking me’

fanziàva (Prd.) tricked

(43) a erem’ auta dice quello che we.were had.to say.INF that which ‘we had to say what they were saying’

diceveno they.said

loro (S.Leu.) they

b illu era venuto priesto (S.Leu.) he was come early ‘he had arrived early’ c s’ era truatu nu bellu self= he.was found a nice ‘he had found himself a nice spot’

postu (S.Leu.) place

This generalization of ESSE to all verb classes and all grammatical persons outside of the present perfect is not limited to these Campanian varieties, but is also widely attested in many dialects which otherwise show a person-based alternation in the present perfect. In particular, in those central and southern dialects reported to display person-driven auxiliary splits, such splits typically only operate in the present perfect, but not in other paradigms such as the pluperfect indicative/subjunctive and the conditional perfect where one of the two auxiliaries is generalized.30 For example, in the eastern Abruzzese dialect of Arielli (D’Alessandro and Roberts 2010) there is a classic 1/2 (HABERE) vs 30

Ledgeway (2000: 204-5), Manzini and Savoia (2005 II: 681–729), D’Alessandro and Ledgeway (2010b: 208–9).

344

From Latin to Romance

3 (ESSE) split in the present perfect (44a), whereas in the conditional perfect, for example, HABERE alone is found (44b). Analogously, in the north-eastern Abruzzese dialect of Bellante (Manzini and Savoia 2005 II: 686) we find an identical person alternation in the present perfect (45a), whereas in other compound paradigms such as the pluperfect (45b) it is the ESSE auxiliary which has generalized. (44)

a so / si /a viste I.am you.are have.3 seen ‘I/you/(s)he have/has seen/come’

/ menute (Arl.) come

b avesse / avisse / avesse I.would.have you.would.have he.would.have ‘I/you/he would have seen/come’ (45) a sɔ /i /a I.am you.are he.has ‘I/you/he slept/came’ b ɣɛrə / ɣirə / ɣɛrə I.was you.were he.was ‘I/you/he had slept/come’

dərmətə slept dərmətə slept

viste seen

/ menute (Arl.) come

/ mənitə (Blt.) come / mənitə (Blt.) come

The second dimension of variation that we must consider concerns modal conditioning. In a number of early dialects of southern Italy (Ledgeway 2000: 301 n. 22; 2003a; 2009a: 600–14; Formentin 2001), as well as in old Spanish (Stolova 2006), there operates a traditional active/stative split in the system of perfective auxiliation, whereby transitives/unergatives (whether reflexive or otherwise) typically align with HABERE and unaccusatives (whether reflexive or otherwise) with ESSE (46a–c). Nonetheless, the same texts show a progressive extension of the active auxiliary HABERE to unaccusative syntax, replacing in part the traditional stative auxiliary ESSE. However, the replacement is not random but shows a gradual encroachment of HABERE on ESSE in accordance with a sensitivity to a realis/irrealis modal distinction (for a discussion of similar Germanic facts, see McFadden and Alexiadou 2006a,b,c; 2010: 394–99). In particular, the initial spread of HABERE with unaccusatives in early texts appears quite consistently to affect only those clauses marked as [+irrealis], typically containing a verb in the subjunctive or conditional (46a'–c').31

31 The role of irrealis modality is also apparent in a number of modern Romance varieties such as Andalusian and Latin American Spanish and some Calabrian dialects (Ledgeway 2000: 205–6). Although these varieties have since otherwise generalized HABERE to all verb classes (i.a–b), they may exceptionally employ auxiliary ESSE ‘be’ in past counterfactuals (ii.a–c):

The rise and fall of alignments (46)

345

a erano fuyute a li templi (ONap., LDT 74.27–8) they.were fled to the temples ‘they had run to the temples’ a' ben che avessero puro foyuto altough they.had.sbjv even fled de via (ONap., LDT 238.29–30) of away ‘although they had run far away’

per for

b li pili ià li the hairs already to.him= ‘he had already lost his hair’

caduti (OSic., LDSG 117.25.2) fallen

b' si killa dirrupa avissi if that cliff had.SBJV ‘if that cliff had collapsed’

(i)

a

(ii)

a

eranu were

luongo long

spacio space

caduta (OSic., LDSG 178.16.25) fallen

han llegado / muerto / llorado (Sp.) they.have arrived died cried ‘they have arrived / died / cried’ statu fora / jutu / accattatu (S.And.) b àiu I.have been outside gone bought ‘I have been away / gone / bought’ si fueras hecho algo desde que if you.were.SBJV done something since that ‘if you had done something since arriving in Cordoba’

¡No actues como si fueras not you.do.SBJV as if you.were.SBJV ‘Don’t act as if you had done something wrong!’ c si fussa statu / jutu (S.And.) if I.were been gone ‘If I had been / gone’ b

hecho done

llegastes you.arrived

a to

Córdoba (And.) Cordoba

algo something

malo! (Mex.) bad

A different reflex of irrealis modality is observable in the compound pluperfect indicative forms of many northern Sardinian dialects (Pisano 2010), where a doubly compound form consisting in the reduplication of the auxiliary in its participial form is exceptionally licensed in counterfactual hypothetical (iii.a) and optative (iii.b) clauses: si bːi ˈviaz iɬˈtaðu ˈvenːiðu ˈɖue nom bi ˈvia iɬˈtaðu aɲˈɖaðu (Car.) if there= you.were been come you not there= I.was been gone ‘if you had come, then I wouldn’t have gone there’ b no aˈi ˈapːiu mai ˈkretːiu k ˈɛre ˈrːikːu (Nuo.) not I.had had never believed that he.was rich ‘I would never have guessed that he was so rich!’

a







(iii)

346

From Latin to Romance c Si el sieruo que es fuydo mora mucho en if the servant that is fled remains much in casa de algun omne (OSp., FJ) house of some man ‘if the servant who has fled stays a long time in some man’s house’ c' si ladrones que furtan de dia & de if thieves that steal of day and of ouissen entrado (OSp., GE IV) had.SBJV entered ‘if thieves who steal by day and night had got in’

noche night

Finally, we need to mention Romanian, a language which by the time of its earliest textual attestations (1521) had already lost the active/stative split in perfective auxiliation otherwise generally attested in other Romance varieties (Loporcaro 2007: 180). In its place, Romanian shows a robust alternation of HABERE and fi ‘be’ (< FIERI ‘to become’), the local Daco-Romance variant of ESSE, in accordance with a finiteness split (Alboiu and Motapanyane 2000b: 16). More specifically, HABERE is restricted to the finite, inflected forms of the present perfect (47), whereas fi ‘be’ surfaces in all remaining non-finite contexts (48a–c). (47)

am / ai /a / am / at¸i / au mâncat / plecat (Ro.) I.have you.have (s)he.has we.have you.have they.have eaten left ‘I/you/(s)he/we/you/they have(/has) eaten/left’

(48)

a înainte de a fi mâncat / plecat citeam ziarul (Ro.) before of to be.INF eaten left I.read newspaper=the ‘before having eaten/left, I was reading the newspaper’ b fîind mâncat / plecat, am citit be.GER eaten left I.have read ‘having eaten/left, I read the newspaper’

ziarul (Ro.) newspaper=the

c vor / ar fi they.will / they.would be.INF ‘they will/would have eaten/left’

/ plecat (Ro.) left

mâncat eaten

Particularly revealing in this respect is the case of the subjunctive: uniquely for Romanian, the verb fi ‘be’ presents distinct (present) subjunctive forms for all six persons (albeit with syncretism in the 3rd person), as shown in (49a). Nonetheless, when the present perfect of the subjunctive is employed, it is not these highly inflected forms that are selected (nor inflected forms of HABERE, which

The rise and fall of alignments

347

we might have legitimately expected in such a relatively finite context), but, rather, a non-inflected form of ‘be’, namely fi, for all six grammatical person-number combinations (49b). (49)

a nu cred să fiu / fii / fie / fim not they.believe that.SBJV I.be.SBJV you.be.SBJV (S)he.be.SBJV we.be.SBJV /fit¸i / fie aici (Ro.) you.be.SBJV they.be.SBJV here ‘they don’t believe that I am/you are/(s)he is/we are/you are/they are here’ b nu cred să fi mâncat / plecat (Ro.) be eaten left not they.believe that.SBJV ‘they don’t believe that I/you/(s)he/we/you/they have(/has) eaten/left’

The inflectional contrast witnessed in (49a–b) is highly significant, insofar as it underlines the key role played by m(orphological)-finiteness (cf. Ledgeway 2007c) in determining the distribution of the Romanian auxiliaries. Indeed, this view is further supported by those substandard varieties of Romanian which, alongside of (or in place of) the standard synthetic pluperfect formation (e.g. mâncasem/ plecasem ‘I had eaten/left’), display an analytic surcomposé-style paradigm. In the latter there occur two auxiliaries, one inflected and the other in participial form, which are both, in turn, followed by the lexical participle. Given what we have seen above, it will then not come as a surprise to learn that the first inflected auxiliary surfaces as HABERE and the second non-finite auxiliary is spelt out as a form of ‘be’ (namely, fost ‘been’). (50) am / ai /a / am / at¸i I.have you.have (s)he.has we.have you.have mâncat / plecat (coll. Ro.) eaten left ‘I/you/(s)he/we/you/they had eaten/left’

/ au they.have

fost been

In conclusion, we see that the series of Romance auxiliary alternations determined by tense, modality, and finiteness reviewed above underlie a (partial) reversal in the earlier active/stative alignment of the verbal system, itself once an innovation, in favour of a new nominative/accusative alignment. In this new alignment the auxiliary system no longer formally marks an opposition between AGENT and UNDERGOER subjects, but simply indexes the nominative argument at the expense of the accusative argument, irrespective of its AGENT or UNDERGOER status. 7.4.1.2 Participle agreement In those same varieties which have long abandoned the original active/stative split in favour of a nominative/accusative pattern that privileges the generalization of a single auxiliary, typically (but not exclusively) HABERE,

348

From Latin to Romance

participle agreement too has generally seen a concomitant attrition (though for the various exceptions, see Loporcaro 1998: 8–12), such that in modern Portuguese, Spanish, and Sicilian, for instance, agreement has been lost in all cases except with the passive participle: (51)

a la manzana la the.FSG apple.FSG it.F= ‘I had eaten the apple’

había I.had

comido/**-a (Sp.) eaten.MSG/FSG

b Ana ha vuelto/**-a (Sp.) Ana has returned.MSG/FSG ‘Ana has returned’ c la carta fue escrita/**-o (Sp.) the.FSG letter.F was written.FSG/MSG ‘the letter was written’ The previous pattern which licensed stative-oriented agreement controlled by O and SO, as still preserved to varying degrees in modern French, Italian, and Occitan (cf. 20a–c), is thus lost. Significant in this respect is the situation found in the modern standard Catalan of Barcelona, where auxiliary ESSE has been lost from the active paradigms as in Portuguese and Spanish, but participial agreement has been residually preserved in conjunction with (typically feminine) third-person accusative clitics:32 (52)

a l’ Alícia ja ha pujat(**a) (Bar.) the Alícia already has ascended.MSG(FSG) ‘Alícia has already gone up’ b aquells llibres ja els those.MPL books.M already them.MPL= ‘those books, I’ve already read them’ c aquella revista ja l’ that.FSG magazine.F already it.F= ‘that magazine, I’ve already read it’

he I.have

he I.have

d aquelles revistes ja les magazines.F already them.FPL= those.F ‘those magazines, I’ve already read them’

llegit(s) (Bar.) read.M(PL)

llegida (Bar.) read.FSG he I.have

llegides (Bar.) read.FPL

As La Fauci (1988: 102–3) perceptively observes, what we have here is not a residue of the former active/stative alignment in the participle, but, rather, a new, albeit rather restricted, nominative/accusative alignment. In particular, the participle now simply marks certain types of O (cf. 52b–d), but never A or S (cf. 52a). 32

Solà (1993: 73–86), Smith (1995), Moll (1997: 135), Loporcaro (1998: 149–53).

The rise and fall of alignments

349

An even more radical and apparently unique development in participle agreement is highlighted by D’Alessandro and Roberts (2005; 2010) for the eastern Abruzzese dialect of Arielli. In this dialect, the auxiliary system in the present perfect operates on a classic person split (ESSE: 1/2 vs HABERE: 3)—itself a nominative/accusative orientation replacing the earlier active/stative alignment—but the participle system displays neither an active/stative nor a nominative/accusative split. Rather, the active participle simply shows participial agreement by way of metaphonetic (viz. tonic vowel) alternation with any plural nominal, be it the subject (53a) or the object (53b): (53)

a seme magnite / **magnate we.are eaten.MPL / eaten.SG ‘we have eaten the biscuit’

lu the.M

b si magnite / **magnate you.are.SG eaten.MPL / eaten.SG ‘you have eaten the biscuits’

biscotte (Arl.) biscuit.M

li the.MPL

biscutte (Arl.) biscuits.M

7.4.2 Sentence: word order Moving now finally to the sentence, here too there are significant signs of new nominative/accusative alignments. Above in }7.3.3 we noted how in the southern Romània word order typically conveys an active/stative alignment that places A and SA in preverbal position (54a–b) and O and SO in postverbal position (54a,c): (54) a yo he minchau asabelas I have eaten many ‘I’ve eaten lots of meatballs’ b yo no en I not of.it= ‘I haven’t slept’

he have

dormiu slept

almondigas (Ara.) meatballs mica (Ara.) not

c en a chaminera creman troncos in the fireplace burn.3PL trunks ‘oak logs are burning in the fireplace’

de of

caxigo (Ara.) oak

In many Romance varieties of the northern Romània, by contrast, this differential word order has largely been replaced by a generalized SV(O) order, in which all subjects, including those of the SO type, now obligatorily occur in preverbal position (La Fauci 1988: 57–8; 1997: 29; Vincent 1988: 62), as illustrated in the French examples in (55a–b). This grammaticalization of the preverbal position as the dedicated ‘subject’ position, originally arising from the reanalysis of frequently fronted subjects

350

From Latin to Romance

within a V2 syntax (cf. }3.3.2, }4.4.1, }5.4.2.4), eventually leads to the reanalysis of preverbal subject pronouns as (obligatory) subject clitics (cf. }6.3.2) and the apparent reversal in the pro-drop parameter (55c). (55)

a la ‘the

marine navy

b le ‘the

navire ship

c il ‘it

/** /

a has

a has a has

coulé sunk

le the

navire (Fr.) ship’

coulé (Fr.) sunk’

coulé (Fr.) sunk’

There thus arises in the northern Romània an unresolved conflict in the markedness status of O(bjects). Postverbal O(bjects) can be considered marked in terms of a nominative/accusative word order typology which places O alone after the verb, but are to be considered unmarked in terms of the active/stative orientation of the perfective verb system, which formally identifies O with SO in terms of auxiliary selection and participle agreement (La Fauci 1988: 58). Consequently, the progressive restriction in the distribution of auxiliary ESSE and participle agreement, as witnessed in the recent history of French, can now be viewed as the gradual dominance of a nominative/accusative orientation over an earlier active/stative orientation. In contrast to this northern Romance system of subject marking, varieties from the southern Romània (including southern Occitan varieties) preserve the original active/stative word order, but tend to overlay this earlier alignment with an innovative (but typologically conservative) nominative/accusative alignment through a system of prepositional object marking (La Fauci 1997: 55–6; Zamboni 1998: 130).33 In particular, highly particularized animate participants, which otherwise typically map onto the subject relation (namely, A and SA), are differentially marked by a reflex of PER ‘through’ in Romanian (> pe ‘on’) (56a) and elsewhere by AD ‘at, to’ (56b–e) whenever they assume the O(bject) function. In short, southern Romance systematically contrasts O, whenever it is a potential candidate for subject status, with A and SA/O in accordance with a classic accusative/nominative orientation.

Cf. }6.2, }6.3.1. See also Meyer-Lübke (1899: }50), Kalepky (1913), Reichenkron (1951), Meier (1948), Rohlfs (1969: }}632, 639; 1971), Diaconescu (1970), Martín Zorraquino (1976), Villar (1983), Nocentini (1985), Pensado (1985; 1995), Stimm (1986), Green (1988a: 106), Bossong (1991), Zamboni (1992), Jones (1993: 65–8; 1995), Trumper (1996: 354–5), Sornicola (1997; 2011: 35–42), Vincent (1997b: 209), Torrego (1998; 1999), Ledgeway (2000: 20–1; 2009a: 831–42), Fiorentino (2003). 33

The rise and fall of alignments (56)

a l-am văzut him=I.have seen ‘I have seen him’

pe on

el (Ro.) him

b les monges no estimen the nuns not like ‘the nuns don’t like the girls’ c l’ aimi a him= I.love to ‘I love my father’

mon my

351

a to

les the

nenes (Cat.) girls

paire (Pyrenean Occ.) father

d Pedro coroou rainha à sua amante Pedro crowned queen to.the his mistress ‘Pedro crowned his dead mistress queen’

morta (Pt.) dead

e chiamàu a Micheli (SCal.) he.called to Micheli ‘he called Micheli’ Together with the generalization of a single perfective auxiliary and the loss of participle agreement, the rise of the prepositional accusative in southern Romània therefore represents a consistent typological development towards an ever expanding nominative/accusative-oriented syntax at the expense of an ever dwindling active/ stative syntax (La Fauci 1988: 60–3).

7.5 Concluding remarks The preceding discussion has demonstrated that in the passage from Latin to Romance there has been, to some extent, a largely still unresolved oscillation between two coexistent morphosyntactic alignments. Starting from a largely nominative/accusative orientation, most robustly attested in the nominal system, there is considerable evidence to suggest that Latin and subsequently Romance increasingly developed a series of novel active/stative splits at the level of the verbal group, the sentence, and ultimately the nominal group. Though no modern Romance variety is today entirely without any overt reflexes of this active/stative opposition in some part of its grammar, many (and, in particular, those of the southern Romània) have since lost a number of reflexes of the original active/stative alignment. In their place have emerged a number of novel nominative/accusative orientations which, although clearly morphosyntactic innovations, effectively constitute a return to the conservative typological orientations of Latin, highlighting an ongoing tension

352

From Latin to Romance

between two opposing and incoherent argument-marking orientations. This gives the deceptive impression of Latin-Romance continuity, whereas at bottom, as we have seen, this is only an apparent case of uninterrupted Latin-Romance continuity which masks an intermediate stage, only in part inferable from documented sources, of an active/stative orientation.

References Abeillé, Anne, and Godard, Danièle (1999). ‘La position de l’adjectif épithète en français: Le poid des mots’, Recherches linguistiques de Vincennes 28: 9–32. ——--——--(2003). ‘Les prédicats complexes dans les langues romanes’ in Danièle Godard (ed.), Les langues romanes: Problèmes de la phrase simple. Paris: Éditions de la CNRS, 125–84. Abels, Klaus (2003). Successive Cyclicity, Anti-locality, and Adposition Stranding. University of Connecticut, Storrs: unpublished thesis. Adams, James (1971). ‘A Type of Hyperbaton in Latin Prose’, Proceedings of the Cambridge Philological Society 197: 1–16. ——--(1976a). ‘A Typological Approach to Latin Word Order’, Indogermanische Forschungen 81: 70–99. ——--(1976b). The Text and Language of a Vulgar Latin Chronicle (Anonymus Valesianus II), Bulletin Supplement No 36. University of London: Institute of Classical Studies. ——--(1977). The Vulgar Latin of the Letters of Claudius Terentianus. Manchester: Manchester University Press. ——--(1991). ‘Some Neglected Evidence for Latin HABEO with Infinitive: The Order of the Constituents’, Transactions of the Philological Society 89: 131–96. ——--(1992). ‘British Latin: Notes on the Language, Text and Interpretation of the Bath Curse Tablets’, Britannia 23: 1–26. ——--(1994a). ‘Wackernagel’s Law and the Position of Unstressed Pronouns in Classical Latin’, Transactions of the Philological Society 92: 103–78. ——--(1994b). Wackernagel’s Law and the Placement of the Copula Esse in Classical Latin. Cambridge: The Cambridge Philological Society. ——--(2003). Bilingualism and the Latin Language. Cambridge: Cambridge University Press. ——--(2005). ‘The Accusative + Infinitive and Dependent Quod-/Quia-clauses: The Evidence of Non-literary Latin and Petronius’, in Sander Kiss, Luca Mondin, and Giampaolo Salvi (eds), Latin et langues romanes: Études de linguistique offertes à József Herman à l’occasion de son 80ème anniversaire. Tübingen: Niemeyer, 195–206. ——--(2007). The Regional Diversification of Latin 200 BC – AD 600. Cambridge: Cambridge University Press. ——--(forthc. a). ‘Late Latin’, in James Clackson (ed.), A Companion to the Latin Language. Oxford: Blackwell. ——--(forthc. b). ‘The Definite Article’, in Social Variation and the Latin Language. Cambridge: Cambridge University Press. ——--(forthc. c). ‘Hiatus’, in Social Variation and the Latin Language. Cambridge: Cambridge University Press. ——--(forthc. d). ‘Habeo-constructions’, in Social Variation and the Latin Language. Cambridge: Cambridge University Press.

354

References

Adams, James (forthc. e). ‘The Nominative and Accusative’, in Social Variation and the Latin Language. Cambridge: Cambridge University Press. ——--(forthc. f). ‘Word Order’, in Social Variation and the Latin Language. Cambridge: Cambridge University Press. Adams, Marianne (1987). Old French, Null Subjects, and Verb Second Phenomena. UCLA: unpublished thesis. Aebischer, Paul (1948). ‘Contribution à la protohistoire des articles ille et ipse dans les langues romanes’, Cultura neolatina 8: 181–203. Agard, Frederick (1958). Structural Sketch of Romanian. Baltimore: Linguistic Society of America. Agbayani, Brian, and Golston, Christ (2010). ‘Phonological Movement in Ancient Greek’, Lingua 86: 133–67. Alarcos Llorach, Emilio (1994). Gramática de la lengua española. Madrid: Espasa. Albano Leoni, Federico, Krosbakken, Eleonara, Sornicola, Rosanna, and Stromboli, Carolina (eds) (2001). Dati empirici e teorie linguistiche. Atti del XXXIII congresso internazionale di studi della Società di linguistica italiana, Napoli, 28–30 ottobre 1999. Rome: Bulzoni. Alboiu, Gabriela, and Motapanyane, Virginia (eds) (2000a). Comparative Studies in Romanian Syntax. Amsterdam: Elsevier. ——--——--(2000b). ‘The Generative Approach to Romanian Grammar. An Overview’, in Alboiu and Motapanyane (eds), 1–48. Alcover, Antoni (1905). ‘L’article “es” i “sa” dins la toponímia catalana’, Bolletí del diccionari de la llengua catalana 2: 347–60. Alexiadou, Artemis, and Anagnostopoulou, Elena (1998). ‘Parametrizing AGR: Word Order, Verb-movement and EPP Checking’, Natural Language and Linguistic Theory 16: 491–539. Alfonsi, Tommaso (1932). Il dialetto còrso nella parlata balanina. Livorno: Giusti. Alfonzetti, Giovanna (1998). ‘Passato prossimo e passato remoto: Dimensioni di variazione’, in Ruffino (ed.), II, 27–37. Alibèrt, Loís (1976). La gramatica occitana. Montpellier: Centre d’Estudis Occitans. Alisova, Tatiana (1967). ‘Studi di sintassi italiana. II. Le posizioni dell’aggettivo nel gruppo sintattico del sostantivo’, Studi di filologia italiana 25: 250–313. ALlA (= Academia de la Llingua Asturiana) (1995). La llingua asturiana. Xixón: Academia de la Llingua Asturiana. ——--(2001). Gramática de la llingua asturiana. 3rd edn. Oviedo: Academia de la Llingua Asturiana. Alkire, Ti, and Rosen, Carol (2010). Romance Languages: A Historical Introduction. Cambridge: Cambridge University Press. Allen, W. Sidney (1964). ‘Transitivity and Possession’, Language 40: 337–43. ——--(1978). Vox Latina: A Guide to the Pronunciation of Classical Latin. Cambridge: Cambridge University Press. Allières, Jacques (2001). Manuel de linguistique romane. Paris: Champion. Alonso, Damoso (1958). ‘Metafonía y neutro de materia en España’, Zeitschrift für romanische Philologie 74: 1–24. Alonso Garrote, Santiago (1947). El dialecto vulgar leonés hablado en Maragatería y tierra de Astorga. Madrid: CSIC.

References

355

Alsina, Alex (1992). ‘On the Argument Structure of Causatives’, Linguistic Inquiry 23: 517–55. ——--(1996). The Role of Argument Structure in Grammar. Stanford: CSLI. Alvar, Manuel (1948). El habla del Campo de Jaca. Salamanca: CSIC. Álvarez, Rosário, Monteagudo, Henrique, and Regueira, Xosé Luís (1986). Gramática galega. Vigo: Galaxia. Álvarez Pedrosa, Juan Antonio (1988). ‘Estudio comparado del orden de palabras en inscripciones jurídicas arcaicas, griegas y latinas’, Revista española de lingüística 18: 109–28. Amacker, René (1989). ‘Sur l’ordre des termes dans la proposition latine’, in Calboli (ed.), 485–501. Ambrosini, Riccardo (1969). ‘Usi e funzioni dei tempi storici nel siciliano antico’, Bollettino del Centro di studi filologici e linguistici siciliani 10: 141–18. Anderson, Eric (1979). ‘The Development of the Romance Future Tense: Morphologization II and a Tendency towards Analysis’, Papers in Romance 1: 21–35. Aranovich, Raúl (2007). ‘Incorporation, Pronominal Arguments, and Configurationality in Fijian’. University of California: unpublished manuscript. Arias Cabal, Álvaro (1999). El morfema de ‘neutro de materia’ en asturiano. Santiago de Compostela: Universidad. Arnaud, François, and Morin, Gabriel (1920). Le langage de la Vallée de Barcelonnette. Paris: Champion. Arnholdt, Karl (1916). Die Stellung des attributiven Adjektivs im Italienischen und Spanischen. Greifswald: Von Bruncken. Arquint, Jachen (1964). Vierv ladin. Tusan: Lia rumantscha. Ascoli, Graziadio Isaia (1873). ‘Saggi ladini’, Archivio glottologico italiano 1: 1–556. Ashdowne, Richard (2004). ‘Re(de)fining Address: An Overlooked French Phenomenon’, Oxford University Working Papers in Linguistics, Philology and Phonetics 9: 1–35. Atanasov, Petar (2002). Meglenoromâna astăzi. Bucharest: Editura Academiei. Austin, Peter (2001). ‘Word Order in a Free Word Order Language: The Case of Jiwarli’, in Jane Simpson, David Nash, Mary Laughren, Peter Austin, and Barry Alpher (eds), Forty Years On: Ken Hale and Australian Languages. Canberra: Pacific Linguistics, 305–24. ——--and Bresnan, Joan (1996). ‘Non-configurationality in Australian Aboriginal Languages’, Natural Language and Linguistic Theory 14: 215–68. Authier, Jean-Marc (1992). ‘Iterated CPs and Embedded Topicalization’, Linguistic Inquiry 23: 329–36. Avolio, Francesco (1995). Bommèsprə: Profilo linguistico dell’Italia centro-meridionale. San Severo: Gerni. ——--(1996). ‘Il “neutro di materia” nei dialetti centro-meridionali: Fonti, dati recenti, problemi aperti’, Contributi di filologia dell’Italia mediana 10: 291–337. Avram, Laris, and Hill, Virginia (2007). ‘An Irrealis BE Auxiliary in Romanian’, in Raúl Aranovich (ed.), Split Auxiliary Systems: A Cross-linguistic Perspective. Amsterdam/ Philadelphia: Benjamins, 47–64. Ayres-Bennett, Wendy (1996). A History of the French Language through Texts. London: Routledge. ——-- and Carruthers, Janice (2001). Problems and Perspectives: Studies in the Modern French Language. London: Longman.

356

References

Azevedo, Milton (2005). Portuguese: A Linguistic Introduction. Cambridge: Cambridge University Press. Bachmann, Iris (2008). ‘Norm and Variation in Brazilian TV: The Case of Hidden Clitics’, paper presented at XXXVI Romance Linguistics Seminar, Trinity Hall, Cambridge. ——--(2011). ‘Norm and Variation on Brazilian TV Evening News Programmes: The Case of Third-person Direct-object Anaphoric Reference’, Bulletin of Hispanic Studies 88: 1–20. ——--(forthc.) ‘Romance Creoles’, in Maiden, Smith, and Ledgeway (eds). Bader, Françoise (1962). La formation des composés nominaux du latin. Paris: Les Belles Lettres. Badia i Margarit, Antonio (1951). Gramática histórica catalana. Barcelona: Noguer. ——--(1962). Gramática catalana (2 vols). Madrid: Gredos. ——--(1991). ‘Le catalan. Evolución lingüística interna I. Grámatica’, in Holtus, Metzeltin, and Schmitt (eds), 127–52. ——--(1995). Gramàtica de la llengua catalana: Descriptiva, normativa, diatòpica, diastràtica. Barcelona: Biblioteca universitària Edicions Proa. Bailey, Laura (2008). The EPP across Languages with Special Reference to Latin. University of Newcastle: unpublished thesis. Baker, Mark (1985). ‘The Mirror Principle and Morphosyntactic Explanation’, Linguistic Inquiry 16: 373–416. ——--(1988). Incorporation. Chicago: University of Chicago Press. ——--(1991). ‘On some Subject/Object Non-asymmetries in Mohawk’, Natural Language and Linguistic Theory 9: 537–76. ——--(1995). The Polysynthesis Parameter. Oxford: Oxford University Press. ——--(2001). ‘Configurationality and Polysynthesis’, in Haspelmath, König, Oesterreicher, and Raible (eds), 1433–41. ——--(2003). ‘Agreement, Dislocation, and Partial Configurationality’, in Andrew Carnie, Heidi Harley, and Mary Ann Willie (eds), Formal Approaches to Function. Amsterdam/ Philadelphia: Benjamins, 107–34. ——--(2008). ‘The Macroparameter in a Microparametric World’, in Theresa Biberauer (ed.), The Limits of Syntactic Variation. Amsterdam/Philadelphia: Benjamins, 351–74. Bakkum, Gabriël (2009). The Latin Dialect of the Ager Faliscus. 150 Years of Scholarship (2 vols). Amsterdam: Amsterdam University Press. Baldi, Philip (1979). ‘Typology and the Indo-European Prepositions’, Indogermanische Forschungen 84: 49–61. ——--and Pierluigi Cuzzolin (eds) (2009). New Perspectives on Historical Latin Syntax. Volume 1. Syntax of the Sentence. Berlin/New York: Mouton de Gruyter. ——--——--(eds) (2010a). New Perspectives on Historical Latin Syntax. Volume 2. Constituent Syntax: Adverbial Phrases, Adverbs, Mood, Tense. Berlin/New York: Mouton de Gruyter. ——--——--(eds) (2010b). New Perspectives on Historical Latin Syntax. Volume 3. Constituent Syntax: Quantification, Numerals, Possession, Anaphora. Berlin/New York: Mouton de Gruyter. ——--and Nuti, Andrea (2010). ‘Possession’, in Baldi and Cuzzolin (eds), 3, 239–387. Baldinger, Kurt (1968). ‘Post- und Prädeterminierung im Französische’, in Festschrift Walther von Wartburg zum 80. Geburtstag. Tübingen: Niemeyer, 87–106.

References

357

Baltin, Mark, and Collins, Chris (eds) (2001). The Handbook of Contemporary Syntactic Theory. Oxford: Blackwell. Banniard, Michel (1992). VIVA VOCE. Communication écrite et communication orale du IVe au IXe siècle en Occident latin. Paris: Institut des Etudes Augustiniennes. ——--(1995). ‘Ille et son système: Chronologie du développement (IIIe–VIIIe s)’, in Callebat (ed.), 313–21. ——--(forthc.) ‘The Transition from Latin to the Romance Languages’, in Maiden, Smith, and Ledgeway (eds). Baños Baños, José (ed.) (2009a). Sintaxis del latín clásico. Madrid: Liceus. ——--(2009b). ‘Subordinación completiva’, in Baños Baños (ed.), 523–62. ——--and Cabrillana, Concepción (2009). ‘Orden de palabras’ in Baños Baños (ed.), 679–708. Barbosa, Pilar (1997). ‘Subject Positions in the Null Subject Languages’, Seminarios de linguística 1: 39–63. Bartoli, Matteo (1906). Das Dalmatische. Altromanische Sprachreste von Veglia bis Ragusa und ihre Stellung in der Apennino-Balkanischen Romania (2 Bde). Vienna: Hölder. ——--(1925). Introduzione alla neolinguistica: Principi, scopi, metodi. Geneva: Olschki. ——--(1929). ‘La norma linguistica dell’area maggiore’, Rivista di filologia e d’istruzione classica 57: 333–45. ——--(1933). ‘Le norme neolinguistiche e la loro utilità per la storia dei linguaggi e dei costumi’, Atti della Società italiana per il progresso delle scienze 21: 157–67. Bastardas Parera, Juan (1953). Particularidades sintácticas del latín medieval. Barcelona: Escuela de filología. Battye, Adrian, and Hintze, Marie-Anne (1992). The French Language Today. London: Routledge. ——--and Roberts, Ian (eds) (1995). Clause Structure and Language Change. Oxford: Oxford University Press. Bauer, Brigitte (1995). The Emergence and Development of SVO Patterning in Latin and French. Oxford: Oxford University Press. ——--(2001a). ‘Variability in Word Order: Adjectives and Comparatives in Latin, Romance, and Germanic’, Southwest Journal of Linguistics 20: 19–50. ——--(2001b). ‘Syntactic Innovation in Latin Poetry? The Origins of the Romance Adverbial Formation in -ment(e)’, in Arpád Orbán and Marc van der Poel (eds), Ad Literas: Latin Studies in Honour of J. H. Brouwers. Nijmegen: Nijmegen University Press, 29–43. ——--(2003). ‘The Adverbial Formation in mente in Vulgar and Late Latin: A Problem in Grammaticalization’, in Solin, Leiwo, and Halla-aho (eds), 439–57. ——--(2006). ‘“Synthetic” vs “Analytic” in Romance: The Importance of Varieties’, in Randall Gess and Deborah Arteaga (eds), Historical Romance Linguistics: Retrospective and Perspectives. Amsterdam/Philadelphia: Benjamins, 287–304. ——--(2008). ‘Nominal Apposition in Vulgar and Late Latin: At the Cross-roads of Major Language Changes’, in Wright (ed.), 42–50. ——--(2009a). ‘Word Order’, in Baldi and Cuzzolin (eds), 241–316. ——--(2009b). ‘Strategies of Definiteness in Latin: Implications for Early Indo-European’, in Vit Bubenik, John Hewson, and Sarah Rose (eds), Grammatical Change in Indo-European Languages. Papers Presented at the Workshop on Indo-European Linguistics at the XVIIIth International Conference on Historical Linguistics, Montreal, 2007. Amsterdam/ Philadelphia: Benjamins, 71–87.

