Periphrasis and Inflexion in Diachrony: A View from Romance (Oxford Studies in Diachronic and Historical Linguistics) 9780198870807, 0198870809

Citation preview

Periphrasis and Inflexion in Diachrony

OX F OR D ST U DI E S I N DIAC H RON IC A N D H I STOR IC A L L I NG U I ST IC S General Editors Adam Ledgeway and Ian Roberts, University of Cambridge Advisory editors Cynthia L. Allen, Australian National University; Ricardo Bermúdez-Otero, The University of Manchester; Theresa Biberauer, University of Cambridge; Charlotte Galves, University of Campinas; Geoff Horrocks, University of Cambridge; Paul Kiparsky, Stanford University; David Lightfoot, Georgetown University; Giuseppe Longobardi, University of York; George Walkden, University of Konstanz; David Willis, University of Oxford recently published in the series 41 Variation and Change in Gallo-Romance Grammar Edited by Sam Wolfe and Martin Maiden 42 Phonetic Causes of Sound Change The Palatalization and Assibilation of Obstruents Daniel Recasens 43 Syntactic Features and the Limits of Syntactic Change Edited by Jóhannes Gı´sli Jónsson and Thórhallur Eythórsson 44 Romance Object Clitics Microvariation and Linguistic Change Diego Pescarini 45 The Diachrony of Differential Object Marking in Romanian Virginia Hill and Alexandru Mardale 46 Noun-Based Constructions in the History of Portuguese and Spanish Patrı´cia Amaral and Manuel Delicado Cantero 47 Syntactic Change in French Sam Wolfe For a complete list of titles published and in preparation for the series, see pp. 486–90

Periphrasis and Inflexion in Diachrony A View from Romance Edited by

ADAM LED GEWAY JOHN CHARLES SMITH NIGEL VINCENT

1

3 Great Clarendon Street, Oxford, OX2 6DP, United Kingdom 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. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © editorial matter and organization Adam Ledgeway, John Charles Smith, and Nigel Vincent 2022 © the chapters their several authors 2022 The moral rights of the authors have been asserted Impression: 1 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, by licence 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 work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2021939171 ISBN 978-0-19-887080-7 DOI: 10.1093/oso/9780198870807.001.0001 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

In honour of Martin Maiden in the year of his sixty-fifth birthday

Contents Series preface List of figures and maps List of tables Abbreviations and typographic conventions List of contributors

xiv xv xvi xx xxvi

Introduction: Maiden, morphology, and more Adam Ledgeway, John Charles Smith, and Nigel Vincent

1

PART I. THE STAT U S OF PE R I PH R ASI S A N D I N F L E X ION 1. Periphrasis and inflexion: Lessons from Romance Adam Ledgeway and Nigel Vincent 1.1 Introduction 1.2 Theoretical issues 1.2.1 Analysis vs periphrasis 1.2.1.1 Lexical theories 1.2.1.2 Inferential theories 1.2.2 Formal issues in modelling periphrases 1.2.2.1 Intersectivity 1.2.2.2 Non-compositionality 1.2.2.3 Distributed exponence 1.2.2.4 Compound periphrases 1.2.2.5 Gradience 1.2.3 Paradigmaticity

1.3 Diachrony 1.3.1 1.3.2 1.3.3 1.3.4

Grammaticalization and directionality Compositionality and univerbation Attestation and reconstruction Language contact

1.4 Conclusions

2. The boundaries of inflexion and periphrasis John Charles Smith 2.1 Introduction 2.2 The boundaries of inflexion 2.2.1 Preliminary issues 2.2.2 Case study 1: gender

11 11 12 12 15 18 21 21 27 31 32 34 37 42 42 51 54 56 59

61 61 62 62 66

viii

contents 2.2.3 Case study 2: number 2.2.4 Case study 3: vocatives

2.3 The dividing line between inflexion and periphrasis 2.4 The boundaries of periphrasis 2.4.1 2.4.2 2.4.3 2.4.4 2.4.5 2.4.6 2.4.7 2.4.8 2.4.9 2.4.10

Introduction Intersectivity Range of forms Distributed exponence Contiguity Non-compositionality Fuzziness Structural and sociolinguistic variables Grammaticalization Intersectivity as a variable

2.5 Conclusion

67 68 71 73 73 75 79 80 81 83 83 84 87 88 90

PA RT I I . PE R I PHR ASI S 3. Layering and divergence in Romance periphrases Nigel Vincent and Max W. Wheeler 3.1 Introduction 3.2 Forms of habere ‘have’ 3.3 Forms of the go verb 3.4 Reinforcement of the lexical verb 3.5 Loss of inflexion 3.6 Conclusion 4. The go-future and go-past periphrases in Gallo-Romance: A comparative investigation Sandra Paoli and Sam Wolfe 4.1 Introduction 4.2 The go-future in French 4.2.1 From Latin to old French 4.2.2 Middle to modern French 4.2.3 Contemporary developments

4.3 Old Occitan 4.4 Discussion 4.5 Concluding remarks

5. The tornare-periphrasis in Italo-Romance: Grammaticalization ‘again’! Mair Parry 5.1 Introduction 5.1.1 Grammaticalization

5.2 Italo-Romance verbal periphrases with tornare

93 93 96 108 118 120 121

123 123 126 126 129 131 132 137 144

145 145 145 146

contents 5.2.1 Italo-Romance data (nineteenth–twentieth centuries) 5.2.1.1 Infinitival periphrases 5.2.1.2 Paratactic or serial periphrases

5.3 The grammaticalization of tornare 5.3.1 Pragmatic and semantic extension, leading to desemanticization 5.3.2 Decategorialization 5.3.3 Cliticization and erosion?

5.4 Conclusion

6. Periphrases and irregular paradigms in Italo-Romance Silvio Cruschina 6.1 Introduction 6.2 Motion and progressive double inflexion constructions 6.3 The N-pattern as a periphrastic morphome 6.4 The Sicilian modal periphrasis 6.5 Conclusions

ix 148 148 154 155 160 162 166 167

169 169 171 177 183 187

PA RT I I I . AU X I L IAT ION 7. Auxiliary selection in Italo-Romance and inflexional classes Xavier Bach and Pavel Štichauer 7.1 Introduction 7.2 Auxiliary selection and inflexional classes 7.2.1 Content paradigm, form paradigm, and realizations 7.2.2 Segregated inflexional classes and compound tenses

7.3 Mixed perfective auxiliation systems in Italo-Romance 7.4 Mixed systems in reflexives and heteroclisis 7.4.1 7.4.2 7.4.3 7.4.4

Reflexives with the BBH-BBH pattern Heteroclisis with default marking Heteroclisis by overabundance Problematic cases of heteroclisis

7.5 Conclusions 7.5.1 Theoretical problems 7.5.2 Diachronic considerations

8. The morphological nature of person-driven auxiliation: Evidence from shape conditions Michele Loporcaro 8.1 Introduction 8.2 Perfective auxiliation: syntax vs morphology 8.3 Perfective auxiliation depending on the phonology in three dialects of Apulia 8.4 A shape condition on the distribution of ‘has’/‘is’

193 193 194 195 197 199 202 203 204 206 208 210 210 211

213 213 214 219 229

x

contents

8.5 From complementary distribution via overabundance to heteroclisis and suppletion 8.6 The limits of morphology 8.7 Conclusion

230 233 236

PA RT I V. A NA LYSI S VS SY N T H E SI S 9. The loss of analyticity in the history of Romanian verbal morphology Adina Dragomirescu, Alexandru Nicolae, and Rodica Zafiu 9.1 Introduction 9.2 Periphrastic forms in old Romanian 9.2.1 Periphrases with the lexical verb in the infinitive 9.2.2 Periphrases with the lexical verb in the past participle 9.2.3 Periphrases with the lexical verb in the gerund/present participle

9.3 Analytic forms in modern Romanian 9.4 Analytic forms in (Daco-)Romanian dialectal varieties 9.4.1 Periphrases with the lexical verb in the infinitive 9.4.2 Periphrases with the lexical verb in the past participle forms 9.4.3 Periphrases with the lexical verb in the gerund/present participle

9.5 An account of the partial loss of analyticity in Romanian verbal morphology 9.5.1 Demise of a subset of periphrastic forms 9.5.2 Diachronically stable analytic formations 9.5.3 The relevance of analytic formations which were preserved dialectally

9.6 Conclusions: the loss of analyticity in a wider perspective

10. The relation between inflexional and analytic marking of obliques in Romanian Gabriela Pană Dindelegan and Oana Uță Bărbulescu 10.1 Introduction: marking of oblique functions in modern standard Romanian 10.2 Old Romanian 10.2.1 Distribution of inflexional and analytic markers: an overview 10.2.2 Analytic markers: distribution and competition 10.2.3 Mixed structures 10.2.4 Quantifiers 10.2.4.1 Homogeneous constructions vs heterogeneous constructions 10.2.4.2 Mixed constructions

241 241 243 243 246 252 255 258 258 259 261 262 262 268 269 270

272

272 276 276 279 286 287 290 293

contents

10.3 Modern Romanian 10.4 Conclusion

11. A diachronic perspective on polymorphism, overabundance, and polyfunctionalism Rosanna Sornicola 11.1 Polymorphism and polyfunctionalism 11.2 Polymorphism and overabundance: on the history of the terms 11.2.1 Overabundance

11.3 On representing the relations between form and function 11.3.1 11.3.2 11.3.3 11.3.4

The relation between form and function Polymorphism and free variation Suppletion and syncretism Polymorphism and polyfunctionalism: paradigmatic and syntagmatic relations 11.3.5 Polymorphism and polyfunctionalism in diachrony

11.4 Polymorphism and polyfunctionalism in late-medieval documents from Italy 11.5 Inflexional systems, polyfunctionalism, and polymorphism 11.6 Influence of phonetic factors 11.7 Atrophization 11.8 Overextension of the stem 11.9 Overextension of inflexion 11.10 Preservation of morphological properties of class III nouns 11.11 Differentiation and hypodifferentiation of paradigm cells 11.12 Interchangeability of forms and merging of paradigm cells 11.13 Syntactic factors 11.14 Hypercharacterization of grammatical relations 11.15 Polyfunctionalism of number and syntactic function 11.16 Towards a diachronic model of polymorphism and polyfunctionalism

xi 294 300

305 305 308 310 311 311 312 314 316 317 318 319 322 322 323 323 324 325 328 329 330 330 331

PART V. I N F L E X ION A N D I T S I N T E R FAC E S 12. Thematic and lexico-aspectual constraints on V–S agreement: Evidence from northern Italo-Romance Delia Bentley and Michela Cennamo 12.1 Introduction 12.2 Theoretical preliminaries: lexico-aspectual structure and thematic roles 12.3 The case of Emilian-Romagnol 12.3.1 Our survey 12.3.2 Cross-dialectal variation 12.3.2.1 Agreement variation with [–state] verbs 12.3.2.2 Agreement variation with [+state] verbs 12.3.2.3 Qualitative analysis: summary

335 335 338 340 340 341 342 349 355

xii

contents 12.3.3 Quantitative analysis

12.4 Theoretical considerations and conclusions

13. Conditioned epenthesis in Romance Mark Aronoff and Lori Repetti 13.1 Introduction 13.1.1 Intrusive vowels 13.1.2 Phonological epenthesis

13.2 Case studies in Romance languages 13.2.1 13.2.2 13.2.3 13.2.4 13.2.5

Brazilian Portuguese San Marino Paduan Italian Catalan varieties

13.3 Accounts of the phenomena 13.3.1 Allomorphic solutions 13.3.2 Epenthesis 13.3.3 Interim conclusion: morphologically conditioned epenthetic segment quality

13.4 Other types of insertion 13.4.1 Catalan stem extenders 13.4.2 Italian 13.4.3 Spanish antesuffixes

13.5 Findings

14. Koinéization and language contact: The social causes of morphological change in and with Portuguese Tom Finbow and Paul O’Neill 14.1 Introduction 14.2 Koinéization in Romance 14.3 Koinéization in the history of Portuguese 14.3.1 Development of second-person plural pronominal and verbal forms in European Portuguese 14.3.2 Second-person singular non-deferential address in unplanned discourse of two Brazilian Portuguese varieties

14.4 Brazilian Portuguese as morphosyntactic model for indigenous languages 14.4.1 Old Tupi and Portuguese 14.4.1.1 Demonstratives and pronouns 14.4.1.2 Portuguese influence on Yẽgatú verb typology 14.4.1.3 Grammaticalization of rãm 14.4.1.4 Grammaticalization of su ‘go’

14.5 Conclusion

References and bibliographical abbreviations Index

356 360

362 362 363 364 365 365 366 368 369 370 371 371 372 373 376 376 377 378 379

381 381 381 384 384 390 394 395 397 398 404 406 411

413 467

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 focusing 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

List of figures and maps 1.1. Grammaticalization of go forms

45

1.2. Grammaticalization of come forms

47

1.3. Grammaticalization of return forms

50

12.1. Incidence of three effects according to [±pro] predictor

357

12.2. Incidence of three effects according to [±def] predictor

357

12.3. Incidence of three effects according to [±state] predictor

358

12.4. Incidence of three effects according to lexical aspect

358

13.1. Factors conditioning the insertion of semantically vacuous material

379

Map 8.1. The towns in the province of Bari (Apulia) whose dialects are mentioned in the chapter

237

List of tables 1.1. Inflexional paradigm of Romanian prieten ‘friend’

19

1.2. Italian present and past subjunctive paradigms

22

1.3. Romanian (3sg) imperfective and perfective paradigms serv- ‘serve’

26

1.4. Aspectual pseudo-coordinate periphrases with stare ‘stand’ in Apulia

36

1.5. Defective pseudo-coordinate periphrases with stare ‘stand’ in Pugliese (Andriani 2017:ch.5)

38

1.6. Dimensions of language structure

51

1.7. Portuguese future and conditional forms of run

52

3.1. Proto-Romance paradigm variants of habere

96

3.2. Modern Romance present indicative paradigms derived from habere*

101

3.3. Spanish hemos / (hemos + habemos) (%)

104

3.4. Growth of Catalan hem, heu, with respect to havem, haveu, over the last two centuries

106

3.5. Suppletive stems in the go verb across Romance

109

3.6. Paradigm split in modern Catalan go

110

3.7. Catalan synthetic perfective, 1st conjugation, trobar ‘find’, old (A) and new (B) 113 3.8. Forms of the present of the go verb in Guardiol and Aragonese

117

4.1. ‘Non-past’ and ‘past’ interpretation with first and second persons

141

6.1. 3sg indicative forms of Latin capere ‘take’

170

6.2. The DIC forms for go (V1) and take (V2) in Sicilian

173

6.3. The DIC forms for go (V1) + V2 in the dialects of Putignano and Lecce

174

6.4. stare ‘stand’ paradigms

175

6.5. The DIC paradigm of (j)ìri ‘go’ + pigliari ‘take’ in Mussomeli, Sicily

177

6.6. Full and defective paradigms in some Pugliese dialects

178

6.7. Progressive with maˈnʤε ‘eat’ in the dialect of Conversano

178

6.8. Present motion DIC in Salentino varieties

180

6.9. Past motion DIC in Salentino varieties

180

6.10. Paradigm of passari ‘pass, come by’ + pigliari ‘take’ in Mussomeli, Sicily

180

6.11. Present forms of aviri ‘have’ in Sicilian (Mussomeli)

184

6.12. Preterite forms of aviri ‘have’ in Sicilian

185

xvi

list of tables

6.13. Preterite paradigms in Sicilian and Italian

186

6.14. Defective AICo preterite paradigm + go

186

7.1. An abstract paradigm of a noun inflecting for number and case

195

7.2. Two idealized paradigms for two classes of lexemes

196

7.3. Two idealized paradigms for two classes of lexemes with two different distributional patterns of syncretism (grey shading)

196

7.4. The content paradigm for the present perfect

198

7.5. Two form paradigms for the classes of lexemes that select the auxiliary h(ave) and b(e)

198

7.6. Two realized paradigms for the verbs mangiare ‘eat’ and arrivare ‘arrive’

199

7.7. The present perfect of the verbs mangià ‘eat’ and scì ‘go’ in the variety of Bari (Andriani 2017:158–160)

200

7.8. The present perfect of the verbs wash, wash oneself, and come in the variety of Altomonte (Calabria, southern Italy)

202

7.9. Heteroclisis in reflexives following the pattern BBH–BBH in the variety of Castelletto Merli (data from Manzini and Savoia 2005:II:650f.)

203

7.10. Heteroclisis in reflexives/unergatives in the variety of Bitetto (data from Manzini and Savoia 2005:II:725f.)

205

7.11. Heteroclisis in reflexives in the variety of Popoli (data from Manzini and Savoia 2005:II:688f.)

207

7.12. Heteroclisis in unergatives with respect to unaccusatives/reflexives and transitives in the variety of Poggio Imperiale (Terranova) (data from Manzini and Savoia 2005:II:720f.) 209 8.1. Auxiliary selection in Italian and Spanish

216

8.2. Unattested auxiliary systems

217

8.3. Implicational scale of auxiliary systems

217

8.4. Implicational scale of auxiliary systems

218

8.5. Ruvo and Bitetto

220

8.6. Ruvese (Manzini and Savoia 2005:II:724f.)

221

8.7. Gravinese (Manzini and Savoia 2005:III:29f.)

221

8.8. Bitettese (Manzini and Savoia 2005:II:725f.)

222

8.9. Bitettese (consultant 1)

222

8.10. Bitettese (consultant 2)

222

8.11. The verb go in some Romanian dialects of Maramures¸ (Maiden 2004b:242)

230

8.12. Compiere/compire ‘fulfil’

232

8.13. Compiere (present indicative): from overabundance to heteroclisis

232

10.1. Definite declension of modern Romanian feminine and masculine nouns

273

list of tables

xvii

10.2. Distribution of inflexional and prepositional markers of oblique functions in old Romanian 279 10.3. Distribution of inflexional and prepositional markers of oblique functions in modern Romanian 294 11.1. Polymorphism of Latin domus ‘house’

306

11.2. Latin noun declensions: singular

315

11.3. Latin noun declensions: plural

316

11.4. Inflexional paradigms of classes I and II

321

12.1. State-based vs non-state-based lexico-aspectual classes

339

12.2. Significance of [±pro] in one-level logistic regression with [±pro] and [±state] 359 12.3. Significance of [±state] in one-level logistic regression with [±pro] and [±state]

359

13.1. San Marino verbs

367

13.2. Types of epenthesis

379

14.1. Different second-person singular possessive pronouns in the Toledo phase and a historical explanation of their forms (Tuten 2003:204–213)

383

14.2. Comparison of Catalan, Portuguese, Galician, and Spanish possessive pronouns

383

14.3. Comparison of second-person plural address (‘you are’) in standard Spanish, Galician, and Portuguese, and in the speech typical of Seville 385 14.4. Second-person plural forms of address for European Portuguese with respective verbal agreement and corresponding pronouns

386

14.5. Second-person plural forms of address for Galician with the respective verbal agreement and corresponding pronouns

387

14.6. New você paradigm for second-person singular non-deferential address in the spoken Brazilian Portuguese of the city of São Paulo

390

14.7. New tu-paradigm for second-person singular non-deferential address in spoken Brazilian Portuguese of the city of Porto Alegre

391

14.8. Standard European Portuguese second-person singular non-deferential address

391

14.9. Standard European Portuguese second-person singular deferential address

391

14.10. Percentages of the use of subject pronouns and verbal agreement in a number of cities in Brazil 393 14.11. Old Tupi demonstratives (Rodrigues 2010:27, adapted)

397

14.12. Old Tupi sentence with active verbs, first-, second-, or third-person agent and third-person patient or reciprocal or reflexive complements

399

14.13. Old Tupi stative verb

400

14.14. Old Tupi transitive verb with third-person Agent and first- or second-person Patient 400

xviii

list of tables

14.15. Old Tupi transitive verb with first-person Agent and second-person Patient

401

14.16. Old Tupi transitive verb with second-person Agent and first-person Patient

401

14.17. Yẽgatú transitive verb with first-, second-, or third-person Agent and Patient 402 14.18. Yẽgatú stative verbs

404

14.19. Uses of su in nineteenth-century Yẽgatú

410

Abbreviations and typographic conventions * ** # ?

Ø > < ≈ ; ∖ ˈ = ː . 1/2/3 A abl Abr. acc ACC ACH act ACT adj adn adv ag AGk. agr Agr(P) AICo AIS

Alt. arg

unattested or reconstructed form or usage ungrammatical form or usage contextually inappropriate form or usage substandard / non-standard form or usage zero (null), covert form becomes, yields; precedes comes from, derives from freely varying with ambiguous/alternative meaning (in glosses) initial pluriform relational mutation (i.e., R1, R2, R3) (primary) word stress cliticized to long / lengthened syllable boundary first / second / third person subject of a transitive clause ablative Abruzzese (dialect group of Abruzzo, upper south-eastern Italy) accusative accomplishments achievements active (voice) form activities adjective adnominal case form adverbal case form agentive Ancient Greek agreement (morphology) agreement (phrase) aviri a ‘have to’ + infinitive’ construction Atlante Italo-Svizzero (Atlante linguistico e etnografico dell’Italia e della Svizzera meridionale / Sprach- und Sachatlas Italiens und der Su¨dschweiz, 1928–1940) Altamurano (central Pugliese dialect of Altamura, upper south-eastern Italy) argument

xx

abbreviations and typographic conventions

Aro.

Aromanian (Daco-Romance dialects spoken in Greece, Albania, Bulgaria, Serbia, and the Republic of Macedonia) auxiliary switch aspect auxiliary perfective auxiliary be Barese (central-eastern Pugliese dialect of Bari, upper south-eastern Italy) Bitettese (central Pugliese dialect of Bitetto, upper south-eastern Italy) Bonorvese (dialect of Bonorva, north-western Sardinia) Brazilian Portuguese (i) central (ii) consonant circa Catalan causative clitic climbing chapter Cicero

AS asp Aux B Bar. Bit. Bon. BrPt. C c. Cat. caus CC ch. Cic. Att. CICA cl Cld. CoLFIS com comp compl cond cop CORDE Cos. Crg. Crs. Cst. CTILC dat def dem DIC dim dir. trans. dist DM DOM

Epistulae ad Atticum Corpus Informatitzat del Català Antic clitic Calderarese (central Emilian dialect of Calderara di Reno, north-eastern Italy) Corpus e Lessico di Frequenza dell’Italiano Scritto comitative complementizer (position) completive conditional copular Corpus Diacrónico del Español Cosentino (northern Calabrian dialect of city of Cosenza, extreme southwest of Italy) Correggese (northern Emilian dialect of Correggio, north-eastern Italy) Corsican Castilian Corpus Textual Informatitzat de la Llengua Catalana dative definite(ness) demonstrative Doubly Inflected Construction diminutive direct transitive clause distal Distributed Morphology differential object marking (or marker)

abbreviations and typographic conventions D(P) ds E Eng. EuPt. excl f f. Fer. Flo. foc Fr. Fuèc fut GdB GdR gen ger Glc. Grv. Gsc. H hab imnt imp incl ind indef indir. trans. indir. unerg. inf intr ivsbl ipfv IPP irr Isc. It. JG K(P) L2 Lat.

xxi

determiner (phrase) different subject (i) east (ii) event time English European Portuguese exclusive feminine and following page Ferrarese (northern Emilian dialect of the city of Ferrara, north-eastern Italy) Florentine focus French morphomic stem for Romance synthetic future and conditional future Guilhem de la Barra Girart de Roussillon genitive gerund Galician (Ibero-Romance language of north-western Spain) Gravinese (central Pugliese dialect of Gravina, upper south-eastern Italy) Gascon perfective auxiliary have habitual imminent(ial) imperative inclusive indicative indefinite indirect reflexive transitive clause indirect reflexive unergative clause infinitive intransitive invisible imperfective (aspect) Istoria Petri et Pauli irrealis Ischitano (Campanian dialect spoken on island of Ischia in the Bay of Naples, upper south-west of Italy) Italian Lo Jutgamen General Case (phrase) second language Latin

xxii Lec. lex lit. Liz. Lmb. loc Log. m MDP1 MDP2 Mid Mil. Mod Mod. Mus. n N n. Nap. Nct. neg nlzr nom N(P) NumP O obj obl Occ. ocl OED OVI p p. part pass Pcn. PF PFM pfv

abbreviations and typographic conventions Leccese (southern Salentino dialect of the city of Lecce, extreme southeast of Italy) lexical literally Lizzanese (south-eastern Emilian dialect of Lizzano in Belvedere) Lombard (dialect group of Lombardy, central northern Italy) locative Logudorese (dialect group of Logudoro, north-western Sardinia) masculine La Passion provençale du manuscrit de Didot Mystère rouergat de la Passion Middle Milanese modern Modenese (northern Emilian dialect of the city of Modena, north-eastern Italy) Mussomelese (central Sicilian dialect of Mussomeli, extreme south of Italy) neuter north(ern) (foot)note Neapolitan Nocetano (north-western Emilian dialect of Noceto, north-eastern Italy) negator nominalizer nominative noun (phrase) number phrase (i) object (ii) old objective case (form) oblique case (form) Occitan object clitic Oxford English Dictionary Opera del Vocabolario Italiano patient page partitive (article) passive Picernese (Basilicatese Gallo-Italic dialect of Picerno, upper southern Italy) Phonological Form Paradigm Function Morphology perfective (aspect)

abbreviations and typographic conventions Pie. pl plpf Pnt. poss pot P(P) pred pro prog proj prox prs Prs. prt pst Pt. ptcp purp PYTA q Q R R1/2/3 rcp RdA refl Reg. rel rep res.p.nmlz retr. Ro. Ruv. S

SA Sal. sbj

xxiii

Piedmontese (dialect group of Piedmont, north-western Italy) plural pluperfect Pontolliese (north-western Emilian dialect of Ponte dell’Olio, northeastern Italy) possessive (form) potential preposition(al phrase) predicate pronominal progressive projected future proximal present tense Persicetano (central Emilian dialect of S. Giovanni in Persiceto, northeastern Italy) preterite past Portuguese participle purposive perfecto/pretérito y tiempos afines (= Romance continuants of Latin perfective forms) question / interrogative form quantifier reference time relational prefix 1/2/3 reciprocal form Le Roman d’Arles reflexive Reggiano (central-northern Emilian dialect of Reggio Emilia, northeastern Italy) relative / relativizer repetition resultative patient nominalizer retroherent clause Romanian Ruvese (central Pugliese dialect of Ruvo, upper south-eastern Italy) (i) postverbal argument of intransitive clause (ii) south(ern) (iii) the moment of speech intransitive (Actor/Agent) subject of an unergative clause Salentino (dialect group of Salento, southern Apulia, extreme southeast of Italy) subject (form)

xxiv sbjv scl sg Slv. SO Sp. Srd. ss ST stat sub SV TAM T(P) trans. trsl TT tv UG unacc. unerg. UT V V1 V2 voc v(P) V(P) V(-)S

abbreviations and typographic conventions subjunctive subject clitic singular Slavonic intransitive (Theme/Undergoer) subject of an unaccusative clause Spanish Sardinian same subject (i) situation time (ii) states stative subordinator subject–verb order tense, aspect, and mood tense (phrase) transitive clause translative case topic time thematic vowel Universal Grammar unaccusative clause unergative clause utterance time (i) verb (ii) vowel initial verb in paratactic structure second verb in paratactic structure vocative light verb (phrase) verb (phrase) verb–subject order

In addition to the conventional use to indicate Latin etyma, in this volume we use small capitals for lexemes (e.g., go, aller) and italics for individual word forms (e.g., go, went, allons, va). We also use small capitals to refer to the cross-linguistic classes of come, go, etc. verbs and in glossing the grammaticalized reflexes of such items.