358

References

Bauer, Brigitte (2011). ‘Word Formation’, in Maiden, Smith, and Ledgeway (eds), 532–63. Bec, Pierre (1967). La langue occitane. Montpellier: PUF. ——--(1973). Manuel pratique d’occitan moderne. Paris: Picard. Beckmann, Gustav (1963). Die Nachfolgekonstruktionen des instrumentalen Ablativs in Spätlatein und im Französischen (Beihefte zur Zeitschrift für romanische Philologie Band 106). Tübingen: Niemeyer. Belletti, Adriana (1990). Generalized Verb Movement. Turin: Rosenberg & Sellier. ——--(2001a). ‘Inversion as Focalization’, in Aafke Hulk and Jean-Yves Pollock (eds), Subject Inversion in Romance and the Theory of Universal Grammar. Oxford: Oxford University Press, 60–90. ——--(2001b). ‘Agreement Projections’, in Baltin and Collins (eds), 483–510. ——--(ed.) (2004a). Structures and Beyond: The Cartography of Syntactic Structures. Volume 3. Oxford: Oxford University Press. ——--(2004b). ‘Aspects of the Low IP Area’, in Rizzi (ed.), 16–51. ——--(2005a). ‘Extended Doubling and the VP Periphery’, Probus 17: 1–35. ——--(2005b). ‘(Past-)Participle Agreement’, in Martin Everaert and Henk van Riemsdijk (eds), Blackwell Companion to Syntax. Volume 3. Oxford: Blackwell, 493–521. ——--and Rizzi, Luigi (1988). ‘Psych-verbs and Theta Theory’, Natural Language and Linguistic Theory 6: 291–352. Benincà, Paola (1988). ‘L’ordine degli elementi della frase: Costruzioni con ordine marcato degli elementi’, in Renzi (ed.), 129–94. ——--(ed.) (1994a). La variazione sintattica: Studi di dialettologia romanza. Bologna: il Mulino. ——--(1994b). ‘Un’ipotesi sulla sintassi delle lingue romanze medievali’, in Benincà (ed.), 177–94. ——--(1994c). ‘La sintassi dei clitici complemento nelle lingue romanze medievali’, in Benincà (ed.), 213–45. ——--(1995). ‘Complement Clitics in Medieval Romance: The Tobler-Mussafia Law’, in Battye and Roberts (eds), 325–44. ——--(1996). ‘La struttura della frase esclamativa alla luce del dialetto padovano’, in Benincà, Cinque, De Mauro, and Vincent (eds), 23–43. ——--(2001). ‘The Position of Topic and Focus in the Left Periphery’, in Cinque and Salvi (eds), 39–64. ——--(2002). ‘Strane e misteriose somiglianze sintattiche fra le antiche varietà italiane’, in Marcato (ed.), 235–45. ——--(2006). ‘A Detailed Map of the Left Periphery of Medieval Romance’, in Raffaella Zanuttini, Hector Campos, Elena Herberger, and Paul Portner (eds), Crosslinguistic Research in Syntax and Semantics. Negation, Tense, and Clausal Architecture. Washington: Georgetown University Press, 53–86. ——--(2010). ‘L’ordine delle parole e la struttura della frase: La periferia sinistra’, in Salvi and Renzi (eds), I, 27–59. ——--Cinque, Guglielmo, De Mauro, Tullio, and Vincent, Nigel (eds) (1996). Italiano e dialetto nel tempo: Saggi di grammatica per Giulio C. Lepschy. Rome: Bulzoni.

References

359

Benincà, Paola, and Munaro, Nicola (eds) (2010). Mapping the Left Periphery: The Cartography of Syntactic Structures, Volume 5. Oxford: Oxford University Press. Benincà, Paola, and Poletto, Cecilia (2004). ‘Topic, Focus, and V2: Defining the CP Sublayers’, in Rizzi (ed.), 52–75. Bennett, Charles (1910). Syntax of Early Latin. I. The Verb. Boston: Allyn and Bacon. ——--(1914). Syntax of Early Latin. II. The Cases. Boston: Allyn and Bacon. Bentley, Delia (2000a). ‘I costrutti condizionali in siciliano: Un’analisi diacronica’, Revue romane 35: 3–20. ——--(2000b). ‘Semantica e sintassi nello sviluppo dei costrutti condizionali: Il caso del siciliano’, Revue romane 35: 163–76. ——--(2006). Split Intransitivity in Italian. Berlin/New York: Mouton de Gruyter. ——--(2007). ‘Relazioni grammaticali e ruoli pragmatici: Siciliano e italiano a confronto’, in Bentley and Ledgeway (eds), 48–62. ——--and Eyþórsson, Þórhallur (2001). ‘Alternation According to Person in Italo-Romance’, in Laurel Brinton (ed.), Historical linguistics 1999. Amsterdam/Philadelphia: Benjamins, 63–74. ——--and Ledgeway, Adam (eds) (2007). Sui dialetti italoromanzi: Saggi in onore di Nigel B. Vincent (The Italianist 27, Special supplement 1). Norfolk: Biddles. Benveniste, Émile (1968). ‘Mutations of Linguistic Categories’, in Wilfred Lehmann and Yakov Malkiel (eds), Directions for Historical Linguistics: A Symposium. Austin: University of Texas, 83–94. Benzing, Joseph (1931). ‘Zur Geschichte von ‘ser’ als Hilfszeitwort bei den intransitiven Verben im Spanischen’, Zeitschrift für romanische Philologie 51: 385–460. Berlinck, Rosane de Andrade (1989). ‘A construção V+SN no português do Brasil: Uma visão diacrônica do fenômeno da ordem’, in Fernando Tarallo (ed.), Fotografias sociolingüísticas. Campinas: Pontes, 95–112. Bernat i Baltrons, Francesc (2007). Un estudi de dialectologia catalana al segle XIX: Les notes de Manuel Milà i Fontanals sobre el maonès. Barcelona: Abadia de Montserrat. Bernstein, Judy (1991). ‘DPs in French and Walloon: Evidence for Parametric Variation in Nominal Head Movement’, Probus 3: 101–26. ——--(1993). Topics in the Syntax of Nominal Structure Across Romance. CUNY: unpublished thesis. Berruto, Gaetano (1998). ‘Sulla posizione prenominale dell’aggettivo in italiano’, in Giuliano Bernini, Pierluigi Cuzzolin, and Piera Molinelli (eds), Ars linguistica: Studi offerti a Paolo Ramat. Rome: Bulzoni, 95–105. Bertocci, David, and Damonte, Federico. (2007). ‘Distribuzione e morfologia dei congiuntivi in alcune varietà salentine’, in Federico Damonte and Jacopo Garzonio (eds), Studi sui dialetti della Puglia. Padua: Unipress, 3–28. Bertocchi, Alessandra (1989). ‘The Role of Antecedents of Latin Anaphors’, in Calboli (ed.), 441–61. Bianchi, Patricia, De Blasi, Nicola, and Librandi, Rita (eds) (1993). I’ te vurria parlà. Storia della lingua a Napoli e in Campania. Napoli: Tullio Pironti.

360

References

Biberauer, Theresa, Holmberg, Anders, and Roberts, Ian (2008). ‘Structure and Linearization in Disharmonic Word Orders’, in Charles Chang and Hannah Haynie (eds). Proceedings of the 26th West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Proceedings Project, 96–104. ——--——--——--(2009). ‘Linearization and the Architecture of Grammar: A View from the Final-Over-Final Constraint’, Vincenzo Moscati and Emilio Servidio (eds). StiL - Studies in Linguistics (Proceedings of XXXV Incontro di Grammatica Generativa) 3: 78–91. ——--——--——--(2010). ‘A Syntactic Universal and its Consequences’. University of Cambridge: unpublished manuscript. ——--Newton, Glenda, and Sheehan, Michelle (2009a). ‘Limiting Synchronic and Diachronic Variation and Change: The Final-over-Final Constraint’, Language and Linguistics 10: 701–43. ——--——--——--(2009b). ‘The Final-over-Final Constraint and Predictions for Diachronic Change’, Toronto Working Papers in Linguistics 31: 1–17. ——-- and Sheehan, Michelle (in press). ‘Disharmony, Antisymmetry, and the Final-over-Final Constraint’, to appear in Valmala Elguea and Myriam Extebarria (eds). Ways of Structure Building. Cambridge: Cambridge University Press. ——--——--and Newton, Glenda (2010). ‘Impossible Changes and Impossible Borrowings’, in Anna Breitbarth, Chris Lucas, Sheila Watts, and David Willis (eds), Continuity and Change in Grammar. Amsterdam/Philadelphia: Benjamins, 35–60. ——--and Roberts, Ian (2008). ‘Cascading Parameter Changes: Internally-driven Change in Middle and Early Modern English’, in Þorhallur Eyþórsson (ed.), Grammatical Change and Linguistic Theory. Amsterdam/Philadelphia: John Benjamins, 79–133. Bichakjian, Bernard (1982). ‘La genèse de la subordination de l’indo-européen au français’, in Quirinus Ignatius, Maria Mok, Ina Spiele, and Paul Verhuyck (eds), Mélanges de linguistique, de littérature et de philologie médiévales. Leiden: Brill, 5–20. ——--(1987). ‘The Evolution of Word Order: A Paedomorphic Explanation’, in Anna Giacalone Ramat, Anna, Onofrio Carruba, and Giuliani Bernini (eds), Papers from the 7th International Conference on Historical Linguistics. Amsterdam/Philadelphia: Benjamins, 87–107. Blasco Ferrer, Eduardo (1984a). Grammatica storica del catalano e dei suoi dialetti con speciale riguardo all’algherese. Tübingen: Narr. ——--(1984b). Storia linguistica della Sardegna. Tübingen: Niemeyer. ——--(1986). La lingua sarda contemporanea. Cagliari: Edizioni della Torre. ——--(1988). ‘Sardisch: Interne Sprachgeschichte I. Grammatik’, in Holtus, Metzeltin, and Schmitt (eds), 836–53. ——--(1994). Ello, ellus. Nuoro: Poliedro. Bobaljik, Jonathan (2002). ‘The Rich Agreement Hypothesis in Review’, in Yasuki Abe (ed.), Complex Predicates and Argument Structure. Nanzan University, Nagoya: Research report for the Ministry of Education Grant-in-Aid for Scientific Research, 63–109. Bolkestein, Machtelt (1989). ‘Parameters in the Expression of Embedded Predications in Latin’, in Calboli (ed.), 3–35. ——--(2001). ‘Random Scrambling? Constraints on Discontinuity in Latin Noun Phrases’, in Moussy, Dangel, Fruyt, Nadjo, and Sznajder (eds), 245–58.

References

361

Bonazzi, Giulio (ed.) (1997). Il condaghe di San Pietro di Silki: Testo logudorese inedito dei secoli XI–XIII. Sassari: Dessì. Borer, Hagit (1984). Parametric Syntax: Case Studies in Semitic and Romance Languages. Dordrecht: Foris. Born, Joachim (1992). ‘Leonesisch’ in Holtus, Metzeltin, and Schmitt (eds), 693–700. Bošković, Želikjo (2005a). ‘On the Locality of Left Branch Extraction and the Structure of NP’, Studia Linguistica 59: 1–45. ——--(2005b). ‘Left Branch Extraction, Structure of NP, and Scrambling’, in Joachim Sabel and Mamoru Saito (eds), The Free Word Order Phenomenon: Its Syntactic Sources and Diversity. Berlin/New York: Mouton de Gruyter, 13–73. ——--(2008). ‘What Will You Have, DP or NP?’, North East Linguistic Society 37: 101–14. ——--(2010). ‘On NPs and Clauses’. University of Connecticut: unpublished manuscript. ——--and Gajewski, Jon (forthc.). ‘Semantic Correlates of the NP/DP Parameter’, North East Linguistic Society 39. Bosque, Ignacio (1996). ‘On Specificity and Adjective Position’, in Javier Gutiérrez-Rexach and Luis Silva-Villar (eds), Perspectives on Spanish Linguistics. Volume 1. UCLA: Department of Linguistics, 1–13. ——--and Demonte, Violeta (eds) (1999). Gramática descriptiva de la lengua española. I. Sintaxis básica de las clases de palabras. II. Las constructions sintácticas. Relaciones temporales, aspectuales y modales. III. Entre la oración y el discurso. Morfología. Madrid: Espasa. ——--and Picallo, Carme (1996). ‘Postnominal Adjectives in Spanish DPs’, Journal of Linguistics 32: 349–85. Bossong, Georg (1991). ‘Differential Object Marking in Romance and Beyond’, in Wanner and Kibbee (eds), 143–71. Bouchard, Denis (1998). ‘The Distribution and Interpretation of Adjectives in French: A Consequence of Bare Phrase Structure’, Probus 10: 139–83. ——--(2002). Adjectives, Number and Interfaces: Why Languages Vary? Elsevier: Amsterdam. Bourciez, Edouard (1956). Eléments de linguistique romane. 4th edn. Paris: Klincksieck. Bourgain, Pascale (2005). Le latin medieval. Turnhout: Brepols. Bourova, Vaira (2005). ‘A la recerche du “conditionnel latin”: Les constructions “infinitif + forme de habere” examinées à partir d’un corpus éléctronique’, in Claus Pusch, Johannes Kabatek, and Wolfgang Raible (eds), Korpora und diachrone Sprachwissenschaft. Tübingen: Narr, 303–16. ——--(2007). ‘Les constructions latines infinitif + habebam vs. infinitif + habui et le développement du conditionnel roman’, in David Trotter (ed.), Actes du XXIVe congrès international de linguistique et de philologie romanes, Aberystwyth 2004. Tome II. Section 4: Les langues perdues et les langues retrouvées – politique linguistique dans la Romania minor. Section 5: La langue des textes anciens (des origines au XVIIIème siècle). Section 6: L’anglonormand. Section 7: Latín tardío. Tübingen: Niemeyer, 461–74. ——--(2008), ‘Les participles futurs en -urus/-ndus combinés avec esse au passé en latin tardif: Un conditionnel non abouti?’, in Wright (ed.), 271–80. ——--and Tasmowski, Liliane (2007). ‘La préhistoire des futurs romans – ordre de constituants et sémantique’, Cahiers Chronos 19: 25–41.

362

References

Bresnan, Joan (2001). Lexical-Functional Syntax. Oxford: Blackwell. Brito, Ana Maria (1993). ‘Aspects de la syntaxe du SN en portugais et en français’, Línguas e literaturas (Revista da Faculdade de Letras do Porto) 10: 25–53. Brown, Paul, Joseph, Brian, and Wallace, Rex (2009). ‘Questions and Answers’, in Baldi and Cuzzolin (eds), 489–530. Bruni, Francesco (1984). L’italiano. Elementi di storia della lingua e della cultura. Turin: UTET. Brunot, Ferdinand (and Bruneau, Charles) (1905–53). Histoire de la langue française des origines à 1900 (13 tomes). Paris: Colin. Buck, Carl Darling [1933] 1979. A Grammar of Oscan and Umbrian. Hildesheim: Olm. Bulhart, Vinzenz (1936–42). ‘Habeo’, Thesaurus Linguae Latinae, V1.3.2454.12–2458.18. Leipzig: Teubner. Buridant, Claude (2000). Grammaire nouvelle de l’ancien français. Paris: SEDES. Burzio, Luigi (1986). Italian Syntax. Dordrecht: Reidel. Busquet Isart, Núria (2010). El parlar salat: Descripció, àmbit geogràfic i ús. Universitat de Barcelona: unpublished thesis. Butt, John, and Benjamin, Carmen (1994). A New Reference Grammar of Modern Spanish. 2nd edn. London: Arnold. Cabrillana, Concepción (1993a). ‘Posiciones relativas a la ordenación de constituyentes (I). Estudios de la posición de sujeto, objeto y verbo en latín’, Habis 24: 249–66. ——--(1993b). ‘Panorama de los estudios sobre el orden de palabras en latín’, Revista de filología clásica 7: 223–54. ——--(1999). ‘Type of Text, Pragmatic Function, and Constituent Order: A Comparative Study of the Mulomedicina Chironis and the Peregrinatio Egeriae’, in Hubert Petersmann and Rudolf Ketteman (eds), Latin vulgaire – latin tardif V. Actes du Ve colloque international sur le latin vulgaire et tardif Heidelberg, 5–8 septembre 1997. Heidelberg: C. Winter, 319–30. Calabrese, Andrea (1993). ‘The Sentential Complementation of Salentino: A Study of a Language without Infinitival Clauses’, in Adriana Belletti (ed.), Syntactic Theory and the Dialects of Italy. Turin: Rosenberg & Sellier, 28–98. Calboli, Gualtiero (1978). ‘Die Entwicklung der klassischen Sprachen und die Beziehung zwischen Satzbau, Wortstellung und Artikel’, Indogermanische Forschungen 83: 197–216. ——--(1983a). ‘Problemi di grammatica latina’, Aufstieg und Niedergang der Römischen Welt 29: 111–77. ——--(1983b). ‘The Development of Latin (Cases and Infinitive)’, in Pinkster (ed.), 41–57. ——--(ed.) (1989a). Subordination and Other Topics in Latin. Proceedings of the Third Colloquium on Latin Linguistics, Bologna, 1–5 April 1985. Amsterdam/Philadelphia: Benjamins. ——--(1989b). ‘Subordination and Opacity’, in Calboli (ed.), 39–64. ——--(2009). ‘Latin Syntax and Greek’, in Baldi and Cuzzolin (eds), 65–193. ——--(ed.) (1995). Latin vulgaire – latin tardif IV. Actes du 4e colloque international sur le latin vulgaire et tardif. Caen, 2–5 septembre 1994. Hildesheim/Zurich/New York: OlmsWeidmann. Campos, Héctor, and Zampini, Mary (1990). ‘Focalization Strategies in Spanish’, Probus 2: 47–64.

References

363

Campos, Odette, and Rodrigues, Ângela (1992). ‘Flexão nominal: Indicação de pluralidade no sintagma nominal’, in Rodolfo Ilari (ed.), Gramática do português falado. II. Níveis de análise lingüística. Campinas: Editora da Unicamp, 111–34. Canale, Michael, Mougeon, Raymond, and Belanger, Monique (1978). ‘Analogical Levelling of the Auxiliary être in French’, in Maragarita Suñer (ed.), Studies in Romance Linguistics. Georgetown: Georgetown University Press, 41–61. Cano González, Ana María (1981). El habla de Somiedo (Occidente de Asturias). Santiago: Universidad de Santiago. ——--(1992). ‘Asturiano. Evolución lingüística interna’, in Holtus, Metzeltin, and Schmitt (eds), 652–80. Capidan, Theodor (1932). Aromânii. Dialectul aromân. Studiu lingvistic. Bucharest: Academia Română. Carballo Calero, Ricardo (1981). ‘Um exemplo de gerúndio flexional’, in Problemas da língua galega. Lisboa: Sá da Costa, 129–33. Cardinaletti, Anna (1997). ‘Subjects and Clause Structure’, in Haegeman (ed.), 33–63. ——--(2004). ‘Toward a Cartography of Subject Positions’, in Rizzi (ed.), 115–65. ——--and Giusti, Giuliana (2001). ‘“Semi Lexical” Motion Verbs in Romance and Germanic’, in Norbert Corver and Henk van Riemsdijk (eds), Semi-lexical Categories. Berlin: De Gruyters, 371–414. ——--——--(2003). ‘Motion Verbs as Functional Heads’, in Tortora (ed.), 31–49. ——--and Repetti, Lori (2010). ‘Proclitic vs Enclitic Pronouns in Northern Italian Dialects and the Null-subject Parameter’, in D’Alessandro, Ledgeway, and Roberts (eds), 119–134. Carrera, Micaela (1983). ‘Orden de palabras en un texto latín alto-medieval’, Revista de la sociedad española de lingüística 13: 63–89. Cecchetto, Carlo, and Oniga, Renato (2002). ‘Consequences of the Analysis of Latin Infinitival Clauses for the Theory of Case and Control’, Lingue e linguaggio 1: 151–89. Cennamo, Michela (1999). ‘Late Latin Pleonastic Reflexives and the Unaccusative Hypothesis’, Transactions of the Philological Society 97: 103–50. ——--(2001a). ‘L’extended accusative e le nozioni di voce e relazione grammaticale nel latino tardo e medievale’, in Valeria Viparelli (ed.), Ricerche linguistiche tra antico e moderno. Naples: Liguori, 3–27. ——--(2001b). ‘L’inaccusatività in alcune varietà campane: Teorie e dati a confronto’, in Albano Leoni, Krosbakken, Sornicola, and Stromboli (eds), 427–53. ——--(2009). ‘Argument Structure and Alignment Variations and Changes in Late Latin’, in Jóhanna Bardal and Shobhana Chelliah (eds), The Role of Semantic, Pragmatic, and Discourse Factors in the Development of Case. Amsterdam/Philadelphia: Benjamins 307–46. Chafe, Wallace (1976). ‘Givenness, Contrastiveness, Definiteness, Subjects, Topics and Point of View’, in Charles Li (ed.), Subject and Topic. New York: Academic Press, 25–56. Charpin, François (1977). L’idée de phrase grammaticale et son expression en latin. Lille: Atelier de reproduction des thèses; Paris: Librairie Honoré Champion. ——--(1989). ‘Etude de syntaxe énonciative: L’ordre des mots et la phrase’, in Calboli (ed.), 503–20. Chinellato, Paolo and Garzonio, Jacopo (2002). ‘La particella “o” in due varietà di toscano: Un’analisi di interfaccia tra sintassi e pragmatica’, in Marcato (ed.), 305–9.

364

References

Chitoran, Ioana (2002b). ‘The Phonology and Morphology of Romanian Diphthongization’, Probus 14: 205–46. Chomsky, Noam (1986). Barriers. Cambridge, MA: MIT Press. ——--(1995). ‘Bare Phrase Structure’, in Gert Webelhuth (ed.), Government and Binding Theory and the Minimalist Program. Oxford: Blackwell, 383–439. ——--(2000). ‘Minimalist Inquiries’, in Roger Martin, David Michaels, and Juan Uriagereka (eds), Step by Step: Minimalist Syntax in Honor of Howard Lasnik. Cambirdge, MA: MIT Press, 89–155. ——--(2001). ‘Derivation by Phase’, in Michael Kenstowicz (ed.), Kale Hale: A Life in Language. Cambridge, MA: MIT Press, 1–52. Christol, Alain (1994). ‘Ipse: “articloïde” ou article dans la Peregrinatio’, LALIES. Actes des sessions de linguistique et de littérature d’Aussois 13: 143–53. Cinque, Guglielmo (1990). Types of A’-dependencies. Cambridge, MA: MIT Press. ——--(1994). ‘On the Evidence for Partial N-Movement in the Romance DP’, in Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi, and Raffaella Zanuttini (eds), Paths Towards Universal Grammar: Studies in Honor of Richard S. Kayne. Washington D.C.: Georgetown University Press, 85–110. ——--(1999). Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford: Oxford University Press. ——--(ed.) (2002). The Structure of DP and IP: The Cartography of Syntactic Structures. Volume 1. Oxford: Oxford University Press. ——--(2004a). ‘A Phrasal Movement Analysis of the Romanian DP’, in Ana-Maria Minut¸ and Eugen Munteanu (eds), Studia Linguistica et Philologica in Honorem D. Irimia. Iasi: Editura Universităt¸ii ‘A.I.Cuza’, 129–42. ——--(2004b). ‘“Restructuring” and Functional Structure’, in Belletti (ed.), 132–91. ——--(2006). Restructuring and Functional Heads: The Cartography of Syntactic Structures. Volume 4. Oxford: Oxford University Press. ——--(2010). The Syntax of Adjectives. Cambridge, MA: MIT Press. ——--and Giusti, Giuliana (eds) (1995). Advances in Rumanian Linguistics. Amsterdam/ Philadelphia: Benjamins. ——--and Salvi, Giampaolo (eds) (2001). Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi. Amsterdam: Elsevier. Clackson, James (2007). Indo-European Linguistics: An Introduction. Cambridge: Cambridge University Press. ——--and Horrocks, Geoffrey (2007). The Blackwell History of the Latin Language. Oxford: Blackwell. Clark, Robin, and Roberts, Ian (1993). ‘A Computational Approach to Language Learnability and Language Change’, Linguistic Inquiry 24: 299–345. Coleman, Robert (1971). ‘The Origin and Development of Latin habeo + Infinitive’, Classical Quarterly 21: 215–32. ——--(1987). ‘Latin and the Italic Languages’, in Bernard Comrie (ed.), The World’s Major Languages. Oxford: Oxford University Press, 180–202. ——--(1991). ‘Latin Prepositional Syntax in Indo-European Perspective’, in New Studies in Latin Linguistics. Amsterdam/Philadelphia: Benjamins, 323–38.

References

365

Colon, Germà. (1961). ‘Le parfait périphrastique catalan “va + infinitif ” ’, Actas do IX congreso internacional de lingüística românica. Volumen I. Lisbon: Centro de Estudos Filológicos, 165–76. Como, Paola (1999). ‘Usi e funzioni del tipo verbale fo- a Monte di Procida’, Rivista italiana di dialettologia 23: 113–36. Comrie, Bernard (1985). ‘Causative Verb Formation and Other Verb-deriving Morphology’, in Tim Shopen (ed.), Language Typology and Syntactic Description. Volume 3. Grammatical Categories and the Lexicon. Cambridge: Cambridge University Press, 309–48. ——--(1989). Language Universals and Linguistic Typology. 2nd edn. Oxford: Blackwell. Contreras, Heles (1963). ‘Una clasificación morfo-sintáctica de las lenguas románicas’, Romance Philology 16: 261–8. ——--(1991). ‘On the Position of Subjects’, in Susan Rothstein (ed.), Perspectives on Phrase Structure: Heads and Licensing. San Diego: Academic Press, 63–79. Cornilescu, Alexandra (1992). ‘Remarks on the Determiner System of Romanian: The Demonstratives al and cel’, Probus 4: 189–260. ——--(2000). ‘The Double Subject Construction in Romanian’, in Alboiu and Motapanyane (eds), 83–133. ——--(2006). ‘Modes of Semantic Combinations: NP/DP Adjectives and the Structure of the Romanian DP’, in Doetjes and González (eds), 43–69. Coromines, Joan (1989–97). Onomasticon Cataloniae (8 vols). Barcelona: Curial. Corver, Norbert (1992). ‘Left Branch Extraction’, in North East Linguistic Society 22: 67–84. ——--and Henk van Riemsdijk (eds) (1994a). Studies on Scrambling: Movement and Nonmovement Approaches to Free Word-Order Phenomena. Berlin/New York: Mouton de Gruyter. ——--——--(1994b). ‘Introduction: Approaches to and Properties of Scrambling’, in Corver and Riemsdijk (eds), 1–15. Coseriu, Eugenio (1987). ‘Le latin vulgaire et le type linguistique roman (A propos de la thèse de Humboldt: “Es sanken Formen, nicht aber die Form”)’, in Herman (ed.), 53–64. ——--[1971] (1988). ‘Der romanische Sprachtypus. Versuch einer neuen Typologisierung der romanischen Sprachen’, in Jörn Albrecht (ed.), Energeia und Ergon. Sprachliche Variation – Sprachgeschichte – Sprachtypologie. Band I. Schfriften von Eugenio Coseriu (1965–1987). Tübingen: Narr, 207–24. Costa, João (ed.) (2000a). Portuguese Syntax. Oxford: Oxford University Press. ——--(2000b). ‘Word Order and Discourse-configurationality in European Portuguese’, in Costa (ed.), 94–115. Cremona, Joseph (1970). ‘L’axe nord-sud de la Romania et la position du toscan’, in Alexandru Rosetti (ed.), Actele celui de-al XII-lea Congres internat¸ional de lingvistică şi filologie romanică. Bucharest: Editura Academiei, 155–9. Crisma, Paola (1993). ‘On Adjective Placement in Romance and Germanic Event Nominals’, Rivista di grammatica generativa 18: 61–100. ——--(1996). ‘On the Configurational Nature of Adjectival Modification’, in Karen Zagona (ed.), Grammatical Theory and Romance Languages: Current Issues in Linguistic Theory 133. Amsterdam/Philadelphia: Benjamins, 59–71.

366

References

Crisma, Paola, and Gianollo, Chiara (2006). ‘Where did Romance N-raising Come from? A Parallel Study of Parameter Resetting in Latin and English’, in Doetjes and González (eds), 71–93. Croft, William (1990). Typology and Universals. Cambridge: Cambridge University Press. Cruschina, Silvio (2006). ‘Informational Focus in Sicilian and the Left Periphery’, in Frascarelli (ed.), 363–85. ——--(2008). Discourse-related Features and the Syntax of Peripheral Positions: A Comparative Study of Sicilian and Other Romance Languages. University of Cambridge: unpublished thesis. ——--(2010). ‘Fronting as Focalization in Sicilian’, in D’Alessandro, Ledgeway, and Roberts (eds), 247–60. Cuzzolin, Pierluigi (1994a). Sull’origine della costruzione dicere quod: Aspetti sintattici e semantici. Florence: La Nuova Italia. ——--(1994b). ‘On Sentential Complementation after Verba Affectuum’, in Herman (ed.), 201–10. ——--(1995). ‘A proposito di sub vos placo e della grammaticalizzazione delle adposizioni’, Archivio glottologico italiano 80: 122–43. Cyrino, Sonia (1996). ‘O objeto nulo no português brasileiro’, Documentação de estudos em lingüística teórica e aplicada 12: 221–38. ——--(1997a). ‘Objeto indireto nulo e dative shift’, Estudos lingüísticos – Anais do XXVI seminário do GEL, Campinas, SP, 466–71. ——--(1997b). ‘Considerações sobre o objeto nulo e elipse’, Anais do I Encontro do CELSUL, 1, Florianópolis, SC, 595–603. ——--(1998). ‘O objeto indireto nulo no português brasileiro’, Signum – estudos da linguagem 1: 35–54. ——--(2001). ‘O objeto nulo no português do Brasil e no português de Portugal’, Boletim da ABRALIN 25: 173–81. ——--(2003). ‘Para a história do português brasileiro: A presença do objeto nulo e a ausência de clíticos’, Letras de hoje 38: 31–47. ——--(2006). ‘Anáfora do complemento nulo na historia do português brasileiro’, in Tânia Lobo, Ilza Ribeiro, Zenaide Carneiro, and Norma Almeida (eds), Para a história do português brasileiro. Volume VI: Novos dados, novas análises. Tomo I. Salvador: EDUFBA, 45–72. ——--(2010). ‘On the Loss of Verb Movement in Brazilian Portuguese’, paper presented at DiGS XII, Queens’ College, University of Cambridge. D’Alessandro, Roberta (2010). ‘The Verbal Domain: TP-VP Structure and Auxiliaries’ in D’Alessandro, Ledgeway, and Roberts (eds), 27–38. ——--(2011). ‘Agreement, Ergativity, Dubling and the Parametrization of Probes’. University of Leiden: unpublished manuscript. ——--and Di Felice, Claudio (2010). ‘The Diachrony of Abruzzese Complementation’, paper presented at the X Incontro di dialettologia, University of Bristol. ——--and Ledgeway, Adam (2010a). ‘At the C-T Boundary: Investigating Abruzzese Complementation’, Lingua 120: 2040–60.

References

367

——--——--(2010b). ‘The Abruzzese T-v System: Feature Spreading and the Double Auxiliary Construction’, in D’Alesssandro, Ledgeway, and Roberts (eds), 201–9. ——--Ledgeway, Adam, and Roberts, Ian (eds) (2010). Syntactic Variation: The dialects of Italy. Cambridge: Cambridge University Press. ——--and Roberts, Ian (2005). ‘Split Ergativity in Abruzzese and the Null-subject Parameter’, paper presented at Going Romance 2005, Utrecht. ——--——--(2008). ‘Movement and Agreement in Italian Past Participles and Defective Phases’, Linguistic Inquiry 39: 477–91. ——--——--(2010). ‘Past Participle Agreement in Abruzzese: Split Auxiliary Selection and the Null-subject Parameter’, Natural Language and Linguistic Theory 28: 41–72. D’Hulst, Yves (2004). ‘French and Italian Conditionals: From Etymology to Representation’, in Jacqueline Guéron and Jacqueline Lecarme (eds), The Syntax of Tense and Aspect: Proceedings of the International Conference on the Expression of Tense and Aspect. Cambridge, MA: MIT Press, 181–201. ——--Coene, Martine, and Tasmowski, Liliane (2000). ‘Last Resort Strategies in DP: Article Reduplication in Romanian and French’, in Alboiu and Motapanyane (eds), 135–75. D’Introno, Francesco (2001). Sintaxis generativa del español: Evolución y análisis. Madrid: Cátedra. Dahl, Östen (1979). ‘Typology of Sentence Negation’, Linguistics 17: 79–106. Dahmen, Wolfgang (1989). ‘Rumänisch: Arealinguistik III. Meglenorumänisch’, in Holtus, Metzeltin, and Schmitt (eds), 436–47. Dalbera Stefanaggi, Marie-José (2001). Essais de linguistique corse. Ajaccio: Editions Alain Piazzola. Damonte, Federico (2002). ‘The Syntax and Semantics of Salentino ku’, paper presented at the XXVIII Incontro di Grammatica Generativa, Lecce. ——--(2005). ‘La diffusione della particella “mi” in alcune varietà messinesi: Problemi di metodo’, in Gianna Marcato (ed.), Dialetti in città. Atti del convegno, Sappada/Plodn (Belluno). Padua: Unipress, 237–42. ——--(2006a). ‘Differenze generazionali nell’uso del congiuntivo presente in salentino’, in Gianna Marcato (ed.), Giovani, lingue e dialetti: Atti del convegno Sappada/Plodn (Belluno), 29 giugno – 3 luglio 2005. Padua: Unipress, 237–42. ——--(2006b). ‘Complementatori e complementi congiuntivi in alcuni dialetti sardi’, Quaderni di lavoro dell’ASIt 6: 71–95. ——--(2010). ‘Matching Moods: Mood Concord between CP and IP in Salentino and Southern Calabrian Subjunctive Complements’, in Benincà and Munaro (eds), 228–56. Danckaert, Lieven (2010). ‘Left Edge Fronting in Latin Embedded Clauses’, paper presented at the XII Diachronic Generative Syntax Meeting, Queens’ College, University of Cambridge. Daniliuc, Laura, and Daniliuc, Radu (2000). Descriptive Romanian Grammar. Munich: Lincom Europa. Dardano, Maurizio, and Frenguelli, Gianluca (eds) (2004). SintAnt. La sintassi dell’italiano antico. Atti del Convegno internazionale di studi. Rome: Aracne. Dardel, Robert de, and Wüest, Jacob (1993). ‘Les systèmes casuels du protoroman: Les deux cycles de simplification’, Vox Romanica 52: 25–65.