List of contributors Mark Aronoff, Stony Brook University Xavier Bach, University of Oxford Delia Bentley, The University of Manchester Michela Cennamo, University of Naples Federico II Silvio Cruschina, University of Helsinki Adina Dragomirescu, ‘Iorgu Iordan—Alexandru Rosetti’ Institute of Linguistics of the Romanian Academy and University of Bucharest Tom Finbow, University of São Paulo Adam Ledgeway, University of Cambridge Michele Loporcaro, University of Zurich Alexandru Nicolae, ‘Iorgu Iordan—Alexandru Rosetti’ Institute of Linguistics of the Romanian Academy and University of Bucharest Paul O’Neill, University of Sheffield Gabriela Pană Dindelegan, ‘Iorgu Iordan—Alexandru Rosetti’ Institute of Linguistics of the Romanian Academy and University of Bucharest Sandra Paoli, University of Oxford Mair Parry, University of Bristol Lori Repetti, Stony Brook University John Charles Smith, University of Oxford, St Catherine’s College Rosanna Sornicola, University of Naples Federico II Pavel Štichauer, Charles University Prague Oana Uță Bărbulescu, University of Oxford Nigel Vincent, The University of Manchester Max W. Wheeler, University of Sussex Sam Wolfe, University of Oxford Rodica Zafiu, ‘Iorgu Iordan—Alexandru Rosetti’ Institute of Linguistics of the Romanian Academy and University of Bucharest

Maiden, morphology, and more Adam Ledgeway, John Charles Smith, and Nigel Vincent

1 Martin Maiden In the dedicatory note to his chapter in this volume, Michele Loporcaro references the comment by Spencer and Zwicky (1998:1): ‘Morphology has been called “the Poland of Linguistics”—at the mercy of imperialistically minded neighbours’. He goes on to call Martin Maiden ‘the Woodrow Wilson of morphology’, an allusion to the latter’s central role in re-establishing the borders of Poland as part of the Treaty of Versailles. We will let our readers judge the appropriateness of the comparison, but we should note that a concern with the borders and bounds of the components of language has been with Martin since his earliest contributions to the field. In this connection it is worth recalling the final words of his first book, in turn a revised version of his 1986 doctoral thesis: The abruptness of frontiers is frequently an idealization invoked by those who seek stability or precision, and imposed on situations which may, in fact, be fluid and ill-defined. In reality, frontiers are often zones of turbulence, fraught with local alliances and skirmishes. Morphonology is such a frontier zone. Turbulent and refractory it may prove to be: but it is there. (Maiden 1991:263) The phenomenon under investigation here is metaphony, a sound change whereby in some Italian dialects a final high vowel causes the raising or diphthongization of a preceding stressed vowel, yielding alternations such as Isc. [surd, surdə, sawrd, sawrd] ‘deaf.msg/mpl/fsg/fpl’ beside It. sordo, -i, -a, -e. The effect of these changes is to shift the marking of gender and number from the inherited Latin suffixal system, which is continued in the corresponding Italian forms, to a pattern based partly or wholly on vowel alternations inside the stem. Patterns like this, similar to the umlaut alternations in Eng. goose/geese, have always been— and continue to be—a problem for approaches that prefer to see word structure as

Adam Ledgeway, John Charles Smith, and Nigel Vincent, Maiden, morphology, and more. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Adam Ledgeway, John Charles Smith, and Nigel Vincent (2022). DOI: 10.1093/oso/9780198870807.003.0001

2

adam ledgeway, john charles smith, and nigel vincent

involving concatenations of morphemes, whether in the Item-and-Arrangement or Item-and-Process variant models favoured by so-called American structuralists, by generative phonologists, and currently by advocates of Distributed Morphology (Anderson 2018 and references there). The case of metaphony is also an excellent example of two persistent and intertwined strands in Martin’s research over the years: the analysis of Romance dialectal material and the exploration of its theoretical consequences. It is not surprising therefore that his work has appeared in a wide range of both Romance and general linguistics journals. Shortly after the publication of his thesis comes an article in Journal of Linguistics in which he develops a line of argument and articulates a principle that has guided his thinking ever since: I wish to propose that […] the ‘irregularity’ inherent in allomorphy can be appreciated not as basically ‘inert’ deviation from a natural isomorphic relationship between meaning and form, but as an abstract structural property of morphological systems. (Maiden 1992:285) In other words, morphology is independent and has principles of its own. The task is then not to deny the possibility of such principles but to identify them and investigate the way they link with other components of language. Central to this latter goal has been his development of Aronoff ’s concept of the morphome. The original insight formulated in Aronoff (1992; 1994) related to individual forms, as for example English so-called past participles such as opened, shot, written, and sung. These can be used in both passive and active periphrases— he has sung the song, the song has been sung—but no matter how irregular the verb, the forms are identical in both contexts. This generalization is therefore purely morphological and not dependent on either phonology or syntax. This observation is, however, exclusively synchronic and open to the charge that it is a structural coincidence with no intrinsic psychological reality for speakers. Martin’s contribution is twofold: first, to show that morphomes can persist and indeed be generalized over time and thus must have been real for the language learner, and, second, that the concept can be extended to groups of forms defined over paradigms rather than applying to individual forms. A simple example is the so-called N-pattern, in which for some verbs there is a shared stem for the singular persons and the third person plural of the present and a different stem in the first and second persons plural as, for example, Fr. je meurs, tu meurs, il/elle meurt, ils/elles meurent ‘I/you (sg)/(s)he/they die(s)’ vs nous mourons, vous mourez ‘we/you (pl) die’. The original motive force for this distribution lies in the different stress contours of the Latin etyma, but what Martin shows is how this then becomes a template for other changes, such as the suppletive stem alternation of

maiden, morphology, and more

3

the go verb: Fr. je vais, tu vas, il/elle va, ils/elles vont vs nous allons, vous allez, or the defective pattern of Sp./Pt. abolir ‘abolish’ which only has forms for the first- and second-person plural cells of the present tense, although it has a full paradigm in other tenses (see Maiden 2018a:ch.6 for further discussion and exemplification). Patterns such as these in turn argue for the paradigm as an independent theoretical construct rather than simply an amalgam of morphemically defined forms, so it is no surprise that contributions like Maiden (2004b) and Maiden and O’Neill (2010) have become staples of the general literature. The converse of suppletion is also possible, that is to say rather than two sets of forms coming together within a single paradigm, it is also possible for an individual lexical item to share inflexions from distinct paradigms in the phenomenon known as heteroclisis, as exemplified for Romanian in Maiden (2009). Once again historians of Romance and theoretical morphologists are equal beneficiaries of Martin’s insights and analyses. Interesting too is the way the independence of morphology has implications for a topic such as folk etymology, which has often been considered a historical backwater. Maiden (2008; 2020a), however, demonstrates its relevance to a theory of the structural independence of linguistic form. For instance, It. battisuocera ‘cornflower’ appears to consist of batti ‘beat’ + suocera ‘mother-in-law’ but serves as a way of making formally transparent albeit semantically opaque its etymological source in Lat. baptisecula. The same goes for Eng. wheatear even though the bird in question does not eat wheat and birds do not have ears! The same commitment to exploring the internal drivers behind shifts in linguistic form also leads to a degree of caution when it comes to adducing contact as the cause of change. As he argues in Maiden (in press), it is not that contact is not a possible source of innovation, but we should be sure that we have first explored all other possible avenues. In the early part of his career Martin pursued these goals mainly through further synchronic and diachronic work on Italian and Italo-Romance, including a booklength history of the Italian language, the translated version of which became a standard pedagogical and reference work in Italy and elsewhere (Maiden 1995; 1998). In later research, he broadened his scope to include Ibero-Romance—the PYTA (perfecto y tiempos afines) morphome, as the name implies, is based principally on Spanish data—and Gallo-Romance. More recently the focus has been on Romanian in a string of important articles on both nominal and verbal patterns, once again with both diachronic and diatopic perspectives to the fore and in close collaboration with colleagues in the region, as can be seen in Maiden (2021) and the work referenced there. And while we are on that side of the Adriatic, we should not omit mention of his analysis of the small body of surviving Dalmatian data, as witness his chapter in the Oxford Guide to the Romance Languages, now revised as Maiden (2020b). This last-mentioned volume highlights another important aspect of Martin’s contributions to the field, namely his involvement in conceiving and bringing to

4

adam ledgeway, john charles smith, and nigel vincent

fruition large reference projects including The Cambridge History of the Romance Languages (2 vols, 2011–2013), The Oxford Guide to the Romance Languages (2016) and the forthcoming Cambridge Handbook of Romance Linguistics. Each of these has a different focus and each reflects a different dimension of his interests: the first self-evidently diachronic, the second encompassing the full spread of Romance in all its structural, historical, and sociological diversity, and the third, like the present volume, aimed at general linguists with a view to demonstrating the theoretical lessons that can be learned from the evidence provided by the Romance languages. They are all, of course, collaborative works—involving among others many of the contributors to the present volume. They sit beside other multi-authored volumes such as Maiden et al. (2011) and Cruschina et al. (2013) deriving from externally funded research projects which have involved not only international colleagues but also some of his numerous doctoral students and postdoctoral researchers, the support and guidance for whom represents yet another dimension of his contribution to the field.

2 Morphology It is clear, then, that morphology and the associated autonomy of form lie at the heart of Martin’s thinking and writing. The same cannot always be said for the place of morphology within linguistic theory. It had, as we have noted, been a core part of what is commonly called American structuralism, as chronicled in Anderson (2018), but it did not have an obvious place within Chomsky-inspired generative grammar. In this context, the observation by Barbara Partee (2000:xiv), writing about the history of the linguistics department at the University of Massachusetts at Amherst in the 1970s and the way the sub-disciplines came into existence within research and teaching, is instructive: ‘At the beginning, it was just syntax, phonology, and historical, then some semantics, then language acquisition’. Then, as an afterthought, she adds: ‘There was always morphology, too, but it never had a distinct “slot” in the curriculum’. Nor, she might have added, within the dominant theoretical framework of the day, despite the work of scholars, mainly European, such as Bierwisch, Dressler, Kiefer, and Wurzel. However, by the time Martin entered the field, things were changing. In the wake of the seminal work by Matthews (1972), various a-morphous (Anderson 1992) or separationist (Beard 1995) frameworks were being formulated. In addition, the approach now known as Paradigm Function Morphology (PFM) was being developed (what Stump 2018b calls PFM1). For the place of the morphome within PFM, see Stump (2016:ch.8) and the references there. In this context, Martin’s exploration of the paradigmatic and diachronic dimensions of Aronoff ’s idea fell on fertile ground and his work has been regularly cited ever since. This of course is not to say that individual morphomic accounts may not be contested, as with the stem alternations

maiden, morphology, and more

5

in Surmiran (Romansh) that have been an ongoing matter of debate between Martin and Stephen Anderson (see most recently Maiden 2017 and references there). The Maiden–Aronoff connection is also key in terms of the geography of the discipline. Accounts such as Anderson (2018) emphasize the divide between Europe and the USA, understandable given the centrality of ideas emerging in particular from the generative tradition in the latter half of the twentieth century. But the transatlantic link established by Martin’s adoption and development of the morphome concept is part of a broader reconfiguration of thinking, where the division is not one based on the location of the researchers but rather on their conceptual foundations. If we stick with acronyms, it is DM vs PFM, with both models having active advocates not only on both sides of the Atlantic but all around the world. In terms of Stump’s typology of theories of morphology, both DM and PFM are ‘realizational’, that is to say ‘a word’s association with a particular set of morphosyntactic properties licenses the introduction of those properties’ inflectional exponents’ (Stump 2001:2). The difference lies in the fact that DM is ‘lexical’ since words can be broken down into their constituent parts, the morphemes, whereas PFM is ‘inferential’ since the morphosyntactic features associated with a given word may be inferred from it even if the form itself is not transparent. And as Stump (2001:12) goes on to note, in general, ‘the assumptions of inferential-realizational morphology presuppose a very limited interface between inflectional morphology and syntax’. By contrast, DM insists on a close integration of syntax and morphology and permits, within as well as between words, the deployment of all the standard operations of a derivational syntax: tiered functional heads; null categories and movement. The realizational part occurs at the end of these operations via the mechanism of Late Insertion after operations such as impoverishment, fission, and fusion and at the expense of the older Lexical Integrity Hypothesis (Anderson 2018:12). In that sense, proponents of DM can still be said to recognize a domain of morphology, even if, as the name implies, the mechanisms involved do not all cluster within a single component or module. These debates are ongoing and lead Anderson to conclude his short history with the observation that ‘morphology is alive and well as a distinct field of study within contemporary linguistics’ (2018:14). But such optimism may be premature if we accept the logic of nanosyntax (Caha 2018) and, in even more extreme form, that of Collins and Kayne (2020). The latter defend the thesis that: ‘Morphological generalizations are accounted for in terms of syntactic operations and principles. There is no morphological component in UG. There are no post-syntactic morphological operations.’ On this view, to pursue the geopolitical analogy, morphology is less the Poland of linguistics than the Prussia—a historical state that has not only been partitioned between syntax and phonology but which, unlike Poland, has been abolished entirely!