368

References

De Blasi, Nicola (1986). Libro de la destructione de Troya: Volgarizzamento napoletano trecentesco da Guido delle Colonne. Rome: Bonacci. ——--(2006). Profilo linguistico della Campania. Bari: Laterza. De Cat, Cécile, and Demuth, Katherine (2008). The Bantu–Romance Connection: A Comparative Investigation of Verbal Agreement, DPs, and Information Structure. Amsterdam/Philadelphia: Benajmins. de Jong, Jan (1983). ‘Word Order within Latin Noun Phrases’, in Pinkster (ed.), 131–44. ——--(1986). ‘Hyperbaton en informatiestruktuur’, Lampas 19: 323–31. ——--(1994). ‘Word order in Cato’s De Agricultura’, in Herman (ed.), 91–101. de Jonge, Casper (2006). Between Grammar and Rhetoric: Dionysius of Halicarnassus on Language, Linguistics, and Literature. University of Leiden: unpublished doctoral thesis. ——--(2007). ‘From Demetrius to Dik: Ancient and Modern Views on Greek and Latin Word Order’, in Rutger Allan and Michel Buijs (eds), The Language of Literature. Linguistic Approaches to Classical Texts. Leiden/Boston: Brill, 211–32. De la Villa, Jesús (2010). ‘Numerals’, in Baldi and Cuzzolin (eds), 3, 175–238. De Melo, Wolfgang (2010). ‘Possessive Pronouns in Plautus’, in Dickey and Chahoud (eds), 71–99. De Sutter, M. (1986). ‘A Theory of Word Order within the Latin Noun Phrase, based on Cato’s De agri cultura’, in Carl Deroux (ed.), Studies in Latin Literature and Roman History. Volume IV. Brussels: Latomus, 171–83. Decat Nascimento, Maria Beatriz (1989). ‘Construções de tópicos em português: Uma abordagem diacrônica à luz do encaixamento no sistema pronominal’, in Fernando Tarallo (ed.), Fotografias sociolingüísticas. Campinas: Pontes, 113–39. DeGraff, Michel (1997). ‘Verb Syntax in, and beond, Creolization’, in Haegeman (ed.), 69–94. Del Vecchio, Tommaso (1989). ‘Observations sur l’ordre des mots dans le latin archaïque’, in Calboli (ed.), 541–58. Delbrück, Berthold (1900). Vergleichende Syntax der indogermanischen Sprachen. Band 3. Strasbourg: Trübner. Demonte, Violeta (1995). ‘Dative Alternation in Spanish’, Probus 7: 5–30. ——--(1999). ‘A Minimal Account of Spanish Adjective Position and Interpretation’, in Jon Franco, Alazne Landa, and Juan Martín (eds), Grammatical Analyses in Basque and Romance Linguistics. Papers in Honor of Mario Saltarelli. Amsterdam/Philadelphia: Benjamins, 45–76. ——--and Fernández Soriano, Olga (2007). ‘La periferia izquierda oracional y los complementantes del español’, in Juan Cuartero and Martine Emsel (eds), Vernetzungen: Kognition, Bedeutung, (kontrastive) Pragmatik. Frankfurt: Peter Lang, 133–47. ——--——--(2009). ‘Force and Finiteness in the Spanish Complementizer System’, Probus 21: 23–50. Désirat, Claude, and Hordé, Tristan (1967). La langue française au 20è siècle. Paris: Bordas. Devine, Andrew, and Stephens, Laurence (1999). Discontinuous Syntax: Hyperbaton in Greek. Structured Meaning and Information. Oxford: Oxford University Press. ——--——--(2006). Latin Word Order: Structured Meaning and Information. Oxford: Oxford University Press.

References

369

Diaconescu, Paula (1970). ‘Acuzativul cu pre în textele traduse din secolul al XVI-lea’, in Paula Diaconescu (ed.), Structură şi evolut¸ie în morfologia substantivului românesc. Bucharest: Editura Academiei, 259–63. Dias, Augusto Epiphanio da Silva (1918). Syntaxe histórica portuguesa. Lisbon: Livraria Clássica. Dickey, Eleanor, and Chahoud, Anna (eds) (2010). Colloquial and Literary Latin. Cambridge: Cambridge University Press. Diez, Friedrich (1836–8). Grammatik der romanischen Sprachen (3 Bde). Bonn: Weber. ——--(1874). Grammaire des langues romanes. Livre II: Flexion. Paris: Franck. Dimitrova-Vulchanova, Mila, and Giusti, Giuliana (1998). ‘Quantified Noun Phrase Structure in Bulgarian’, in Jindrich Toman (ed.), Formal Approaches to Slavic Linguistics. Anna Arbor: Michigan Slavic Publications, 333–60. Dixon, Robert (1994). Ergativity. Cambridge: Cambridge University Press. Dobrovie-Sorin, Carmen (1987). Syntaxe du roumain: Chaînes thématiques. Université de Paris 7: unpublished thesis. ——--(1994). The Syntax of Romanian. Berlin/New York: Mouton de Gruyter. Doetjes, Jenny, and González, Paz (eds) (2006). Romance Languages and Linguistic Theory 2004. Selected Papers from ‘Going Romance’, Leiden, 9–11 December 2004. Amsterdam/ Philadelphia: Benjamins. Doria, Mario (1989). ‘Dalmatico: Storia linguistica interna’, in Holtus, Metzeltin, and Schmitt (eds), 522–30. Draeger, Anton (1882). Syntax und Stil des Tacitus. Lepizig: B.G. Tuebner. Dressler, Wolfgang (1971). ‘Über die Rekonstruktion der indogermanischen Syntax’, Zeitschrift für vergleichende Sprachforschung 85: 5–22. Dryer, Matthew (1988a). ‘Object-Verb Order and Adjective-Noun Order: Dispelling a Myth’, Lingua 74: 185–217. ——--(1988b). ‘Universals of Negative Position’, in Machael Hammond, Edith Moravcsik, and Jessica Wirth (eds), Studies in Syntactic Typology. Philadelphia/Amsterdam, John Benjamins, 93–124. ——--(1992). ‘The Greenbergian Word Order correlations’, Language 68: 81–138. ——--(2005a). ‘Order of Object and Verb’, in Haspelmath, Dryer, Gil, and Comrie (eds), 338–41. ——--(2005b). ‘Order of Adposition and Noun Phrase’, in Haspelmath, Dryer, Gil, and Comrie (eds), 346–9. ——--(2005c). ‘Relationship between Order of Object and Verb and Order of Adjective and Noun’, in Haspelmath, Dryer, Gil, and Comrie (eds), 394–97. ——--(2009). ‘The Branching Direction Theory of Word Order Correlations Revisited’, in Scalise, Magni, and Bisetto (eds), 185–207. ——--and Haspelmath, Martin (eds) (2011). World Atlas of Language Structures Online. Munich: Max Planck Digital Library. Duarte, Inês (1996). ‘A topicalização em português europeu: Uma análise comparativa’, in Inês Duarte and Isabel Leira (eds), Actas do congresso internacional sobre o português. Volumen 1. Lisbon: Edições Colibri, 327–60. Dupuis, Fernande (1989). L’expression du sujet dans les subordonnées en ancien français. Université de Montréal: unpublished thesis.

370

References

Durante, Marcello (1981). Dal latino all’italiano moderno. Bologna: Zanichelli. Egidi, Francesco (1965). Dizionario dei dialetti piceni fra Tronto e Aso. Montefiore dell’Aso. Elcock, William (1975). The Romance Languages. Revised with a New Introduction by J. N. Green. London: Faber. Elerick, Charles (1989a). ‘Gapping, Preemptive Markedness, and Word Order in Latin’, in Calboli (ed.), 559–71. ——--(1989b). ‘Word Order in Caesar: SOV/V-1’. University of Texas at El Paso: unpublished manuscript. ——--(1992). ‘Latin Word Order: Living on the Edge’, The Classical World 86: 21–24. ——--(1994). ‘Phenotypic Linearization in Latin, Word Order Universals, and Language Change’, in Herman (ed.), 67–73. Elvira, Javier (1994). ‘Uno en español antiguo’, Verba. Anuario galego de filoloxía 21: 167–82. Emonds, Joseph (1978). ‘The Verbal Complex V’-V in French’, Linguistic Inquiry 9: 151–75. ——--(1985). A Unified Theory of Syntactic Categories. Dordrecht: Foris. England, John (1982). ‘Ser and aver with the Past Participles of Intransitive Verbs in the Works of Don Juan Manuel’, in Don Juan Manuel: VII centenario. Universidad de Murcia & Academia Alfonso X el Sabio, 117–33. Engver, Karl (1972). Place de l’adverbe déterminant un infinitif dans la prose du français contemporain (Studia Romanica Upsaliensia 7). Stockholm: Almqvist & Wiksell. Ernout, Alfred, and Thomas, François ([1953] 1993). Syntaxe latine. Paris: Klincksieck. Ernst, Gerhard, Gleßgen, Martin-D., Schmitt, Christian, and Schweickard, Wolfgang (eds) (2003–6). Romanische Sprachgeschichte: Ein internationales Handbuch zur Geschichte der romanischen Sprachen und ihrer Erforschung (2 Bde). Berlin/New York: Mouton de Gruyter. Ewert, Alfred (1978). The French Language. London: Faber. Fabra, Pompeu (1956). Gramàtica catalana. Barcelona: Teide. Fagyal, Zsuzsanna, Kibbee, Douglas, and Jenkins, Fred (2006). French: A Linguistic Introduction. Cambridge: Cambridge University Press. Fankhänel, Herbert (1938). Verb und Satz in der lateinschen Prosa bis Sallust. Leipzig: Vogel. Fedele, Emily (2010). Verb Movement and Functional Heads in Standard Italian and the Dialects of Italy. University of Cambridge: unpublished thesis. Feix, Joseph (1934). Wortstellung und Satzbau in Petrons Roman. University of Breslau: unpublished thesis. Ferraresi, Gisella, and Goldbach, Maria (2003). ‘Some Reflections on Inertia: Infinitive Complements in Latin’, in Nicole Baumgarten, Caludia Böttger, Markus Motz, and Julia Probst (eds), Übersetzen, Interkulturelle Kommunikation, Spracherwerb und Sprachvermittlung – das Leben mit mehreren Sprachen. Festschrift für Juliane House zum 60. Geburtstag. Zeitschrift für Interkulturellen Fremdsprachenunterricht 8: 240–51 (available online at: ). Fiorentino, Giuliana (2003). ‘Prepositional Objects in Neapolitan’, in Giuliana Fiorentino (ed.), Romance Objects: Transitivity in Romance Languages. Berlin/New York: Mouton de Gruyter, 117–51. Fischer, Anton (1908). Die Stellung der Demonstrativpronomina bei lateinischen Prosaikern. Tübingen: Heckenhauerschen Buchhandlung.

References

371

Fleischman, Suzanne (1982). The Future in Thought and Language: Diachronic Evidence from Romance. Cambridge: Cambridge University Press. Fontana, Josep (1993). Phrase Structure and the Syntax of Clitics in the History of Spanish. University of Pennsylvania: unpublished thesis. ——--(1997). ‘On the Integration of Second Position Phenomena’, in van Kemenade and Vincent (eds), 207–49. Formentin, Vittorio (1996). ‘Flessione bicasuale del pronome relativo in antichi testi italiani centro-meridionali’, Archivio glottologico italiano 81: 133–76. ——--(ed.) (1998). Loise De Rosa, Ricordi. Rome: Salerno. ——--(2001). ‘L’ausiliazione perfettiva in antico napoletano’, Archivio glottologico italiano 86: 79–117. ——--(2002). ‘Tra storia della lingua e filologia: Note sulla sintassi della “Cronica” d’Anonimo romano’, Lingua e stile 37: 203–50. Forteza i Cortès, Tomàs (2008). Gramática de la lengua catalana (2 volums), edited by Maria Pilar Perea. Barcelona/Palma: Publicaions de l’Abadia de Montserrat/Edicions UIB. Fortson, Benjamin (2008). Language and Rhythm in Plautus: Synchronic and Diachronic Studies. Berlin/New York: Mouton de Gruyter. Foubert, Frédéric (2007). ‘L’hyperbate dans les Res gestae Alexandri Macedonis de Julius Valerius’, in Purnelle and Denooz (eds), 55–63. Foulet, Lucien (1923). Petite syntaxe de l’ancien français. 2nd edn. Paris: Champion. Fraenkel, Eduard (1928). Iktus und Akzent im lateinischen Sprachvers. Berlin: Weidmann. ——--(1932). ‘Kolon und Satz: Beobachtungen zur Gliederung des antiken Satzes I’, Nachrichten der Göttinger Gesellschaft der Wissenschaften, Philosophisch-historische Klasse, 197–213. ——--(1933). ‘Kolon und Satz: Beobachtungen zur Gliederung des antiken Satzes II’, Nachrichten der Göttinger Gesellschaft der Wissenschaften, Philosophisch-historische Klasse, 319–54. ——--(1964). ‘Nachtrage zu “Kolon und Satz, II”’, in Kleine Beiträge zur klassischen Philologie I. Rome: Edizioni di storia e letteratura, 131–39. ——--(1965). ‘Noch einmal Kolon und Satz’, Bayerische Akademie der Wissenschaften, Philosophisch-historische Klasse, Sitzungsberichte, 65.2: 1–73. ——--(1968). Leseproben aus Reden Ciceros und Catos. Rome: Edizloni di Storia e Letteratura. Frank, Anette (2003). ‘Projecting LFG F-Structures from Chunks – or Non-configurationality from a Different View Point’, in Miriam Butt and Tracy Holloway King (eds), Proceedings of the LFG03 Conference University at Albany, State University of New York. CSLI online publications (available at: http://csli-publications.stanford.edu). Frascarelli, Mara (ed.) (2006). Phases of Interpretation. Berlin/New York: Mouton de Gruyter. ——--and Hinterhölzl, Roland (2007). ‘Types of Topics in German and Italian’, in Susanne Winkler and Kerstin Schwabe (eds), On Information Structure, Meaning and Form. Amsterdam/Philadelphia: Benjamins, 87–116. Friedrich, Paul (1975). Proto-Indo-European Syntax: The Order of Meaningful Elements. Journal of Indo-European Studies. Monograph 1. Butte: Montana College of Mineral Science and Technology. Fruyt, Michèle (2003). ‘Anaphore, cataphore et déixis dans l’Itinerarium d’Egérie’, in Solin, Leiwo, and Halla-aho (eds), 99–119.

372

References

Fruyt, Michele and Orlandini, Anna (2008). ‘Some Cases of Linguistic Evolution and Grammaticalisation in the Latin Verb’, in Wright (ed.), 230–7. Fugier, Huguette (1983). ‘Le syntagme nominal en latin classique’, in Hildegard Temporini and Wolfgang Haase (eds), Aufstieg und Niedergang der römischen Welt. II.29.1. Berlin: De Gruyter: 212–69. ——--(1994). ‘Le verbe “incorpore”-t-il ses compléments?’, in Herman (ed.), 75–90. Galves, Charlotte (1993). ‘O enfraquecimento da concordância no português brasileiro’, in Ian Roberts and Mary Kato (eds), Português brasileiro: Uma viagem diacrônica. Campinas: Editora da Unicamp, 387–408. Ganzoni, Gian Paul (1983). Grammatica ladina: Grammatica sistematica dal rumantsch d’Engiadin’ Ota per scolars e creschüts da lingua rumauntscha e tudas-cha. Samedan: Lia Rumantscha. García Arias, Xosé Lluis (2003). Gramática histórica de la lengua asturiana. Oviedo: Academia de la Llingua Asturiana. García de Diego, Vicente (1946). Manual de dialectología española. Madrid: Instituto de Cultura Hispánica. ——--(1951). Gramática histórica española. Madrid: Gredos. García de la Fuente, Olegario (1983). ‘Orden de palabras en hebreo, griego, latín y romanceamiento castellano medieval de Joel’, Emerita 51: 41–61, 185–213. García-Hernández, Benjamín (ed.) (1998). Estudios de lingüística latina. Actas del IX coloquio internacional de lingüística latina, Universidad autonoma de Madrid, 14–18 de abril de 1997. Madrid: Ediciones Clásicas. García Sanchidrián, María Luisa (1994). ‘El orden de palabras en las cartas de San Braulio’, in Actas del VIII congreso español de estudios clásicos. Madrid: Sociedad española de estudios clásicos, 549–53. García Santos, Juan Felipe (1992). ‘Extremeño’, in Holtus, Metzeltin, and Schmitt (eds), 701–8. Gartner, Theodor (1883). Raetoromanische Grammatik. Heilbronn: Sammlung Romanischen Grammatiken. ——--(1910). Handbuch der rätoromanischen Sprache und Literatur. Halle a. S.: Niemeyer. Garzonio, Jacopo (2004). ‘Le frasi interrogative non-standard in fiorentino’, Rivista italiana di dialettologia 28: 219–35. ——--(ed.) (2010). Studi sui dialetti della Sicilia (Quaderni di lavoro ASIt n.11). Padua: Unipress. Gerola, Berengario (1949–50). ‘Aspetti della sintassi del nominativo e dell’accustivo nel tardo latino’, Atti dell’Istituto Veneto di SS.LL.AA 108: 207–36. Giacalone Ramat, Anna, and Paolo Ramat (eds) (1998). The Indo-European Languages. London: Routledge. Gianollo, Chiara (2005). Constituent Structure and Parametric Resetting in the Latin DP: A Diachronic Study. University of Pisa: unpublished thesis. ——--(2006). ‘Tracing the Value of Syntactic Parameters in Ancient Languages: The Latin Nominal Phrase’, in Jones-Bley, Huld, Della Volpe, and Robbins Dexter (eds), 142–63. ——--(2007). ‘The Internal Syntax of the Nominal Phrase in Latin: A Diachronic Study’, in Purnelle and Denooz (eds), 65–80.

References

373

Gierling, Diana (1997). ‘Clitic Doubling, Specificity and Focus in Romanian’, in James Black and Virginia Motapanyane (eds), Clitics, Pronouns and Movement. Amsterdam/ Philadelphia: Benjamins, 63–85. Gil, David (1987). ‘Definiteness, Noun Phrase Configurationality, and the Count-Mass Distinction’, in Eric Reuland and Alice ter Meulen (eds), The Representations of (In)definiteness. Cambridge, MA: MIT Press, 254–69. Gildersleeve, Basil, and Lodge, Gonzalez ([1895] 1997). Latin Grammar. Bristol: Bristol Classical Press. Giorgi, Alessandra (2010). ‘La struttura del sintagma nominale’, in Salvi and Renzi (eds), I, 275–94. ——--and Longobardi, Giuseppe (1991). The Syntax of Noun Phrases: Configuration, Parameters and Empty Categories. Cambridge: Cambridge University Press. ——--and Pianesi, Fabio (1996). ‘Verb Movement in Italian and Syncretic Categories’, Probus 8: 137–60. ——--——--(1997). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Giurgea, Ion, and Remberger, Eva-Maria (2009). ‘Postverbal Subjects in Romance Null-Subject Languages: Information-structural Conditions and Variation’, paper presented at Going Romance XXIII, Nice. Giusti, Giuliana (1993). ‘Enclitic Articles and Double Definiteness: A Comparative Analysis of Nominal Structure in Romance and Germanic’, The Linguistic Review 11: 231–55. ——--(1997). ‘The Categorial Status of Determiners’, in Haegeman (ed.), 95–123. ——--(2002). ‘The Functional Structure of Noun Phrases: A Bare Phrase Structure Approach’, in Cinque (ed.), 54–90. ——--(2006). ‘Parallels in Clausal and Nominal Periphery’, in Frascarelli (ed.), 163–84. ——--(2010a). ‘Il sintagma aggettivale’, in Salvi and Renzi (eds), I, 593–616. ——--(2010b). ‘Le espressioni di quantità’, in Salvi and Renzi (eds), I, 377–400. ——--and Oniga, Renato (2006). ‘La struttura del sintagma nominale latino’, in Oniga and Zennaro (eds), 71–99. ——--——--(2007). ‘Core and Periphery in the Latin Noun Phrase’, in Purnelle and Denooz (eds), 81–95. Glatigny, Michel (1967). ‘La place des adjectifs épithètes dans deux oeuvres de Nerval’, Le français moderne 35: 201–20. Golumbia, David (2004). ‘The Interpretation of Nonconfigurationality’, Language and Communication 24: 1–22. Gonzaga, Manuela (2004). ‘The Structure of DP in European Portuguese – Evidence from Adjectives and Possessives’, Harvard Working Papers in Linguistics 10: 19–49. González i Planas, Francesc (2009). ‘Cartografia de les pocisions de subjecte en gascó’, Linguistica occitana 7: 83–99. González Rolán, Tomás (1993). ‘Sobre el origen del giro “habeo + participio”: ¿Innovación o pervivencia?’, Helmantica 44: 517–26. Graffi, Giorgio (1994). Sintassi. Bologna: il Mulino. ——--(1996). ‘Alcune riflessioni sugli imperativi italiani’, in Benincà, Cinque, De Mauro, and Vincent (eds), 143–48.

374

References

Grandgent, Charles (1905). An Outline of the Phonology and Morphology of Old Provençal. Boston: Heath. ——--(1907). An Introduction to Vulgar Latin. Boston: Heath. ——--(1909). An Outline of the Phonology and Morphology of Old Provençal. 2nd edn. Boston: Heath. Greco, Paolo (2007). Accusativus cum infinitivo e subordinate completive con quod, quia e quoniam in alcune cronache latine dell’Italia centro-meridionale (secoli X–XII). University of Naples: unpublished thesis. Green, John (1977). ‘How Free is Word Order in Spanish?’, in Harris (ed.), 7–32. ——--(1982). ‘The Status of the Romance Auxiliaries of Voice’, in Vincent and Harris (eds), 97–138. ——--(1987). ‘The Evolution of Romance Auxiliaries: Criteria and Chronology’, in Harris and Ramat (eds), 255–67. ——--(1988a). ‘Spanish’, in Harris and Vincent (eds), 79–130. ——--(1988b). ‘Romance Creoles’, in Harris and Vincent (eds), 420–73. ——--(2006). ‘The North-South Axis of Romance: Contact Reinforcing Typology’, in Lepschy and Tosi (eds), 73–86. Greenberg, Joseph (1963). ‘Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements’, in Universals of Languages. Cambridge, MA: MIT Press, 58–90. ——--(1966). ‘Language Universals’, in Thomas Sebeok (ed.), Current Trends in Linguistics: III. Theoretical Foundations. The Hague: Mouton, 61–112. Grohmann, Kleanthes (2000). Prolific Peripheries: A Radical View from the Left. University of Maryland, College Park: unpublished thesis. ——--(2003). Prolific Domains: On the Anti-locality of Movement Dependencies. Amsterdam/ Philadelphia: Benjamins. Grosu, Alexander (1988). ‘On the Distribution of Genitive Phrases in Rumanian’, Linguistics 26: 931–49. ——--(1994). Three Studies in Locality and Case. London: Routledge. Guasti, Maria Teresa (1993). Causative and Perception Verbs: A Comparative Approach. Turin: Rosenberg & Sellier. ——--(1996). ‘Semantic Restriction in Romance Causatives and the Incorporation Approach’, Linguistic Inquiry 27: 97–110. Gupton, Timothy (2010). The Syntax-Information Structure Interface: Subjects and Clausal Word Order in Galician. University of Iowa: unpublished thesis. Gut¸ia, Ioan (1967). Grammatica romena moderna. Rome: Bulzoni. Gutiérrez Ordóñez, Salvador (1994). Estructuras comparativas. Madrid: Arco. Habinek, Thomas (1985a). ‘Prose Cola and Poetic Word Order: Observations on Adjectives and Nouns in the Aeneid’, Helios 12: 51–66. ——--(1985b). The Colometry of Latin Prose. Berkeley/Los Angeles/London: University of California Press. Haden, Ernest (1973). ‘French Dialect Geography’, in Thomas Sebeok (ed.), Current Trends in Linguistics. Volume 10: Linguistics in North America. The Hague: Mouton, 422–44. Haegeman, Liliane (ed.) (1997). The New Comparative Syntax. London: Longman.

References

375

Haida, Roman (1928). Die Wortstellung in der Peregrinatio ad Loca Sancta. University of Breslau: unpublished thesis. Haiman, John (1988). ‘Rhaeto-Romance’, in Harris and Vincent (eds), 351–90. ——--and Benincà, Paola (1992). The Rhaeto-Romance Languages. London: Routledge. Hale, Ken (1981). On the Position of Warlpiri in a Typology of the Base. Bloomington, IN: Indiana University Linguistics Club. ——--(1982). ‘Preliminary Remarks on Configurationality’, North East Linguistic Society 12: 86–96. ——--(1983). ‘Warlpiri and the Grammar of Non-configurational Languages’, Natural Language and Linguistic Theory 1: 5–47. ——--(1992). ‘Basic Word Order in Two “Free Word Order” Languages’, in Doris Payne (ed.), Pragmatics of Word Order Flexibility. Amsterdam/Philadelphia: Benjamins, 63–82. ——--(1994). ‘Core Structures and Adjunctions in Warlpiri Syntax’, in Corver and van Riemsdijk (eds), 185–219. Hale, William, and Buck, Carl ([1903] 1994). A Latin Grammar. Tuscaloosa, AL: University of Alabama Press. Hall, Robert A. (1968). ‘“Neuters”, Mass-nouns, and the Ablative in Romance’, Language 44: 480–6. Harbert, Wayne (2007). The Germanic Languages. Cambridge: Cambridge University Press. Harder, Andreas (1998). ‘La declinazione dei verbi in un dialetto di transizione nelle Marche’, in Ruffino (ed.), V, 389–99. Harris, Alice, and Campbell, Lyle (1995). Historical Syntax in Cross-Linguistic Perspective. Cambridge: Cambridge University Press. Harris, Martin (1970). ‘The Verbal Systems of Latin and French’, Transactions of the Philological Society 69: 62–90. ——--(1971). ‘The History of the Conditional Complex from Latin to Spanish: Some Structural Considerations’, Archivum Linguisticum 2: 25–33. ——--(ed.) (1977). Romance Syntax: Synchronic and Diachronic Perspectives. University of Salford. ——--(1978). The Evolution of French Syntax: A Comparative Approach. London: Longman. ——--(1982). ‘The “Past Simple” and the “Present Perfect” in Romance’, in Vincent and Harris (eds), 42–70. ——--(1988). ‘French’, in Harris and Vincent (eds), 209–45. ——--and Ramat, Paolo (eds) (1987). The Historical Development of Auxiliaries. Berlin/New York: Mouton de Gruyter. ——--and Vincent, Nigel (eds) (1988). The Romance Languages. London: Croom Helm. Haspelmath, Martin, Dryer, Matthew, Gil, David, and Comrie, Bernard (eds) (2005). World Atlas of Language Structures. Oxford: Oxford University Press. Haspelmath, Martin, König, Ekkehard, Oesterreicher, Wulf, and Raible, Wolfgang (eds) (2001). Language Typology and Language Universals: An International Handbook. Volume 2. Berlin/New York: Mouton de Gruyter.

376

References

Haudry, Jean (1973). ‘Parataxe, hypotaxe et corrélation dans la phrase latine’, Bulletin de la Société de linguistique 68: 147–86. Haverling, Gerd (2010). ‘Actionality, Tense, and Viewpoint’, in Baldi and Cuzzolin (eds), 2, 277–523. Hawkins, John (1983). Word Order Universals. New York: Academic Press. ——--(1990). Explaining Language Universals. Oxford: Blackwell. ——--and Gilligan, Gary (1988). ‘Prefixing and Suffixing Universals in Relation to Basic Word Order’, Lingua 74: 219–59. Heine, Bernd (1993). Auxiliaries. Oxford: Oxford University Press. Herman, József (1966). ‘Recherches sur l’évolution grammaticale du latin vulgaire: Les emplois “fautifs” du nominatif ’, Acta Classica Universitatis Scientiarum Debrecen 2: 109–12. ——--(1985). ‘La disparition de la déclinaison latine et l’évolution du syntagme nominal’, in Touratier (ed.), 345–60. ——--(ed.) (1987a). Latin vulgaire – latin tardif. Actes du Ier colloque international sur le latin vulgaire et tardif (Pécs, 2–5 septembre 1985). Tübingen: Niemeyer. ——--(1987b). ‘La disparition de -s et la morphologie dialectale du latin parlé’, in Herman (ed.), 97–108. ——--(1989). ‘Accusativus cum infinitivo et subordonné à quod, quia en latin tardif – Nouvelles remarques sur un vieux problème’, in Calboli (ed.), 133–52. ——--(1990). ‘Recherches sur l’ordre des mots dans les plus anciens texts français en prose’, in Du latin aux langues romanes. Etudes de linguistique historique. Tübingen: Niemeyer, 234–85. ——--(ed.) (1994). Linguistic Studies on Latin. Selected papers from the 6th International Colloquium on Latin Linguistics (Budapest, 23–27 March 1991). Amsterdam/Philadelphia: Benjamins. ——--(1995). ‘Les ardoises wisigothiques et le problème de la différenciation territoriale du latin’, in Callebat (ed.), 63–76. ——--(1996a). ‘The End of the History of Latin’, Romance Philology 49: 364–82. ——--(1996b). ‘Remarques sur l’histoire du futur latin et sur la préhistoire du futur roman’, in Rodie Risselada, Jan R. de Jong, and Machtelt Bolkestein (eds), On Latin: Linguistic and Literary Studies in Honour of Harm Pinkster. Amsterdam: Gieben, 57–70. ——--(ed.) (1998a). La transizione dal latino alle lingue romanze. Atti della tavola rotonda di linguistica storica Università Ca’ Foscari di Venezia 14–15 giugno 1996. Tübingen: Niemeyer. ——--(1998b). ‘La chronologie de la transition: Un essai’, in Herman (ed.), 5–25. ——--(2000). Vulgar Latin. Pennsylvania: The Pennsylvania State University Press. Hewson, John, and Bubenik, Vit (2006). From Case to Adposition: The Development of Configurational Syntax in Indo-European Languages. Amsterdam/Philadelphia: Benjamins. Hill, Virginia (2002). ‘Complementizer Phrases in Romanian’, Rivista di linguistica 14: 223–48. ——--(2007). ‘Romanian Adverbs and the Pragmatic Field’, The Linguistic Review 24: 61–86. Hilty, Gerold (1991). ‘Problemas de metafonía en asturiano’, Lletres asturianes 42: 7–15. Hinojo, Gregorio (1985). ‘El orden de palabras en el Satiricón’, in José Melena (ed.), Symbolae Ludovico Mitxelena Oblatae. 1. Vitoria: Universidad País Vasco Victoriaco Vasconum, 245–54.

References

377

——--(1986). ‘El orden de palabras en la Peregrinatio Aetheriae’, Studia Zamorensia 7: 81–7. ——--(2002). ‘El orden de palabras en el latín medieval’, in Maurilio Pérez González (ed.), Actas del III congreso hispánico de latín medieval. Leon: Universidad de León, 627–35. Hofmann, Johannes, and Szantyr, Anton (1965). Lateinische Grammatik. II. Lateinische Syntax und Stilistik. Munich: Beck. Holtus, Günter, Metzeltin, Michael, and Schmitt, Christian (eds) (1988). Lexikon der romanistischen Linguistik. Band IV. Italienisch, Korsisch, Sardisch. Tübingen: Niemeyer. ——--——--——--(eds) (1989). Lexikon der romanistischen Linguistik. Band III. Die einzelnen romanischen Sprachen und Sprachgebiete von der Renaissance bis zur Gegenwart. Rumänisch, Dalmatisch/Istroromanisch, Friaulisch, Ladinisch, Bündnerromanisch. Tübingen: Niemeyer ——--——--——--(eds) (1990). Lexikon der romanistischen Linguistik. Band V, 1. Französisch. Tübingen: Niemeyer. ——--——--——--(eds) (1991). Lexikon der romanistischen Linguistik. Band V, 2. Okzitanisch, Katalanisch. Tübingen: Niemeyer. ——--——--——--(eds) (1992). Lexikon der romanistischen Linguistik. Band VI, 1. AragonesischNavarresisch, Spanisch, Asturianisch-Leonesisch. Tübingen: Niemeyer. ——--——--——--(eds) (1996). Lexikon der romanistischen Linguistik. Band II, 1. Latein und Romanisch. Historisch-vergleichende Grammatik der romanischen Sprachen. Tübingen: Niemeyer. Hopper, Paul, and Traugott, Elizabeth (1993). Grammaticalization. Cambridge: Cambridge University Press. Howe, Chad, and Schwenter, Scott (2008). ‘Variable Constraints on Past Reference in Dialects of Spanish’, in Maurice Westmoreland and Juan Antonio Thomas (eds), Selected Proceedings of the 4th Workshop on Spanish Sociolinguistics. Somerville, MA: Cascadilla Proceedings Project, 100–8. Hualde, José Ignacio (1992). Catalan. London: Routledge. ——--Olarrea, Antxon, and Escobar, Anna María (2001). Introducción a la lingüística hispánica. Cambridge: Cambridge University Press. Huang, James (1994). ‘More on Chinese Word Order and Parametric Theory’, in Barbara Lust, Margarita Suñer, and John Whitman (eds), Syntactic Theory and First Language Acquisition: Cross-Linguistic Perspectives. Volume 1: Heads, Projections, and Learnability. Hillsdale: Laurance Erlbaum Associates, 15–35. Huber, Joseph (1933). Altportugiesisches Elementarbuch. Heidelberg: Carl Winters Universitätsbuchhandlung. Humboldt, Wilhelm von (1836). Über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluß auf die geistige Entwickelung des Menschengeschlechts. Berlin: Druckerei der Königlichen Akademie der Wissenschaftern. Iannace, Gaetano (1983). Interferenza linguistica ai confini fra Stato e Regno: Il dialetto di San Leucio del Sannio. Ravenna: Longo. Ishii, Toru (1999). ‘Cyclic Spell-out and the That-t Effects’, in Sonya Bird, Andrew Carnie, Jason Haugen, and Peter Norquest (eds), Proceedings of the West Coast Conference on Formal Linguistics 18. Somerville: Cascadilla, 220–31. Ive, Antonio (1886). ‘L’antico dialetto di Veglia’, Archivio glottologico italiano 9: 114–87.

378

References

Jaeggli, Osvlado (1981). Topics in Romance Syntax. Dordrecht: Foris. Janson, Tore (1979). Mechanisms of Language Change in Latin. Stockholm: Almqvist & Wiksell. Jelinek, Eloise (1984). ‘Empty Categories, Case, and Configurationality’, Natural Language and Linguistic Theory 2: 39–76. Jespersen, Otto (1917). Negation in English and Other Languages. Copenhagen: A.F. Host. Jones, Mari (2001). Jersey Norman French: A Linguistic Study of an Obsolescent Dialect. Oxford: Blackwell. Jones, Michael (1988a). ‘Sardinian’, in Harris and Vincent (eds), 314–50. ——--(1988b). ‘Auxiliary Verbs in Sardinian’, Transactions of the Philological Society 86: 173– 203. ——--(1992). ‘Infinitives with Specified Subjects in Sardinian’, in Laeufer and Morgan (eds), 295–309. ——--(1993). Sardinian Syntax. London: Routledge. ——--(1995). ‘The Prepositional Accusative in Sardinian: Its Distribution and Syntactic Repercussions’, in Smith and Maiden (eds), 37–75. ——--(1996). Foundations of French Syntax. Cambridge: Cambridge University Press. ——--(1997). ‘Sardinian’, in Maiden and Parry (eds), 376–84. Jones, Samantha (2008). Extraordinary Agreement in Ripano: A Constraint-Based Approach. University of Cambridge: unpublished B.A. thesis. Jones-Bley, Karlene, Huld, Martin E., Della Volpe, Angela, and Robbins Dexter, Miriam (eds) (2006). Proceedings of the Seventeenth Annual UCLA Indo-European Conference. Los Angeles October 27–28, 2005 (Journal of Indo-European Studies Monograph Series 52). Washington DC: Institute for the Study of Man. Jordana, Cèsar August (1933). El català i el castellà comparats. Barcelona: Barcino. Joseph, John (1991). ‘Latin Prepositions and Romance Syntax’, in Wanner and Kibbee (eds), 187–99. ——--(1992). ‘The Gascon Enunciative as [sic] Syntactic Solution’, in Laeufer and Morgan (eds), 481–95. Juge, Matthew (2006). ‘Morphological Factors in the Grammaticalization of the Catalan “Go” Past’, Diachronica 232: 313–39. Kabatek, Johannes (forthc.). ‘Koines and Scriptae’, in Maiden, Smith, and Ledgeway (eds). Kaiser, Georg (1999). ‘A ordem das palavras e a posição do verbo finito no português antigo’, in Pál Forenc (ed.), Actas do congresso internacional por motivo dos vinte anos do portoguês no ensino superior. Budapest: Departamento de língua e literatura portuguesas da Faculdade de Letras da Universidade Eötvös Loránd de Budapeste, 248–61. ——--(2002). Die Verbstellung und Verbstellungswandel in den romanischen Sprachen. Tübingen: Niemeyer. ——--(2002–3). ‘Di Verb-Zeit-Stellung in Rätoromanischen: Ein typologischer Vergleich’, Ladinia 26–7: 313–34. Kalepky, Theodor (1913). ‘Präpositionale Passivobjeckte im Spanischen, Portugiesischen und Rumänischen’, Zeitschrift für romanische Philologie 37: 358–64. Karlsen, Epsen (2001). ‘The Accusatiuus cum Infinitiuo and Quod Clauses in the Revelaciones of St. Bridget of Sweden’, in Moussy, Dangel, Fruyt, Nadjo, and Sznajder (eds), 371–84.

References

379

Karlsson, Keith (1981). Syntax and Affixation: The Evolution of MENTE in Latin and Romance. Tübingen: Niemeyer. Kayne, Richard (1975). French Syntax: The Transformational Cycle. Cambridge, MA: MIT Press. ——--(1991). ‘Romance Clitics, Verb Movement and PRO’, Linguistic Inquiry 22: 647–86. ——--(1993). ‘Towards a Modular Theory of Auxiliary Selection’, Studia Linguistica 47: 3–31. ——--(1994). The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Keil, Heinrich (1868). Grammatici latini ex recensione Henrici Keilii. Vol. V. Leipzig: Teubner. Kemenade, Ans van, and Vincent, Nigel (eds) (1997). Parameters of Morphosyntactic Change. Cambridge: Cambridge University Press. Kessler, Brett (1995). ‘Discontinuous Constituents in Latin’. Washington University in St. Louis: unpublished manuscript (available at: ). Kiss, Katalin (1987). Configurationality in Hungarian. Dordrecht: Reidel. ——--(1998). ‘Identificational Focus versus Informational Focus’, Language 74: 245–73. ——--(2008). ‘Free Word Order, (Non)Configurationality, and Phases’, Linguistic Inquiry 39: 441–75. Klausenburger, Jurgen (2000). Grammaticalization: Studies in Latin and Romance Morphosyntax. Amsterdam/Philadelphia: Benjamins. Koch, Monica (1974). ‘A Demystification of Syntactic Drift’, Montreal Working Papers in Linguistics 3: 63–114. Koeneman, Olaf, and Zeijlstra, Hedde (forthc.). ‘Resurrecting the Rich Agreement Hypothesis: Weak Isn’t Strong Enough’, to appear in Duk-Ho An and Soo-Yeon Kim (eds), Movement in Minimalism. Proceedings of the 12th Seoul Conference on Generative Grammar. Koll, Hans-Georg (1965). ‘Zur Stellung des Verbs im spätantiken und frühmittelalterlichen Latein’, Mittellateinisches Jahrbuch 2: 241–72. Konneker, Beverly Hill (1975). ‘Word Order Change in Italic’, in Robin Grossman, James San, and Timothy Vance (eds), Papers from the Eleventh Regional Meeting of the Chicago Linguistic Society. Chicago: The Linguistic Society, 366–70. Koschwitz, Eduard (1894). Grammaire historique de la langue des Félibres. Paris: Greifswald. Kramer, Johannes (1989). ‘Areallinguistik II. Aromunisch’, in Holtus, Metzeltin, and Schmitt (eds), 423–35. Kroch, Anthony (1989). ‘Reflexes of Grammar in Patterns of Language Changes’, Journal of Language and Variation and Change 1: 199–244. Kroll, Wilhelm (1912). ‘Der lateinische Relativsatz’, Glotta 3: 1–18. ——--(1918). ‘Anfangsstellung des Verbums im Lateinischen’, Glotta 9: 112–23. Kühner, Raphael, and Stegmann, Carl (1912–14). Ausführliche Grammatik der lateinischen Sprache (2 vols). Hannover: Hahnsche Buchhandlung. La Fauci, Nunzio (1988). Oggetti e soggetti nella formazione della morfosintassi romanza. Pisa: Giardini (English translation: La Fauci, Nunzio (1994). Objects and Subjects in the Formation of Romance Morphosyntax. Bloomington, Indiana: IULC). ——--(1991). ‘La continuità nella diversità formale: Aspetti di morfosintassi diacronica romanza’, in Vincenzo Orioles (ed.), Innovazione e conservazione nelle lingue. Pisa: Giardini, 135–58.

380

References

La Fauci, Nunzio (1992). ‘Capitoli di morfosintassi siciliana antica: Tassonomia dei costrutti medi e ausiliari perfettivi’, in Studi linguistici e filologici offerti a Girolamo Caracausi. Palermo: Centro di studi filologici e linguistici siciliani, 185–220. ——--(1997). Per una teoria grammaticale del mutamento morfosintattico: Dal latino verso il romanzo. Pisa: ETS. ——--(1998). ‘Riflettendo sul mutamento morfosintattico: Nel latino, verso il romanzo’, in Ramat and Roma (eds), 519–45. ——--(2004). ‘Armonia differenziale dell’ausiliazione perfettiva nel volgare di Dante’, in Dardano and Frenguelli (eds), 237–52. ——--(2006). ‘Dinamiche sistematiche: Perifrasi perfettive e futuro sintetico: Dal latino al romanzo’, in Oniga and Zennaro (eds), 101–31. ——--(2011a). Relazioni e differenze: Questioni di linguistica razionale. Palermo: Sellerio. ——--(2011b). ‘Tempo e tipi: Dinamiche morfosintattiche latino-romanze’, in Relazioni e differenze: Questioni di linguistica razionale. Palermo: Sellerio, 93–128. ——--(2011c). ‘Armonia differenziale nel volgare di Dante’, in Relazioni e differenze: Questioni di linguistica razionale. Palermo: Sellerio, 153–67. Laenzlinger, Christopher (2005a). ‘French Adjective Ordering: Perspectives on DP-internal Movement Types’, Lingua 115: 645–89. ——--(2005b). ‘Some Notes on DP-internal Movements’, Generative Grammar in Geneva 4: 227–60. Laeufer, Christiane, and Morgan, Terrell (eds) (1992). Theoretical Analyses in Romance Linguistics. Amsterdam/Philadelphia: Benjamins. Lafont, Robert (1967). La phrase occitane: Essai d’analyse systématique. Montpellier: PUF. ——--(1991). ‘Okzitanisch: Interne Sprachgeschichte I. Grammatik-Histoire interne de la langue I. Grammaire’, in Holtus, Metzeltin, and Schmitt (eds), 1–18. Lakoff, George (1972). ‘Another Look at Drift’, in Robert Stockwell and Ronald Macauley (eds), Linguistic Change and Generative Theory. Bloomington, IN: Indiana University Press, 172–98. Lakoff, Robin (1968). Abstract Syntax and Latin Complementation. Cambridge, MA: MIT Press. Lamarche, Jacques (1991). ‘Problems for N -movement to NumP’, Probus 3: 215–36. Lambrecht, Knud (1994). Information Structure and Sentence Form. Cambridge: Cambridge University Press. Landau, Idan (2010). The Locative Syntax of Experiencers. Cambridge, MA: MIT Press. Langslow, David (2000). Medical Latin in the Roman Empire. Oxford: Oxford University Press. ——--(forthc.). ‘Praetor Urbanus – Urbanus Praetor: Some Aspects of Attributive Adjective Placement in Latin’, in Philomen Probert and Andreas Willi (eds), Laws and Rules in IndoEuropean. Oxford: Oxford University Press. Lanly, André (1973). ‘Sur le conditionnel français et roman (à propos d’une remarque de M. Imbs)’, Travaux de linguistique et de littérature 11: 391–99. Lapesa, Rafael (1974). ‘El sustantivo sin actualizador en español’, in Rafael Cano and María Teresa Echenique (eds), Estudios de morfosintaxis histórica. Madrid: Gredos, 436–54. ——--(1975). ‘La colocación del calificativo atributivo en español’, in Homenaje a la memoria de Don Antonio Rodríguez-Moniño 1910–1970. Madrid: Castalia, 329–45.

References

381

——--(1980). Historia de la lengua española. 8th edn. Madrid: Gredos. Laughton, Erich (1970). ‘Review of Eduard Fraenkel’s Leseproben aus Reden Ciceros und Catos’, Journal of Roman Studies 60: 188–94. Lausberg, Heinrich (1956a). Romanische Sprachwissenschaft. I. Einleitung und Vokalismus. The Hague: Mouton. ——--(1956b). Romanische Sprachwissenschaft. II. Konsonantismus. The Hague: Mouton. ——--(1962). Romanische Sprachwissenschaft. III. Formenlehre. The Hague: Mouton. Leal Cruz, Pedro (2003). El español tradicional de La Palma. Gobierno de Canarias. Ledgeway, Adam (1997). ‘Asyndetic Complementation in Neapolitan Dialect’, The Italianist 17: 231–73. ——--(1998). ‘Variation in the Romance Infinitive: The Case of the Southern Calabrian Inflected Infinitive’, Transactions of the Philological Society 96: 1–61. ——--(2000). A Comparative Syntax of the Dialects of Southern Italy: A Minimalist Approach. Oxford: Blackwell. ——--(2003a). ‘L’estensione dell’ausiliare perfettivo avere nell’antico napoletano: Intransitività scissa condizionata da fattori modali’, Archivio glottologico italiano 88: 27–71. ——--(2003b). ‘Linguistic Theory and the Mysteries of Italian Dialects’, in Anna Laura Lepschy and Arturo Tosi (eds), Multilingualism in Italy: Past and Present. Oxford: Legenda, 108–40. ——--(2004a). ‘Il sistema completivo dei dialetti meridionali: La doppia serie di complementatori’, Rivista italiana di dialettologia 27: 89–147. ——--(2004b). ‘Lo sviluppo dei dimostrativi nei dialetti centromeridionali’, Lingua e stile 39: 65–112. ——--(2004c). ‘Extraordinary Agreement: The Case of the Ripano Verb’, Italian Linguistics Round Table, St Catherine’s College, University of Cambridge. ——--(2005). ‘Moving through the Left Periphery: The Dual Complementiser System in the Dialects of Southern Italy’, Transactions of the Philological Society 103: 336–96. ——--(2006). ‘The Dual Complementiser System in Southern Italy: Spirito greco, materia romanza? ’, in Lepschy and Tosi (eds), 112–26. ——--(2007a). ‘La posizione dell’aggettivo nella storia del napoletano’, in Bentley and Ledgeway (eds), 104–25. ——--(2007b). ‘Old Neapolitan Word Order: Some Initial Observations’, in Lepschy and Tosi (eds), 121–49. ——--(2007c). ‘Diachrony and Finiteness: Subordination in the Dialects of Southern Italy’, in Irina Nikolaeva (ed.), Finiteness: Theoretical and Empirical Foundations. Oxford: Oxford University Press, 335–65. ——--(2008a). ‘Satisfying V2: Sì Clauses in Old Neapolitan’. Journal of Linguistics 44: 437–40. ——--(2008b). ‘The Grammaticalisation of Progressive and Andative Aspect in the Dialects of Apulia’, paper presented at the 4th Oxford-Kobe Linguistics Seminar, The History and Structure of the Romance Languages. ——--(2009a). Grammatica diacronica del napoletano (Beihefte zur Zeitschrift für romanische Philologie Band 350). Tübingen: Niemeyer. ——--(2009b). ‘Aspetti della sintassi della periferia sinistra del cosentino’, in Diego Pescarini (ed.), Studi sui dialetti della Calabria (Quaderni di lavoro ASIt n.9). Padua: Unipress, 3–24.

382

References

Ledgeway, Adam (2009c). ‘Case Study in Agreement: Ripatransone’, paper presented at the Department of General Linguistics, Graduate School, University of Zurich. ——--(2010a). ‘The Clausal Domain: CP Structure and the Left Periphery’, in D’Alessandro, Ledgeway, and Roberts (eds), 38–51. ——--(2010b). ‘Subject Licensing in CP: The Neapolitan Double-Subject Construction’, in Benincà and Munaro (eds), 257–96. ——--(2011). ‘Morphosyntactic Typology and Change’, in Maiden, Smith, and Ledgeway (eds), 382–471, 724–34. ——--(forthc. a). ‘La sopravvivenza del sistema dei doppi complementatori nei dialetti meridionali’, in Patrizia Del Puente (ed.), Dialetti: Per parlare e per parlarne. Secondo convegno internazionale di dialettologia. Progetto dell’Atlante linguistico della Basilicata. ——--(forthc. b). ‘Grammaticalization from Latin to Romance’, in Bernd Heine and Heiko Narrog (eds), Handbook of Grammaticalization. Oxford: Oxford University Press. ——--(forthc. c). ‘Reconstructing Complementiser-drop in the Dialects of the Salento: A Syntactic or Phonological Phenomenon?’, in Theresa Biberauer and George Walkenden (eds), Proceedings of the XII Diachronic Generative Syntax Meeting, 14 – 16 July 2010, Queens’ College, University of Cambridge. Oxford: Oxford University Press. ——--(forthc. d). ‘Adverb Agreement and Split Intransitivity: Evidence from Southern Italy’, Archivio glottologico italiano. ——--(forthc. e). ‘When Theory Meets Data: The Case of the Placiti Cassinesi ’, in Sylvia Adamson and Wendy Bennett (eds), Language, Text, & History: Linguistics and Philology in the 21st Century. Oxford: Blackwell. Special Issue of the Transactions of the Philological Society. ——--and Lombardi, Alessandra (2005). ‘Verb Movement, Adverbs and Clitic Positions in Romance’, Probus 17: 79–113. Legate, Julie Anne (2001). ‘The Configurational Structure of a Nonconfigurational Language’, in Pierre Pica (ed.), Linguistic Variation Book. Volume 1 2001. Amsterdam/Philadelphia: Benjamins, 63–99. ——--(2002). Warlpiri: Theoretical Implications. MIT: unpublished thesis. Legendre, Géraldine (2000). ‘Optimal Romanian Clitics: A Cross-linguistic Perspective’, in Alboiu and Motapanyane (eds), 227–64. ——--(2010). ‘A Formal Typology of Person-based Auxiliary Selection in Italo-Romance’, in D’Alesssandro, Ledgeway, and Roberts (eds), 186–200. Lehmann, Christian (1984). Der Relativsatz. Tübingen: Narr. ——--(1985). ‘Latin Case Relations in Typological Perspective’, in Touratier (ed.), 81–104. ——--(1991). ‘The Latin Nominal Group in Typological Perspective’, in Robert Coleman (ed.), New Studies in Latin Linguistics. Amsterdam/Philadelphia: Benjamins, 203–32. Lehmann, Winfred (1972a). ‘Contemporary Linguistics and Indo-European Studies’, Publications of the Modern Language Association of America 87: 976–93. ——--(1972b). ‘Converging Theories in Linguistics’, Language 48: 266–75. ——--(1974). Proto-Indo-European Syntax. Austin: University of Texas Press. ——--(1992). Historical Linguistics. London: Routledge. Lemieux, Monique, and Dupuis, Fernande (1995). ‘The Locus of Verb Movement in Nonasymmetric Verb-second Languages: The Case of Middle French’, in Battye and Roberts (eds), 80–109.

References

383

Lepschy, Giulio (1984). ‘Costruzioni impersonali con se in veneziano’, in Manlio Cortelazzo (ed.), Guida ai dialetti veneti VI. Padua: CLEUP, 69–79. Lepschy, Anna Laura, and Lepschy, Giulio (1988). The Italian Language Today. London: Hutchinson. ——--and Tosi, Arturo (eds) (2006). Rethinking Languages in Contact: The Case of Italian. Oxford: Legenda. ——--——--(eds) (2007). Histories and Dictionaries of the Languages of Italy. Ravenna: Longo. Leumann, Manu, and Hofmann, Johannes (1928). Lateinische Grammatik. Munich: Beck. Lightfoot, David (1979). Principles of Diachronic Syntax. Cambridge: Cambridge University Press. ——--(1991). How to Set Parameters: Arguments from Language Change. Cambridge: Cambridge University Press. ——--and Hornstein, Norbert (eds) (1994). Verb movement. Cambridge: Cambridge University Press. Linde, Paul (1923). ‘Die Stellung des Verbs in der lateinischen Prosa’, Glotta 12: 153–78. Lindsay, Wallace (1907). Syntax of Platus. Oxford: James Parker & Co. Lipski, John (1994). Latin American Spanish. London: Longman. Lisón, Nicolás (2001). El orden de las palabras en los grupos nominales en latín. Zaragoza: Monografías de la Universidad de Zaragoza. Lloyd, Paul (1987). From Latin to Spanish: Historical Phonology and Morphology of the Spanish Language. Philadelphia: American Philosophical Society. Lobo, Maria (2001). ‘On Gerund Clauses of Portuguese Dialects’, in Alexandre Veiga, Víctor M. Longa, and JoDee Anderson (eds), El verbo. Entre el léxico y la gramática. Lugo: Tris Tram, 107–18. Löfstedt, Einar (1933). Syntactica: Studien und Beiträge zur historischen Syntax des Lateins. Zweiter Teil: Syntaktisch-Stilistische Gesichtspunkte und Probleme. Lund: Gleerup. ——--(1956). Syntactica: Studien und Beiträge zur historischen Syntax des Lateins (2 Bde). Lund: Gleerup. ——--(1959). Late Latin. Oslo: H. Aschehoug & Co. Lois, Ximena (1989). Aspects de la syntaxe de l’espagnol et théorie de la grammaire. Université de Paris VIII: unpublished thesis. Lombard, Alf (1974). La langue roumaine. Une présentation. Paris: Klincksieck. Lombardi, Alessandra (1997). The Grammar of Complementation in the Dialects of Calabria. University of Manchester: unpublished thesis. ——--(1998). ‘Calabria greca e Calabria latina da Rohlfs ai giorni nostri: La sintassi dei verbi modali-aspettuali’, in Ramat and Roma (eds), 613–26. ——--(2007). ‘Posizione dei clitici e ordine dei costituenti nella lingua sarda medievale’, in Bentley and Ledgeway (eds), 133–47. ——--and Middleton, Roberta (2004). ‘Alcune osservazioni sull’ordine delle parole negli antichi volgari italiani’, in Dardano and Frenguelli (eds), 553–82. Longa, Víctor (1994). ‘The Galician Inflected Infinitive and the Theory of UG’, Catalan Working Papers in Linguistics 4: 23–44.

384

References

Longo, Beatriz (1998). ‘Perífrases temporais no português falado’, Veredas 2: 9–24. ——--and Souza Campos, Odette (2002). ‘A auxiliaridade: Perífrases de aspecto e tempo no português falado’, in Maria Bernadeta Abaurre and Ângela Rodrigues (eds), Gramática do português falado. Volumen VIII. Campinas: Ed. Unicamp, 445–77. Longobardi, Giuseppe (1994). ‘Reference and Proper Names: A Theory of N-movement in Syntax and Logical Form’, Linguistic Inquiry 25: 609–65. ——--(2001). ‘The Structure of DPs: Some Principles, Parameters and Problems’ in Baltin and Collins (eds), 562–603. López, Luis (2009). A Derivational Syntax for Information Structure. Oxford: Oxford University Press. López Fonseca, Antonio (2009). ‘Modo y modalidad’, in Baños Baños (ed.), 443–68. Loporcaro, Michele (1986). ‘L’infinito coniugato nell’Italia centro-meridionale: Ipotesi genetica e ricostruzione storica’, Italia dialettale 49: 609–65. ——--(1988). Grammatica storica del dialetto di Altamura. Pisa: Giardini. ——--(1997). ‘Puglia and Salento’, in Maiden and Parry (eds), 338–48. ——--(1998). Sintassi comparata dell’accordo participiale romanzo. Turin: Rosenberg & Sellier. ——--(1999). ‘Il futuro CANTARE-HABEO nell’Italia meridionale’, Archivio glottologico italiano 84: 67–114. ——--(2001). ‘La selezione dell’ausiliare nei dialetti italiani: Dati e teorie’, in Albano Leoni, Krosbakken, Sornicola, and Stromboli (eds), 455–76. ——--(2002). ‘Review of Allières 2001’, Romanische Forschungen 114: 345–51. ——--([2003] 2004). ‘The Unaccusative Hypothesis and Participial Absolutes in Italian: Perlmutter’s Generalization Revised’, Rivista di Linguistica/Italian Journal of Linguistics 15: 199–26. ——--(2007). ‘On Triple auxiliation’, Linguistics 45: 173–222. ——--(2010). ‘Per lo studio della morfosintassi dei dialetti lucani: Acquisizioni recenti e nuove prospettive’, paper presented at Dialetti per parlare e parlarne. II convegno internazionale di dialettologia, Potenza – Venosa – Matera. ——--(2011a). ‘Phonological Processes’, in Maiden, Smith, and Ledgeway (eds), 109–54. ——--(2011b). ‘Auxiliaries over Time: Syntax and Semantics’, paper presented at Italian (Dialects) in Diachrony, University of Leiden. ——--and Paciaroni, Tania (forthc.). ‘Four-gender Systems in Indo-European’, Folia Linguistica. ——--and Vigolo, M. Teresa (1995). ‘Ricerche sintattiche sul confine dialettale veneto-trentino in Valsugana: L’accordo del participio passato’, in Emanuele Banfi, Giovanni Bonfadini, Patrizia Cordin, and Maria Iliescu (eds), Italia settentrionale: Crocevia di idiomi romanzi. Tübingen: Niemeyer, 87–101. ——--——--([2002–3] 2005). ‘Accordo per genere del verbo finito in italo-romanzo: L’arco alpino orientale’, Italia dialettale 63–4: 7–32. Lot, Ferdinand (1931). ‘À quelle époque a-t-on cessé de parler latin?’, Archivum Latinitatis Medii Aevii (Bulletin du Cange) 6: 97–159. Lüdtke, Helmut (1974). ‘Die Mundart von Ripatransone – ein sprachtypologisches Kuriosum’, Acta Universitatis Carolinae – Philologica 5: 173–77.

References

385

——--(1976). ‘La declinazione dei verbi in un dialetto di transizione marchigiano-abruzzese’, Abruzzo 14: 79–84. Luís, Ana, and Spencer, Andrew (2005). ‘A Paradigm Function Account of “Mesoclisis” in European Portuguese’, in Geert Booij and Jaap van Marle (eds), Yearbook of Morphology 2004. Dordrecht: Springer, 177–228. Lurà, Franco (1987). Il dialetto del Mendrisiotto: Descrizione sincronica e confronto coll’italiano. Mendrisio and Chiasso: Unione di Banche Svizzere. Luraghi, Silvia (1997). ‘Omission of the Direct Object in Classical Latin’, Indogermanische Forschungen 102: 239–57. ——--(1998). ‘Omissione dell’oggetto diretto in frasi coordinate: Dal latino all’italiano’, in Ramat and Roma (eds), 183–96. ——--(2006). ‘Null Objects in Latin and Greek and the Relevance of Typology for Language Reconstruction’, in Jones-Bley, Huld, Della Volpe, and Miriam Robbins Dexter (eds), 235–56. ——--(2010a). ‘Configurationality, Transitivity, and Valency in Indo-European and Beyond’, paper presented at the conference Variation and Change in Argument Realization, Universities of Naples Federico II & Bergen, Naples and Capri. ——--(2010b). ‘Adverbial Phrases’, in Baldi and Cuzzolin (eds), 2, 19–107. Lyons, Christopher (1986). ‘On the Origin of the Old French Strong-weak Possessive Distinction’, Transactions of the Philological Society 84: 1–41. ——--(1999). Definiteness. Cambridge: Cambridge University Press. Magni, Elisabetta (2000). ‘L’ordine delle parole nel latino pompeiano: Sulle tracce di una deriva’, Archivio glottologico italiano 85: 3–37. ——--(2009). ‘The Evolution of Latin Word (Dis)order’, in Scalise, Magni, and Bisetto (eds), 225–51. ——--(2010). ‘Mood and Modality’, in Baldi and Cuzzolin (eds), 2, 193–75. Maiden, Martin (1991). Interactive Morphonology: Metaphony in Italy. London: Routledge. ——--(1995). A Linguistic History of Italian. London: Longman. ——--(1996). ‘Ipotesi sulle origini del condizionale analitico come “futuro del passato” in italiano’, in Benincà, Cinque, De Mauro, and Vincent (eds), 149–73. ——--(1998). Storia linguistica dell’italiano. Bologna: il Mulino. ——--(2000). ‘Il sistema desinenziale del sostantivo nell’italo-romanzo preletterario: Ricostruzione parziale a partire dai dialetti moderni (il significato storico di plurali del tipo amici)’, in József Herman and Anna Marinetti (eds), La preistoria dell’italiano. Atti della tavola rotonda di linguistica storica, Università Ca’ Foscari di Venezia 11–13 giugno 1998. Tübingen: Niemeyer, 167–79. ——--(2011a). ‘Morphological Persistence’, in Maiden, Smith, and Ledgeway (eds), 155–215. ——--(2011b). ‘Morphophonological Innovation’, in Maiden, Smith, and Ledgeway (eds), 216–67. ——--and Parry, Mair (eds) (1997). The Dialects of Italy. London: Routledge. ——--and Robustelli, Cecilia (2007). A Reference Grammar of Modern Italian. London: Arnold.

386

References

Maiden, Martin, Smith, John Charles, and Ledgeway, Adam (eds) (2011). The Cambridge History of the Romance Languages. Volume 1: Structures. Cambridge: Cambridge University Press. ——--——--——--(eds) (forthc.). The Cambridge History of the Romance Languages. Volume 2: Contexts. Cambridge: Cambridge University Press. Malkiel, Yakov (1974). ‘Distinctive Traits of Romance Linguistics’, in Dell Hymes (ed.), Language in Culture and Society: A Reader in Linguistics and Anthropology. New York: Harper and Row, 671–88. ——--(1977). ‘Factors in the Unity of Romania’, Romance Notes 18: 263–71. ——--(1978). ‘The Classification of the Romance Languages’, Romance Philology 31: 467–500. Mallinson, Graham (1986). Rumanian. London: Croom Helm. ——--(1988). ‘Rumanian’, in Harris and Vincent (eds), 391–419. Mancarella, Giovan Battista (ed.) (1998). Salento. Monografia regionale della ‘Carta dei Dialetti Italiani’. Lecce: Edizioni del Grifo. Mancini, Anna Maria (1993). ‘Le caratteristiche morfosintattiche del dialetto di Ripatransone (AP), alla luce di nuove ricerche’, in Sanzio Balducci (ed.), I dialetti delle Marche meridionali. Alessandria: Edizioni dell’Orso, 111–36. Manoliu Manea, Maria (1989). ‘Le roumain: Morphosyntaxe’, in Holtus, Metzeltin, and Schmitt (eds), 101–14. Manzini, Maria Rita, and Savoia, Leonardo (2005). I dialetti italiani e romanci: Morfosintassi generativa (3 vols). Alessandria: Edizioni dell’Orso. Maracz, Laszlo, and Muysken, Pieter (eds) (1989). Configurationality: The Typology of Asymmetries. Dordrecht: Foris. Marazzini, Claudio (1994). La lingua italiana: Profilo storico. Bologna: il Mulino. Marcato, Gianna (ed.) (2002). La dialettologia oltre il 2001. Padua: Unipress. ——--and Ursini, Flavia (1998). Dialetti veneti: Grammatica e storia. Padua: Unipress. Marchetti, Pascal (1974). Le corse sans peine. Chennevières-sur-Marne: Assimil. Marouzeau, Jules (1922). L’ordre des mots dans la phrase latine. I. Les groupes nominaux. Paris: Champion. ——--(1938). L’ordre des mots dans la phrase latine. II. Le verbe. Paris: Les Belles Lettres. ——--(1949). L’ordre des mots dans la phrase latine. III. Les articulations de l’énoncé. Paris: Les Belles Lettres. ——--(1953). L’ordre des mots en latin. Volume complémentaire. Paris: Les Belles Lettres. Martín Zorraquino, María Antonia (1976). ‘A + objeto directo en el Cantar de Mio Cid’, in Germán Colón and Robert Kopp (eds), Mélanges offerts à C. Th. Gossen. Bern-Liège: Francke, 554–66. Mattoso Câmara, Joaquim (1972). The Portuguese Language. Chicago/London: University of Chicago Press. Maturi, Pietro (2002). Dialetti e substandardizzazione nel Sannio beneventano. Frankfurt: Lang. Maurer, Theodoro (1968). O infinito flexionado português: Estudo histórico-descritivo. São Paulo: Editora Nacional.

References

387

McFadden, Thomas, and Artemis Alexiadou (2006a). ‘Auxiliary Selection and Counterfactuality in the History of English and Germanic’, in Jutt Hartmann and Molnárfi Lászloacute (eds), Comparative Studies in Germanic Syntax. Amsterdam/ Philadelphia: Benjamins, 237–62. ——--——--(2006b). ‘Counterfactuals and the Rise of HAVE in the History of English’, in John Alderete, Chung hye Han, and Alexei Kochetov (eds), Proceedings of the 24th West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Press, 272–80. ——--——--(2006c). ‘Counterfactuals and the Loss of BE in the History of English’, in Aviad Eilam, Tatjana Scheffler, and Joshua Tauber (eds), Proceedings of the 29th Penn Linguistics Colloquium, Penn Working Papers in Linguistics 12: 251–64. ——--——--(2010). ‘Perfects, Resultatives, and Auxiliaries in Earlier English’, Linguistic Inquiry 41: 389–425. Mclachlan, Kathryn (forthc.). Latin Word Order: The Placement of Dependent Genitives. University of Oxford: D.Phil. thesis. Meier, Harri (1948). ‘Sobre as origens do acusativo preposicional nas línguas românicas’, in Ensaios de filologia românica. Lisbon: Edição da Revista de Portugal, 115–64. Meillet, Antoine (1908). Introduction à l’étude comparée des langues indo-européennes. 2nd edn. Paris: Hachette. ——--([1928] 1977). Esquisse d’une histoire de la langue latine (avec une bibliographie mise à jour et complétée par J. Perrot). Paris: Klincksieck. ——--([1937] 1964). Linguistique historique et générale. Paris: Champion. Menéndez Pidal, Ramón (1956). Orígenes del español: Estado lingüístico de la península ibérica hasta el siglo XI. 4th edn. Madrid: Espasa Calpe. ——--(1966). Manual de gramática histórica española. 12th edn. Madrid: Espasa Calpe. Mensching, Guido (2000). Infinitive Constructions with Specified Subjects: A Syntactic Analysis of the Romance Languages. Oxford: Oxford University Press. ——--and Remberger, Eva-Maria (2010a). ‘Focus Fronting and the Left Periphery in Sardinian’, in D’Alessandro, Ledgeway, and Roberts (eds), 261–76. ——--——--(2010b). ‘La periferia sinistra romanza: Topicalizzazione, focalizzazione e interrogazione in sardo’, in Maria Iliescu, Heidi Siller-Runggaldier, and Paul Danler (eds), Actes du XXVe congrès international de linguistique et de philologie romanes VII. Tübingen: Niemeyer, 189–97. Mereu, Lunella (2004). La sintassi delle lingue del mondo. Rome/Bari: Laterza. Merlo, Clemente (1906–7). ‘Dei continuatori del lat. ille in alcuni dialetti dell’Italia centromeridionale’, Zeitschrift für romanische Philologie 30: 11–25; 31: 157–63. ——--(1917). ‘L’articolo determinativo nel dialetto di Molfetta’, Studi romanzi 14: 69–99. Messing, Gordon (1972). ‘The Romance Collective Neuter and the Survival of the Latin Ablative’, Glotta 50: 255–62. Meyer-Lübke, Wilhelm (1890). Grammatik der romanischen Sprachen. I. Lautlehre. Leipzig: Reisland. ——--(1894). Grammatik der romanischen Sprachen. II. Formenlehre. Leipzig: Reisland. ——--(1899). Grammatik der romanischen Sprachen. III. Syntax. Leipzig: Reisland. ——--(1901). Einführung in das Studium der romanischen Sprachwissenschaft. Heidelberg: Winter.

388

References

Meyer-Lübke, Wilhelm (1902). Grammatik der romanischen Sprachen. IV. Register. Leipzig: Reisland. Miller, Gary (1975). ‘Indo-European: VSO, SOV, SVO, or All Three?’, Lingua 37: 31–57. Möbitz, Otto (1924). ‘Die Stellung des Verbums in den Schriften des Apuleius’, Glotta 13: 116–26. Mohrmann, Christine (1952). ‘Les formes du latin “vulgaire”: Essai de chronologie et de systématisation de l’époque augustéenne aux langues romanes’, in Actes du premier congrès de la Fédération internationale des études classiques. Paris: Klincksieck, 207–20. Molinelli, Piera (1998). ‘Dai casi alle preposizioni in latino: Analisi sociolinguistica e spiegazione tipologica’, in Ramat and Roma (eds), 147–66. Moll, Francesc de B. (1993). El parlar de Mallorca. Mallorca: Moll. ——--(1997). Gramàtica catalana, referida especialment a les Illes Balears. Palma de Mallorca: Editorial Moll. ——--(2006). Gramàtica històrica catalana. Valencia: Publicacions de Universitat de València. Monachesi, Paola (2005). The Verbal Complex in Romance: A Case Study in Grammatical Interfaces. Oxford: Oxford University Press. Morais, Maria Aparecida (2003). ‘EPP generalizado, sujeito nulo e línguas de configuração discursiva’, Letras de hoje 38: 71–98. Moran, Josep (2007). ‘L’article en els topònims: teoria i pràctica’, Bulletí homenatge a Andreu Romà i Espí 106–7: 126–33. Moreno Hernández, Antonio (1989). ‘Tipología lingüística y orden de la palabras en el latín de Terencio’, Actas del VII congreso español de estudios clásicos. Volumen I. Madrid: Universidad Complutense, 523–8. Morimoto, Yukiko (2001). ‘Deriving the Directionality Parameter in OT-LFG’, in Miriam Butt and Tracy Holloway King (eds), Proceedings of the LFG01 Conference, University of Hong Kong. CSLI online publications (available at ). Morin, Annick (2005). ‘Syntaxe de la particule que en gascon’, Linguistica occitana 4: 60–7. Moro, Andres (1993). I predicati nominali e la struttura della frase. Padua: Unipress. Motapanyane, Virginia (1989). ‘La position du sujet dans une langue à l’ordre SVO/VSO’, Rivista di grammatica generativa 14: 75–103. Moussy, Claude, Dangel, Jacqueline, Fruyt, Michèle, Nadjo, Léon, and Sznajder, Lyliane (eds) (2001). De lingua latina nouae questiones. Actes du Xe colloque international de linguistique latine Paris-Sèvres, 19–23 avril 1999. Louvain/Paris/Sterling: Editions Peeters. Mulder, Hotze (1991). ‘Object Continuity and Elliptical Objects in Latin’, Romance Linguistics and Literature Review 4: 12–26. Muldowney, Mary (1937). Word Order in the Works of St. Augustine. The Catholic University of America, Washington, DC: unpublished thesis. Muller, Henri (1929). A Chronology of Vulgar Latin. Halle: Niemeyer. ——--(1939). ‘On the Origin of French Word Order’, The Romantic Review 30: 52–62. ——--and Taylor, Pauline (1932). A Chrestomathy of Vulgar Latin. Boston/New York: Heath. Munaro, Nicola (2002). ‘Splitting up Subject Clitic-Verb Inversion’, in Claire Beyssade, Reineke Bok-Bennema, Frank Drijkoningen, and Paola Monachesi (eds), Romance Languages and Linguistic Theory 2000. Amsterdam/Philadelphia: Benjamins, 233–52.