6

adam ledgeway, john charles smith, and nigel vincent

3 More In the course of his career, Martin has been the deserved recipient of many honours and tributes including election as a Fellow of the British Academy (2003), as a Member of the Academia Europaea (2018), and as a Corresponding Fellow of the Accademia della Crusca (2019). In 2013 he was awarded an honorary doctorate by the University of Bucharest and the following year was appointed ‘Commander’ in the ‘National Order for Faithful Service’ of Romania, by decree of the president of Romania, for services to the Romanian language in Britain. The chapters that follow offer more recognition of the importance and impact of his work, in the form of a series of case studies covering a wide range of Romance languages and dialects and linked to major theoretical issues in the domain of periphrasis and inflexion. Our starting point is an overview chapter jointly authored by Ledgeway and Vincent, which sets out some of the main issues, both in regard to the specifics of Romance and the more general questions relating to theories of language structure and change. This is followed by the chapter by Smith, who discusses the often blurred and fuzzy boundaries of inflexion and periphrasis, drawing on Romance data. There then comes a group of four chapters which examine different aspects of Romance periphrases and their diachronic profiles. Vincent and Wheeler investigate developments in form and the way verbs such as have and go sometimes do and sometimes do not display formal differences to match their shifts from lexical to grammatical content. By contrast, Paoli and Wolfe explore the semantic issues, looking in detail at the different histories of the go verb + infinitive in French and Occitan. Parry also focuses principally on changes in meaning, but with respect to the rather less studied motion verb tornare ‘turn, return’ and its development in north-western Italian dialects. With Cruschina, we move down to the south of Italy and back to go and have, with the difference that the changes are now explored from a theoretical perspective that seeks to unite insights from Maiden’s concept of the morphome with the framework of Distributed Morphology. A classic context in which have has grammaticalized alongside be involves the perfective periphrasis where Romance varieties, and in particular Italo-Romance varieties, show a great deal of diversity. These are studied here in two complementary ways: by Bach and Štichauer in terms of inflexional classes and by Loporcaro with special regard to the boundary between morphology and syntax. From ItaloRomance we move to Daco-Romance in the contributions of Dragomirescu, Nicolae, and Zafiu, and of Pană Dindelegan and Uță Bărbulescu, both of which address aspects of the traditional synthesis/analysis question. The former show that there is by no means a single directionality and that in some non-standard varieties emerging periphrases have, so to speak, stopped in their tracks and synthetic forms have taken over the same grammatical functions. The latter examine nominal marking and the complex interactions of prepositional and case marking

maiden, morphology, and more

7

of nominal arguments. There are, of course, also complex relations within the synthetic part of grammar and it is these which Sornicola explores, taking us back in time to a body of medieval Latin texts and considering the concept of overabundance in the broader context of the history of linguistic theory. With Bentley and Cennamo we return to Italy, this time to examine the question of verbal agreement in dialects where the relevant constraints need to be stated in terms of topic and theme rather than in terms of grammatical arguments such as subject and object. At the other side of grammar, so to speak, Aronoff and Repetti investigate the links—and boundaries!—between morphology and phonology. Finbow and O’Neill then bring the present volume to a conclusion with their discussion of the relation between internal change and the effects of language contact, which they study with respect to interaction between related dialects in Portuguese as contrasted with contacts between Portuguese and unrelated languages in Brazil. It goes without saying that the issues addressed by all our contributors are ongoing ones, but of one thing we can be sure. Whenever they are discussed in the future, among the scholars cited there will always be found the name of Martin Maiden.

PART I

THE STATUS OF PERIPHRASIS AND INFLEXION

1 Periphrasis and inflexion Lessons from Romance Adam Ledgeway and Nigel Vincent

1.1 Introduction It is no accident that evidence from the Romance languages has played such a central role in general linguistic debates, particularly when, as in many chapters in the present volume, the concern is the relation between structure and change. It is hard to imagine a better test bed: a large family of languages showing various degrees of structural relatedness and rich historical records, in some cases covering more than a millennium, together with a well-documented source language going back another thousand or so years provides an ideal context for the investigation of patterns of change and the way they interact with existing structures. At the same time, the sociohistorical circumstances of these developments mean that there are frequent opportunities to investigate the role of external factors, including contact between more or less closely related members of the family, contact with a variety of non-Romance languages, and even contact between the members of the family and their historical ancestor, Latin. And yet no body of historical data, however rich and varied, is ever complete. There are always gaps, inconsistencies, and puzzles, and it is at this point that general linguistic theory can step in and help to complete the picture. It also comes as no surprise that, in this context, one of the issues that has attracted most attention has been the connection between periphrasis and inflexion, in turn part of the broader range of topics that constitute the traditional and somewhat vexed question of the relation between analysis and synthesis. Latin had rich systems of verbal and nominal inflexion that interacted in various ways with existing periphrastic patterns and with structures that incorporated the ingredients of incipient Romance periphrases. These structures are of interest not only to those whose principal focus lies in exploring and explaining the development of the Romance languages but also to those whose concern is with the broader theoretical implications for our understanding of language structures and the ways

Adam Ledgeway and Nigel Vincent, Periphrasis and inflexion. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Adam Ledgeway and Nigel Vincent (2022). DOI: 10.1093/oso/9780198870807.003.0002

12

adam ledgeway and nigel vincent

they can change. The Romance languages thus provide a locus where history and theory can interact to mutual benefit. Our aim in this overview is to set out the issues, both theoretical—what models are most appropriate to describe and explain the Romance data?—and empirical— what sources of data are available and what issues arise in exploiting them? We begin with the criteria that serve to distinguish analytic patterns in general before going on to examine the specific properties of periphrastic constructions. It is a natural step from there to consider the interface between periphrasis and inflexion and the different ways they are modelled in current theoretical work. From synchrony, we move to diachrony and the role of grammaticalization in the genesis of new periphrases and inflexions, and interaction on the one hand between internally and externally motivated change and on the other between reconstructed and attested patterns. We conclude with some general reflections on the relation between synchrony and diachrony before opening the door to the individual case studies that constitute the core of this volume.

1.2 Theoretical issues 1.2.1 Analysis vs periphrasis Although, superficially at least, there is unmistakable evidence in certain areas of the grammar for the greater analyticity of Romance in contrast to the typical synthetic tendencies of Latin (Meillet 1936:116–121; Bourciez 1956:23; Harris 1978:15f.; Schwegler 1990; Posner 1996:156f.), this traditional contrast between Latin and Romance in terms of a synthetic–analytic split proves problematic on several grounds, as does the fundamental typological distinction on which it rests (cf. Schwegler 1990:4f.; Bauer 1995:10f., 138, 166; Vincent 1997a:99f., 105; Ledgeway 2012:ch.2; 2017a). Indeed, Humboldt (1836:288f.) observed that, although individual inflexional forms were lost in the passage from Latin to Romance, inflexion as a whole was not (‘Es sanken Formen, nicht aber die Form’; cf. also Vincent 2016a:40). This highlights how the Latin–Romance transition cannot be characterized tout court in terms of a shift from synthetic to analytic structures since Romance continues to display extensive syntheticity in both the nominal and verbal domains (see also Chapter 9 in this volume).1 Consequently, the diachronic 1 Coseriu (1987) reinterprets the traditional synthetic–analytic dichotomy as a distinction between ‘internal’ and ‘external’ structures, respectively (cf. Coseriu [1971] 1988:211, 224). Accordingly, 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. Within this approach, what sets Latin apart from Romance is its failure to make any formal distinction between relational and non-relational functions, while Romance follows a new formative principle of mapping relational and non-relational functions respectively onto externally determined (viz. analytic) and internally determined (viz. synthetic) structures (see Ledgeway 2012:26–28).

periphrasis and inflexion

13

study of Romance offers the linguist a privileged data set in which to explore and model the changing relationships between inherited inflexional forms and a series of competing analytic exponents. Yet, while it is commonplace in treatments of Romance to concentrate on the typological shift from the predominantly synthetic structures of Latin towards the characteristically analytic structures of Romance, considerably less attention has been devoted to those instances where the two intersect.2 Thus, although at a certain level of superficial grammatical analysis it might prove descriptively adequate to place all Romance multi-word grammatical structures under the general heading of ‘analytic’, it is instructive to separate out those instances where the emergence of analytic exponents serves not so much a substitutive as an additive and complementary function, entering into a direct formal and functional contrast with synthetic exponents. By way of illustration, consider the examples in (1)–(2). (1) a. statim Roma-m profectus est. (Lat.) at.once Rome-acc set.out.ptcp.nom.msgbe.prs.ind.3sg ‘at once he left for Rome.’ (Cic., Att. 12.18.1) b. A plecat la Roma. (Ro.) have.prs.ind.3sg leave.ptcp to Rome. ‘(S)he has left for Rome.’ (2) a. epistulam Malli-o dedi (Lat.) letter.acc.fsg Mallius-dat.msg give.pfv.ind.1sg ‘I have given the letter to Mallius’ (Cic., Att. 1.16.16) b. Leam dat scrisoarea them.dat= have.prs.1sg give.ptcp letter.def.fsg.nom-acc la trei prieteni. (Ro.) to three friends.m.nom-acc ‘I gave the letter to three friends.’ Ostensibly, the pairs in (1)–(2) both involve a contrast between a Latin inflexional case form—accusative marking the directional complement in (1a) and dative the Recipient in (2a)—and a corresponding Romance analytic expression involving the use of the preposition la ‘to’ in the Romanian examples (1b, 2b). Given these parallels, examples such as (1b) and (2b) are standardly lumped together as canonical cases of analyticity. Yet, such a broad-brush approach misses a fundamental difference between the two cases, as becomes clear when examples such as (2c) are introduced.

2 Notable exceptions include Vincent (1987; 2011; 2016a:§4.2.3), Bo¨rjars, Vincent, and Chapman (1997), Taylor (2011).

14

adam ledgeway and nigel vincent

(2) c. Le-am dat scrisoarea them.dat=have.prs.1sg give.ptcp letter.def.fsg.nom-acc prieteni-lor. (Ro.) friends.m-def.gen-dat.pl ‘I gave the letter to the friends.’ Now the Recipient is specific and correspondingly appears in the definite form, which in Romanian is realized by means of an inflexional suffix, viz. -lor < illorum ‘that’ (cf. Ledgeway 2017b), which simultaneously encodes a binary case opposition (cf. prieteni-i ‘friends.m-def.nom-acc.pl’).3 This example therefore highlights how, unlike in examples such as (1b), where a prepositional phrase constitutes the only option in conjunction with directional complements, the use of the same preposition in marking Recipients represents but one of a set of available analytic (2b) and inflexional (2c) markers of ‘dative’. In the standard language at least (Maiden 2016a:101), these markers form part of a complementary system whose distribution is determined, among other things, by the availability or otherwise of an initial element (e.g., definite noun or quantifier, but not cardinal numerals like trei ‘three’) within the nominal group able to bear inflexional case marking (Iorga Mihail 2013:147; Stan 2013a:264f.).⁴ While we continue to refer to examples such as (1b) under the generic label of ‘analysis’, we shall refer to cases such as (2b), where analyticity and inflexion intersect as ‘periphrasis’ (cf. also Anderson’s 2011:99 distinction between ‘analytical expression’ and ‘grammatical periphrasis’).⁵ Irrespective of the theoretical approach that one adopts, an inescapable conclusion from contrasts such as (2b–c) is that a proper understanding of periphrasis must make reference to both morphology and syntax (cf. Vincent 1987:251). On the one hand, the distribution of the preposition la as a marker of Recipients falls out as a consequence of the idiosyncratic limits placed on the morphological marking of dative, insofar as specific nominal categories such as cardinals simply lack the relevant inflexional morphology, i.e., there are no such forms as **treior ‘three-gen-dat.pl’ (though for examples in earlier stages of the language, see the discussion in §10.2.4.1 in this volume). Periphrastic marking of Recipients can therefore be understood as the output of a Paninian Principle introduced as a last-resort mechanism whenever inflexional morphology is blocked. On the other hand, there is no getting away from the fact that the preposition la in (2b) combines with the relevant nominal to form a prepositional phrase (though see the discussion in relation to examples (5)–(7) in §1.2.1.2), an operation which is clearly part of the syntax which builds constituent structure and determines, for example,

3 For further detailed discussion of the relationship between periphrastic and synthetic marking within the Romanian nominal group, see Chapter 10 in this volume. ⁴ Cf. also the discussion of definiteness marking via inflexions and free-standing articles in Danish in Spencer and Popova (2015:200f.). ⁵ Pace Spencer and Popova (2015:197) who take ‘analytic’ to be a synonym of ‘periphrasis’.

periphrasis and inflexion

15

the head-initial linearization of la and its complement and ensures that the latter is case-marked accusative under subcategorization by its governing prepositional head. We thus see that periphrasis is firmly situated at the interface between morphology and syntax, a characterization which, as noted by Brown et al. (2012:272), ‘is often problematic for linguistic models since they have to accommodate not only straightforwardly morphological and straightforwardly syntactic phenomena but also phenomena that seem to share properties of both’. Where theories differ is how they draw the line between the two, if indeed they draw one, and the concomitant balance they strike between morphology and syntax in modelling periphrasis. Following Stump’s (2001:15; 2018a:56f.) taxonomy, we distinguish here between lexical and inferential approaches to inflexional morphology.

1.2.1.1 Lexical theories Broadly speaking lexical theories, whether incremental like Lieber’s (1992) or realizational such as Distributed Morphology (DM; cf. Halle and Marantz 1993; 1994), assume that the association of particular word forms and their morphosyntactic properties arises in the syntax. As such there is no line to be drawn between morphology and syntax since morphology is simply syntax. This is encapsulated in the leading claim of DM that morphology is ‘syntax all the way down’ (Marantz 1997), with the atomic units of syntax taken to be morphemes which are subject to the same generative principles that manipulate words into larger phrasal hierarchical constituents (cf. also §2.2.1 in this volume). On this view, periphrasis ought to be ‘grist for the mill of Distributed Morphology’ (Kiparsky 2005:113) since it appears to offer a natural way of blending morphological and syntactic behaviours within a unified framework. Indeed, Embick (2007:3) goes so far as to claim that ‘alternations between “words” and larger syntactic objects (phrases) seem to be directly compatible with a syntactic approach to morphology’. For example, the difference between (2b–c) would presumably be modelled in terms of the application or otherwise of a post-syntactic PF process of Fusion (Halle and Marantz 1993:116) on a common syntactic derivation prior to Vocabulary Insertion when the various feature bundles (viz. abstract morphemes dubbed ‘listemes’) are given phonological exponence.⁶ On this view, the contrast witnessed in (2b–c) highlights how morphosyntactic variation in dative-marking through the formal alternation between periphrastic la ‘to’ and inflexional -lor crucially depends on whether K(ase) is realized in a ⁶ In other cases the amalgamation of individual functional heads and their associated features may arise in the narrow syntax through such operations as head movement (cf. Baker’s 1985 Mirror Principle). For instance, the contrast between the French synthetic and periphrastic future expressions (il) parler-a ‘(he) speak.inf-fut.3sg’ vs (il) va parler ‘(he) go.prs.ind.3sg speak.inf’ can be understood as the respective outputs of the move vs merge operations. In the former case the lexical root—instantiated, at least historically, by the infinitive—undergoes V-movement to left-adjoin to the T-related functional head realizing affixal future features (together with person/number Agr(eement)), thereby yielding a complex head which is readily interpreted as a synthetic expression at PF, namely [T* [V parler-] [T -a]] (cf. Roberts and Roussou 2002; 2003:48–58; Roberts 2009). In the latter case, by contrast, the features of the T-related ‘future’ head are lexicalized by direct merger of a free-standing

16

adam ledgeway and nigel vincent

scattered or syncretic fashion (cf. Giorgi and Pianesi 1997). As illustrated in (3a),⁷ whenever lexical material such as the cardinal trei intervenes between the K° and N° nodes, as happens when the nominal is not definite and hence remains in situ, then these two positions are independently projected and the two heads are realized in a scattered fashion with the K° head lexicalized by la ‘to’. When, however, the two heads are adjacent, as happens whenever the nominal is definite and lexicalizes the D° position under N-movement, then a syncretic, complex K/D head obtains under Fusion, in which both case and definiteness are now inextricably bound together and phonologically spelt out in the inflexional suffix -lor realized on the nominal head (3b). (3)

(a)

(b)

KP

K la

K+D prietenilor

DP NumP

D Num trei

K/DP

NP

NP

N prietenilor

N prieteni

The theory therefore predicts that unfused structures are realized as periphrastic datives, whereas the resulting fused structures are spelt out as synthetic datives.⁸ In some contexts the surface output is variable, as in the case of quantifiers which may optionally undergo Fusion with K (4a), an option favoured in higher registers of the language, to produce synthetic forms such as puțin-or ‘few-gen-dat.pl’ alongside periphrastic la puțini ‘to few.nom-acc.mpl’. This variability in the application of the post-syntactic PF process also explains substandard periphrastic structures such as la cei ‘to the.nom-acc.mpl’ in (4b), where, in contrast to the standard synthetic variant celor ‘the.gen-dat.pl’, Fusion fails to apply to the freestanding definite article cei merged under Dº, itself a periphrastic form of the article employed just in those structures where an intervening cardinal blocks N-to-D raising.

auxiliary such that the lexical infinitive remains in situ (viz. [T’ [T va]…[v-VP parler]]) and, in the absence of post-syntactic PF rules, a periphrastic expression obtains (cf. also discussion in Andriani, Groothuis, and Silvestri 2020:§2). ⁷ For ease of exposition, in (3a–b) we informally represent the features of terminal nodes with their surface PF lexicalizations. ⁸ A similar line of reasoning in terms of post-syntactic ‘merger’ operations is proposed in Embick (2000) and Embick and Noyer (2001), and in Embick (2007) to account for synthetic and periphrastic alternations in the Latin perfect and the English comparative/superlative, respectively.

periphrasis and inflexion

(4)

17

a. Dau scrisoarea la puțini give.prs.1sg letter.def.fsg.nom-acc to few.nom-acc.mpl /puținor prieteni. (Ro.) few.gen-dat.pl friends.m ‘I’ll give the letter to a few friends.’ ? b. Dau scrisoarea la cei give.prs.1sg letter.def.fsg.nom-acc to the.nom-acc.mpl /celor doi prieteni. (Ro.) the.gen-dat.pl two.m friends.m ‘I’ll give the letter to the two friends.’

To sum up, la represents the scattered spell-out (cf. Giorgi and Pianesi 1997) of a single [Kase] feature in contrast to its syncretic spell-out in the feature bundle [Kase, Definite] (⇒ -lor, celor) or [Kase, Q] (⇒ puținor). However, it is not immediately clear what, if any, independent syntactic motivation there is for positing post-syntactic operations such as Fusion and other PF readjustment rules other than to get the morphology to fit the syntax by manipulating and rearranging the hierarchical structure generated by the narrow syntax (cf. Vincent 1987:251; Stewart and Stump 2007:384f.).⁹ Apart from being unconstrained in this way, processes such as Fusion also face an additional timing problem in that the premise of their application rests on a ‘look-ahead’ principle (cf. Siddiqi 2018:155f.): fused forms obtain just in those cases when the Vocabulary contains an item such as prietenilor which requires two heads to fuse, but this information is not available at the time of Fusion, which follows syntax but precedes Vocabulary Insertion. By the same token, variation between fused and unfused variants such as puținor/celor and la puțini/? la cei proves at best hit-and-miss, with the theory apparently hedging its bets between which of the two available Vocabulary Items might be inserted. More worrying, however, for our immediate concerns are the consequences that such approaches have for our understanding of periphrasis. To put it bluntly, periphrasis has no place in a theory like DM, which simply fails to recognize it as a distinct category at all. Rather, it is forced to treat cases like (2b) under the allembracing general heading of ‘analysis’, indiscriminately lumping it together with examples such as (1b), for they both represent cases in which Fusion has not applied and, crucially, in which there is no a priori expectation for Fusion to have failed to apply in one or other of the two cases. In short, the system simply outputs objects, irrespective of whether the morphophonological packaging of the relevant set of terminal nodes surfaces in a one- or multi-word form. On this view, periphrasis turns out to be rather unspectacular, just another undifferentiated case of analysis. Yet, we have seen in §1.2.1 that there are very good reasons to keep examples like (2b) separate from examples such as (1b) and failure to do so misses some non-trivial empirical and theoretical generalizations. ⁹ For an overview of such post-syntactic processes and DM in general, see Bobaljik (2017) and Siddiqi (2018).

18

adam ledgeway and nigel vincent

1.2.1.2 Inferential theories In contrast to lexical approaches, where word forms and their associated morphosyntactic properties are mediated entirely within the syntax, inferential theories, whether incremental (e.g., Steele 1995) or realizational such as Paradigm Function Morphology (PFM; Stump 1991; 2001; 2016; 2018b; Stewart and Stump 2007), formalize such associations within an autonomous system of morphology. Above we saw in relation to contrasts such as (2b–c) that our understanding of periphrasis crucially relies on the assumption that the domain of inflexion is not limited to the conventional building blocks and units of traditional morphological analysis, but should be interpreted in such a way as to be able to also accommodate multi-word sequences whose forms and distribution are generally taken to fall within the realm of syntax. In our view, such a state of affairs is only compatible with inferential–realizational models of the Word-and-Paradigm type since these are able to integrate both single synthetic forms and multi-word expressions with the morphosyntactic realization of individual paradigm cells (cf. Blevins, Ackerman, and Malouf 2018:273f.). Key therefore to our definition of periphrasis is the role of the inflexional paradigm and the relations that hold between the interdependent exponents—whether synthetic or analytic—of individual cells of a given lexeme.1⁰ Indeed, the notion of paradigm figures prominently in most definitions of periphrasis, witness Haspelmath’s (2000:655) now classic formulation as ‘a situation in which a multi-word expression is used in place of a single word in an inflectional paradigm’ (cf. also Vincent 1987:240; Stump 2001:43; Stewart and Stump 2007:408; Anderson 2011:143). On this view, the inflexional and periphrastic forms in (2b–c) can now be laid out as part of a structured inflexional paradigm along the lines of Table 1.1. With very few exceptions, the paradigmatic relations of form and function illustrated in Table 1.1 are representative of all masculine nouns, where the distribution of individual forms can be stated in terms of a network of intersecting features relating to (in)definiteness, number (singular vs plural), and case (nominative–accusative vs genitive–dative) (Maiden 2016a:100f.). In most of the cells the exponents of the available feature combinations are synthetic, the exception being those cells realizing the pairing genitive–dative and indefinite, where genitive and dative are marked, respectively, by a and la in combination with the unmarked nominative–accusative indefinite forms.11

1⁰ Of course, paradigmaticity plays no role in configuration-based lexical theories such as DM which do not recognize the paradigm in their analyses of such phenomena as synthetic/analytic alternations (Embick 2000; 2007) and suppletion (Pomino and Remberger 2019). In short, to recognize the paradigm within DM would involve a special case of the problem of transderivationality which has always bedevilled derivational models since it is hard to see how the derivation of one structure can ‘look across’ to see if it is compatible with the derivation of another structure. 11 Although genitive and dative are syncretic when realized synthetically (e.g., -lui, -lor), they are separated out in their periphrastic realizations, namely a vs la, respectively. Given our focus here on dative, we ignore in what follows the periphrastic genitive marker a, though it should be noted that its distribution is conditioned by exactly the same factors as those operating on la.

periphrasis and inflexion

19

Table 1.1 Inflexional paradigm of Romanian prieten ‘friend’

Nom–Acc Gen–Dat

Indefinite Singular Plural prieten prieten-i a/la prieten a/la prieteni

Definite Singular Plural prieten-ul prieteni-i prieten-ului prieteni-lor

Within their paradigmatic representation, these periphrastic exponents can therefore be interpreted as a ‘repair strategy’ for language-specific deficiencies in the morphological realization of particular feature combinations which find no formal expression in the synthetic morphology of the language (cf., by contrast, the well-formedness of Lat. (in)definite amic-i/-orum ‘friend. -gen.msg/-gen.pl’). Confirmation that the periphrastic realizations in Table 1.1 are effectively part and parcel of the inflexional paradigm is evidenced by a number of considerations. First, the cells involved do not represent accidental gaps in the paradigm of particular defective lexemes which are otherwise regularly realized by synthetic exponents in the inflexional paradigms of other lexemes belonging to the same category. Rather, as the examples in (5)–(6) illustrate, the paradigmatic pattern characterizes all masculine nouns of the same inflexional class without exception. (5)

a. Îi vorbește popor-/comitet-/grup-ului. (Ro.) dat.3sg= speak.prs.ind.3sg people/committee/group-def.gen-dat.msg ‘(S)he speaks to the people/committee/group.’ b. Îi vorbește la tot dat.3sg= speak.prs.ind.3sg to all.msg popor-/comitet-/grup-ul. (Ro.) people/committee/group-def.nom-acc.msg ‘(S)he speaks to all the people/committee/group.’

(6) a. Le vorbește dat.3pl= speak.prs.ind.3sg studenți-/muncitori-/musafiri-lor. (Ro.) students.m/workers.m/guests.m-def.gen-dat.pl ‘(S)he speaks to the students/workers/guests.’ b. Le vorbește la niște dat.3pl= speak.prs.ind.3sg to some studenți/muncitori/musafiri. (Ro.) students/workers/guests.nom-acc.mpl ‘(S)he speaks to some students/workers/guests.’ Second, just like the synthetic members of the inflexional paradigm (5a, 6a), the periphrastic exponents can also be doubled by a dative clitic (5b, 6b), although

20

adam ledgeway and nigel vincent

prepositional objects introduced by la do not otherwise allow clitic doubling (cf. Cornilescu 1995:13), witness the ungrammaticality of (7). (7)

(**Le) recurge la niște dat.3pl= resort.prs.ind.3sg to some studenți/muncitori/musafiri. (Ro.) students/workers/guests.nom-acc.mpl ‘(S)he has recourse to some students/workers/guests.’

Finally, it is standardly assumed that one of the constraints operating on coordination is the requirement that both conjuncts be of equal status and, in the case of DP coordination, that both DP conjuncts bear the same case form. Consequently, coordinate structures such as (8a), where a single nominal functions as the complement of two coordinated verbs, the first subcategorizing for an accusative DP and the second for an oblique PP headed by la ‘to’, prove ill-formed since the complement cannot simultaneously meet the conflicting categorial and case requirements of both verbs. Against this backdrop, the grammaticality of examples such as (8b), in which inflexional and periphrastic datives are felicitously coordinated, highlights that we are indeed dealing with the same type of grammatical beast, irrespective of its surface synthetic or analytic form, to which the grammar is seemingly blind. (8)

a. **Citeam și mă uiteam (la) read.pst.ipfv.1sg and me.acc= look.pst.ipfv.1sg to un sms. (Ro.) a.nom-acc.msg sms.nom-acc.msg ‘I was reading and looking at a text message.’ b. Le vorbeam studenților și dat.3pl= speak.pst.ipfv.1sg students.def.gen-dat.pl and la niște prieteni. (Ro.) to some friends.nom-acc.mpl ‘I was speaking to the students and some friends.’

In conclusion, these brief considerations highlight how periphrastic forms are fully integrated into otherwise predominantly synthetically based inflexional paradigms. Despite their superficial analytic appearance, they have been shown to pattern in many respects (e.g., clitic doubling, coordination), not with canonical phrasal prepositional constituents formed in the syntax, but, rather, with the monolectal forms that constitute the atomic objects of inflexional morphology with which they share key properties. This highlights how all that is analytic is not necessarily syntactic, but can equally be morphological at the appropriate level of abstraction. At the same time, it is not at all clear how lexical-based models such as DM, which do not make reference to paradigms, would be able to handle and make

periphrasis and inflexion

21

sense of the paradigmatically predictable synthetic–periphrastic alternations in (5a, 6a) and (5b, 6b) and the concomitant categorial and case distinctions highlighted in the distribution of clitic doubling and coordination. In particular, it is hard to see how the relevant synthetic–periphrastic alternations can be accounted for other than by stipulating a post-syntactic process of Fusion precisely and only in those cases where synthetic datives are not available. Similar problems face the clitic doubling and coordination contrasts in (5)–(7) and (8a–b). In the former case one might have to appeal to an abstract distinction, the motivation for which is not particularly strong in relation to these clitic doubling facts, between, say, structural and inherent (or lexical) Cases, on the assumption that the dative clitics îi/le are a spell-out of structural, not inherent Case. Following this same line of reasoning, it could be argued that when marking structural dative, la heads a KP (cf. 3a–b), in contrast to its uses in inherent/lexical Case contexts such as (7, 8a), where it heads a fully fledged PP, and hence is amenable to coordination in (8b) with a synthetic dative since they both instantiate (K/)DP constituents.