References

389

——--(2003). ‘On Some Differences between Exclamative and Interrogative Wh-phrases in Bellunese: Further Evidence for a Split-CP hypothesis’, in Tortora (ed.), 137–51. ——--(2004). ‘Computational Puzzles of Conditional Clause Preposing’, in Anne Marie Di Sciullo and Rodolfo Delmonte (eds), UG and External Systems. Amsterdam/Philadelphia: Benjamins, 73–94. ——--(2010). ‘Criterial Conditions for Wh-structures: Evidence from Wh-exclamatives in Northern Italian Dialects’, in D’Alessandro, Ledgeway, and Roberts (eds), 292–311. Nagore Laín, Francho (2001). ‘La conjugación verbal compuesta en aragonés medieval’, Alazet 13: 65–88. Naudeau, Olivier (1979). ‘ “Za, sa, cha, ca” articles occitans et le latin IPSAM: Un cas de dépendance de la morpho-syntaxe par rapport à la phonétique’, Romance Philology 32: 397–410. Nichols, Johanna (1986). ‘Head-marking and Dependent-marking Grammar’, Language 62: 56–119. Nisbet, Robin (2000). ‘The Word Order of Horace’s Odes’, in James Adams and Roland Mayer (eds), Aspects of the Language of Latin Poetry. Oxford: Oxford University Press, 135–54. Nocentini, Alberto (1985). ‘Sulla genesi dell’oggetto preposizionale nelle lingue romanze’, in Studi linguistici e filologici per Carlo Alberto Mastrelli. Pisa: Pacini, 299–311. ——--(1990). ‘L’uso dei dimostrativi nella Peregrinatio Egeriae e la genesi dell’articolo romanzo’, in Atti del convegno internazionale sulla Peregrinatio Egeriae. Arezzo: Accademia Petrarca di lettere e arti e scienze, 137–58. ——--(2001). ‘La genesi del futuro e del condizionale sintetico romanzo’, Zeitschrift für romanische Philologie 117: 367–401. Norberg, Dag (1943). Syntaktische Forschungen auf dem Gebiete des Spätlateins und des frühen Mittellateins. Uppsala: Lundqvist; Leipzig: Harassowitz. ——--(1944). Beiträge zur spätlateinischen Syntax. Uppsala: Almqvist & Wiksell. Nordlinger, Rachel (1998). Constructive Case: Evidence from Australian Languages. Stanford, CA: Centre for the Study of Language and Information. Nouvel, Alain (1975). L’occitan sans peine. Chennevières-sur-Marne: Assimil. Nunes, José Joaquim (1989). Compêndio de gramática histórica portuguesa. Lisbon: Clássica. Nuñez, Salvador (1991). Semántica de la modalidad en latín. Granada: Universidad de Granada. ——--(1998). ‘Habere + participio de perfecto y la cateogría de la anterioridad en latín y en las lenguas rómanicas’, in García-Hernández (ed.), 199–204. Oliviéri, Michèle (2009). ‘Typology or Reconstruction: the Benefits of Dialetology for Diachronic Analysis’, paper presented at Going Romance XXIII, Nice. Oniga, Renato (1998). I composti nominali latini: Una morfologia generativa. Bologna: Pàtron. ——--(2004). Il latino: Breve introduzione linguistica. Milan: Franco Angeli. ——--and Zennaro, Luigi (eds) (2006). Atti della Giornata di linguistica latina, Venezia, 7 maggio 2004. Venice: Cafoscarina. Orinsky, Kurt (1923). ‘Die Wortstellung bei Gaius’, Glotta 12: 83–100. Ostafin, David (1986). Studies in Latin Word Order: A Transformational Approach. University of Connecticut: unpublished thesis.

390

References

Paciaroni, Tania (2009). ‘Sull’ausiliazione perfettiva in maceratese’, in Bruno Moretti, Elena Pandolfi, and Matteo Casoni (eds), Linguisti in contatto: Ricerche di linguistica italiana in Svizzera. Bellinzona: OLSI, 37–58. Pádua, Maria da Piedade Canaes e Mariz (1960). A ordem das palavras no português arcaico. Coimbra: Instituto des Estudos Românicos. Palmer, Frank (2001). Mood and Modality. Cambridge: Cambridge University Press. Palmer, Leonard ([1954] 1990). The Latin Language. Bristol: Bristol Classical Press. Panchón, Federico (1986). ‘Orden de palabras en latín (César, BG.; Cicerón Pro Milone)’, Studia zamorensia 7: 213–29. Panhuis, Dirk (1980). ‘Gapping in Latin’, The Classical Journal 75: 229–41. ——--(1982). The Communicative Perspective in the Sentence: A Study of Latin Word Order. Amsterdam/Philadelphia: Benjamins. ——--(1984). ‘Is Latin an SOV Language?’, Indogermanische Forschungen 89: 140–59. Paoli, Sandra (2002). ‘Il doppio “che” nei dialetti piemontesi’, in Marcato (ed.), 231–36. ——--(2003a). COMP and the Left-Periphery: Comparative Evidence from Romance. University of Manchester: unpublished thesis. ——--(2003b). ‘Mapping out the Left Periphery of the Clause: Evidence from North-Western Italian Varieties’, in Quer, Schroten, Scorretti, Sleeman, and Verheugd (eds), 267–85. ——--(2005). ‘COMP: A Multi-talented Category. Evidence from Romance’, in Laura Brugè, Giuliana Giusti, Nicola Munaro, Walter Schweikert, and Giuseppina Turan (eds), Contributions to the XXX Incontro di Grammatica Generativa. Venice: Cafoscarina, 185–202. ——--(2007). ‘The Fine Structure of the Left Periphery: COMPs and Subjects. Evidence from Romance’, Lingua 117: 1057–79. ——--(2010). ‘In Focus: An Investigation of Information and Contrastive Constructions’, in D’Alessandro, Ledgeway, and Roberts (eds), 277–91. Parascandola, Vittorio (1976). Vèfio: Folk-glossario del dialetto procidano. Napoli: Alfredo Guida Editore. Parkinson, Jennie (2009). A Diachronic Study into the Distribution of Two Italo-Romance Synthetic Conditional Forms. University of St Andrews: unpublished thesis. Parkinson, Stephen (1988). ‘Portuguese’, in Harris and Vincent (eds), 131–69. Parrino, Flavio (1967). ‘Su alcune particolarità della coniugazione nel dialetto di Ripatransone’, L’Italia dialettale 30: 156–66. Parry, Mair (1997). ‘Negation’, in Maiden and Parry (eds), 179–85. ——--(2007). ‘The Interaction of Semantics and Syntax in the Spread of Relative che in the Early Vernaculars of Italy’, in Bentley and Ledgeway (eds), 200–19. ——--and Lombardi, Alessandra (2007). ‘The Interaction of Semantics, Pragmatics and Syntax in the Spread of the Articles in the Early Vernaculars of Italy’, in Lepschy and Tosi (eds), 77–97. Payne, John (1985). ‘Negation’, in Timothy Shopen (ed.), Language Typology and Syntactic Description. Volume 1: Clause Structures. Cambridge: Cambridge University Press, 197–242. Penney, John (1999). ‘Archaism and Innovation in Latin Poetic Syntax’, in James Adams and Roland Mayer (eds), Aspects of the Language of Latin Poetry. Oxford: Oxford University Press, 249–68.

References

391

——--(2005). ‘Connections in Archaic Latin Prose’, in Reinhardt, Lapidge, and Adams (eds), 37–51. Penny, Ralph (1970). ‘Mass Nouns and Metaphony in the Dialects of North-Western Spain’, Archivum Linguisticum 1: 21–30. ——--(2000). Variation and Change in Spanish. Cambridge: Cambridge University Press. ——--(2002). A History of the Spanish Language. 2nd edn. Cambridge: Cambridge University Press. Pensado, Carmen (1985). ‘La creación del objeto directo preposicional y la flexión de los pronombres personales en las lenguas románicas’, Revue roumaine de linguistique 30: 123–58. ——--(1986). ‘Inversion de marcage et perte du système casuel en ancien français’, Zeitschrift für romanische Philologie 102: 271–96. ——--(ed.) (1995). El complemento directo preposicional. Madrid: Visor Libros. Pensalfini, Rob (2004). ‘Towards a Typology of Configurationality’, Natural Language and Linguistic Theory 22: 359–408. Perea, Maria Pilar (2005). ‘El parlar salat’, in Antoni M. Alcover: Dialectòleg, gramàtic i polemista. Barcelona: Publicacions de l’Abadia de Montserrat, 69–83. Pérez Saldanya, Manuel, and Hualde, José Ignacio (2003). ‘On the Origin and Evolution of the Catalan Periphrastic Preterit’, in Pusch and Welsch (eds), 47–60. Perrochat, Paul (1926). ‘Sur un principe d’ordre des mots: La place du verbe dans la subordonnée’, Revue des études latines 4: 50–60. ——--(1932). Recherches sur la valeur et l’emploi de l’infinitif subordonné en latin. Paris: Les Belles Lettres. Petersmann, Hubert (1977). Petrons Urbane Prosa. Vienna: Böhlaus. Peverini, Claudia (2008). ‘Subject-Verb Non-agreement in Marchigiano’, paper presented at the XXXVI Romance Linguistics Seminar, Trinity Hall, Cambridge. Pieroni, Silvia (1999). ‘Non-promotional Objects in Late Latin’, Verbum 21: 117–29. ——--(2001). ‘Subject Properties, Zero-Anaphoras and Semantic Roles’, in Moussy, Dangel, Fruyt, Nadjo, and Sznajder (eds), 539–52. ——--(2010). ‘Deixis and Anaphora’, in Baldi and Cuzzolin (eds), 3, 389–501. Pinkster, Harm (1969). ‘A, B and C Coordination’, Mnemosyme 22: 258–67. ——--(ed.) (1983). Latin Linguistics and Linguistic Theory. Proceedings of the 1st International Colloquium on Latin Linguistics Amsterdam, April 1981. Amsterdam/Philadelphia: Benjamins. ——--(1987). ‘The Strategy and Chronology of the Development of Future and Perfect Tense Auxiliares in Latin’, in Harris and Ramat (eds), 193–223. ——--(1989). ‘Some Methodological Remarks on Research on Future Tense Auxiliaries in Latin’, in Calboli (ed.), 311–26. ——--(1990a). Latin Syntax and Semantics. London: Routledge. ——--(1990b). ‘The Development of Cases and Adpositions in Latin’, in Harm Pinkster and Inge Genee (eds), Unity in Diversity: Papers Presented to Simon C. Dik on his 50th Birthday. Dordrecht: Foris, 195–209. ——--(1990c). ‘La coordination’, L’information grammaticale 46: 8–13.

392

References

Pinkster, Harm (1991). ‘Evidence for SVO in Latin?’, in Roger Wright (ed.), Latin and the Early Romance Languages in the Middle Ages. London: Routledge, 69–82. ——--(1992). ‘Notes on the Syntax of Celsius’, Mnemosyne 45: 513–24. ——--(1993a). ‘Tipo di testo e variazione linguistica in latino’, in Gerold Hilty (ed.), Actes du XXe Congrès international de linguistique et philologie romanes. Université de Zurich (6–11 avril 1992). Tome II, section II, Analyse de conversation. Tübingen: Narr, 645–50. ——--(1993b). ‘Chronologie et cohérence de quelques évolutions latines et romanes’, in Gerold Hilty (ed.), Actes du XXe congrès international de linguistique et philologie romanes. Tome III, section IV, Typologie des langues romanes. Tübingen: Narr, 239–50. ——--(1996). ‘Linguistique latine et pragmatique’, Lengauje en contexto 64: 245–60. ——--(2005a). ‘Changing Patterns of Discontinuity in Latin’, paper presented at the 13th International Colloquium on Latin Linguistics, Facultés Universitaires Saint-Louis, Brussels. ——--(2005b). ‘The Language of Pliny the Elder’, in Reinhardt, Lapidge, and Adams (eds), 239–56. ——--(2006). ‘Review of Maria Selig, Die Entwicklung der Nominaldeterminanten im Spätlatein (ScriptOralia, 26), Tübingen 1992 (Narr)’, Mittellateinisches Jahrbuch 31: 143–49. Pires, Acrisio (2002). ‘Cue-based Change: Inflection and Subjects in the History of Portuguese Infinitives’, in David Lightfoot (ed.), Syntactic Effects of Morphological Change. Oxford: Oxford University Press, 143–50. Pîrvulescu, Michaela, and Roberge, Yves (2000). ‘The Syntax and Morphology of Romanian Imperatives’, in Alboiu and Motapanyane (eds), 295–312. Pisano, Simone (2009). ‘Il futuro e il condizionale analitici in alcune varietà sarde moderne: Genesi di marche grammaticali da forme verbali lessicalmente piene’, Bollettino di studi sardi 2: 147–66. ——-- (2010). ‘L’utilizzo dei tempi sovraccomposti in alcune varietà sarde’, Lingua e stile 45: 123–31. Plangg, Guntram (1989). ‘Ladinisch: Interne Sprachgeschichte I. Grammatik’, in Holtus, Metzeltin, and Schmitt (eds), 646–67. Poletto, Cecilia (1993). La sintassi dei pronomi soggetto nei dialetti settentrionali. Padua: Unipress. ——--(1995). ‘The Diachronic Development of Subject Clitics in Northern Eastern Italian Dialects’, in Battye and Roberts (eds), 295–324. ——--(2000). The Higher Functional Field. Oxford: Oxford University Press. ——--(2001). ‘Complementizer Deletion and Verb Movement in Standard Italian’, in Cinque and Salvi (eds), 265–86. ——--(2005). ‘Sì and e as CP Expletives in Old Italian’, in Montserat Batllori, Maria-Luïsa Hernanz, Carme Picallo, and Francesc Roca (eds), Grammaticalization and Parametric Variation. Oxford: Oxford University Press, 206–35. ——--(2006). ‘Parallel Phases: A Study on the High and Low Left Periphery of Old Italian’, in Frascarelli (ed.), 261–94. ——--(2007). ‘Tra tutto e niente: Ordini OV in italiano antico’, Laboratorio sulle varietà romanze antiche 1: 33–48. ——--(2010). ‘La struttura della frase’, in Salvi and Renzi (eds), I, 60–75. ——--(forthc.). Word Order in Old Italian. Oxford: Oxford University Press.

References

393

——--and Zanuttini, Raffaella (2000). ‘Marking New Information in the Left Periphery: The Case of pa in Central Rhaeto-Romance’, paper presented at the VI Giornata di Dialettologia, University of Padua. ——--——--(2003). ‘Making Imperatives: Evidence from Central Rhaetoromance’, in Tortora (ed.), 175–206. Politzer, Robert (1958). ‘On Late Latin Word Order’, Symposium 12: 178–82. Pollock, Jean-Yves (1989). ‘Verb Movement, Universal Grammar, and the Structure of IP’, Linguistic Inquiry 20: 365–424. Polo, Chiara (2004). Word Order in Latin, Italian and Slovene between Morphology and Syntax. Padua: Unipress. ——--(2005). ‘Latin Word Order in Generative Perspective: An Explanatory Proposal within the Sentence Domain’, in Katalin Kiss (ed.), Universal Grammar in the Reconstruction of Ancient Languages. Berlin/New York: Mouton de Gruyter, 373–427. ——--(2006). ‘L’ordine delle parole nel latino di Petronio: Una proposta di analisi’, in Oniga and Zennaro (eds), 203–26. Pontes, Eunice (1987). O tópico no português do Brasil. Campinas: Pontes. Pope, Mildred (1952). From Latin to Old French, with Especial Consideration of Anglo-Norman. Manchester: Manchester University Press. Porzio Gernia, Maria Luisa (1986). ‘Latin Declension: A Theoretical and Methodological Approach’, in Gaultiero Caboli (ed.), Papers on Grammar. II. Bologna: Editrice, 1–18. Posner, Rebecca (1984). ‘Double Negatives, Negative Polarity, and Negative Incorporation in Romance: A Historical and Comparative View’, Transactions of the Philological Society 82: 1–26. ——--(1985a). ‘Post-verbal Negation in Non-standard French: A Historical and Comparative View’, Romance Philology 39: 170–97. ——--(1985b). ‘L’histoire de la négation et la typologie romane’, in Actes du XVIIème congrès international de linguistique et philologie romanes (Aix-en-Provence, 29 aoüt – 3 septembre 1983). Volume 2: Linguistique comparée et typologie des langues romanes. Aix-en-Provence: Université de Provence, 265–71. ——--(1996). The Romance Languages. Cambridge: Cambridge University Press. Postal, Paul (1971). Crossover Phenomena. New York: Holt, Reinhart, and Winston. Pottier, Bernard (1961). ‘Sobre el concepto de “verbo auxiliar”’, Nueva revista de filología hispánica 15: 325–31. Pountain, Christopher (1982). ‘*ESSERE/STARE as a Romance Phenomenon’, in Vincent and Harris (eds), 139–60. ——--(1998). ‘Nuevo enfoque de la posición del adjetivo atributivo’, in Ruffino (ed.) II, 697–708. ——--(2001). A History of the Spanish Language through Texts. London: Routledge. ——--(forthc.). ‘Latin and the Written Structure of Romance’, in Maiden, Smith, and Ledgeway (eds). Powell, Jonathan (2010). ‘Hyperbaton and Register, in Cicero’, in Dickey and Chahoud (eds), 163–85. Pozas-Loyo, Julia (2008). ‘On the Evolution of un in Medieval and Classical Spanish’, paper presented at the XXXVI Romance Linguistics Seminar, Trinity Hall, Cambridge.

394

References

Price, Glanville (1962). ‘The Negative Particles pas, mie, and point in French’, Archivum Linguisticum 14: 14–34. ——--(1971). The French Language: Present and Past. London: Arnold. ——--(1986). ‘Aspects de l’histoire de la négation en français’, in Actes du XVIIème congrès international de linguistique et philologie romanes (Aix-en-Provence, 29 aoüt – 3 septembre 1983). Volume 2: Morphosyntaxe des langues romanes. Aix-en-Provence: Université de Provence, 569–75. ——--(2000). Encyclopedia of European Languages. London: Blackwell. Purnelle, Gérald, and Denooz, Jospeh (eds) (2007). Ordre et cohérence en latin. Communications présentées au 13e colloque international de linguistique latine (BruxellesLiège, 4–9 avril 2005). Geneva: Droz. Pusch, Claus, and Welsch, Andreas (eds) (2003). Verbalperiphrasen in den (ibero-)romanischen Sprachen. Hamburg: Helmut Buske. Quer, Josep, Schroten, Jan, Scorretti, Mauro, Sleeman, Petra, and Verheugd, Els (eds) (2003). Romance Languages and Linguistic Theory 2001. Amsterdam/Philadelphia: Benjamins. Rabella i Ribas, Joan (1998). Un matrimoni desavingut i un gat metzinat: Procés criminal barceloní del segle XIV. Barclona: Institut d’Estudis Catalans. Radatz, Hans-Ingo (2001). Die Semantik der Adjektivstellung. Tübingen: Niemeyer. Raiskila, Pirjo (1989). ‘Periphrastic Use of habere in Tertullian’, in Calboli (ed.), 209–17. Ramat, Paolo (1980). ‘Zur Typologie des pompejanischen Lateins’, in Günter Brettschneider and Christian Lehmann (eds), Wege zur Universalienforschung. Tübingen: Narr, 187–91. ——--(1984). ‘Per una tipologia del latino pompeiano’, in Linguistica tipologica. Bologna: il Mulino, 137–42. ——--(1987). ‘Introductory paper’, in Harris and Ramat (eds), 3–19. ——--(1994). ‘On Latin Absolute Constructions’, in Herman (ed.), 259–68. ——--and Roma, Elisa (eds) (1998). Sintassi storica. Atti del XXX congresso internazionale della Società di linguistica italiana, Pavia, 26–28 settembre 1996. Rome: Bulzoni. Ramsden, Herbert (1963). Weak Pronoun Position in the Early Romance Languages. Manchester: Manchester University Press. Raposo, Eduardo (1986). ‘On the Null Object in European Portuguese’, in Osvaldo Jaeggli and Carmen Silva-Corvalán (eds), Studies in Romance Linguistics. Dordrecht: Foris, 373–90. ——--(1987). ‘Case Theory and Infl-to-Comp: The Inflected Infinitive in European Portuguese’, Linguistic Inquiry 18: 85–109. Ravier, Xavier (1991). ‘L’occitan. Les aires linguistiques’, in Holtus, Metzeltin, and Schmitt (eds), 80–105. Reichenkron, Günter (1951). ‘Das präpositionale Akkusativobjekt im ältesten Spanisch’, Romanische Forschungen 63: 342–97. Reiner, Edwin (1968). La place de l’adjectif épithète en français. Vienna and Stuttgart: Wilhelm Braumüller. Reinhardt, Tobias, Lapidge, Michael, and Adams, James (eds) (2005). The Language of Latin Prose. Oxford: Oxford University Press. Remberger, Eva-Maria (2006). Hilfsverben: Eine minimalistiche Analyse am Beispiel des Italienischen und Sardischen. Tübingen: Niemeyer.

References

395

Renzi, Lorenzo (1976). ‘Grammatica e storia dell’articolo italiano’, Studi di grammatica italiana 5: 5–42. ——--(1985). Nuova introduzione alla filologia romanza. Bologna: il Mulino. ——--(ed.) (1988). Grande grammatica italiana di consultazione. Volume 1: La frase. I sintagmi nominale e preposizionale. Bologna: il Mulino. ——--(1997). ‘The Stucture of the Noun Phrase’, in Maiden and Parry, 162–70. ——--(2010). ‘L’articolo’, in Salvi and Renzi (eds), I, 295–347. Renzi, Lorenzo, and Salvi, Giampaolo (eds) (1991). Grande grammatica italiana di consultazione II. I sintagmi verbale, aggettivale, avverbiale. La subordinazione. Bologna: il Mulino. ——--Salvi, Giampaolo, and Cardinaletti, Anna (2001). Grande grammatica italiana di consultazione. III. Tipi di frase, deissi e formazione delle parole. Bologna: il Mulino. ——--and Vanelli, Laura (1983). ‘I pronomi soggetto in alcune varietà romanze’, in Paola Benincà, Michele Cortelazzo, Aldo Prosdocimi, Laura Vanelli, and Alberto Zamboni (eds), Studi in onore di Giovan Battista Pellegrini. Pisa: Pacini, 121–45. Ribeiro, Ilza (1995). ‘Evidence for a Verb-second Phase in Old Portuguese’, in Battye and Roberts (eds), 110–39. Ricca, Davide (2010). ‘Adverbs’, in Baldi and Cuzzolin (eds), 2, 109–91. Richter, Michael (1983). ‘A quelle époque a-t-on cessé de parler latin en Gaule? A propos d’une question mal posée’, Annales: Économies, sociétés, civilisations 38: 439–48. Rigau, Gemma, and Prieto, Pilar (2007). ‘The Syntax-Prosody Interface: Catalan Interrogative Sentences Headed by que’, Journal of Portuguese Linguistics 6: 29–59. Rivero, María-Luisa (1994a). ‘Clause-structure and V-movement in the Languages of the Balkans’, Natural Language and Linguistic Theory 12: 63–120. ——--(1994b). ‘Negation, Imperatives and Wackernagel Effects’, Rivista di linguistica 6: 39–66. Rizzi, Elena, and Molinelli, Piera (1994). ‘Latin and Greek Compared: Word Order in a Bilingual Papyrian Text’, in Herman (ed.), 113–27. Rizzi, Luigi (1982). Issues in Italian Syntax. Dordrecht: Foris. ——--(1986). ‘Null Objects in Italian and the Theory of pro’, Linguistic Inquiry 17: 501–57. ——--(1997). ‘The Fine Structure of the Left Periphery’, in Liliane Haegeman (ed.), Elements of Grammar. Dordrecht: Kluwer, 281–337. ——--(2001). ‘On the Position “Int(errogative)” in the Left Periphery of the Clause’, in Cinque and Salvi (eds), 287–96. ——--(ed.) (2004). The Structure of CP and IP: The Cartography of Syntactic Structures. Volume 2. Oxford: Oxford University of Press. Roberge, Yves (1990). The Syntactic Recoverability of Null Arguments. Montreal/London: McGill-Queen’s University Press. Roberts, Ian (1985). ‘Agreement Parameters and the Development of English Modal Auxiliaries’, Natural Language and Linguistic Theory 3: 21–58. ——--(1993a). Verbs and Diachronic Syntax: A Comparative History of English and French. Dordrecht: Kluwer. ——--(1993b). ‘A Formal Account of Grammaticalisation in the History of Romance Futures’, Folia linguistica historica 13: 219–58.

396

References

Roberts, Ian (1994). ‘Two Types of Head Movement in Romance’, in Lightfoot and Hornstein (eds), 207–42. ——--(1997). ‘Directionality and Word Order Change in the History of English’, in van Kemenade and Vincent (eds), 397–426. ——--and Roussou, Anna (2003). Syntactic Change: A Minimalist Approach to Grammaticalization. Cambridge: Cambridge University Press. Roehrs, Dorian, and Labelle, Marie (2003). ‘The Left Periphery in Child French: Evidence for a Simply Split CP’, in Quer, Schroten, Scoretti, Sleeman, and Vergheugd (eds), 279–94. Rögnvaldsson, Eiríkur. (1995). ‘Old Icelandic: A Non-configurational Language?’, NorthWestern European Language Evolution 26: 3–29. Rohlfs, Gerhard (1966). Grammatica storica della lingua italiana e dei suoi dialetti: Fonetica. Turin: Einaudi. ——--(1968). Grammatica storica della lingua italiana e dei suoi dialetti. Turin: Einaudi. ——--(1969). Grammatica storica della lingua italiana e dei suoi dialetti: Sintassi e formazione delle parole. Turin: Einaudi. ——--(1970). Le gascon: Étude de philologie pyrénéenne. Tübingen: Niemeyer. ——--(1971). ‘Autour de l’accusatif prépositionnel dans les langues romanes’, Revue de linguistique romane 35: 312–33. ——--(1975). Rätoromanisch. Munich: Beck. Rohrbacher, Bernhard (1997). Morphology-Driven Syntax. Amsterdam/Philadelphia: Benjamins. Roncaglia, Aurelio (1965). La lingua dei trovatori. Rome: Edizioni dell’Ateneo. Ronjat, Jules (1913). Essai de syntaxe des parlers provençaux modernes. Macon: Protat Frères. ——--(1930–41). Grammaire istorique [sic] des parlers provençaux modernes (4 tomes). Montpellier: Société des Langues Romanes. Rooryck, Johan (1992). ‘Romance Enclitic Ordering and Universal Grammar’, Linguistic Review 9: 219–50. Rosén, Hannah (1994). ‘The Definite Article in the Making, Nominal Constituent Order, and Related Phenomena’, in Herman (ed.), 131–50. ——--(1998). ‘Latin Presentational Sentences’, in Benjamín García-Hernández (ed.), Estudios de lingüística latina. Madrid: Ediciones Clásicas, 723–42. ——--(1999). Latine loqui: Trends and Directions in the Crystallization of Classical Latin. Munich: Fink. ——--(2009). ‘Coherence, Sentence Modification, and Sentence-part Modification’, in Baldi and Cuzzolin (eds), 317–441. Rosenkranz, Bernhard (1933). ‘Die Stellung des attributiven Genitivs im Italischen’, Indogermanische Forschungen 51: 131–39. Rosetti, Alexandru (1968). Istoria limbii române de la origine pînă în secolul al XVII-lea. Bucharest: Editura pentru literatură. Ross, John Robert (1967). Constraints on Variables in Syntax. MIT: unpublished thesis. ——--(1970). ‘Gapping and the Order of Constituents’, in Manfred Bierwisch and Karl Heidolph (eds), Progress in Linguistics. The Hague: Mouton, 249–59. Rovai, Francesco (2005). ‘L’estensione dell’accusativo in latino tardo e medievale’, Archivio glottologico italiano 90: 54–89.

References

397

Rowlett, Paul (1998). Sentential Negation in French. Oxford: Oxford University Press. ——--(2007). The Syntax of French. Cambridge: Cambridge University Press. Rubio, Lisardo ([1972] 1982). ‘Estructura del estilo indirecto en latín y en castellano: Problemas de traducción’, in Lisardo Rubio, Introducción a la sintaxis estructural del latín. Barcelona: Ariel, 257–70. Rudin, Catherine (1988). ‘On Multiple Questions and Multiple Wh-fronting’, Natural Language and Linguistic Theory 6: 445–501. Ruffino, Giovanni (ed.) (1998a). Atti del XXI congresso internazionale di linguistica e filologia romanza. Centro di studi filologici e linguistici siciliani, Università di Palermo 18–24 settembre 1995. Volume I, sezione 1, Grammatica storica delle lingue romanze. Tübingen: Niemeyer. ——--(ed.) (1998b). Atti del XXI congresso internazionale di linguistica e filologia romanza. Centro di studi filologici e linguistici siciliani, Università di Palermo 18–24 settembre 1995. Volume II, sezione 2, Morfologia e sintassi delle lingue romanze. Tübingen: Niemeyer. ——--(ed.) (1998c). Atti del XXI congresso internazionale di linguistica e filologia romanza. Centro di studi filologici e linguistici siciliani, Università di Palermo 18–24 settembre 1995. Volume V, sezione 5, Dialettologia, geolinguistica, sociolinguistica. Tübingen: Niemeyer. Russell, Donald (2001). Quintilian. The Orator’s Education. Books 1–2. Edited and Translated by Donald A. Russell. Cambridge, MA: Harvard University Press. Saeed, John (1996). ‘Head-marking and Pronominal Clitics in Somali’, in Richard Hayward and I.M. Lewis (eds), Voice and Power: The Culture of Language in North-East Africa. Essays in Honour of B. W. Andrzejewski (African languages and cultures, supplement 3). London: SOAS, 37–49. Saltarelli, Mario (1999). ‘An Integrated Theory of Adjectives: The Alternating Adjectives of Romance’, in Pierrette Bouillon and Evelyne Viegas (eds), Description des adjectifs pour le traitement informatique. Proceedings of TALN, Cargese, 45–56. ——--(2007). ‘Mass Nouns and the Structure of the Nominal Phrase: Evidence from the Marsican Dialects’, paper presented at the 2nd Cambridge Dialect Syntax Meeting, University of Cambridge. Salvi, Giampaolo (1982). ‘Sulla storia sintattica della costruzione romanza habeo + part. perf.’, Revue romane 17: 118–33. ——--(1985). ‘L’infinito con l’articolo’, in Annalisa Franchi De Bellis and Leonardo Savoia (eds), Sintassi e morfologia della lingua italiana d’uso. Rome: Bulzoni, 243–68. ——--(1987). ‘Restructuring in the Evolution of Romance Auxiliaries’, in Harris and Ramat (eds), 225–36. ——--(1997). ‘Ladin’, in Maiden and Parry (eds), 286–94. ——--(2000). ‘La formazione del sistema V2 delle lingue romanze antiche’, Lingua e stile 35: 665–92. ——--(2004). La formazione della struttura di frase romanza. Ordine delle parole e clitici dal latino alle lingue romanze antiche. Tübingen: Niemeyer. ——--(2005). ‘Some Firm Points on Latin Word Order: The Left Periphery’, in Katalin É. Kiss (ed.), Universal Grammar in the Reconstruction of Ancient Languages. Berlin/New York: Mouton de Gruyter, 429–56. ——--(2011). ‘Morphosyntactic Persistence’, in Maiden, Smith, and Ledgeway (eds), 318–81.

398

References

Salvi, Giampaolo and Renzi, Lorenzo (eds) (2010). Grammatica dell’italiano antico (II volumi). Bologna: il Mulino. Sandfeld, Kristian, and Olsen, Hedvig (1936). Syntaxe roumaine. I. Emploi des mots à flexion. Copenhagen: Munksgaard. ——--——--(1960). Syntaxe roumaine. II. Les groupes de mots. Copenhagen: Munksgaard. ——--——--(1962). Syntaxe roumaine. III. Structure de la proposition. Copenhagen: Munksgaard. Sankoff, Gillian, and Thibault, Pierrette (1980). ‘The Alternation between the Auxiliaries avoir and être in Montréal French’, in Ulrich Sankoff (ed.), The Social Life of Language. Philadelphia: University of Pennsylvania, 295–310. Santorini, Beatrice (1995). ‘Two Types of Verb Second in the History of Yiddish’, in Battye and Roberts (eds.), 53–79. Saralegui, Carmen (1992). ‘Aragonés-Navarro’, in Holtus, Metzeltin, and Schmitt (eds), 37–54. Sasse, Hans-Jürgen (1977). ‘Gedanken über Worstellungsveränderung’, Papiere zur Linguistik 9: 82–142. Sauzet, Patrick (1989). ‘Topicalisation et prolepse en occitan’, Revue des langues romanes 93: 235–73. Savoia, Leonardo, and Maiden, Martin (1997). ‘Metaphony’, in Maiden and Parry (eds), 15–25. Scalise, Sergio, Magni, Elisabetta, and Bisetto, Antonietta (eds) (2009). Universals of Language Today. Berlin: Springer. Scarano, Antonietta (1999). Gli aggettivi qualificativi in italiano: Uno studio su corpora di italiano scritto e parlato. Department of Linguistics, University of Florence: unpublished thesis. ——--(2005). ‘Aggettivi qualificativi, italiano parlato e articolazione dell’informazione’, in Elisabeth Burr (ed.), Tradizione e innovazione: Atti del VI convegno della Società di linguistica e filologia italiana. Florence: Cesati, 277–92. Schachter, Paul (1977). ‘Reference-related and Role-related Properties of Subjects’, in Peter Cole and Jerrold Sadock (eds), Grammatical Relations (Syntax and Semantics 8). New York: Academic Press, 279–306. Scherer, Anton (1975). Handbuch der lateinischen Syntax. Heidelberg: Carl Winter. Schifano, Norma (2011). Verb-movement in Italian, French and Spanish: A Survey from Written Sources. Universities of Cambridge/Venice: unpublished thesis. Schlegel, August Wilhelm von (1818). Observations sur la langue et la littérature provençales. Paris: Librairie grecque-latine-allemande. Schlieben-Lange, Brigitte (1971). Okzitanisch und Katalanisch. Tübingen: Narr. Schmidt, Wilhelm (1926). Die Sprachfamilien und Sprachenkreise der Erde. Heidelberg: Winter. Schmitt, Cristina (1998). ‘Lack of Iteration: Accusative Clitic Doubling, Participial Absolutes and Have + Agreeing Participles’, Probus 10: 243–300. Schneider, Nicolaus (1912). De verbi in lingua latina collocatione. Münster: Monasterii Guestfalorum. Schsler, Lene (1984). La déclinaison bicasuelle de l’ancien français. Odense: Odense Universitetsforlag. ——--(2000). ‘Le statut de la forme zéro du complément d’object direct en français moderne’, Études romanes 47: 105–29.