1.2.2 Formal issues in modelling periphrases Having reviewed some basic theoretical assumptions about the nature of periphrasis and its formal and functional relationship to inflexion and analyticity more generally, we now turn to explore a number of key issues that have figured prominently in the theoretical literature in defining and modelling inflexional periphrasis. In particular, our aim here will be to investigate some of the central ideas and claims that have emerged in the formal definition and identification of periphrasis as a distinct linguistic category by considering a selection of Romance examples which have not previously figured in the relevant literature. The Romance evidence, it will be shown, brings to light some significant theoretical questions and challenges which force us to reassess and refine in a number of ways our understanding of periphrasis and the interfaces between morphology, syntax, and semantics.

1.2.2.1 Intersectivity As briefly noted above, a fundamental criterion in the definition of inflexional periphrasis in paradigmatic approaches is intersectivity (cf. also §2.4.2 in this volume), a notion which allows us to distinguish between ‘pure’ cases of analyticity, which fall entirely within the realm of syntax, and those incidences of ‘superficial’ analyticity, namely periphrasis, which can only be appropriately defined in relation to the syntax–morphology interface (Ackerman and Stump 2004; Vincent 2011:425; Brown et al. 2012:246; Spencer and Popova 2015:204). Essentially, intersectivity arises whenever within a given paradigm the realization of

22

adam ledgeway and nigel vincent

permissible combinations of different grammatical features results in the integration of cells which, for the most part, are realized by monolectal synthetic expressions with one or more cells whose exponence is exceptionally realized by multi-word periphrastic expressions, producing what Haspelmath (2000:656) calls ‘paradigm symmetry’. As observed in the Romanian examples above (cf. Table 1.1), within the nominal domain intersectivity is commonly determined by the convergence of core nominal morphosyntactic features such as (in)definiteness, number, gender, and case, but arguably not by purely morphological features such as declensional class.12 Within the verbal domain, by contrast, the relevant intersecting parameters that yield structured synthetic–periphrastic alternations relate to person, number and, in particular, tense, aspect, mood, and voice. Person and number, whether individually or together, are less strong determinants of intersectivity as witnessed by the paucity of relevant Romance examples. However, one plausible candidate is the Italian present and past subjunctive paradigms set out in Table 1.2. Although Italian arguably represents an exemplary case of a pro-drop language, it exceptionally requires the use of the overt second-person singular subject pronoun tu ‘you’ in the present and past subjunctive (9a–b), where, in contrast to non-subjunctive contexts, the use of an overt pronoun does not license the typical focus or topic–shift interpretations (Benincà 1993:258–262; Cardinaletti 1997:51f.; Cordin 1988:540; Ledgeway 2016b:216).

Table 1.2 Italian present and past subjunctive paradigms Present 1sg 2sg 3sg 1pl 2pl 3pl

cant- ‘sing’ cant-i **(tu) cant-i cant-i cant-iamo cant-iate cant-ino

vend- ‘sell’ vend-a **(tu) vend-a vend-a vend-iamo vend-iate vend-ano

Past serv- ‘serve’ serv-a **(tu) serv-a serv-a serv-iamo serv-iate serv-ano

cant- ‘sing’ cantass-i **(tu) cantass-i cantass-e cantass-imo canta-ste cantass-ero

vend- ‘sell’ vendess-i **(tu) vendess-i vendess-e vendess-imo vende-ste vendess-ero

serv- ‘serve’ serviss-i **(tu) serviss-i serviss-e serviss-imo servi-ste serviss-ero

12 There are a not insignificant number of native (e.g., ianuarie ‘January’, luni ‘Monday’, b ‘letter b’, Ion) and non-native (e.g., dingo ‘dingo’, euro, miss ‘beauty contestant’, yoga) nouns in Romanian which are invariant and hence inflexionless (cf. Ledgeway 2017b:§2.2). Unlike most nouns (cf. Table 1), their definite genitive–dative forms are in most (but not all) cases therefore supplied by the unbound prenominal determiner lui, e.g., începutul lui ianuarie ‘the beginning of January’, a potential case of periphrasis. However, even if one were to treat such invariant nouns as constituting a distinct declensional class from those to which the majority of regular inflecting nouns belong, the relevant forms would still not involve any notion of intersectivity because, as inflexionless nouns, they do not belong to a paradigm. Rather, they represent a genuine case of inflexional defectivity.

periphrasis and inflexion

(9)

23

a. Non sapevano che pro1/**2/3sg cant-i neg=know.pst.ipfv.3pl that pro1/**2/3sg sing-prs.sbjv.sg/ -assi. (It.) -pst.sbjv.1-2sg b. Non sapevano che tu cant-i neg=know.pst.ipfv.3pl that you.sg sing-prs.sbjv.sg/ -assi. (It.) -pst.sbjv.1-2sg ‘They didn’t know that I/you/(s)he sing(s)/sang.’

As Table 1.2 illustrates, there is an obvious functionalist explanation for this distribution, in that in the present there is syncretism in the synthetic exponents of the paradigm between all three persons in the singular and between the firstand second-persons singular in the past. Obligatory use of the pronoun therefore serves to (partially) disambiguate reference between the singular persons and thereby reduce the computational burden placed on the hearer. Even if such functionalist explanations might lie behind the emergence of this usage, synchronically they are untenable, especially since there are many other Romance varieties with similar subjunctive paradigms where such surface inflexional ambiguity is readily tolerated (Manasantivongs 2005). For instance, the overt pronoun is still required even if no such ambiguity arises, as happens in cases like (10a), where the subjunctive verb is inherently reflexive and second-person singular reference is already unambiguously marked by the coreferential reflexive clitic. Similarly, the pronoun cannot be omitted in (10b), where the matrix verb is firstperson singular and the subjunctive verb is past, even though the latter can only be interpreted as second-person singular on account of the obviation effect. (10)

a. Vuole/Voleva che **(tu) ti want.prs.ind/pst.ipfv.ind.3sg that you.sg yourself.sg= penta/ pentissi. (It.) repent.prs.sbjv.sg repent.pst.sbjv.1-2sg ‘(S)he wants/wanted you to repent.’ b. Volevo che **(tu) ti want.pst.ipfv.ind.1sg that you.sg yourself.sg= pentissi. (It.) repent.pst.sbjv.1-2sg ‘I wanted you to repent.’

Rather, the distribution of the overt pronoun is clearly driven by syntax rather than by functionalist considerations, although we see once again that we must also make reference to morphology, since the relevant two-word form is clearly integrated into the inflexional paradigm of the subjunctive, where it spells out the intersection of the features second-person singular and subjunctive. Indeed, the

24

adam ledgeway and nigel vincent

morphological nature of this periphrastic expression is supported by the observation that its presence within the inflexional paradigm of the present subjunctive gives rise to a potentially morphomic distribution of forms whereby the secondperson singular (e.g., tu canti) is formally contrasted with the first-/third-person singular (e.g., canti). In this way, distinctive marking of the second-person singular is responsible for placing together first/third singular in a single syncretic category which otherwise lacks any syntactic or semantic motivation, rather than say first/second-person singular (the discourse participants) vs third-person singular, as happens in the inflexional morphology of the past subjunctive (viz. -i vs -e). In other respects too, there are reasons to treat the combination tu+Vsbjv increasingly as a single morphological unit, rather than as the output of the syntax. First, in northern and central varieties of Italian, where linguistic prestige is highest, the second-person singular nominative pronoun is not tu as in the official standard language and in southern varieties, but has been replaced by the erstwhile tonic oblique pronoun te (11a), which also occurs in all varieties in ‘strong’ positions such as exclamative vocatives and under coordination (11b). As a result, the only context in which northern and central speakers use tu (on a par with southern speakers) is the subjunctive (although here too they use te if the subject is given any particular prominence), a distribution which eliminates tu from the paradigm of pronouns and relegates it to a greatly weakened preverbal second-person singular subjunctive marker more akin to an inflexional element than an autonomous piece of phrasal syntax. (11)

a. Io rimango, ma tu/te che I remain.prs.ind.1sg but you.nom/obl what fai? ((S)It/N-CIt.) do.prs.ind.2sg ‘I’m staying, but what are you going to do?’ b. Povero te /**tu! // Io e te /**tu … (It.) poor.msg you.obl nom I and you.obl nom ‘Poor you! / I and you…’

Finally, a similar conclusion comes from examples of complementizer deletion in subjunctive contexts (Wanner 1981; Scorretti 1994; Giorgi and Pianesi 1997; 2004; Poletto 2001; Giorgi 2010:43–47). For most speakers, complementizer deletion also entails deletion of the preverbal subject position, such that canonical preverbal subjects such as lui ‘he’ and Ida in (12a) are obligatorily realized in postverbal position under complementizer deletion in (12b). The exception, however, is tu, which uniquely survives under complementizer deletion, a distribution which reinforces our conclusion that it must be treated as part of the inflexional structure of the verb, rather than as a tonic pronoun since there is ostensibly no available position for preverbal subjects in contexts such as (12b).

periphrasis and inflexion

(12)

25

a. Penso che tu / lui / Ida sia in grado di think.prs.ind.1sg that you he Ida be.prs.sbjv.sg in position of aiutarlo. (It.) help.inf=him b. Penso tu/**lui/**Ida sia lui/Ida in think.prs.ind.1sg you he Ida be.prs.sbjv.sg he Ida in grado di aiutarlo. (It.) position of help.inf=him ‘I think (that) you/he/Ida are/is able to help him.’

Our example of the Italian subjunctive also highlights another useful distinction in relation to the morphological space across which intersectivity may range. In this case, the relevant intersection of features occurs at the level of the ‘local’ paradigm, in the sense that the alternation between synthetic and periphrastic forms arises specifically within the individual sub-paradigms of the present and past subjunctive without any need for us to make reference to the wider verb system. In other cases, by contrast, the distribution of synthetic and periphrastic expressions can only be stated in terms of the ‘extended’ paradigm (cf. also reference to the ‘extended paradigmatic set’ in Vincent 2011:424). Classic examples are the Romance periphrastic perfects (Vincent 1987:238f.; 2011:430–432; Taylor 2011:405–407; Brown et al. 2012:271 n.11) such as in northern regional Italian, where the relevant intersection between imperfective (13a) and perfective (13b) aspect cuts across distinctions of person/number (13c), tense (13d), mood (13e), and finiteness (13f) to hold across all individual sub-paradigms of the extended active verb system. (13)

a. Parlo ‘speak.prs.ipfv.ind.1sg’, parlavo ‘speak.pst.ipfv.ind.1sg’, … (NIt.) b. Ho parlato ‘have.prs.ipfv.ind.1sg speak.ptcp’, avevo parlato ‘have.pst.ipfv.ind.1sg speak.ptcp’, …(NIt.) c. Parlano/parlavano ‘speak.prs/pst.ipfv.ind.3pl’ vs Hanno/avevano parlato ‘have.prs/pst.ipfv.ind.3pl speak.ptcp’, … (NIt.) d. Parlerò ‘speak.fut.ipfv.1sg’ vs Avrò parlato ‘have.fut.ipfv.1sg speak.ptcp’, … (NIt.) e. Parli/parlassi ‘speak.prs/pst.ipfv.sbjv.1sg’ vs Abbia/avessi parlato ‘have.prs/pst.ipfv.sbjv.1sg speak.ptcp’ (NIt.) f. Parlare ‘speak.inf.ipfv’ vs Aver parlato ‘have.inf.ipfv speak.ptcp’, … (NIt.)

As noted by Blevins (2015:88), typically the members of a nominal paradigm (viz. declensional system) are organized into a single network of forms differentiated by specific featural combinations based on number, gender, case, and

26

adam ledgeway and nigel vincent

(in)definiteness (cf. Table 1.1), whereas verbal paradigms (viz. conjugations) introduce a further level of complexity in that individual verbal lexemes are organized not just in terms of person and number distinctions, but are also generally structured into multiple sub-paradigms based on such distinctions as tense, aspect, mood, and voice. It follows that the relevant intersection that produces synthetic– periphrastic splits typically applies locally within the global paradigm in the case of the nominal domain (cf. Table 1.1), but across different (sets of) sub-paradigms belonging to the extended paradigm in the case of the verb (cf. 13a–f). While this is undoubtedly the usual state of affairs, our example of the Italian subjunctive in Table 1.2 is, however, revealing since it highlights that in the verbal domain the relevant intersection can also occur ‘locally’ within individual sub-paradigms without reference to the extended verbal paradigm. Finally, we briefly consider cases of so-called anti-periphrasis (Haspelmath 2000:658f.), where the canonical distributional balance between synthetic (the majority) and periphrastic (the minority) expressions within the paradigm is reversed. By way of example, consider the partial Romanian paradigms in Table 1.3. Table 1.3 Romanian (3sg) imperfective and perfective paradigms serv- ‘serve’ Ipfv Pfv

Present servește a servit

Past servea servise

Future va servi va fi servit

Conditional ar servi ar fi servit

Infinitive a servi a fi servit

Subjunctive să servească să fi servit

In contrast to the forms of the imperfective, which are either synthetic (present, past) or periphrastic (cf. future/conditional auxiliaries va/ar, infinitival/subjunctive markers a/să), the forms of the perfective are, on a par with the (northern) Italian forms reviewed above, periphrastic, consisting of auxiliary (a/fi ‘has/be’) and perfect participle (servit), with the notable exception of the past perfective form (the so-called pluperfect), which, in the standard language at least, is synthetic (cf. servise; see also Chapter 9 in this volume). We thus see that the synthetic past perfective sub-paradigm enters into a formal opposition with all the other available perfective sub-paradigms, which are periphrastic, leading us to question whether the paucity of available monolectal forms warrants defining all the remaining perfective forms as examples of intersective periphrasis (cf. also discussion in §2.4.2 in this volume). Indeed, Spencer and Popova (2015:207) prefer in such cases to talk of ‘overdifferentiation’ for morphosyntactic properties in a particular sub-paradigm which are otherwise realized syntactically elsewhere in the extended paradigm. It is doubtful, however, that such a characterization is correct. In the prototypical case, overdifferentiation is taken to arise where a particular grammatical distinction fails to extend beyond a restricted number of lexemes (cf. Brown 2007), whereas the distribution of the Romanian synthetic pluperfect knows no such restrictions but, rather,

periphrasis and inflexion

27

is a fully productive category extended to all verbal lexemes with a perfective stem (e.g., mâncasem/putusem/mersesem/plânsesem/servisem … ‘I had eaten/been able/gone/cried/served…’). Furthermore, it should be remembered that we are dealing here with two intersective relationships. The first operates at the level of the extended paradigm between imperfective sub-paradigms, realized both synthetically and periphrastically, and perfective sub-paradigms predominantly realized periphrastically. It is across this latter local network of sub-paradigms that the second and most relevant intersection operates where the synthetic nature of the pluperfect represents an aberration from the otherwise predominantly periphrastic realization of the perfect. We thus see that the intersection of the features perfective and past exceptionally sets off this sub-paradigm of the perfect from all other perfective sub-paradigms. The fact, however, remains, that, irrespective of the number of synthetic sub-paradigms and cells that make these up, there is a formal split within the extended paradigmatic set of the perfective that places both synthetic and periphrastic expressions alongside each other as part of a single (sub)system. We conclude therefore that it seems more appropriate within this perspective to treat the Romanian pluperfect as an anti-periphrastic perfect, albeit one which enters into an intersective relationship with the periphrastic perfective forms.

1.2.2.2 Non-compositionality The latter example involving the Romanian pluperfect and its relationship to other perfective sub-paradigms also exemplifies an important semantic aspect of inflexional periphrasis which has frequently been considered definitional, namely non-compositionality (cf. Ackerman and Stump 2004:142; see also the discussion in §2.4.6 in this volume). Arguably, at a certain level of analysis the periphrastic forms of the perfect exemplified in Table 1.3 can said to be compositional, in that the overall meaning of each of them transparently corresponds to the sum of its individual parts; for instance a present form of the auxiliary a in conjunction with the past participle servit straightforwardly gives rise to a present perfect, in that same way the future form of the auxiliary va fi, itself periphrastic, yields a future perfect in conjunction with the participle, and so forth. The synthetic pluperfect servise, by contrast, does not lend itself to such a transparent compositional analysis between form and meaning, but is characterized by a degree of morphological opacity. At most, one can inferentially recognize a perfective stem servì(cf. preterite servì ‘(s)he served’, participle servìt ‘served’),13 whereas it proves more difficult, at least from a purely synchronic perspective, to transparently associate -s- with past tense.1⁴ There thus emerges an asymmetry across the perfective subparadigms between the transparency of the periphrastic forms and the opacity of 13 We indicate here word stress by the use of the grave accent, although in standard orthography word stress is not marked. 1⁴ Historically, the relevant forms of the Romanian synthetic pluperfect continue the Latin pluperfect subjunctive, cf. seruisset (cf. see Chapter 9 in this volume).

28

adam ledgeway and nigel vincent

the synthetic forms, highlighting how non-compositionality represents a frequent, though not a necessary or definitional, characteristic of inflexional morphology. In view of such tendencies, it is natural to assume, if periphrasis is to be defined in terms of its membership of a structured inflexional paradigm, that it too might develop and show varying degrees of non-compositionality. This view is particularly strongly espoused within PFM, where, within a theory of autonomous morphology, the meaning of a periphrasis is not derived from the contentive information associated with the sum of its individual parts but, rather, is specified morpholexically (Ackerman and Stump 2004:142). This contrasts with frameworks such as DM, which struggles to accommodate non-compositionality, since it predicts a strong one-to-one mapping between form and meaning. And indeed non-compositionality is what we find in many cases of Romance periphrasis. Returning to the Romanian forms of the perfect, we have already noted how under its present perfect reading a servit is transparently compositional, but what we have not yet considered is its preterite uses in all but formal written registers, the outcome of a process of aoristic drift (Schaden 2012; Bertinetto and Squartini 2016:944f.; Bossong 2016:70; cf. also §1.3.2), where the relevant past punctual value cannot be transparently inferred from the sum of its parts (cf. also Spencer 2001:283; Taylor 2011:§18.5.1; Vincent 2011:430). Rather, what we have is a textbook case of a non-compositional reading. Interestingly, in south-eastern varieties of Romanian spoken in Oltenia the synthetic preterite (cf. servi) survives alongside the periphrastic perfect (cf. a servit), but, curiously, the respective past punctual and present perfect values associated with these paradigms in other Romance varieties are reversed (Haase 1995:141; Zafiu 2013:59; Bertinetto and Squartini 2016:942; Maiden 2016a:109). Consequently, as illustrated in (14), the periphrastic forms mark pre-hodiernal values, whereas the synthetic forms convey hodiernal readings. (14)

jeri ̯ dimiˈne̯ aʦә m am skuˈlat m am yesterday morning me=have.prs.1sg raise.ptcp me= have.prs.1sg ɨmbrәˈkat m am spәˈla ʃɨ jo. […] tot aˈʃa dress.ptcp me=have.prs.1sg wash.ptcp and I also thus fәˈkui ̯ ʃɨ ̯ ˈazi ̯ dimiˈne̯aʦә. mә skuˈlai ̯ […] ˈdupә kum do.pfv.1sg and today morning me= raise.pfv.1sg after as spuˈsei ̯ […] mulˈsei ̯ ˈvaka […] fәˈkui ̯ mɨŋˈkare. (Câlnic) say.pfv.1sg milk.pfv.1sg cow.def do.pfv.1sg food ‘Yesterday morning I too got up, got dressed, washed. […] I’ve done the same this morning too. I’ve got up […] as I’ve said […] I’ve milked the cow […] I’ve prepared some food.’

Evidence like this supports the view that the mapping between function and form, be it synthetic or periphrastic (pace Corbett 2007b), is often only weakly correlated and at times even entirely opaque as in the Oltenian case. Yet, at

periphrasis and inflexion

29

the same time it would be disingenuous not to recognize that periphrases often do retain something, if not all, of their original compositional semantics, even in later stages of advanced grammaticalization (Vincent 2011; 2015). It is for this reason that—in contrast to Ackerman and Stump (2004:142)—Haspelmath (2000:661), Brown et al. (2012:§5.3), and Spencer and Popova (2015:211f.) reject the idea that non-compositionality is a sufficient, if not a necessary, condition in defining periphrasis. Rather, we must recognize that morphosyntactic noncompositionality is a possible, though not necessary, by-product of periphrasis which often emerges with the shifting balance between syntactic and morphological properties that accompanies the increasing morphologization and concomitant opacity of multi-word expressions as they are integrated into the inflexional paradigm (cf. discussion of gradience in §1.2.2.5). Finally, we return again to the Romanian periphrastic perfective forms reviewed above. Even under its present perfect reading, a servit and the other periphrastic paradigms of the perfect in Table 1.3 are arguably not strictly compositional. In our discussion so far, we have deliberately overlooked the role of the participle in the various periphrastic formations: despite what our compositional analysis would suggest, there is nothing inherently active about the participle. Rather, the Romance participle is essentially neutral with respect to voice (for extensive diachronic discussion, see Vincent 2011:§19.3.2; cf. also Taylor 2011:406f.), as evidenced by its use in the formation of the passive across Romance, including in Romanian (15). (15)

Ion a /este Ion have.prs.ind.3sg be.prs.ind.3sg ‘Ion has/is served.’

servit. (Ro.) serve.ptcp.msg

Given such active/passive contrasts in the interpretation of the participle, Vincent (1982; 2011:429f.) draws on the origins of the two constructions to argue that the voice value of the participle can be straightforwardly derived from the choice of auxiliary, while still retaining the compositionality of the two readings. In particular, selection of have specifically licenses the active value, which, as the output of a Panini Principle, allows the participle to retain its original default passive value in all other contexts, whether employed in isolation (e.g., un client odată servit … ‘a customer once (having been) served…’) or in conjunction with be or one of a series of other possible passive auxiliaries (see Ledgeway 2021:§3). While this oneto-one transparent mapping between the choice of auxiliary and the voice value of the participle is met in most Romance varieties, the Romanian evidence warns us against assuming such isomorphism as a given. With the exception of the present perfect (cf. 15), the other periphrastic forms of the perfect in Table 1.3 all employ the auxiliary be, viz. fi, in accordance with a distribution of the two auxiliaries arranged along parameters of finiteness (Ledgeway 2014). This immediately raises a

30

adam ledgeway and nigel vincent

problem for Vincent’s account, or at the very least forces us to recognize the emergence of non-compositionality as a result of some quite spectacular readjustments in the distribution of the auxiliaries (Ledgeway 2019:348–354), since the value of the participle cannot be inferred from the choice of auxiliary (cf. further discussion §1.2.3). Indeed, the perfective forms with be in Table 1.3 are genuinely ambiguous in that, alongside their active readings (16a), they also readily license transparently passive readings, at least when the subject is masculine singular (16b).1⁵ (16)

Ion Ion a to

va fi servit // fi / ar Înainte de fut.3sg be cond.3sg be serve.ptcp.msg before of fi servit … (Ro.) be serve.ptcp

a. ‘Ion will/would have served // Before having served…’ b. ‘Ion will/would be served // Before being served…’ The result is an asymmetrical distribution of non-compositional and compositional readings of the active and passive, respectively, where, not accidentally, the non-compositional reading arises precisely in the case of the active, where we know that there have been some quite exceptional changes in the distribution of the Romanian perfective auxiliaries (Ledgeway 2014; 2019), unlike in the passive, whose morphosyntax has essentially remained unchanged. This underlines how in large part the emergence of non-compositionality and concomitant opacity is intimately connected to questions of diachrony (cf. Vincent 2015) and, in particular, to the extent to which the relevant periphrasis has been integrated into the inflexional paradigm (cf. §1.3.2). While DM might be able to conveniently circumvent the problem of the contradictory values of the participle, presumably in terms of underspecification for the feature [voice] with the ‘passive’ value treated as the elsewhere form (cf. the treatment of aspect in the Latin participle in -urus in Embick and Halle 2005), there is no natural way to account for the observed correlations between diachrony and non-compositionality.

1⁵ The participle displays the default masculine singular form in the active, but not in the passive, where it agrees with the surface subject for number and gender. Furthermore, there is no ambiguity in the subjunctive, where, in contrast to the invariant form of auxiliary fi ‘be’ employed in the active, the passive requires an inflected form of auxiliary fi ‘be’. Both of these facts are illustrated in (i.a–b). (i) a. b.

Vor să fi servit Ioana. (Ro.) want.prs.ind.3pl that.irr be serve.ptcp.msg Ioana ‘They want Ioana to have served.’ Vor să fie servită Ioana. (Ro.) want.prs.ind.3pl that.irr be.prs.sbjv.3sg serve.ptcp.fsg Ioana ‘They want Ioana to be served.’

periphrasis and inflexion

31

1.2.2.3 Distributed exponence Further to intersectivity and non-compositionality, the third defining property for periphrasis proposed by Ackerman and Stump (2004:144f.) is distributed exponence,1⁶ which arises in cases where different elements of the component parts of the periphrasis are individually responsible for realizing different parts of its content (cf. also §2.4.4). A particularly revealing example comes from the pluperfect in Ariellese (D’Alessandro and Ledgeway 2010), an Italian dialect of eastern Abruzzo, where alongside the expected periphrastic formation consisting of a past form of the auxiliary in conjunction with the participle (17a), we also find a compound periphrasis (see §1.2.2.4) involving the use of two finite auxiliaries in conjunction with the past participle (17b). (17)

a.

Avive parlate. (Arielli, EAbr.) have.pst.ipfv.2sg speak.ptcp

b. Si’ ’vé parlate. (Arielli, EAbr.) be.prs.2sg have.pst.ipfv speak.ptcp ‘You had spoken.’ In (17b) the higher auxiliary references the features of the subject, also in part through a classic person-based split identical to that found in the present perfect (viz. 1/2 ⇒ be vs 3 ⇒ have; cf. D’Alessandro and Roberts 2010; see also Chapters 7 and 8 in this volume), whereas the lower auxiliary surfaces as a past tense form of have, namely ’vé, a reduced and undifferentiated syncretic form of the singular persons (< avé/avive/avè < habeba-m/-s/-t; cf. Verratti 1998:71, 73), which constitutes the sole exponent of imperfective past tense in the construction (arguably together with singular number). Consequently, Ariellese presents an interesting case of feature-spreading in the discontinuous expression of agreement and tense features across two distinct auxiliaries, in contrast to its non-compound variant in (17a), where both are realized syncretically on a single auxiliary. In this way, the contrast between (17a–b) highlights what probably lies at the heart of Ackerman and Stump’s original observation, namely that the features which are distributed in this way should typically involve those which are otherwise expected (in a given language) to be realized in a monolectal expression (Spencer and Popova 2015:214). This leads us to conclude following Vincent (2011:426) that distributed exponence cannot be considered ‘even a mild signal of a possible periphrasis, leave alone a sufficient condition’ (Brown et al. 2012:263), but simply reflects deviation from a language’s syntactic norms (Spencer and Popova 2015:230). Indeed, from a realizational perspective the occurrence of distributed exponence would a priori seem to favour a more transparent and discrete mapping between meaning and 1⁶ Of course, the use of ‘distributed’ here in the sense of a non-continuous morphological realization is not to be confused or equated with the use of the term in the name of the framework ‘Distributed Morphology’.