References

399

Schwegler, Armin (1990). Analyticity and Syntheticity: A Diachronic Perspective with Special Reference to Romance Languages. Berlin/New York: Mouton de Gruyter. Schwenter, Scott, and Silva, Gláucia (2002). ‘Overt vs Null Direct Objects in Spoken Brazilian Portuguese: A Semantic-pragmatic Account’, Hispania 853: 577–86. Seidl, Christian (1995). ‘Le système acasuel des protoromans ibériques et sarde: Dogmes et faits’, Vox Romanica 54: 41–73. Sekerina, Irian (1997). The Syntax and Processing of Scrambling Constructions in Russian. CUNY Graduate Centre: unpublished thesis. Selig, Maria (1992). Die Entwicklung der Nominaldeterminanten im Spätlatein. Tübingen: Niemeyer. Sheehan, Michelle (2008). ‘The Final-Over-Final Constraint and Phase Theory’. The University of Newcastle: unpublished manuscript. ——--(2009). ‘The Final-over-Final Constraint as an Effect of Complement Stranding’, Newcastle Working Papers in Linguistics 15: 104–25. Shopen, Tim (2001). ‘Explaining Typological Differences between Languages: De Facto Topicalisation in English and Warlpiri’, in Jane Simpson, David Nash, Mary Laughren, Peter Austin, and Barry Alpher (eds), Forty Years On: Ken Hale and Australian Languages. Canberra: Pacific Linguistics, 187–97. Sihler, Andrew (1995). A New Comparative Grammar of Greek and Latin. Oxford: Oxford University Press. Simpson, Jane, and Bresnan, Joan (1983). ‘Control and Observation in Warlpiri’, Natural Language and Linguistic Theory 1: 49–64. Sitaridou, Ioanna (2002). The Synchrony and Diachrony of Romance Infinitives with Nominative Subjects. University of Manchester: unpublished thesis. Skårup, Povl (1975). Les premières zones de la proposition en ancien français. Copenhagen: Akademisk Forlag. ——--(1997). Morphologie élémentaire de l’ancien occitan. Copenhagen: Museum Tusculanum Forlag. Slobbe, Bianca (2008). ‘Restructuring and the Development of the Romance Conditional Verb Forms’, in Sylvia Blaho, Luis Vicente, and Mark de Vos (eds), Proceedings of ConSole XII. University of Leiden, 107–23. Smith, John Charles (1993). ‘La desaparición de la concordancia entre participio de pasado y objeto directo en castellano y catalán: Aspectos geográficos e históricos’, in Ralph Penny (ed.), Actas del primer congreso anglo-hispano Volumen I. Lingüística. Madrid: Castalia, 275–85. ——--(1995). ‘Perceptual Factors and the Disappearance of Agreement between Past Participle and Direct Object in Romance’, in Smith and Maiden (eds), 161–80. ——--(1999). ‘Markedness and Morphosyntactic Change Revisited: The Case of Romance Past Participle Agreement’, in Sheila Embleton, John Joseph, and Hans-Josef Niederehe (eds), The Emergence of the Modern Language Sciences: Studies on the Transition from HistoricalComparative to Structural Linguistics in Honour of E. F. K. Koerner. 2: Methodological Perspectives and Applications. Amsterdam/Philadelphia: Benjamins, 203–15. ——--(2005). ‘Some Refunctionalizations of the Nominative-Accusative Opposition between Latin and Gallo-Romance’, in Bernadette Smelik, Rijclof Hofman, Camiel Hamans, and

400

References

David Cram (eds), A Companion in Linguistics: A Festschrift for Anders Ahlqvist on the Occasion of his Sixtieth Birthday. Nijmegen: De Keltische Draak, 269–85. Smith, John Charles (2011). ‘Change and Continuity in Form–Function Relationships’, in Maiden, Smith, and Ledgeway (eds), 268–317. ——--and Maiden, Martin (eds) (1995). Linguistic Theory and the Romance Languages. Amsterdam/Philadelphia: Benjamins. Smith, Nathaniel, and Bergin, Thomas (1984). An Old Provençal Primer. New York/London: Garland. Solà, Jaume (1992). Agreement and Subjects. Universitat Autònoma de Barcelona: unpublished thesis. Solà, Joan (1993). Estudis de sintaxi catalana 2. Barcelona: Edicions 62. ——--(1994). Sintaxi normativa: Estat de la qüestió. Barcelona: Empúries. ——--Lloret, Maria-Rosa, Mascaró, Joan, and Pérez Saldayna, Manuel (eds) (2008). Gramática del català contemporani. Volum 1. Introducció. Fonètica i fonologia. Morfologia. Volum 2. Sintaxi (1–16). Volum 3. Sintaxi (17–31) (amb la col·laboració de Gemma Rigau). Barcelona: Editorial Empúries. Solin, Heikki, Leiwo, Martti, and Halla-aho, Hilla (eds) (2003). Latin vulgaire et latin tardif VI. Actes du VIème colloque international sur le latin vulgaire et tardif, 29 août – 2 septembre 2000 Helsinki. Hildesheim: Olms. Sornicola, Rosanna (1976). ‘Vado a dire, vaiu a ddicu: Problema sintattico o problema semantico’, Lingua nostra 37: 65–74. ——--(1997). ‘L’oggetto preposizionale in siciliano antico e in napoletano antico. Considerazioni su un problema di tipologia diacronica’, Italienische Studien 18: 66–80. ——--(2000). ‘Stability, Variation and Change in Word Order: Some Evidence from the Romance Languages’, in Rosanna Sornicola, Erich Poppe, and Ariel Shisha-Halevy (eds), Stability, Variation and Change of Word-Order Patterns Over Time. Amsterdam/ Philadelphia: Benjamins, 101–18. ——--(2008). ‘Sul problema dei resti di ipse nella Romània’, in Romano Lazzeroni, Emanuele Banfi, Giuliano Bernini, Marina Chini, and Giovanna Marotta (eds), Diachronica et Synchronica: Studi in onore di Anna Giacalone Ramat. Pisa: ETS, 537–62. ——--(2009). ‘Continuités textuelles et discontinuités catégorielles: Le cas de l’article roman’, Travaux de linguistique 59: 25–46. ——--(2011). ‘Romance Linguistics and Historical Linguistics: Reflections on Synchrony and Diachrony’, in Maiden, Smith, and Ledgeway (eds), 1–49. Sorrisi, Fabrizio (2010). ‘Le costruzioni flesse con va + verbo lessicale in palermitano’, in Garzonio (ed.), 127–38. Speas, Margaret (1990). Phrase Structure in Natural Languages. Dordrecht: Kluwer. Spence, Nicol (1971). ‘La survivance des formes du nominatif latin en français’, Revue romane 6: 74–84. Spevak, Olga (2007). ‘La place du focus en latin: Quelques observations’, in Purnelle and Denooz (eds), 191–201. ——--(2008a). ‘Le placement de is, hic et ille chez Grégoire de Tours’, in Wright (ed.), 361–70. ——--(2008b). ‘The Position of Focus Constituents in Latin: A Comparison between Latin and Czech’, Journal of Pragmatics 40: 114–26.

References

401

——--(2009). ‘L’hyperbate en latin tardif ’, paper presented at Latin vulgaire, latin tardif. IXe colloque international, 2–6 septembre 2009. ——--(2010). Constituent Order in Classical Latin Prose. Amsterdam/Philadelphia: Benjamins. Sproat, Richard, and Chinlin, Shih (1988). ‘Prenominal Adjectival Ordering in English and Mandarin’, in James Blevins and Juli Carter (eds), North East Linguistic Society 18: 465–89. ——--——--(1990). ‘The Cross-Linguistics Distribution of Adjectival Ordering Restrictions’, in Carol Georgopoulos and Roberta Ishihara (eds), Interdisciplinary Approaches to Language: Essays in Honor of S-Y. Kuroda. Dordrecht: Kluwer, 565–93. Squartini, Mario (1999). ‘Voice Clashing with Aspect: The Case of Italian Passives’, Rivista di linguistica 11: 341–65. ——--(2003). ‘La grammaticalizzazione di in italiano: Anticausativo o risultativo?’, in Pusch and Wesch (eds), 23–34. ——--and Bertinetto, Pier Marco (2000). ‘The Simple and Compound Past in Romance languages’, in Östen Dahl (ed.), Tense and Aspect in the Languages of Europe. Berlin/New York: Mouton de Gruyter, 385–402. Stammerjohann, Harro (ed.) (1986). Tema-rema in italiano – Theme-Rheme in Italia-ThemaRhema in Italienischen. Tübingen: Narr. Stanovaïa, Lydia (1993). ‘Sur la déclinaison bicasuelle en ancien français (point de vue scriptologique)’, Travaux de linguistique et de philologie 23: 163–82. Stark, Elisabeth (2002). ‘Indefinites and Specificity in Old Italian Texts’, Journal of Semantics 19: 315–332. Stassen, Leon (2000). ‘AND-languages and WITH-languages’, Linguistic Typology 4: 1–54. ——--(2001). ‘Noun Phrase Coordination’, in Haspelmath, König, Österreicher, and Raible (eds), 1105–11. Stati, Sorin (1989). ‘Le roumain: Syntaxe’, in Holtus, Metzeltin, and Schmitt (eds), 114–37. Steele, Susan, Akmajian, Adrian, Jelinek, Eloise, Kitagawa, Chisato, Oehrle, Richard, and Wasow, Thomas (1981). An Encyclopedia of AUX: A Study of Cross-Linguistic Equivalence. Cambridge, MA: MIT Press. Ştefănescu, Ioana (1997). The Syntax of Agreement in Romanian. City University of New York: thesis (distributed as MIT Occasional Papers in Linguistics no 14. Cambridge, MA: MIT Working Papers in Linguistics, MIT, Department of Linguistics). Sten, Holger (1944). Les particularités de la langue portugaise. Copenhagen: Cercle Linguistique. Stewart, Miranda (1999). The Spanish Language Today. London: Routledge. Stimm, Helmut (1977). ‘Ein neuer früher Beleg des synthetischen Futurs’, in Hugo Laitenberger (ed.), Festgabe für Julius Wilhlem zum 80. Geburtstag (Zeitschrift für französische Sprache und Literatura, Beiheft N.F. 5). Wiesbaden: Steiner, 40–52. ——--(1986). ‘Die Markierung des direkten Objekts durch a im Unterengadinischen’, in Günter Holtus and Kurt Ringger (eds), Raetia Antiqua et Moderna, W. Th. Elwert zum 80. Geburtstag. Tübingen: Niemeyer, 407–48. ——--and Linder, Karl Petere (1989). ‘Bündnerromanisch. Interne Sprachgeschichte I. Grammatik’, in Holtus, Metzeltin, and Schmitt (eds), 764–85. Stolova, Natalya (2006). ‘Split Intransitivity in Old Spanish: Irrealis and Negation Factors’, Revue roumaine de linguistique LI 2: 301–20.

402

References

Sugisaki, Koji (2007). ‘The Configurationality Parameter in the Minimalist Program: A View from Child Japanese’, in Heather Caunt-Nulton, Samantha Kulatilake, and I-hao Woo (eds), Proceedings of the 31st Annual Boston University Conference on Language Development. Somerville, MA: Cascadilla Press, 597–608. Suñer, Maragarita (1988). ‘The Role of Agreement in Clitic-doubled Constructions’, Natural Language and Linguistic Theory 6: 391–434. Swartz, Stephen (1987). Measuring Naturalness in Translation by Means of a Statistical Analysis of Warlpiri Narrative Texts with Special Emphasis on Word Order Principles. Pacific College of Graduate Studies: unpublished thesis. ——--(1988). ‘Pragmatic Structure and Word Order in Warlpiri’, Papers in Australian Linguistics 17: 151–66. Sznajder, Lyliane (1998). ‘Verbes transitifs en latin sans objet’, in García-Hernández (ed.), 791–808. Tagliavini, Carlo (1972). Le origini delle lingue neolatine: Introduzione alla filologia romanza. Bologna: Pàtron. Talavera, Francisco (1981). ‘Aspectos vulgares de la Vetus Latina: Análisis especial del orden de palabras en el libro de Rut’, Analecta malacitana 4: 211–27. Tarriño Ruiz, Eusebia (2009a). ‘El adjetivo’, in Baños Baños (ed.), 251–272. ——--(2009b). ‘Formas nominales del verbo’, in Baños Baños (ed.), 469–94. ——--(2009c). ‘Comparativas’, in Baños Baños (ed.), 646–56. Tekavčić, Pavao (1980). Grammatica storica della lingua italiana. Volume 2. Morfosintassi. Bologna: il Mulino. Tesnière, Lucien (1959). Eléments de syntaxe structurale. Paris: Klincksieck. Teyssier, Paul (1980). Histoire de la langue portugaise. Paris: PUF. ——--(1984). Manuel de langue portugaise (Portugal-Brésil). Paris: Klincksieck. Thiella, Anna (2008). ‘Il sintagma nominale negli antichi volgari di area veneta e lombarda’, Laboratorio sulle varietà romanze antiche 2: 1–163. Thielmann, Phili (1885a). ‘Habere mit dem Infinitive und die Entstehung des romanischen Futurms’, Archiv für Lateinische Lexikographie 2: 157–202. ——--(1885b). ‘Habere mit dem Part. Perf. Pass.’, Archiv für Lateinische Lexikographie 2: 372– 423; 509–49. Thomas, Earl (1969). The Syntax of Spoken Brazilian Portuguese. Nashville: Vanderbilt University Press. Torrego, Esperanza (2009a). ‘Coordination’, in Baldi and Cuzzolin (eds), 443–87. ——--(2009b). ‘La oración compleja. De la parataxis a la subordinación. La coordinación’, in Baños Baños (ed.), 495–522. Torrego, Esther (1998). The Dependencies of Objects. Cambridge, MA: MIT Press. ——--(1999). ‘El complemento directo preposicional’, in Bosque and Demonte (eds), II, 1779–807. Tortora, Christina (2002). ‘Romance Enclisis, Prepositions, and Aspect’, Natural Language and Linguistic Theory 20: 725–58. ——--(ed.) (2003). The Syntax of Italian Dialects. Oxford: Oxford University Press.

References

403

——--(2010). ‘Domains of Clitic Placement in Finite and Non-finite Clauses: Evidence from a Piedmontese Dialect’, in D’Alessandro, Ledgeway, and Roberts (eds), 135–49. Touratier, Christian (ed.) (1985). Syntaxe et latin. Actes du IIème congrès international de linguistique latine. Aix-en-Provence, 28–31 mars 1983. Marseille: Université de Provence. Tovar, Antonio (1979). ‘Orden de las palabras y tipología: Una nota sobre el latín’, Euphrosyne 9: 161–71. Trager, George (1932). The Use of the Latin Demonstratives (Especially Ille and Ipse) up to 600 A.D., as the Source of the Romance Article. New York: Publications of the Institute of French Studies. Trumper, John (1996). ‘Riflessioni pragmo-sintattiche su alcuni gruppi meridionali: L’italiano “popolare”’, in Benincà, Cinque, De Mauro, and Vincent (eds), 351–67. Tuttle, Edward (1986). ‘The Spread of ESSE as Universal Auxiliary in Central Italo-Romance’, Medioevo romanzo 11: 229–87. Urit¸escu, Dorin. (2007). ‘Dans la perspective de l’atlas linguistique de la Crişana (I). Le participe passé daco-roumain en -ă: Mythe roumain ou innovation d’une langue romane?’, in Sanda Reinheimer Rîpeanu and Ioana Vintilă-Radulescu (eds), Limba română, limbă romanică. Bucharest: Éditions de l’Académie roumaine, 555–66. Ursini, Flavia (1989). ‘Istroromanzo: Storia linguistica interna’, in Holtus, Metzeltin, and Schmitt (eds), 537–48. Väänänen, Veikko (1966). Le latin vulgaire des inscriptions pompéiennes. 3rd edn. Berlin: Akademie-Verlag. ——--(1982). Introduzione al latino volgare. Bologna: Pàtron. ——--(1987). Le journal-épître d’Egérie: Etude linguistique. Helsinki: Suomalainen Tiedeakatemia. Valesio, Paolo (1968). ‘The Romance Synthetic Future Pattern and its First Attestations’, Lingua 20: 113–61. ——--(1969). ‘The Synthetic Future Again: Phonology and Morphosyntax’, Lingua 24: 181–93. Vallduví, Enric (1988). ‘Sobre el perfet perifràstic del català’, in Philip Rasico and Curt Wittlin (eds), El cinquè col.loqui d’estudis catalans a nord-amèrica. Tampa & St Augustine: l’Abadia de Montserrat, 85–97. Valois, Daniel (1991). ‘The Internal Syntax of DP and Adjectival Placement in French and English’, North East Linguistic Society 21: 367–82. Vance, Barbara (1997). Syntactic Change in Medieval French: Verb-second and Null Subjects. Dordrecht: Kluwer. Vanelli, Laura (1980). ‘A Suppletive Form of the Italian Article and its Phonosyntax’, Journal of Linguistic Research 1: 69–90. ——--(1986). ‘Strutture tematiche in italiano antico’, in Harro Stammerjohann (ed.), TemaRema in Italian. Tübingen: Narr, 248–73. ——--(1987). ‘I pronomi soggetto nei dialetti italiani settentrionali dal Medio Evo a oggi’, Medioevo romanzo 12: 173–211. ——--(1999). ‘Ordine delle parole e articolazione pragmatica nell’italiano antico: La “prominenza” pragmatica della prima posizione nella frase’, Medioevo romanzo 23: 229–46.

404

References

Vanelli, Laura, Renzi, Lorenzo, and Benincà, Paola (1985). ‘Typologie des pronoms sujets dans les langues romanes’, in Actes du XVIIe congrès international de linguistique et philologie romanes. Volume 3: Linguistique descriptive, phonétique, morphologie et lexique. Aixen-Provence: Università di Provence, 163–76. van Reenen, Pieter, and Schsler, Lene (2000). ‘Declension in Old and Middle French: Two Opposing Tendencies’, in John Charles Smith and Delia Bentley (eds), Historical Linguistics 1995. Volume 1: General Issues and Non-Germanic Languages. Amsterdam/Philadelphia: Benjamins, 327–44. van Valin, Robert (1987). ‘The Role of Government in the Grammar of Head-Marking Languages’, International Journal of American Linguistics 53: 371–97. Varvaro, Alberto (forthc.). ‘Latin and the Making of the Romance Languages’, in Maiden, Smith, and Ledgeway (eds). Vecchio, Paola (2010). ‘The Distribution of the Complementizers /ka/ and /ku/ in the North Salentino Dialect of Francavilla Fontana (Brindisi)’, in D’Alessandro, Ledgeway, and Roberts (eds), 312–22. Vennemann, Theo (1974). ‘Topics, Subjects, and Word Order: From SXV to SVX via TVX’, in John Anderson and Charles Jones (eds), Historical Linguistics. Amsterdam: North Holland, 339–76. ——--(1975). ‘An Explanation of Drift’, in Charles Li (ed.), Word Order and Word Order Change. Austin: University of Texas, 269–305. Veny, Joan (1982). Els parlars catalans (Síntesi de dialectologia). Mallorca: Moll. ——--(1991). ‘Katalanisch: Areallinguistik – Áreas lingüisticas’, in Holtus, Metzeltin, and Schmitt (eds), 243–61. ——--(2001). Llengua històrica i llengua estàndard. Valencia: Universitat de València. Vielliard, Jeanne, (1927). Le latin des diplômes royaux et chartes privées de l’époque merovingienne. Paris: Champion. Vikner, Sten (1995). Verb Movement and Expletive Subjects in the Germanic Languages. Oxford University Press. ——--(1997). ‘V-to-I Movement and Inflection for Person in all Tenses’, in Haegeman (ed.), 189–213. Villa-García, Julio (2010). Recomplementation and Locality of Movement in Spanish. University of Connecticut: unpublished second general examination paper. Villangómez i Llobet, Marià (1978). Curs d’iniciació a la llengua. Normes gramaticals: Lectures eivissenques i formentereres. Ibiza: Institut d’Estudis Eivissencs. Villar, Francisco (1983). Ergatividad, acusatividad y género en la familia lingüística indoeuropea. Salamanca: Ediciones Universidad de Salamanca. Vincent, Nigel (1977). ‘Perceptual Factors and Word Order Change in Romance’, in Harris (ed.), 54–68. ——--(1982). ‘The Development of the Auxiliaries habere and esse in Romance’, in Vincent and Harris (eds), 71–96. ——--(1986). ‘La posizione dell’aggettivo in italiano’, in Stammerjohann (ed.), 181–95. ——--(1987). ‘The Interaction of Periphrasis and Inflection: Some Romance Examples’, in Harris and Ramat (eds), 237–56. ——--(1988). ‘Latin’, in Harris and Vincent (eds), 26–78.

References

405

——--(1993). ‘Head- versus Dependent-marking: The Case of the Clause’, in Greville Corbett, Norma Fraser, and Scott McGlashan (eds), Heads in Grammatical Theory. Cambridge: Cambridge University Press, 140–63. ——--(1994). ‘On the Default Case in Latin’, University of Manchester: unpublished mansuscript. ——--(1996). ‘L’infinito flesso in un testo napoletano del Trecento’, in Benincà, Cinque, De Mauro, and Vincent (eds), 389–409. ——--(1997a). ‘Synthetic and Analytic Structures’, in Maiden and Parry (eds), 99–105. ——--(1997b). ‘Prepositions’, in Maiden and Parry (eds), 208–13. ——--(1997c). ‘The Emergence of the D-system in Romance’, in Kemenade and Vincent (eds), 149–69. ——--(1998a). ‘Tra grammatica e grammaticalizzazione: Articoli e clitici nelle lingue (italo)romanze’, in Ramat and Roma (eds), 411–40. ——--(1998b). ‘On the Grammar of Inflected Non-finite Forms (with Special Reference to Old Neapolitan)’, in Irn Korzen and Michael Herslund (eds), Clause Combining and Text Structure, Copenhagen Studies in Language 22: 135–58. ——--(1999). ‘The Evolution of C-structure: Prepositions and PPs from Indo-European to Romance’, Linguistics 37: 1111–53. ——--(2000). ‘Competition and Correspondence in Syntactic Change: Null Arguments in Latin and Romance’, in Susan Pintzuk, George Tsoulas, and Anthony Warner (eds), Diachronic Syntax: Models and Mechanisms. Oxford: Oxford University Press, 25–50. ——--(2005). ‘Come insegnare l’italiano antico’, in Anna Laura Lepschy and Anna Rita Tamponi (eds), Prospettive sull’italiano come lingua straniera. Perugia: Guerra, 33–40. ——--(2006). ‘Il problema del doppio complementatore nei primi volgari d’Italia’, in Alivse Andreose and Nicoletta Penello (eds), Laboratorio sulle varietà romanze antiche: Giornata di lavoro sulle varietà romanze antiche. Padua: University of Padua, 27–42. ——--(2007). ‘Learned vs Popular Syntax: Adjective Placement in Early Italian Vernaculars’, in Lepschy and Tosi (eds), 55–75. ——--Bentley, Delia, and Samu, Borbala (2004). ‘The Syntax of the Complementizer de in Old Sardinian’, paper presented at the XXXII Romance Linguistics Seminar, Trinity Hall, Cambridge. ——--and Harris, Martin (eds) (1982). Studies in the Romance Verb. London: Croom Helm, Vineis, Edoardo (1993). ‘Preliminari per una storia (ed una grammatica) del latino parlato’, in Friedrich Stolz, Albert Debrunner, and Wolfgang Schmid (eds), Storia della lingua latina. 4th edn. Bologna: Pàtron, xxxvii–lviii. ——--(1998). ‘Latin’, in Giacalone Ramat and Ramat (eds), 261–321. Wagner, Max Leopold (1951). La lingua sarda. Storia, spirito e forma. Bern: Francke. Wanner, Dieter (1996). ‘Second Position Clitics in Medieval Romance’, in Aaron Halpern and Arnold Zwicky (eds), Approaching Second: Second Position Clitics and Related Phenomena. Stanford: Center for the Study of Language and Information, 537–78. ——--(1998). ‘Les subordonnées à double complémentateur en roman médiéval’, in Ruffino (ed.), I, 421–33.

406

References

Wanner, Dieter and Kibbee, Douglas (eds) (1991). New Analyses in Romance Linguistics. Amsterdam/Philadelphia: Benjamins. Wartburg, Walther von ([1934] 1971). Évolution et structure de la langue française. Bern: Francke. ——--(1950). Die Ausgliederung der romanischen Sprachräume. Bern: Francke. Watkins, Calvert (1964). ‘Preliminaries to the Reconstruction of Indo-European Sentence Structure’, in Horace Lunt (ed.), Proceedings of the Ninth International Congress of Linguistics. London: Mouton, 1035–45. ——--(1998). ‘Proto-Indo-European: Comparison and Reconstruction’, in Giacalone Ramat and Ramat (eds), 25–73. Weil, Henri ([1844] 1887). The Order of Words in the Ancient Languages Compared with That of the Modern Languages. Boston: Ginn & Co. Westerbergh, Ulla (1956). Chronicon Salernitanum: A Critical Edition with Studies on Literary and Historical Sources and on Language. Stockholm: Almqvist & Wiksell. Wheeler, Max (1988a). ‘Catalan’, in Harris and Vincent (eds), 170–208. ——--(1988b). ‘Occitan’, in Harris and Vincent (eds), 246–78. ——--Yates, Alan, and Dols, Nicolau (1999). Catalan: A Comprehensive Grammar. London: Routledge. Williamson, Janis (1984). Studies in Lakhota Grammar. University of California at San Diego: unpublished thesis. Woledge, Brian (1956). ‘The Plural of the Indefinite Article in Old French’, The Modern Language Review 51: 17–32. Wright, Roger (1982). Late Latin and Early Romance in Spain and Carolingian France. Liverpool: Francis Cairns. ——--(1983). ‘Unity and Diversity among the Romance Languages’, Transactions of the Philological Society 81: 1–22. ——--(2002). A Sociophilological Study of Late Latin. Turnhout: Brepols. ——--(ed.) (2008). Latin vulgaire – latin tardif. Actes du VIIIème colloque international sur le latin vulgaire et tardif (Oxford, 6–9 septembre 2006). Olms-Weidmann, Hildesheim. ——--(forthc.). ‘Periodization’, in Maiden, Smith, and Ledgeway (eds). Wüest, Jakob (1998). ‘Pour une linguistique historique non linéaire: Les formes analytiques du latin’, in Herman (ed.), 87–98. Wurff, Wim van der (1994). ‘Null Objects and Learnability: The Case of Latin’, Working Papers of Holland Institute for Generative Linguistics, University of Leiden, 1: 83–103. ——--(1997). ‘Syntactic Reconstruction and Reconstructability: Proto-Indo-European and the Typology of Null Objects’, in Jacek Fisiak (ed.), Linguistic Reconstruction and Typology. Berlin/New York: Mouton de Gruyter, 337–56. Yates, Alan (1975). Catalan. Kent: Hodder and Stoughton. Zagona, Karen (2002). The Syntax of Spanish. Cambridge: Cambridge University Press. Zamboni, Alberto (1992). ‘Postille alla discussione sull’accusativo preposizionale’, in Lorenzo, Ramón (ed.), Actas do XIX Congreso internacional de lingüística e filoloxía románicas V: Gramática histórica e historia da lingua. A Coruña: Fundación ‘Pedro Barrié de la Maza, Conde de Fenosa’, 787–808.

References

407

——--(1998). ‘Dal latino tardo al romanzo arcaico: Aspetti diacronico-tipologici della flessione nominale’, in Ramat and Roma (eds), 127–46. ——--(2000). Alle origini dell’italiano: Dinamiche e tipologie della transizione dal latino. Rome: Carocci. Zamora Vicente, Alonso (1967). Dialectología española. Madrid: Gredos. Zanini, Chiara, and Damonte, Federico (2010). ‘Ordini marcati dei costituenti in siciliano’, in Garzonio (ed.), 139–53. Zanuttini, Raffaella (1997). Syntactic Properties of Sentential Negation: A Comparative Study of Romance Languages. Oxford: Oxford University Press. Zribi-Hertz, Anne (1984). ‘Prépositions orphélines et pronoms nuls’, Recherches linguistiques 12: 46–91. Zubizarreta, María-Luisa (1985). ‘The Relationship between Morphophonology and Morphosyntax: The Case of Romance Causatives’, Linguistic Inquiry 16: 247–89. ——--(1998). Prosody, Focus, and Word Order. Cambridge, MA: MIT Press.

This page intentionally left blank

Index Abeillé 51, 144 Abels 281 ablative generalized 22 standard of comparison 23 Abruzzese (see also southern Italian dialects) 135, 170, 171, 288 eastern 296 Scanno 104 Academia de la Llingua Asturiana (ALlA) 7, 106 Acadian (French) 341–2 accusative (see also case; case: prepositional accusative) 10, 157–8 extended 48, 328–33 generalized 22, 332 loss of final -M 21–2, 329–30 retention 64 accusative and infinitive construction (AcI) (see complementation) Acquafondata (see also central/southern Italian dialects) 324 active-stative alignment (see alignments) Adams, James 3, 8, 14, 21, 22, 23, 29, 33, 35, 39, 41, 42, 45, 47, 59, 60, 63, 64, 65, 90, 91, 92, 93, 95, 96, 112, 130, 131, 132, 135, 136, 138–40, 153, 158, 186, 189, 191, 192, 193, 195, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 219, 222, 223 225, 226, 227, 228, 229, 230, 231, 232, 233, 234–5, 236, 240, 241, 243, 245, 246, 251, 256, 259, 260, 261, 270, 277, 318–19, 328, 329, 330, 331, 334 Adams, Marianne 66 adjective apocopated forms 54–5 position 25, 28, Latin 37, 41, 42–3, 50–1, 210–13, 270 Romance 49, 50–4, 198–9

early Romance 56 interpretation 52–3 liaison 53–4 restricted prenominal position 55–7, 198 adjunct 37–8, 39, 57–8 adposition 205–7 gerundive complement 252–3 null 240–1, 263 postposition 206–7, 276–7 preposition (see also case) 12, 291 agreeing 308–9 increased use 3, 11, 21–2, 25, 291 nominal origin 241 preverb origin 206–7, 241 adpositional phrase PP (prepositional phrase) 288–9 configurationality 77–80, 193 left periphery 78, 195–6, 258, 259–62, 268–9, 275–7 phase/colon status 261–2 position vs bare case-marked nominal 249–52 rise of head-initiality 40–2, 249–55 adverb circumstantial 164–5 clitic 145 interpolation 19 of manner in men(t(e)) 14, 18–19, 24, 212–13 position 38, 42, 127 cartographic representation 141–50 higher adverb space (HAS) 142–50 lower adverb space (LAS) 142–50 VP 141 scene-setting 166–7 quantificiational 164–5 AdvP (Adverb Phrase) 63 Aebischer 89, 90, 92, 103 Agart 7

410

Index

Agbayani 273 Agnese (see also southern Italian dialects) 325 agreement (see also case) 31, 35, 47, 63, 71, 72, 77, 116, 130, 259, 277, 286, 292–3 adjectival adverbs 339–40 controllers 306–7 on D(eterminer) 290–1 participle 286, 301, 314 active-stative split 295, 307, 326–7 default feminine 295 functional value 132–3 reflexive verbs 321–2 antipassive 308 nominative-accusative split 347–9 weakening 132–3 with plural DP 349 Ripano 299–310 cardinal quantifiers 309 object-verb agreement 305–8 clitic-controlled agreement 305–6 predicative complements 309–10 prepositions 308–9 subject-verb agreement 300–2 gerund 301, 308 infinitive 302, 308 past participle 301 subject-/obect-verb agreement 302–5 matching agreement 302–3 non-matching agreement 303–5 wh-phrases 309 Spec-head 116–17, 118, 119, 292–3 subject-verb 300–2, 338–9 Alarcos Llorach 7 Alboiu 7, 69, 108, 113, 116, 119, 140, 159, 162, 170, 288, 290, 297, 346 Alcover 100, 104 Alexiadou 69, 344 Alfonsi 7 Alfonzetti 15, 184 Alguerés (see also Catalan) 104, 112, 320 Alibèrt 7 alignment 25, 312–52 active-stative 313–14, 316–40 absolute participles 337–8

adjectival adverbs 339–40 classical Latin verbal system 316 distribution of bare plural NPs 339–40 extended accusative 328–33 INDE-cliticization 339–40 participle agreement 326–7 perfective auxiliary selection 130–4, 295, 314, 317–23 late Latin 317–19 Romance 319–23 passive 322–3 reflexive 321–22, 324–5 triple auxiliation 323–6 possessive copula 317–19 relative pronoun 336–7 retention of nominative with animates 334–5 sentential word order 69, 233–4, 335–6 subject-verb agreement 338–9 core participants 312–14 ergative-absolutive 313–14, 332 nominative-accusative 313, 340–51 Latin nominal system 314–15 verbal system 315–16 participle agreement 347–9 perfective auxiliary selection 327, 341–7 determined by finiteness 346–7 Romanian pluperfect surcomposé paradigm 347 generalized HABERE (/TENERE) 341–2 modal conditioning 344–6 person split 323–6, 343–4, 349 tense split 342–4 sentential word order 349–51 Alisova 51, 198 Alkire 4 Allen 21, 318 Allières 69 Alonso 106 Alonso Garrote 7 Alsina 16, 297, 339 Altamurano (see also Pugliese; southern Italian dialects) 322–4

Index Alvar 7 Álvarez 7 Álvarez Pedrosa 214, 225, 228 Amacker 40, 60, 62, 185, 201, 226, 228, 229, 230 Ambrosini 184 Anagnostopoulou 69 analysis / analyticity 3, 241 syntheticity and analyticity 10–29, 30, 31, 81, 197, 198, 286 case 240–2 competing forms 23–4 definition 10 grammaticalization 28–9 internal vs external determination 26–8 morphophonological erosion 21–3, 29, 241 passive 27 prepositions 21–2, 241–2 relational vs non-relational functions 26–8 sentential word order 20–1, 22–3, 28 anaphor binding 75–7 long-distance control of reflexives 76–7 anastrophe (see also fronting) 78, 193, 195–6 Anderson 134 Andria (Terence) 233 Annals (Tacitus) 233 Anonymus Valesianus 222, 225–6, 229, 233 Anthimus 235 AP (Adjective Phrase) 63 left periphery 276 Appicius 226 apposition 38, 39, 48, 89–90 Apuleius 226 Aragonese 7, 17, 69, 89, 111, 176, 177–8, 199–201, 320, 349 Aranovich 71 Arias Cabal 106 Ariellese (see also Abruzzese; southern Italian dialects) 145, 172, 176, 343–4, 349 Arnaud 13, 99 Arnholdt 51 Aromanian 295 Arquint 7

411

Ars gramatica 138, 215 article (see determiner) Ascoli 7 Ashdowne 99 aspect (see also present perfect) andative 126 continuous 2–3, 122 iterative 15–16, 122, 123 present perfect 123, 130–4, 234, 317–26 preterit 314 progressive 122, 125, 126 resultative 123, 130–2, 296 retrospective 122–3 Asturian (see also Ibero-Romance) 7, 55, 83, 97, 106, 112–13, 127, 198, 341 asyndeton (see also complementation) 126, 205 Atanasov 296 Augustine 90 Austin 71, 72, 79, 188 Authier 173 Auvernhat (see also Occitan) 13, 17, 290 auxiliary (AUX) (see also future; passive; present perfect) 2–3, 25, 33–4, 64, 79, 80, 81, 119–40, 182 clitic forms 128–9 conditional 14, 33, 135–8 decategorialization 124–5 future(-in-the-past) 10–11, 12, 13, 14, 24, 34, 122–4, 128–9, 134–40, 192, 225 morphophonological specialization 127–9 passive/middle 11, 12–13, 15, 28, 38, 133, 180, 223–4, 255–8 preterit 122–3, 128–9 perfect 12, 15–16, 18, 19, 24, 122, 123, 125, 128, 130–4, 184–5, 234 selection 121, 295–6, 314 active-stative split 295, 317–23 reflexives 321–2, 324–5 determined by finiteness 346–7 Romanian pluperfect surcomposé paradigm 347 modal conditioning 344–6 person split 296, 323–6, 343–4, 349 tense split 342–4

412

Index

auxiliary (AUX) (cont.) triple auxiliation 323–6 with modal 120–1 position 33, 34, 38, 133, 143–4, 147, 223–4, 255–8 prosodic properties 129–30 Romance class 120–1 semantic weakening 121–4, 136 Aviglianese (see also southern Italian dialects) 106 Avolio 7, 105 Avram 296 Ayres-Bennett 7, 69 Azevedo 4, 7, 69, 74, 159, 290 Bachmann 8, 74 Bader 197 Badia i Margarit 7, 51, 100, 140 Badiot (see also Ladin; Ræto-Romance) 82 Bailey 34, 36, 59, 223, 256 Baker 47, 71, 147, 184, 252, 310, 311 Bakkum 213, 231 Baldi 206, 212, 213, 214, 215, 216, 296, 318 Baldinger 197 Banniard 1, 96, 97 Baños Baños 35, 41, 59, 61, 150, 185, 196, 210, 213, 216, 218, 225, 228, 229, 230, 242 Bantu 287 Barbosa 69 Barceloní 109, 348 Barégeois 174 Bartoli 7, 314 Bastardas Parera 328 Battye 7, 53 Bauer 2, 3, 8, 10, 12, 14, 20, 21, 22, 23, 24, 25, 28, 31, 34, 35, 39, 40, 41, 42, 43, 45, 47, 48, 59, 64, 78, 89, 90, 133, 151, 152, 153, 157, 158, 159, 185, 186, 187, 191, 193, 195, 196, 197, 198, 201, 202, 203, 204, 205, 206, 207, 210, 211, 212, 213, 215, 216, 222, 223, 224, 225, 227, 229, 230, 232, 236, 238, 240, 241, 242, 249, 250, 251, 256, 259, 260, 268, 277, 332 Bearnés (see also Gascon; Occitan) 13, 88, 178 Bec 7, 173

Beckmann 22 Begurenc (see also Catalan) 100, 101, 102 Behaghel’s Law 250 Belanger 341 Bellante (see also Abruzzese; southern Italian dialects) 344 Belletti 76, 81, 140, 141, 145, 149, 163, 183, 260 Bellum Africum 226, 227, 234 Bellunese (see also northern Italian dialects) 334 Benincà 7, 8, 65, 66, 69, 81, 82, 159, 160, 161, 163, 164, 165, 166, 167, 168, 177, 185, 292, 332 Benjamin 7 Bennett 13, 23, 134, 157, 207, 208, 214 Bentley 8, 70, 130, 135, 162, 171, 296, 323, 335, 337, 339 Benveniste 136, 139 Benzing 341 Bergin 138 Berlinck 69 Bernat i Baltrons 7, 100 Bernstein 51, 56 Berruto 51, 199, 210 Bertinetto 15, 184 Bertocci 298 Bertocchi 75 Biberauer 8, 147–8, 204, 221, 238, 241, 242 Bichakjian 3, 23, 42, 59, 78, 157, 202, 205, 225, 240, 251 Bigourdan (see also Gascon; Occitan) 88 Blasco Ferrer 7 Bobaljik 147, 251 Bolkestein 150, 155, 158, 185, 228, 230, 240, 242, 243, 248, 259, 263 Bolognese 17, 293 Borer 184, 284 Born 7 Bošković 31, 34, 72, 279, 280, 281 Bosque 4, 51 Bossong 288, 350 Bouchard 53, 56 Bourciez 3, 4, 10, 21, 26, 136 Bourgain 246 Bourova 14, 34, 135, 139

Index branching parameter (see head parameter; word order) Braulius 213–14, 229, 230 Bresnan 31, 64, 71, 72, 77, 79, 80, 284, 285 bridge verbs (see also Verb Second) 165, 167 Brito 51 Brown 221 Bruneau 4 Bruni 7 Brunot 4 Bu Njem letters 245–6 Bubenik 12, 31, 35, 71, 78, 187, 193, 206, 207 Buck 39, 41, 43, 59, 62, 63, 190, 191, 193, 205, 206, 207, 209, 210, 211, 220, 259 Bulgarian 296 Bulhart 139 Buridant 225 Burzio 16, 297, 339 Busquet Isart 100, 101, 104 Butt 7 Cabrillana 35, 40, 41, 59, 61, 64, 150, 185, 187, 196, 210, 213, 214, 216, 218, 225, 226, 227, 228, 229, 230, 232, 233, 242 Cadaquesenc (see also Catalan) 100, 101, 102, 103 Caesar 22, 42, 45, 60, 214, 227, 228, 231, 243, 252–4, 255 Cairese (see also Piedmontese; northern Italian dialects) 224, 320 Calabrese, Andrea 170 Calabrian (see also southern Italian dialects) 15, 27, 53, 82, 128, 140–1, 144, 145, 147, 177, 298, 334, 335, 344–5 northern 170 southern 170, 184–5, 351 Calboli 78, 96, 157, 158, 205, 225, 240, 242, 246 Campanian (see also southern Italian dialects) 168, 342–3 Campbell 124, 132, 187, 231 Campidanese (see also Sardinian) 82, 104 Campos 159, 290 Canale 341 Cano González 7

413

Capcinese (see also Catalan) 296 Capidan 7 Captivi (Plautus) 232–3 Carballo Calero 293 Cardinaletti 7, 69, 126, 292 Cargeghese (see also Sardinian) 345 Carrera 229 Carruthers 7, 69 case (see also ablative; accusative; alignment; dative; genitive; nominative) 26, 31, 71, 72, 77, 240, 259, 284 binary system 241–2, 292, 314, 332, 333–5 pronominal system 333 Romanian 292 checking 251 erosion/weakening 2–3, 11, 21–3, 64, 241, 251, 277, 290, 332 extended accusative 328–33 loss 2–3, 241–2, 314, 332–3 marking 72, 119, 251 prepositional accusative 288, 314, 350–1 pronominal 14 replaced by prepositions 11, 25, 240–1, 291 retention 314, 332 nominative reinterpreted as predicative case 332 nominative/oblique doublets 335 restricted nominative 328–33 survival 64 ternary system 314, 333 Castrovillarese (see also Calabrian; southern Italian dialects) 325–6 Catalan 3, 7, 11, 14, 18, 20, 25, 48, 52, 53, 56–7, 58, 70, 82, 83, 86, 87, 89, 97, 98, 99, 106–7, 122, 123, 127, 128–9, 135, 140, 141, 167, 169, 176, 201, 232, 290, 299, 314, 327, 334, 335, 351 Balearic 89, 100–03, 109, 110–11, 112, 119, 320, 327, 336, 339–40, 341 Formentera 101 Ibizan 101, 288 Majorcan 100, 101, 102–3, 326–7 Pollença 100

414

Index

Catalan (cont.) Menorcan 101, 103, 212 Costa Brava 89, 100–03 northern 17, 199–201, 296 northwestern 89, 100 Pitiuses 320, 327 substandard 86–7, 100, 140 Catanzarese (see also Calabrian; southern Italian dialects) 177 Cato 64, 153–4, 220, 223, 226, 231, 232 causative 16–17, 297 c-command 46, 75, 76, 77 Cecchetto 240 Celsius 228 Cena Trimalchionis (see also Petronius) 231 Cennamo 8, 48, 284, 296, 323, 328, 329, 331, 336, 338 central Italian dialects (see also Italo-Romance) 55, 66, 105–6, 177, 296, 299, 314, 327 Chafe 166 Charpin 59, 185–6, 189, 247, 264 Chierese (see also Piedmontese; northern Italian dialects) 324 Chinellato 159 Chitoran 14 Chomsky 107, 160, 184, 260, 281 Christol 90, 91, 92, 95 Chronicon of St Benedict of Sant’Andrea del Soratte 244–5 Chronica Monasterii Casinensis 244–5 Chronicon Salernitanum 244–5 Chronicon Vulturnense 244–5 Cicero 42, 45, 60, 130, 139, 153, 203, 214, 226, 227, 228, 231, 243, 252, 255 Cinque 7, 18, 49–50, 51, 52, 53, 56, 81, 115, 140, 141, 143, 144, 145, 148, 149, 167, 185, 192, 194, 201 Clackson 22, 35, 42, 45, 59, 60, 64, 65, 153, 157, 183, 184, 186, 196, 202, 203, 204, 210, 225, 227, 229, 230, 232, 236, 259 Clark 255, 270 classification of the Romance languages (see also Romània) 314

clause structure cartographic representation 141–50 higher adverb space (HAS) 142–50 lower adverb space (LAS) 142–50 clitic (see adverbs; auxiliary; determiners: possessive; pronouns) Coene 113, 116 Coleman 1, 14, 33, 41, 135, 136, 139, 192, 210 Colombian Spanish 184–5 colon/cola (see also phase) 191–2, 256–9, 259–62, 277 Colon 128 Columella 226 comparative 10–11, 14, 23–4, 26, 201, 207–9 analytic with PLUS/AMPLIUS/MINUS 208–9 complementation (see also object) archaic non-configurational pattern 156–8 accusative and infinitive construction (AcI) 10–11, 78, 157–9, 205, 240–9, 255, 286, 298 accusative subject 245–6 avoidance 246 decline 245–6, 270 loss 158–9, 241–2 position 242–9 Christian Latin 243–4 classical Latin 242–3 discontinuous 243, 270 medieval Latin 244–5 asyndeton 126, 205 parataxis 157, 205 finite clause 60–1, 78, 121, 286, 287–8 extraposition 248, 250–2 heavy 250–2 UT/QUOD/QUIA-clause 158, 159, 205, 208, 240, 244, 286, 287–8 position 242–9 hypotaxis 157, 205 non-finite 287–8 position of complement clause 242–9 complementizer (COMP) (see also Verb Second) 10–11, 13, 78, 80, 81, 157, 205, 207, 287

Index complementizer (COMP) (cont.) finite 169, 205, 241, 287–8 dual complementizer system 169–73, 298 QUAM 208 QUOD/QUIA 158, 159, 205, 240, 244 root clauses 173–6 UT 240 head-marking on C(omp) 298–9 in comparatives 23 interrogative 177–8, 298–9 irrealis/subjunctive 170, 173, 298 non-finite prepositional 169, 171, 287–8 null 240–1, 246–9 position 65, 169–73, 240, 248–9 with topic/focus 171–3 recomplementation 171–3, 175 compound nouns 49, 53, 54, 225 Comrie 16, 297, 312 conditional (see also auxiliary; future: -in-the-past) 14, 33, 135–8 Romance 135–7 inflectional 33, 135–7 Confessiones (Augustine) 90 configurational(ity) 24, 71–80, 281–3, 285 clausal domain (CP) 77–80, 153–5, 193–4 constituent structure 30–1, 35–9, 45–6, 61–4, 71, 194–5 emergence 47–8, 235–6 early evidence 185–96 of functional structure 81–180 flat structure 46–7, 61, 75, 77, 79, 80, 179, 195, 282, 311 gradience 77–80 Hale’s tests 71–7 head-/dependent-marking 285, 311 Latin configurational analysis 181–283 non-configurational analysis 30–80 lexocentricity 71, 72, 259 nominal group 30–1 Latin 31–3, 36–47 Romance 47–58 prepositional group 77–80, 150, 193 pronominal argument hypothesis 47, 311

415

root vs embedded clauses 187 sentence 59–79 classical Latin 59–64 late Latin and early Romance 64–8 modern Romance 68–71 verbal group 30–1, Latin 33–5, 38–47 Romance 47–8, 57–8 conjunction (see complementation; complementizer) constituent structure (see configurational(ity)) Contreras 69 coordination 219–21 null objects 73–4 copula 116 possessive 126, 296–7, 317–19 Corese (see also central/southern Italian dialects; Laziale) 327 Cornilescu 108, 113, 115, 140, 292 Coromines 104 Corsican 67–8, 82, 89, 98, 128, 175–6, 206, 287–8 Corver 43, 280 Cosentino (see also Calabrian; southern Italian dialects) 11, 82, 124, 143, 144, 145, 146, 148, 176, 290, 320, 321–2, 340 Coseriu 12, 14, 15, 25, 27, 28 Costa 7, 69 CP (Complementizer Phrase) 150–80 configurationality 77–80, 193–4 Latin evidence 150–6 phase/colon status 260–2 recursion 173 rise of head-initiality 239–48 Romance evidence 158–80 Cremona 314 creoles 7–8 creolization 333 Crisma 40, 41, 50, 51, 210, 211, 216, 219 Croft 307 Cruschina 8, 70, 159, 162, 163, 164, 165, 168, 177, 299 Cuzzolin 78, 158, 207, 240, 241 Cyrino 74, 127, 144

416

Index

D’Alessandro 8, 144, 145, 149, 159, 171, 175, 296, 299, 323, 343, 349 D’Hulst 113, 116, 135 D’Introno 7, 51, 290 Daco-Romance 199, 288, 314, 333, 346 Dahl 221 Dahmen 7, 295, 296 Dalbera Stefanaggi 7 Dalmatian (see also Istro-Romance; Vegliote) 7, 135, 314, 341 Damonte 8, 70, 159, 170, 173, 298 Danckaert 42, 150, 229, 249 Daniliuc, Laura 7 Daniliuc, Radu 7 Dante 261, 322 dative (see also case) replaced by (reflex of ) AD 22 De agricultura (Varro) 226, 232 de Bello Gallico (Caesar) 254 De Blasi 105 De Cat 287 De Dardel 332, 333 de Jong 39, 151, 216, 232, 259, 263 de Jonge 36 De la Villa 39, 85, 96 De Melo 32, 39, 40, 42, 45, 51, 210 De re coquinaria (Apicius) 226 De Sutter 41 Decat 69 definiteness (see also determiners) 35, 90, 108–9 individuation 85, 97–103 Degree-0 Learnability 187 DeGraff 147 Del Vecchio 59, 225, 230 Delbrück 39, 42, 211 demonstrative (see determiner) Demonte 4, 51, 159, 171, 288 Demuth 287 dependent-marking (see head-marking) deponent (see also alignment: active-stative; voice: middle) 33–4, 133, 315–16, 329–33 derivational morphology 25, 197, 225 Désirat 7

determiner (DET) 2, 31–2, 81–119, 182, 199, 284, 290–2 article 31–2, 80, 81 case distinctions 292 definite 89–109 articloid 90, 91, 96 distribution early Romance 96–8, 103 modern Romance 98–103 article salat 100–04 inalienable possessa 98 IPSE-derived forms in Sardinian 104 proper names 99–100, 103, 108–9 Catalan 99–100 proverbs and fixed expressions 97–8 Romanian 20, 28, 108, 199, 241–2, 292 with prepositions 107 emergence and grammaticalization 85, 89–98, 107 honorific 100 ILLE 77, 89–96 IPSE 77, 89–96 late Latin 89–96 demonstrative (Romanian) 113–15 indefinite 82–9 information structure 88–9 plural 82–4 quantifiers 88–9 zero 84, 85–9 proverbs and fixed expressions 87 Left Branch Condition 279–80 neuter 105–7, 300–1 partitive 84–5 possessive 115–18, 212 demonstrative 32, 89–96, 110–11 IPSE-based forms 89 metaphony 106 position in Latin 32–3, 36, 39 position in Romance 32, 110–11, 199 Romanian short forms 110 head-marking on D(et) 290–2 intensifier

Index determiner (DET) (cont.) position 37, 39 locus of agreement 290–1 null 108–9 position 32–3, 34 possessive 32, 111–13 article (Romanian) 115–18 determiner (clitic) vs adjectival (tonic) 111–13 double-marking 288–9 position Latin 32, 37, 39, 40, 51 Romance 33, 111–13 quantifier 36, 39, 84 agreement on cardinal quantifier 309 Devine 32, 34, 35, 36, 39, 40, 41, 42, 46, 47, 59, 71, 78, 151, 152, 155, 157, 187, 193, 195–6, 205, 216, 219, 223, 236, 240, 242, 243, 247, 249, 250, 251, 254, 256, 259, 263, 264, 267, 270, 271, 272 Di Felice 175 Diaconescu 288, 350 Dias 7, 84 Diez 4, 18 differential object marking (see case: prepositional accusative) diminutive suffix 16 Dimitrova-Vulchanova 108 discontinous structures / discontinuity (see fronting; negation) disjunction (see fronting) Dixon 187, 231, 312 Dobrovie-Sorin 7, 70, 113, 116, 145, 159, 170 Dols 7, 99, 100, 101 Doria 7, 69 DP (Determiner Phrase) 32, 72, 80, 82–119, 182–3, 199, 288, 290–1 left periphery 258, 259–62, 263, 264, 267, 268, 269–70, 271, 272, 274, 276, 277–81 Left Branch Condition 277–80 phase/colon status 261–2 Dressler 72 drift 10 Dryer 46, 71, 199, 202, 210, 221, 241, 248–9

417

Duarte 159 Dupuis 65 Durante 7, 328 Dyirbal 285 early Romance (see medieval Romance) edge-fronting (see fronting) Egidi 299 Elcock 4 Elerick 39, 40, 42, 60, 61, 188, 202, 216, 227, 236, 252–4 Elvira 85 embedded clause (see complementation; subordinate clause) Emilian (see also northern Italian dialects) 200 Emonds 141 Engadine (see also Ræto-Romance; Romansh) 334 England 341 English 120 Engver 145 epistemic modality (see also auxiliary; modal verb) 2–3, 140 EPP (Extended Projection Principle) (see also subject) 34, 270 Ernout 12, 39, 41, 45, 59, 62, 63, 130, 151, 157, 159, 190, 191, 193, 195, 196, 206, 207, 209, 213, 214, 219, 221 241 Ernst 4 Ervedosa do Douro (see also Ibero-Romance; Portuguese) 294 Escobar 7 ESSE-fronting (see also Wackernagel) 156, 191, 256–9, 260 evaluative suffix 14 Extremaduran / Extremeño (see also Ibero-Romance) 7, 198 Ewert 7 existential construction 328 extraction (see fronting; Left Branch Condition) Eyþórsson 296, 323 Fabra 7 Fagyal 7, 53

418

Index

Faliscan (see also Italic) 213, 231 Fankhänel 39, 227 Fedele 148 Feix 39, 40, 41, 210 Fernández Soriano 159, 171 Ferraresi 158, 240 finiteness (see also complementation; complementizer; left periphery: Fin(iteness)) 67, 147, 150, 173–7 auxiliary selection in Romanian 346–7 inflected gerund and present/past participles 293–5, 301, 308 inflected infinitive 14–15, 293–4, 302, 308 Fiorentino 288, 350 Firmicus 226 Fischer 39 Fleischman 13, 14, 33, 135, 136, 187 Florentine (see also Italian; Italo-Romance; Tuscan) 177–8 focus / focalization (see also colon/cola; fronting; left periphery; phase; word order) 45, 50, 65, 70, 79, 150–1, 153, 154, 155–6, 188, 192, 216, 256–7, 268 focalizing adverb 189–91 Focus field 159–66, 267–8 contrastive focus 160, 162–5 indefinite quantifier 165 informational focus 160, 162–5 negator NON 256–7 postverbal positions 155 FOFC (Final-over-Final Constraint; see also head parameter; word order) 204, 221, 238–59, 269–70 adpositional phrase 240–2 gerundive complement 252–3 position 249–52 clausal word order 255–8 complementizer phrase 239–42 position 242–9 other harmonization patterns 252–5 roll-up 236–59, 269–70, 272–6, 281–2, 285, 310 partial 272–7 reanalysed as edge-fronting 269–70

Fontana 65 Formentin 105, 322, 334, 336, 341, 344 Forteza i Cortès 100, 101, 102 Fortson 195 Foubert 45, 259, 263 Foulet 42 Fraenkel 45, 191, 259, 260 Francavillese (see also Salentino; southern Italian dialects) 173 Franco-Provençal 242 Frank 71 Frankish 335 Frascarelli 168 Fredegar’s Chronicle 136 French 3, 7, 11, 14, 15–16, 17–18, 19, 20, 25, 27, 32, 33, 48, 52, 53–4, 55, 56, 65, 66, 67, 69, 74, 75, 82, 83, 84, 85, 86, 87, 89, 96, 97, 98, 99, 100, 103, 107, 108, 112, 113, 120, 121, 122, 123, 124, 125, 129, 134, 135, 136, 140–1, 143, 144, 145, 146, 147, 162, 163, 166, 169, 171, 186, 198, 238, 280, 282, 289, 290, 291, 297, 299, 320, 327, 334, 335–6, 337–8, 348, 350 Acadian 341–2 Canadian 341 colloquial 178, 200, 293, 310–11 literary 335 popular 299 Friedrich 203 Friulian (see also Ræto-Romance) 7, 124, 134–5, 148–9, 177, 178, 224 fronting (to left edge) 43–7, 61–4, 71, 72, 281–3, 285, 310 anastrophe 78 colon/phase 259–62, 277 decline of case system 277 definition 258–9 diachronic variation 44–5 (left) edge-fronting (see also Left Branch Condition) 182–3, 192, 236, 258–81 focus-fronting 182–3, 185, 188, 192, 244, 247, 258, 259, 263, 270, 271, 277 topic-fronting 182, 192, 247, 258, 259, 263, 270, 277 Left Branch Condition 277–81

Index fronting (to left edge) (cont.) local 43–4, 264–5 locality 264 loss 277–81 medieval Romance 68, 257, 277 modifier 32, 262–70 partial roll-up 276 multiple 264–5 nominal 270–6 in adpositional phrase 275–7 partial roll-up 272–7 pragmatic interpretation 272 non-local 264–7 pragmatic function 45, 259 reanalysis of roll-up 269–70 prepositional group (see also anastrophe) 43–4, 78–9, 195–6 register 45, 78, 259 Fruyt 16, 90, 91, 92, 94, 95 Fugier 41 functional categories (see adposition; auxiliary; complementizer; determiner; functional structure; I(nfl); left periphery vP) functional structure (see also auxiliary; complementizer/CP; determiner/DP; I(nfl)/IP; 24, 30, 34, 80–180, 182–3, 198, 281, 282–3 emergence 179–80, 310 head-/dependent-marking 285, 289–99, 310 in Latin 183–5, 198 future (see also auxiliary) in-the-past 135–8 Latin 13, 24 periphrastic 12, 13, 34, 134, 135, 192 inflectional 10–11, 13, 134, 225 Romance auxiliary 10–11, 14, 122–4, 128–9, 134–40, 225 particle construction 126 synthetic 33, 135–40 Gaius 223, 226 Gajewski 31, 280 Galician 7, 13, 89, 127, 172, 213, 224, 293–4, 314 Gallese (see also Campanian; southern Italian dialects) 105, 205

419

Gallo-Romance 2, 15, 17, 64, 65, 67, 69, 74, 84, 178, 199–201, 290, 292–3, 314, 332, 333, 335 Galves 74 Ganzoni 322 gapping 60, 227 García Arias 7 García de Diego 7, 184 García de la Fuente 214 García Sanchidrián 214, 229 García Santos 7 Gardenese (see also Ladin; RætoRomance) 54 Gartner 7 Garzonio 159, 177, 299 Gascon (see also Occitan) 13, 14, 20, 88, 89, 97, 173–5 gender marking (see also agreement; determiner: article) nominal 13, 26, 290–1 metaphonetic 14 neuter 105, 300–1 Occitan proper names 13, 103 zero 107 participle 286 verbal (Ripano) 300–8 genitive (see also case; word order) position of dependent genitive 25, 37, 39, 40–1, 42, 80, 207 replaced by (reflex of ) DE 22, 215 Genoese (see also Ligurian; northern Italian dialects) 293, 334, 337 German 65, 120, 148, 250 Germanic 119, 148, 344 Gerola 48, 328, 330, 331 Gianollo 40, 41, 47, 50, 80, 210, 211, 213, 214, 215, 216, 219 Gierling 290 Gil 31 Gildersleeve 39, 41, 42, 43, 51, 53, 190, 193, 195, 205, 206, 207, 210, 213, 220, 268 Gilligan 201 Giorgi 51, 184, 279 Giurgea 69 Giusti 7, 41, 45, 51, 108, 115, 126, 199, 210, 219, 261

420

Index

Glatigny 51 Gleßgen 4 Godard 51,144 Goldbach 158, 240 Golston 273 Golumbia 47, 71 Gonzaga 51 González i Planas 159 González Rolán 131 Graffi 7, 152 grammars in competition 60 grammatical functions/relations (see also object; subject) 61, 71, 72, 84 grammaticalization 28–9, 77, 180 auxiliaries 119, 121–40, 136, 138, 192 decategorialization 124–5 morphophonological specialization 127–9 perfective 130–4, 317–19 reanalysis 131–3 semantic weakening 121–4, 136 definite article 85, 89–98 prepositions 241 preverbal subject position 349–50 Grandgent 3, 7, 19, 21, 45, 62, 100, 196, 203, 224, 225, 241 Grazzanisano (see also Campanian; southern Italian dialects) 168 Greco 158, 205, 240, 242, 244–5, 246 Greek 115, 195 Green 8, 14, 69, 119, 135, 288, 314, 350 Greenberg 64 196, 204, 227 Gregory of Tours 132, 329–30 Grohmann 281 Grosu 108 Guasti 16, 297 Gupton 171 Gutiérrez Ordóñez 214 Guţia 7, 51 Habinek 45, 260, 261, 262 Haden 341 Haida 47, 210, 225 Haiman 7, 65, 82, 332

Hale, Ken 47, 71, 72, 77, 79, 80, 188, 258, 311 Hale, William 39, 41, 43, 59, 62, 63, 190, 191, 193, 205, 206, 207, 209, 210, 220, 259 Hall 105, 106 Harder 299 Harmonic Phenotypic Linearization (see also FOFC) 254 Harris, Alice 124, 132, 135, 136, 187, 231 Harris, Martin 2, 3, 4, 10, 14, 15, 21, 24, 25, 26, 59, 64, 65, 69, 184, 196, 204, 310–11, 332, 334, 335 Haspelmath 221 Haudry 78, 157, 205, 240 Haverling 13, 34, 131, 132, 135, 136 Hawkins 201, 202, 204, 209, 210, 241 head-marking and dependent-marking 25, 283, 284–311 configurationality 285 free word order 310–11 clitic-doubling 310–11, 314 prepositions 291 variation 286–9 split-marking 287 verb phrase 286, 287, 288 C(omp) 298–9 configurationality 311 D(eterminers) 290–2 case: Romance article 292 clitic-doubling 288, 290, 310–11, 314 possessives 288–9 double-marking 288–9, 295, 299 functional categories 285, 289–99 Infl 292–8 auxiliary selection 314, 317–26 active-stative split 295, 317–23 determined by finiteness 346–7 Romanian pluperfect surcomposé paradigm 347 modal conditioning 344–6 person split 296, 323–6, 343–4, 349 tense split 342–4 triple auxiliation 323–6 causative 297 imperative 297–8

Index head-marking (cont.) participle agreement 295, 314, 326–7, 347–9 possessive copula 296–7 subject clitics (see also subject) 292–3 null anaphora 311 Ripano 299–310 cardinal quantifiers 309 object-verb agreement 305–8 clitic-controlled agreement 305–6 predicative complements 309–10 preposition 308–9 subject-verb agreement 300–2 gerund 301, 308 infinitive 302, 308 past participle 301 subject-/obect-verb agreement 302–5 matching agreement 302–3 non-matching agreement 303–5 wh-phrases 309 head parameter (see also FOFC; word order) 24–5, 33, 42, 78, 80, 136, 138, 181–2, 196–258, 281–3 crosscategorial harmonization 216, 242 diaphasic variation 236 Indo-European 202–3, 206, 220, 225 Latin head-final > head-first 202–35 adpositions 205–7 adverbs 222–3 auxiliary + infinitive constructions 234–5 comparatives 207–9 complementizers 205, 207 coordination 219–21 pragmatic principles 220 ESSE + participle 223–4, 256–9, 316, 317 interrogative markers 221 negation 221–2 noun phrase 210–19 AN/NA 210–13 GN/NG 207, 213–19, 224, FOFC 252–5 pragmatic principles 216–17, 236 relatives 209–10 VO/OV 203–4, 225–34

421

diamesic/diaphasic variation 231 register-switching 230 root vs embedded clauses 230–1, 232, 238, 248, 257–8 unaccusative structures 231–4, 335–6, 349 stylistic/pragmatic variation 204, 209 Latin > Romance 235–58, 273 roll-up 236–59, 285, 310 partial 272–7 morphology 197, 225 pragmatically motivated variation 236 rise of head-initiality 238–59 adpositional phrase 240–2, 249–55 complementizer phrase 239–48 Romance counterexamples 198–202 heaviness 60, 250–2 Heine 121, 122 Heir-Apparent Principle 124, 132 Herman 1, 31, 35, 39, 41, 42, 45, 47, 48, 51, 60, 61, 64, 65, 90, 96, 134, 138, 158, 159, 205, 210, 216, 229, 240, 241, 242, 244, 259, 277, 328, 329, 331 Hewson 12, 31, 35, 71, 78, 187, 193, 205, 206, 207 Hill (see also Motapanyane) 159, 296 Hilty 14 Hinojo 214, 225, 228, 229 Hinterhölzl 168 Hintze 7, 53 Hofmann 13, 39, 42, 45, 47, 78, 135, 136, 193, 195, 206, 211, 219, 220 Holmberg 204, 221, 238, 241, 242 Holtus 4 homoeoteleuton 189 Hopper 28 Horace 62 Hordé 7 Hornstein 147 Horrocks 22, 35, 42, 45, 59, 60, 64, 153, 186, 196, 202, 203, 204, 210, 225, 227, 229, 230, 232, 236, 259 Howe 184 Hualde 7, 128 Huang 202, 210 Huber 7, 198, 341

422

Index

Humboldt 12 hyperbaton (see fronting) Ibero-Romance 2, 18, 19, 66, 69, 82, 83–4, 86, 89, 105, 106, 110, 121, 127, 135, 137–8, 171, 175–6, 177, 199, 205, 206, 288, 293–4, 296, 299, 314, 341 northwestern 184 Icelandic 285 ILLE

development into article 77, 89–96 topic marker 77 illocutionary force (see also interrogative; left periphery: Force) 150, 155–6, 175 imperative 125, 152 transitivity distinction 297–8 negated 221–2 Indo-European 22–3, 30, 42, 59, 78, 81, 157, 158, 179, 181, 202–3, 206–7, 225, 240, 242, 330 Indovinello veronese 96 infinitive (see also complementation) inflected 14–15, 293–4, 302, 308 inflectional morphology 15, 63, 107, 140, 251–2 interrogative markers 19, 177–8, 221, 298–9 null objects 73–4 polar 67, 152 V(erb)-raising 178 wh- 164, 193 agreeing wh-phrase 309 multiple wh-fronting 162 IP (Inflectional Phrase) 79, 80, 119–50 head-marking on Infl 292–8 IPSE

development into article 77, 89–96 development into demonstrative 89 position 37, 39 topic marker 77, 90 Irsinese (see also southern Italian dialects) 341 Ishii 281 Istro-Romance (see also Dalmatian; Vegliote) 7, 314

Italian 7, 14, 15, 16–17, 18, 19, 20, 25, 33, 49, 52, 53, 54, 55, 56, 61, 70, 75, 77, 82, 84, 85, 86, 87, 97, 98, 107, 109, 111, 112, 120, 121, 122, 125, 127, 130, 135, 136, 140–1, 142, 143, 144, 145, 146, 147, 148, 149–50, 169, 182, 197, 198, 201, 280–1, 291, 300, 307, 320, 321–2, 324, 327, 334, 335, 336, 348 central-northern varieties 99 northern 109 southern 109 Trentino 99 Italic 204, 206, 211 Italo-Romance 2, 82, 96, 105, 136, 160, 206, 319, 323, 327, 335 central 136 northern 136 Ive 7 Jaeggli 288 Janson 191, 220 Jelinek 47, 71, 310, 311 Jenkins 7, 53 Jèrriais (see also French; Gallo-Romance) 341 Jespersen 17, 200 Jones, Mari 341 Jones, Michael 7, 14, 55, 70, 119, 128, 129, 162, 178, 288, 293, 299, 320, 350 Jones, Samantha 299, 305 Jordana 7 Joseph, Brian 221 Joseph, John 78, 173, 206, 207 Juge 128 Jurament feudal 97 Kabatek 1 Kaiser 65 Kalepky 288, 350 Karlsen 158, 240, 242, 244, 248 Karlsson 14, 212 Kayne 16, 140, 220, 236, 292, 296, 297, 323 Kessler 35, 43, 45, 59, 71, 195, 259, 262, 264 Kibbee 7, 53 Kiss 71, 163 Klausenburger 28

Index Koch 23 Koll 225, 228 Konneker 39, 42, 59, 204, 211 Koschwitz 54 Kramer 295 Kroch 60 Kroll 39, 153, 209, 227 Kühner 39, 42 La Fauci 2, 3, 8, 14, 15, 21, 33, 48, 64, 135, 201, 251, 312, 313, 314, 316, 317, 318, 322, 328, 329, 331, 332, 332, 334, 335, 340, 341, 348, 349, 350 Labelle 159 Ladin (see also Ræto-Romance) 7, 65 Laenzlinger 50 Lafont 7, 54, 83, 173 Lakoff 3, 21, 59, 241, 332 Lamarche 53 Lambrecht 162 Landais 88 Landau 76 Langslow 8, 41, 50, 210, 211, 270 Lanly 135 Lapesa 7, 51, 85 Latin 1–3, 5, 11, 12, 13, 15, 21–2, 23, 25, 27, 28, 30, 31–2, 33, 34, 35–47, 48, 50–1, 71–3, 75–6, 79–80, 90–6, 126, 130–3, 134, 136–7, 138, 150–8, 179–80, 181–283, 284, 286, 287, 289, 291, 297, 310, 311, 314–6, 328–32 African 210 archaic 42, 45, 59, 158, 181–2, 186, 203–4, 209, 213, 222, 230–1, 236, 240, 241, 270, 285 Christian 243–4 classical 39, 41, 42, 44–5, 59–64, 181–2, 186, 203–4, 213, 215–16, 222, 225, 227, 236, 243, 254, 255, 312, 314, 314–16 colloquial 60, 138 early 42, 44–5, 59, 60, 64, 85, 132, 153–4, 203–4, 206, 208, 212, 222, 224, 230, 236 formal 206, 209, 230, 234 formulaic 60, 209, 213–14, 230

423

late 40, 41, 64–8, 89–96, 153, 156, 186, 203–4, 213, 215, 225–6, 233, 236, 243–4, 249, 251, 254, 277, 314, 316–19, 328–9, 330, 336 legal 60, 209, 221, 225, 230 literary (registers) 39, 45, 59, 60, 78, 203–4, 215, 224, 225, 229, 233–4, 285 medical 41 medieval 213–14, 230, 244–5 Merovingian 330 normative (registers) 224, 229, 233, 276 poetry 45, 64, 78, 195–6, 229, 275–7 religious 212, 230 spoken (registers) 59, 136, 230, 231 sub-substandard 96, 330 subliterary 42, 59, 138, 153–4, 215, 223, 229, 231, 245, 246, 249, 251, 270, 277 substandard 136, 138, 156, 246, 330 vulgar 41, 159 written 45, 230, 231 Laughton 260 Lausberg 4, 134 Laziale (see also central/southern Italian dialects) 171–2 Leal Cruz 15, 184 Leccese 126, 141, 321–2 Ledgeway 4, 7, 8, 14, 31, 48, 51, 53, 54, 55, 65, 68, 82, 84, 88, 89, 105, 121, 126, 128, 129, 140, 143, 144, 145, 148, 156, 159, 160, 161, 168, 169, 170, 171, 173, 175, 177, 185, 194, 198, 251, 288, 293, 296, 298, 299, 323, 328, 336, 339, 341, 343, 344, 350 Left Branch Condition (see also fronting) 32, 183, 274, 277–8 left periphery of the clause (see also complementizer; DP; focus: Focus field; fronting; PP; topic: Topic field; vP) 65–6, 150–3, 158–80, 205, 258, 259–62, 264–7, 278 Fin(iteness) 169–76, 173–6 Force 169–73 concessive/conditional clauses 176–7 exclamative clauses 177 lower vP left periphery 258, 259–62, 267, 278 Legate 71 Legendre 145, 296, 323

424

Index

Leges XII Tabularum 59, 209, 211, 214, 220, 222, 223, 225, 228, 230 Lehmann, Christian 41, 46, 72, 78, 157, 193, 205, 206, 210, 240, 286, 310 Lehmann, Winfred 39, 42, 59, 196, 202, 204, 211, 225, 227 Lemieux 65 Lengadocien (see also Occitan) 14, 18, 19, 55, 56, 83, 85, 89, 98, 99, 120, 121, 122, 123, 200, 320, 337–8 Leonese (see also Ibero-Romance) 7, 184–5, 293 Lepschy, Anna Laura 7 Lepschy, Giulio 7, 323 Leumann 78, 193, 195, 206 lexocentricity 71, 72, 285 LFG (Lexical Functional Grammar) 6 liaison 53–4 Libellus a regula sancti benedicti subtractus 229, 230 Lightfoot 147, 187, 225, 231 Ligurian (see also northern Italian dialects) 171, 172, 334 Linde 39, 40, 42, 47, 59, 64, 151, 152, 185, 186, 189, 222, 226, 227, 229, 230, 232, 249, 253 Linder 7 Lindsay 22, 23 Linear Correspondence Axiom 236 Lipski 7, 184, 290 Lisón 214 Livinallonghese (see also Ladin; Ræto-Romance) 334 Livy 214, 226, 229, 243, 254–5 Lloret 7 Lloyd 7 Lobo 293 Lodge 39, 41, 42, 43, 51, 53, 190, 193, 195, 205, 206, 207, 210, 213, 220, 268 Löfstedt 21, 96, 241, 328 Logudorese (see also Sardinian) 172–3, 334 Lois 140, 144 Lombard (see also northern Italian dialects) 138, 334, 337 Lombard, Alf 4, 7, 86, 297, 298 Lombardi 8, 65, 85, 90, 97, 107, 111, 140, 143, 144, 145, 148, 170, 320

Longa 293 Longo 119 Longobardi 50, 51, 107, 199 Loporcaro 8, 14, 70, 105, 135, 293, 295, 296, 298, 299, 307, 319, 321, 322, 323, 324, 327, 337, 341, 346, 348 López Fonseca 135, 159 Lot 2 Lucifer of Cagliari 243–4 Lüdtke 299 Luís 16, 138 Lulese (see also Sardinian) 320 Lurà 323 Luraghi 71, 72, 73, 78, 292 Lyons 31, 75, 80, 81, 85, 90, 92, 107, 111 Macedonian 296 Maceratese (see also central Italian dialects) 324–5 Magni 2, 23, 41, 42, 46, 64, 78, 202, 203, 204, 206, 207, 210, 213, 215, 219, 221, 227, 236, 251 Maiden, Liliana 110 Maiden, Martin 4, 7, 8, 14, 15, 29, 33, 51, 85, 98, 106, 110, 135, 296, 335 Malkiel 1 Mallinson 7 Mancarella 298 Mancini 299, 304, 308 Manoliu Manea 7, 113 Mantovano (see also Emilian; Lombard; northern Italian dialects) 334 Manzini 7, 170, 177, 178, 296, 299, 323, 343, 344 Maracz 71 Marazzini 7 Marcato 341 Marchetti 67, 98 Marouzeau 32, 34, 35, 39, 40, 41, 42, 43, 45, 50, 51, 59, 78, 150, 151, 153, 185, 193, 195, 206, 211, 223, 227, 230, 256, 259 Marsalese (see also Sicilian; southern Italian dialects) 126 Martín Zorraquino 288, 350 Mascaró 7

Index Mattoso Câmara 7 Maturi 55, 339 Maurer 14, 293 McFadden 344 McLachlan 8, 213, 214, 216–18, 236, 255 medieval Romance 65–6, 137, 161–2, 186, 203–4, 314, 316–19 Megleno-Romanian 295, 296 Meier 288, 350 Meillet 1, 10, 30, 35, 39, 59, 62, 71, 230 Melicucco 298 (see also Calabrian; southern Italian dialects) Mendrisiotto (see also northern Italian dialects) 323 Menéndez Pidal 7, 136, 138 Mensching 8, 14, 70, 162, 164, 178, 293, 299 Mereu 64, 71, 77, 79, 80, 284 Merlo 105 mesoclisis (see also clitic) 16, 138 Messinese (see also Sicilian; southern Italian dialects) 170 Messing 105, 106 metaphony 14, 106 Metzeltin 4 Meyer-Lübke 4, 15, 288, 350 Mexican Spanish 345 middle (see voice) middle field 70 Middleton 65 Milanese (see also Lombard; northern Italian dialects) 85, 200 Miller 23, 225 Mirror Principle 147, 251–2 Möbitz 152, 153, 223, 227, 249, 256 modal verb (see also auxiliary) 28, 34, 135, 136, 138–9, 180, 192 cartographic representation 141, 143–4, 192 deontic 123–4, 136 position 234–5 VP-ellipsis 194–5 Mohawk 285 Mohrmann 1 Moldovan (see also Daco-Romance) 117 Molinelli 3, 21, 22, 64, 210, 241, 244

425

Moll 7, 100, 320, 348 Monachesi 16, 19, 128, 129, 138, 143, 144, 145 Monteagudo 7 Montpelhierenc (see also Occitan) 13 Morais 74 Moran 103 Moreno Hernández 228, 230 Morimoto 221 Morin 13, 159 Moro 69 Motapanyane (see also Hill) 7, 69, 70, 108, 113, 116, 119, 140, 159, 162, 170, 288, 290, 297, 346 Mougeon 341 movement (see also N(oun)-movement; raising; V(erb)-movement) A- vs A'-movement 270 Chain Uniformity Condition 195–6, 271 Head Movement Constraint 114 Mulder 72 Muldowney 39, 64, 223 Muller 3, 21, 96, 226, 241 Mulomedicina Chironis 226, 229, 230, 329 Munaro 81, 159, 164, 176, 185 Muro Lucano (see also southern Italian dialects) 14 Muysken 71 Nagore Laín 320 Naudeau 89 Navajo 285 negation 12–13, 199–201 discontinuous 17–18 Jespersen’s cycle 17, 199–200 NegP 200–1 partitive article 84 position 127, 221–2, 256 root clauses 173–4 with imperative 221–2 Neapolitan (see also Campanian; southern Italian dialects) 14–15, 51, 53, 54, 55, 56, 66, 68, 84, 86, 88–9, 105, 106, 109, 112, 120, 121, 123, 128, 167, 170, 173, 178, 200, 293–5, 334, 336–6

426

Index

Nemorense (see also Laziale; central/ southern Italian dialects) 105 Newton 204, 221, 238, 241 Nichols 201, 284, 285, 286, 287, 288, 292, 310, 311 Niçois (see also Occitan) 13 Nisbet 62 Nocentini 14, 33, 89, 92, 135, 288, 350 nominative (see also alignment; case) 119, 174, 286–7, 288 loss of final -S 21 plural in -AS/-OS 330–1 restricted 328–33 retention 64 with animates animates 334–5 non-configurational(ity) (see configurational(ity)) 30–80 Norberg 1, 328, 330 Nordlinger 6, 285 northern Italian dialects (see also ItaloRomance) 66, 137–8, 314 noun incorporation 71 N(oun)-movement 56–7, 108–9, 112, 113, 114, 115, 117–18, 182–3, 219, 242, 285 absence in Latin 182–3, 219 NP (Noun Phrase) 32–3, 35, 47–58, 63, 210–19 bare 85–9 cartographic representation 56–7 null anaphora/arguments (see also object; subject) 71–5, 147, 252, 289, 300–1, 311 number marking (see also agreement) nominal 13, 26, 54, 290–1 collective plurals 82–4 metaphonetic 14 non-count 105–7 Occitan doubly-marked inflectional plurals 13 Spec-head 116–17 zero 107 Nuorese (see also Sardinian) 125, 345 Nouvel 55 Nunes 7 Nuñez 34, 132

Nursino (see also central Italian dialects; Umbrian dialect) 105 Nuti 212, 213, 214, 215, 216, 296, 318 object (see also complementation) heavy 250–2 null 71–5, 289, 311 position 38, 42, 57–8, 62 bare case-marked vs prepositional 249–52 raising 64 Occitan (see also the names of the individual Occitan varieties) 3, 7, 11, 13, 18–19, 33, 53, 55, 66, 68, 82, 83, 86, 87, 89, 96, 97, 99, 100, 103, 110–11, 122, 123, 135, 137, 177, 208, 224, 292, 295, 298–9, 314, 324, 327, 334, 335, 348 Pyrenean 351 southeastern 84 Olarrea 7 Oliviéri 292 Olotí (see also Catalan) 324 Olsen 4, 7, 51, 110 Oniga 8, 25, 31, 39, 41, 42, 43, 45, 59, 60, 64, 71, 72, 78, 150, 157, 185, 193, 196, 197, 202, 204, 205, 207, 210, 216, 218, 219, 224, 225, 229, 230, 236, 240, 241 Oribasius 132 Orinsky 153, 223 Orlandini 16 orthography 19–20, 21–2 Oscan (see also Italic) 42, 186, 206, 215 Ossalais (see also Occitan) 174 Ostafin 59, 71, 150, 195 OV/VO (see head parameter; word order) Ovid 226, 228 Paciaroni 105, 324 Padua 69 Paduan (see also northern Italian dialects) 177, 295 Palafrugellenc (see also Catalan) 100, 101 Palladius 226 Palmarino 178 Palmer, Frank 221 Palmer, Leonard 78, 157, 240, 241

Index Palmero Spanish (see also Spanish) 16 Panchón 59, 214, 228 Panhuis 36, 41, 42, 45, 59–60, 187, 202, 203, 204, 225, 227, 229, 230 Paoli 159, 162, 163, 164, 171 Papyrus Bononiensis 226, 244 Parascandola 342 Parkinson, Jennie 135 Parkinson, Stephen 7 Parrino 299 Parry 4, 7, 8, 17, 85, 90, 97, 107, 199, 336 past participle agreement (see agreement) passive (see also auxiliary; voice) 2–3, 12–13, 27, 136, 315–16 analytic Latin 28, 33–4, 133–4, 317–19 position of ESSE 28, 34, 38, 133, 191, 223–4, 255–9, 260 Romance 121–2, 125–6, 322–3, 322–3 deontic 123–4 extension of ESSE to imperfective 11, 15 impersonal passive 27, 328 subject case-marking 328–33 positions 213–4 Payne 221 Penney 78, 195, 206, 221 Penny 7, 15, 106, 184, 332 Pensado 48, 288, 328, 331, 350 Pensalfini 71 Perea 100 Peregrinatio Aetheriae 22, 64, 65, 90, 91–5, 153, 214, 226, 229, 233, 244 Pérez Saldayna 7, 128 Perrochat 39, 158, 159, 205, 227, 240 person and number marking (see also agreement) 13–14 head-marking 286, 292–3, 296 null subjects 34, 67, 71–74, 174, 252, 300–1, 311 Ripano 300–2 personal a (see case: prepositional accusative) Pescasserolese (see also Abruzzese; southern Italian dialects) 104–5

427

Petersmann 234 Petronius 153, 214, 226, 227, 231, 245 Peverini 338–9 phase (see also colon/cola) 191, 259–62, 277 Phase Impenetrability Condition (PIC) 242, 264–5, 274 phasehood: absence/presence of articles 280–1 Pianesi 184 Picallo 51 Piedmonese (see also northern Italian dialects) 17, 84, 293, 296, 334, 336–7 Pieroni 8, 72, 76–7, 90, 94, 328 Pinkster 8, 12, 13, 14, 21, 22, 23, 24, 31, 33, 35, 39, 40, 41, 42, 45, 51, 59, 60, 62, 63, 64, 89, 90, 130, 131, 134, 135, 136, 150, 151, 155, 185, 191, 203, 210, 211, 219, 226, 227, 228, 229, 230, 231, 232, 242, 246, 259, 263, 318 Pires 14, 293 Pîrvulescu 297 Pisano 128 Placiti cassinesi 96 Plangg 7 Plautus 42, 45, 51, 59, 60, 75, 85, 92, 132, 208, 209, 209, 211, 216, 220, 222, 223, 228, 230, 232, 233, 235 Pliny (the Younger) 226 plural (see agreement; number marking) Poirinese (see also Piedmontese; northern Italian dialects) 178 Poletto 8, 68, 159, 160, 161, 163, 164, 165, 167, 168, 260, 277, 279, 292 Politzer 226 Pollock 140, 141 Polo 59, 60, 71, 150, 152, 153, 155, 187, 228, 249, 250, 251, 272 polysynthetic languages 310 Pompeiano (see also Campanian; southern Italian dialects) 323 Pompeius 138–40, 215, 234 Pompey Inscriptions 229, 251 Pontes 69 Pope 7 Portuguese 7, 14, 15, 16, 18, 19, 20, 52, 53, 54, 55, 56, 57–8, 66, 69, 82, 83, 84, 86, 87, 89, 97,

428

Index

99, 106–7, 110, 111, 122, 124, 125, 126, 127, 135, 163, 196–7, 206, 289, 293–4, 297, 314, 341, 348, 351 Brazilian 74–5, 112–13, 123, 135, 144, 289, 290–1, 334–5 European 16, 74, 99, 112, 293–4 literary 137–8, 225 Porzio Gernia 202, 236 Posner 4, 10, 17, 106, 199 possessive (see determiner) Postal 76 Pottier 119 Pountain 1, 7, 51, 119, 199, 210 Póvoa de Atalaia (see also Portuguese) 294 Powell 43, 45, 62, 258, 263, 267, 268, 271 Pozas-Loyo 85 preposition/PP (prepositional phrase; see adposition/adpositional phrase) prepositional accusative (see case) present perfect 2–3, 184–5, 234 auxiliary selection 121, 130–4, 295–6, 314 active-stative split 295, 317–26 determined by finiteness 346–7 Romanian pluperfect surcomposé paradigm 347 modal conditioning 344–6 person split 296, 323–6, 343–4, 349 tense split 342–4 triple auxiliation 323–6 periphastic 12, 15–16, 18, 19, 24, 122, 123, 125, 130–4, 184–5, 234 synthetic 15–16, 27, 184–5 with modal 120–1 preterit (see also auxiliary) loss vs retention 314 preverb (see adposition) Price 1, 7, 17, 65, 82, 85, 199, 332 Prieto 299 pro-drop (see null anaphora; object: null; subject: null) Procidano (see also Campanian; southern Italian dialects) 342–3 pronoun case distinctions 14, 333

clitic 81, 259 object 77, 96, 284 agreement controller 305–8 climbing 120, 127 doubling 288, 290, 310–11, 314 enclisis 161–2 INDE-cliticization 339–40 proclisis 161–2 Tobler-Mussafia Law 161–2 resumptive 161, 166 subject 292–3, 349–50 with postposition 206–7 Wackernagel (see also colon/cola; phase) 63–4, 140, 191–2, 260 Proto-Indo-European 39, 206–7, 209, 211, 241 Provençal (see also Occitan) 54, 66, 67, 99, 167, 168, 184–5, 323 Pugliese (see also southern Italian dialects) 15, 126 northern 170 psych-predicate 76 quantifier (see determiner) Quintilian 31, 36, 39, 193, 203 Rabella i Ribas 103 Radatz 51, 53 Ræto-Romance (see also Friulian; Ladin; Romansh) 7, 15, 65, 82, 84, 164–5, 176, 242, 292–3, 314, 332, 333, 334 raising (see also N(oun)-movement; V(erb)-movement) subject-raising 117 Raiskila 138 Ramat 31, 59, 78, 81, 124, 157, 205, 229, 240, 242 Ramsden 185, 230 Raposo 14, 74, 293 Ravier 89, 173 recomplementation (see complementizer) Regueira 7 Reichenkron 288, 350 Reiner 51 relative clauses 139, 209–10 active-stative distinction 336–7

Index relative clauses (cont.) correlatives 157, 205 V-finality 139, 225 Remberger 8, 69, 70, 119, 162, 164, 178, 299 Renzi 7, 65, 84, 86, 89, 90, 92, 97, 196, 197, 199, 200, 292 Repetti 292 restructuring 127 resultative 15–16, 130–2, 296 Revelaciones 244, 248 Rhodanien (see also Occitan) 54 Ribeiro 65 Ricca 14, 189, 212, 222 Rich Agreement Hypothesis 147, 251–2 Richter 2 Riemsdijk, van 43 Rigau 299 Ripano (see also central Italian dialects) 299–310 Ritmo Laurenziano 137 Rivero 152 Rizzi, Elena 210, 226, 244 Rizzi, Luigi 74, 76, 81, 141, 150, 159, 163, 169, 177, 183, 280 Roberge 292, 297 Roberts 8, 23, 65, 66, 135, 138, 147–8, 149, 170, 194, 204, 221, 238, 241, 242, 255, 270, 296, 311, 323, 343, 349 Robustelli 7, 51 Rodrigues 290 Rögnvaldsson 71 Roehrs 159 Rohlfs 7, 55, 69, 84, 85, 89, 104, 105, 138, 159, 170, 173, 175, 177, 288, 296, 298, 299, 323, 334, 350 Rohrbacher 147 Romanesco (see also central/southern Italian dialects; Laziale) 321–2 Romània (see also classification of the Romance languages) northern 314, 319–20, 332, 333, 340, 341, 349, 350 southern 314, 320, 332, 333, 340, 350, 351 Romanian 2–3, 7, 14, 15, 20, 28, 35, 51, 52, 53, 69–70, 74, 82, 84, 86, 87, 89, 98, 99, 107,

429

108, 111, 112, 113–18, 124, 126, 127, 128–9, 134–5, 140–1, 143, 145, 146, 147, 162, 163, 166, 170, 199, 205, 221, 241–2, 288, 290, 291, 292, 293, 297–8, 298, 314, 334, 346–7, 350–1 colloquial 108, 110–11 substandard 347 western 295 Romansh (see also Ræto-Romance) 7 Roncaglia 7 Ronjat 7, 55, 84, 88, 99 Rooryck 152 Rossellonès (see also Catalan) 13, 17 Rosen 4 Rosén 89, 90, 152, 190, 191, 213, 232 Rosenkranz 42, 215 Rosetti 7 Ross 35, 42, 59, 227, 264, 277 Rouergat (see also Occitan) 17 Roussou 170 Rovai 328, 330, 334 Rovignese (see also Istro-Romance) 334 Rowlett 4, 7, 50, 51, 56, 74, 144, 145, 159, 201 Rubio 39, 78, 193, 196, 205 Rule of Chrodegan, The 96 Rustius Barbarus 245 Russell 31 Saeed 292, 311 Saint Cyprian 243–4 Salentino (see also southern Italian dialects) 99, 170, 173, 178, 205, 298 Sallust 214, 218, 226, 227, 231, 243, 255 San Leucio del Sannio (see also Campanian; southern Italian dialects) 342–3 Saltarelli 51, 55, 104 Salvi 7, 8, 22, 27, 34, 35, 40, 42, 49, 54, 59, 63, 64, 65, 66, 70, 87, 130, 150, 151, 153, 154, 155, 156, 159, 160, 161, 185, 186, 187, 191, 192, 199, 223, 248, 249, 250, 251, 256, 257, 317, 333 Salvian of Marseille 244 Samu 171

430

Index

San Leonardo (see also Ladin; Ræto-Romance) 164–5, 167 Sandfeld 4, 7, 51, 110 Sanfeliciano (see also Laziale; central/ southern Italian dialects) 105 Sankoff 341 Sant’Andrea (see also Calabrian; southern Italian dialects) 345 Santorini 173 Sapir 10 Saralegui 7 Sardinian (see also Campidanese; Logudorese; Nuorese) 3, 7, 11, 55, 65, 69–70, 89, 104, 124, 128–9, 134–5, 145, 162, 164, 166, 171, 178, 198, 224, 293–4, 299, 314, 334 northern 345 Sasse 3, 21, 23, 251 Sassoferratese (see also central Italian dialects) 166, 338–9 Sauzet 173 Savoia 7, 14, 170, 177, 178, 296, 299, 323, 343, 344 Scarano 51 Schachter 76 Scherer 131 Schifano 140, 147, 149 Schlegel, von 10 Schlieben-Lange 7 Schmidt 196, 204 Schmitt 4, 288 Schneider 152 Schsler 74, 333 Schwegler 3, 10, 12, 17, 19, 199 Schweickard 4 Schwenter 74, 184 scrambling (see fronting) Seidl 334 Sekerina 195 Selig 90, 92 Senatus consultum de Bacchanalibus 59, 214, 223, 225, 228, 230 Seneca (the Younger) 226 Sheehan 204, 221, 238, 241, 242

Shih 53, 115 Shopen 79 Sicilian (see also southern Italian dialects) 14, 69–70, 124, 162, 163, 168, 170, 177, 334, 341, 345, 348 Sihler 241 Silva 74 Simpson 71 Sitaridou 14, 293 Skårup 65, 162 Slavonic languages 34, 162 Slobbe 135, 136 Smith, John Charles 4, 8, 132, 333, 334, 335, 348 Smith, Nathaniel 138 Solà, Jaume 70 Solà, Joan 7, 348 Sornicola 8, 65, 95, 96, 97, 126, 288, 333, 350 Sorrisi 126 southern Italian dialects 54, 55, 66, 82, 89, 100, 104–5, 105–6, 112, 134–5, 143–4, 160, 170–3, 177, 205, 288, 296, 298, 299, 314, 327, 334, 341, 344–5 extreme south 184–5 Souza Campos 119 Spanish 3, 7, 15, 16, 18, 19, 20, 28, 51, 52, 53, 54, 55, 56, 68, 69–70, 75, 82, 83, 84, 86, 87, 89, 97, 98, 99, 100, 103, 106–7, 108, 110–11, 120, 121, 122, 123, 125, 126, 129, 134, 135, 136, 138, 140–1, 142, 143, 144, 145, 146, 147, 148, 162, 165, 171, 172, 176, 177, 182–3, 198, 199, 224, 225, 278–8, 290, 312–13, 314, 321–2, 323, 334, 335, 340, 341, 344–6, 348 Andalusian 344–5 Canarian 184 colloquial 284 Colombian 184–5 Latin American 184, 344–5 Mexican 345 Palmero 16 Speas 47, 71, 311 Spence 334 Spencer 16, 138

Index Spevak 34, 35, 41, 45, 72, 150, 155, 190, 191, 206, 210, 212, 216, 219, 221, 223, 227, 232, 243, 259 split intransitivity (see alignment; unaccusative/unaccusativity) Sproat 53, 115 Squartini 15, 121, 184 Stanovaïa 334 Stark 88–9 Stati 7, 51, 69 Ştef ănescu 113, 116 Stegmann 39, 42 Sten 7 Stephens 32, 34, 35, 36, 39, 40, 41, 42, 46, 47, 59, 71, 78, 151, 152, 155, 157, 187, 193, 195–6, 205, 216, 219, 223, 236, 240, 242, 243, 247, 249, 250, 251, 254, 256, 259, 263, 264, 267, 270, 271, 272 Stewart 7 Stimm 7, 136, 288, 350 Stolova 341 Strasbourg Oaths, The 96, 113, 136, 186, 280 Stravinschi 110 subjacency (see also Left Branch Condition) 280 bounding nodes 280–1 subject 77 clitic 292–3, 314, 349–50 clitic vs tonic 67–8 expletive 34, 72, 74, 174 focalized 174 LOCATIVE 130–1 null 34, 71–74, 174, 252, 300–1, 311 embedded contexts (Corsican) 67–8 left-dislocated 69 V2 contexts 66–7 -object asymmetries 72, 75–7, 194–5 of predication 168 position 38, 62, 68–9, 335–6 preverbal absence 174 dedicated 80, 119, 292–3, 314, 349 focused pronominal 70 indefinite/negative quantified 70

431

postverbal 90, 335–6 unaccusative/middle structures 231–4 raising 64 topicalization 68–9, 174 UNDERGOER 68–9, 133–4, 231–4, 295, 313, 315–16, 318, 326, 328–9 subjunctive (see also complementation; complementizer) 286 complementizer 170, 173, 286, 298 erosion 298 inflected and non-inflected forms of Romanian fi 346–7 marked through consonantal doubling on Salentino verb vs indicative contrast 87, 298 subordinate clause (see also complementation) 65, 66, 287, 298–9 null subjects 66–8 V-final position 42, 185–7 Suetonius 231 Sugiaski 71 Suñer 288 superlative 14, 114 Surselvan (see also Ræto-Romance; Romansh) 123, 134–5, 225, 332 Swartz 79, 188 Syntactic Extraposition (see also topic(alization)) 168 synthesis / syntheticity (see analytic / analyticity) Szantyr 13, 39, 42, 45, 47, 135, 136, 211, 219, 220 Sznajder 72 Tacitus 226, 231, 233 Tagliavini 4 Talvera 229 Tarantino (see also Pugliese; southern Italian dialects) 334 Tarriño Ruiz 39, 41, 130, 131, 132, 207, 210 Tasmowski 14, 34, 113, 116, 135, 139 Taylor 3, 21, 241 Tekavčić 3, 7, 10, 11, 334 tense marking with aspect and mood 13–14, 33–4

432

Index

tense marking (cont.) vs aspect 34 T(ense) 34 marking 148 Terence 223, 228, 230, 233 Terentianus, Claudius 223, 226, 229, 231, 245, 246 Tertullian 136, 138, 153 Tesnière 310 Teyssier 7, 74, 112 that-trace effect 280 thetic sentence (see word order: sentential: presentative) Thibault 341 Thiella 51, 318 Thielmann 131, 135, 139 Thomas, Earl 7, 74, 112, 290 Thomas, François 12, 39, 41, 45, 59, 62, 63, 130, 151, 157, 159, 190, 191, 193, 195, 196, 206, 207, 209, 213, 214, 219, 221, 241 Tobler-Mussafia Law (see pronoun: clitic) Tomida femina 96 topic(alization) (see also fronting: focus-; topic-) 36, 59, 65, 76, 77, 79, 94–5, 150, 150–1, 154, 192, 205, 268 of subject 68–9, 90 Topic field 159–62, 166–8, 267–8 Frame subfield 166–7 hanging topic 161, 166–7 scene-setting adverbs 166–7 multiple topics 166 Theme subfield 166–8 aboutness topic 168 clitic left-dislocation (ClLD) 167–8 familiar topic 168 left-dislocated topic 69, 161 subject of prediation 168 Torrego 219, 220, 221, 251, 288, 350 Tortora 7, 140 Tossenc (see also Catalan) 100, 101, 102 Tovar 59, 196 TP (Tense Phrase) 34 Trager 96 Traugott 28

Trebisaccese (see also Calabrian; southern Italian dialects) 341 Trentino (see also northern Italian dialects) 325–6, 334 Triestino (see also northern Italian dialects) 162, 163 Trumper 288, 350 Turinese (see also Piedmontese; northern Italian dialects) 162, 290 Tuscan (see also Italian; Italo-Romance) 18, 61, 65, 66, 67, 74, 86, 88, 97, 135, 171, 260–1, 279–80, 290, 292–3, 299, 321 Tuttle 130, 296, 317, 323 Umbrian dialect (see also central Italian dialects) 137 Umbrian language (see also Italic) 42, 206, 215 unaccusative/unaccusativity (see also alignment: active-stative; deponent; passive; voice: middle) 133–4, 315–16, 328–9 auxiliary selection 130–4, 295, 314, 316, 317–26, 322, 324, 341–2, 344–5 subject positions 231–4, 335–6, 349 verb 76 Uriţescu 295 Ursini 7, 341 Väänänen 3, 21, 39, 41, 43, 47, 48, 62, 64, 89, 159, 211, 225, 229, 241, 277, 328, 330, 331 Valéian 99 Valencian (see also Catalan) 99, 101, 103, 109 Tarbener 101, 103, 104 Vall de Gallinera 101, 104 Valesio 14, 33, 135 Vallader (see also Ræto-Romance; Romansh) 322, 332 Vallduví 128 Valois 53 Valsuganotto (see also northern Italian dialects) 326 van Reenen 333 van Valin 284, 285, 286 Vance 65, 173

Index Vanelli 65, 162, 199, 292 Varro 42, 214, 226, 230, 231, 255 Varvaro 1, 8, 96, 97, 104, 136, 137 Vecchio 170, 173 Vegliote (see also Dalmatian; Istro-Romance) 52, 199, 341 Venetian (see also northern Italian dialects) 14, 123, 323, 334, 341–2 Veneto (see also northern Italian dialects) 334 Vennemann 3, 21, 204, 332 Veny 7, 17, 100, 104, 199, 296, 320, 323 V(erb)-movement 127 140–50, 182, 200, 252, 285 absence in Latin 182, 219 ESSE-fronting 156, 191, 256–9, 259, 260 finite auxiliaries 143–4 finite lexical verbs 140–2, 144 Germanic vs Romance contrast 148 infinitive 145–6 participles active 144–5, 149 passive 149–50 rich Agr(eement) 147 summary 146 T(ense)-marking 148 temporal, aspectual and modal interpretation 148–50 to C(omp) (see also Verb Second) 150–6 pragmatically-motivated 151–2 subordinate clauses 153 syntactically-motivated 152–6 Verb Second (V2) 23, 65–8, 78, 80, 81, 162, 164–5, 169, 173, 201–2, 314, 349–50 emergence 257–8 Latin 150–6, 158, 159 null subjects 66–7 Vergil 193 Victor Vitensis 226 Vie de Saint Alexis 85 Vielliard 330 Vigolo 299 Vikner 147, 173 Villa-García 171 Villalaghese 105 Villangómez i Llobet 7, 100, 101, 327

433

Villar 288, 332, 335, 350 Vincent 4, 8, 10, 12, 14, 16, 20, 29, 31, 32, 33, 34, 35, 39, 40, 41, 42, 43, 45, 46, 47, 50, 51, 53, 59, 60, 61, 63, 65, 69, 71, 72, 75, 78, 79, 81, 89, 90, 95, 107, 119, 130, 131, 136, 150, 153–4, 157, 161, 171, 179, 180, 191, 193, 195, 196, 198, 199, 202, 203, 206–7, 210, 211, 223, 225, 227, 229, 230, 236, 240, 256, 268, 284, 286, 287, 288, 290, 292, 293, 295, 298, 310, 317, 349, 350 Vindolanda letters 245–6 Vineis 1, 59, 61, 229 Vitruvius 228 vocative 99 voice (see also passive) middle 27, 133–4, 315–16 participle agreement 286, 307, 316, 326–7 possessive copula 317–19 subject case-marking 328–33 positions 231–4 vP (lower) left periphery 259–62, 264–5, 271, 273, 278 phase/colon status 260–2 VP (Verb Phrase) 33–5, 47, 47–8, 57–8, 63, 69–70, 72, 75–7, 119, 286 ellipsis 194–5 fronting 75 subject-object asymmetries 72, 75–7, 194–5 Vulgata/Vulgate 214, 229, 230, 233 Wackernagel elements (see also colon/cola; phase) 63–4, 191–2, 260 adpositions 78 clause connectors 63, 154, 191 colon 259–62 ESSE 156, 191, 256–9, 260 pronouns 63, 191–2, 260 Wagner 7 Wallace 221 Walloon (see also Gallo-Romance) 56 Wanner 161, 171 Warlpiri 64, 71, 72, 79, 80, 188 Wartburg, von 7, 25, 196, 197, 314

434

Index

Watkins 39, 42, 59, 202, 203, 227 Weil 36, 59, 196, 197 Welsh 308 Westerbergh 328 Wheeler 7, 13, 54, 55, 83, 99, 100, 101, 173 Williamson 71 Woledge 82 word order (see also configurational(ity); head parameter; left periphery) adjective (AN/NA order) 25, 210–13, 270 Latin 37, 41, 42–3, 50–1, 270 Romance 49, 50–4, 198–9 early Romance 56 interpretation 52–3 liaison 53–4 restricted prenominal position 55–7, 198 adjunct 57–8 adverb 25, cartographic representation 141–50 Latin 38, 42–3, 222–3 degree and focalizing 189–91 Romance 127 auxiliary + infinitive 138–9, 192, 234–5 auxiliary + participle 130–4, 136, 223–4, 255–8, 307, 317–26 determiner 32–3 finite complement clauses 60–1 free 31, 35, 42, 43, 59, 61, 68, 71, 187, 188, 192, 202, 224, 281, 310–11 genitive (GN/NG order) 25, 37, 39, 40–1, 80, 207, 213–19, 224 FOFC 252–5 pragmatic principles 216–17 Latin fixed orders 187–86, 189–94 Latin nominal group 218–19 pragmatically-motivated 35–6, 45, 48, 59, 61, 68, 71, 150, 182–3, 188, 222, 258–81, 281–2 sentential 2–3, 20–1, 22–23, 28, 30–1, 35–6, 150, 285 active-stative 69, 233–4, 335–6 archaic Latin 59, 77–8

classical Latin 59–64, 150, 185–283 pragmatically neutral order 187–8 early Romance 64–8 emergence of definite article 90 FOFC 255–8 late Latin 64–8 modern Romance 68–71 VSO 69–70 presentative 151, 155–6 root vs embedded clauses 185–7, 230–1, 232, 238, 248, 257–8 unaccuastive/middle structures 231–4 VSO in colloquial French 311 Universal Base Hypothesis 236 V-final 39, 41–2, 59–60, 64, 69–70, 139, 203, 225–31, 233, 253 diamesic/diaphasic variation 231 embedded clauses 185–7, 230 prestige order 229–31, 233 text type 226 V-initial (see also Verb Second) 150–3 V-medial 39–40, 41–2, 60, 64, 230–1 diamesic/diaphasic variation 231 Wright 1 Wüest 10, 13, 23, 24, 29, 34, 332, 333 Wurff, van der 72 Yates 7, 99, 100, 101 Zagona 4, 7, 69, 70, 140, 159, 288, 339 Zamboni 2, 3, 10, 21, 42, 48, 89, 90, 130, 133, 159, 197, 288, 297, 312, 314, 316, 317, 318, 328, 331, 332, 333, 334, 335, 340, 350 Zamora Vicente 7, 184 Zampini 159 Zanini 70 Zanuttini 17, 152, 159, 164, 199, 200, 201 Zribi-Hertz 74 Zubizarreta 16, 69, 297