32

adam ledgeway and nigel vincent

forms, whereas we have seen in §1.2.2.2 that a strong hallmark of periphrasis is growing non-compositionality. Irrespective of the status of distributed exponence, it strikes us that this might be an area where a framework such as DM is particularly well placed to model the relevant facts, since it has a powerful post-lexical machinery which can readily manipulate the distribution and (multiple) exponence of the relevant features. In particular, while examples such as (17a) arguably involve the PF spell-out of bundled features brought together in the narrow syntax through head movement of the auxiliary through Tense and Agr heads (cf. note 6), (17b) is to be interpreted as the output of a post-syntactic process of ‘Fission’ (Bobaljik 2017:§3.4; Siddiqi 2018:148), which splits a single node and its features into two or more nodes in the morphological representation (for an analysis along these lines, see D’Alessandro and Ledgeway 2010). In principle, distributed exponence is not incompatible with WP approaches either, which can otherwise readily handle morphological examples of distributed and/or extended exponence such as Pt. pãozinho ‘roll’ (< pão ‘bread’ + dim -zinho), which in the plural is marked not only through 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’. By allowing one or more morphosyntactic properties of a given cell to be expressed by realization rules that may apply in more than one block (Stump 2018a:65; 2018b:289), a framework such as PFM is able to handle both distributed and extended exponence and, by implication, similar cases involving periphrastic cells of the paradigm.

1.2.2.4 Compound periphrases A common, though relatively understudied, aspect of periphrasis is compound periphrasis.1⁷ By way of example, in (18) the auxiliary of a given verbal periphrasis, the passive in (18a) and the progressive in (18b), intersects, in turn, with the grammatical feature perfect, which itself is realized periphrastically, namely era (> era estat) aidat and so (> so istatu) travallande. (18)

a.

Qu’ era estat aidat per un that be.pst.ipfv.ind.3sg be.ptcp.msg help.ptcp.msg by a amic. (Gsc.) friend ‘He had been helped by a friend.’ (Darrigrand 1974:117) b. So istatu travallande. (Srd.) be.prs.ind.1sg be.ptcp.msg work.ptcp ‘I’ve been working.’ (Jones 1993:141)

The embedding of one periphrasis under another in this way presents challenges for standard definitions of periphrasis and, in particular, for the question of how 1⁷ See, however, the extensive discussions of the so-called temps surcomposés (‘double compound tenses’) in Vincent (2011:430–432; 2014:12f.; 2015:111f.).

periphrasis and inflexion

33

the boundary between syntax and morphology is to be drawn. On the one hand, recursion of the type that we see here is canonically taken to be a hallmark of syntax, hence offering us a simple way out if we go down the avenue of a DM-style analysis where morphology is ultimately syntax all the way down. On the other, we have adduced incontrovertible evidence above for the view that periphrasis should be understood in terms of a paradigmatic convergence between syntax and morphology. This therefore requires us to accommodate compound periphrases within a theory of inflexional morphology, ultimately to be interpreted as the result of the cross-categorization of morphosyntactic and morphosemantic features (Brown et al. 2012:269f.). Essentially, this amounts to treating compound periphrasis, not as simplex periphrasis embedded within syntax, but, rather, as complex multi-word expressions which realize two or more intersective featural relationships within a given cell of the paradigm. This leads us to expect compound periphrasis to unite in greater or lesser measure aspects of syntactic and morphological properties so as to display exactly the same characteristics (e.g., intersectivity, non-compositionality, distributed exponence) that we have identified above for cases of simplex periphrasis. A priori there is no reason to believe that this is not the case. For example, we have already observed how the examples in (18a–b) combine a primary intersection of features, e.g., passive and third-person singular or progressive and first-person singular, with a subsequent intersective relation with the feature perfect. At the same time, both examples display distributed and extended exponence with, for example, the person features of the subject encoded on the finite auxiliary and its gender specification on the participial auxiliary and, in the case of (18b), also on the lexical participle. Furthermore, although the result in (18a–b) is arguably compositional, there are many examples of compound periphrasis where this is not the case, witness the Romanian contrast in (19a–b). (19)

Ion a fost servit. (Ro.) Ion have.prs.ind.3sg be.ptcp.msg serve.ptcp.msg a. ‘Ion has been served.’ b. ? ‘Ion had served.’

Demonstrably, the structure in (19) transparently delivers a present perfective passive interpretation, as indicated by its translation in (19a). However, for many speakers in a large dialectal area (including North Moldavia, Maramureș, Crișana, Transylvania, Banat, and Muntenia), as well as in earlier stages of the written language, the structure in (19) is also compatible with the substandard active pluperfect reading given in (19b).1⁸ While the passive variant in (19a) licenses a 1⁸ For further details, see Densusianu ([1938] 1961:144), Nevaci and Todi (2009:142), Dragomirescu (2012:208f.), Pană Dindelegan (2012:568), Zafiu (2013:34f.), Ledgeway (2014:20f.), and Chapter 9 in this volume.

34

adam ledgeway and nigel vincent

one-to-one form–function compositional reading, the same cannot be said of the variant in (19b), where the same structural template with the exact same lexicalizations yields an active non-compositional interpretation which cannot be directly read off the sum of its component parts, exactly along the lines of what we observed in (16a–b).

1.2.2.5 Gradience If periphrasis is to be modelled in terms of structured paradigms, as maintained here, then a reasonable expectation is that periphrastic expressions should share other typical properties of inflexional morphology, as well as syntax. Indeed, in the canonical case periphrasis ‘is canonical syntax but within the paradigmatic organization of canonical morphology’, such that ‘some properties of periphrastic constructions follow from their syntactic nature and some from their morphological nature’ (cf. Brown et al. 2012:245). Rarely, if ever, however, are actual cases so canonical, with individual examples exhibiting varying degrees of morphological and syntactic interdependence, the balance of which is often subject to change over time (cf. Vincent 2011:424), especially as original multi-word expressions are increasingly integrated within the inflexional paradigm as their grammatical ingredients progressively move down the word > clitic > affix cline (Hopper and Traugott 2003:7), a typical effect of grammaticalization (see §1.3.1). By way of example, consider the contrasts in (20)–(21). (20)

(21)

a.

Son benguts / An be.prs.ind.3pl come.ptcp.mpl have.prs.ind.3pl las fenestras. (Périgord Occ.) the.fpl windows.f (Miremont 1976:54f.) b. Han venido /Han have.prs.ind.3pl come.ptcp.msg have.prs.ind.3pl las ventanas. (Sp.) the.fpl windows.f ‘They have come / They have closed the windows.’

barradas close.ptcp.fpl

cerrado close.ptcp.msg

a.

A trôp travalhat. (Périgord Occ.) have.prs.ind.3sg too.much work.ptcp.msg (Miremont 1976:147) b. Ha (**demasiado) trabajado demasiado. (Sp.) have.prs.ind.3sg too.much work.ptcp.msg too.much ‘(S)he has been working too much.’

All Romance varieties show, with varying degrees of grammaticalization and different shades of meaning, a perfective periphrasis built on the auxiliary be and/or have in conjunction with the participle, a construction which ultimately has its roots in Latin (for in-depth discussion and bibliography, see Ledgeway 2012:130–134). As the example in (20a) illustrates, Occitan, together with French and many Italo-Romance varieties, including Italian, shows a rather conservative

periphrasis and inflexion

35

behaviour, in that the periphrasis is still transparently organized around an original active–stative alignment in which the syntax is sensitive to the semantic nature of the available arguments (Ledgeway 2012:319–326). Thus, the auxiliary alternates in accordance with the semantic nature of the surface subject, namely, A/SA ⇒ have vs SO ⇒ be (Vincent 1982; Loporcaro 2016:802f., 812f.; Ledgeway 2019:364–366), and the participle continues to agree with (various subclasses of) stative arguments, viz. O or SO (Smith 1991; 1996; 1999; Loporcaro 1998:64–78; 2016:§49.2.3; Manzini and Savoia 2005:II:§5.1; Ledgeway 2012:3326f.). The Spanish example (20b), by contrast, which is representative of Ibero-Romance as a whole as well as modern Romanian, displays a more innovative behaviour in line with a core nominative–accusative alignment, such that auxiliary alternation has given way to the generalization of have and participle agreement has disappeared entirely. We thus observe in the contrast (20a–b) a progressive desyntacticization of the perfective construction in a process of increasing morphologization (Joseph and Janda 1988; Joseph 2008). The latter strips away the former alternations of the syntax proper in favour of non-alternating forms in line with the growing cohesion of the composite parts of the periphrasis, as it moves along a cline from periphrasis to inflexion and the borders between the two become increasingly fuzzy. This process of univerbation manifested in the increased integration between the individual elements of the periphrasis is further visible in the contrast between (21a) and (21b), where the adjacency of auxiliary and participle may still be interrupted in Occitan by the interpolation of elements such as adverbs, but not in Spanish, where the effects of desyntacticization have been greater (cf. also the discussion in §2.4.5 in this volume). Ultimately, in the latter case the fusion between an albeit still fully inflected auxiliary and participle gives rise to a distribution and behaviour of the periphrasis more akin in many respects to monolectal expressions. Similar developments in the shifting balance between syntactic and morphological properties, arguably reaching a more advanced stage of desyntacticization and morphologization, are also observable in the progressive periphrasis of many dialects of Apulia. In these varieties the progressive originally involves a pseudocoordinate structure in which a reflex of stare ‘stand’ is coordinated by means of the conjunction ac ‘and’ (itself often subject to surface deletion, though still signalled by initial consonantal lengthening of the following word) with a finite lexical predicate, as illustrated in Table 1.4. As shown in Ledgeway (2016a) and Andriani (2017:ch 5), the forms of stare exhibit varying degrees of inflexional attrition (cf. also Manzini and Savoia 2005:I:688–701). In particular, it is possible to identify three patterns, which display increasing degrees of inflexional attrition as we move southwards across the region from Pugliese to Salentino dialects. For instance, in the Pugliese dialect of Putignano (province of Bari) in (a) in Table 1.4 stare displays a near full inflexional paradigm (albeit with syncretisms in the second-/third-person singular and first-/second-person plural), whereas in the more southerly Pugliese dialect

36

adam ledgeway and nigel vincent

of Martina Franca (province of Taranto) in (b) in Table 1.4 the paradigm shows reduced agreement with distinctive forms, based on the stem stɔ(-), in just the firstperson singular and the third-person plural according to a morphomic U-pattern widespread across Romance (cf. Maiden 2011a; 2016b:§43.2.3). More drastic are Salentino dialects such as Leccese in (c) in Table 1.4, where stare has lost all agreement and is now reduced to an invariable aspect marker occurring as a bare stem, a clear counterexample to Spencer and Popova’s (2015:208) claim that periphrasis is not expected to produce uninflectability (cf. also discussion in §1.2.3 and §3.5).1⁹

Table 1.4 Aspectual pseudo-coordinate periphrases with stare ‘stand’ in Apulia 1sg 2sg 3sg 1pl 2pl 3pl

(a) Putignano ‘do’ stok a fˈfattsә ste fˈfaʃә ste fˈfaʃә sta ffaˈʃeimә sta ffaˈʃeitә ston a fˈfaʃәnә

(b) Martina Franca ‘call’ stɔ cˈcεmә stε cˈcεmә stε cˈcεmә stε ccaˈmεmә stε ccaˈmεtә stɔnә (a) cˈcamәnә

(c) Lecce ‘lose’ sta pˈpεrdu sta pˈpεrdi sta pˈpεrdε sta ppεrˈdimu sta pˈpεrˈditi sta pˈpεrdεnu

These synchronic cross-dialectal patterns of inflexional attrition may also be read vertically as evidence to reconstruct different diachronic stages in the progressive desyntacticization of the periphrasis. In particular, this morphological reflex highlights an important change in the nature of stare, inasmuch as a full inflexional paradigm constitutes the hallmark of what it means to belong to the lexical category of verb in Romance. Consequently, any attrition in the inflexional paradigms of stare can be taken to represent a weakening in its defining verbal characteristics and, at the same time, to signal a concomitant change in its category from auxiliary verb to functional prefixal marker. Indeed, the prefixal nature of stare in the final stages of this development is clearly revealed by the fact that, while in some dialects object clitic pronouns may still surface between invariant sta and the finite lexical verb (e.g., Scorrano sta te visciu ‘sta.prog you= I.see (= I can see you)’), in more progressive dialects clitics must precede sta, which cannot be separated from the finite verb by any intervening material (e.g., Matino ci sta penzu ‘to.it= sta.prog I.think (= I’m thinking about it)’).

1⁹ Another example of Romance uninflectability already seen is the Romanian invariant be stem fi used for all persons in the active perfect subjunctive (cf. Table 3), in contrast to the inflected forms found in the present passive (see note 15 for examples).

periphrasis and inflexion

37

Examples like these, together with those in (20)–(21), serve to remind us that periphrasis is not an immutable, absolute category. Rather, it must be conceived of as a fluid label that can be equally applied to a range of multi-word expressions showing varying degrees of greater or lesser core syntactic and morphological properties, a view championed in approaches such as Canonical Typology (Brown, Chumakina, and Corbett 2013). Given this gradient interpretation of periphrasis (cf. similar remarks about analyticity in Vincent 1997a:100; Ledgeway 2012:§2.2.2; 2017a:§3.2), we can readily accommodate both synchronic variation across a range of unrelated examples of periphrasis, as well as diachronic variation in related examples of periphrasis within the same language (family). Ultimately, changes in the balance between syntactic and morphological properties in the compositional make-up of individual periphrases, whereby we note over time a tendency towards desyntacticization and concomitant growing morphologization, are to be understood as a direct effect of paradigm membership. As members of a paradigm in which they realize one or more cells, the relevant periphrases enter into direct opposition with synthetic inflexional forms with which they come to be inextricably bound both formally and functionally. A priori growing morphologization is therefore expected as a natural consequence of this paradigmatic interdependence which aligns originally syntactic elements alongside synthetic inflexional elements within the domain of morphology. By contrast, in configuration-based lexical approaches to morphology such as DM the relationship between periphrases and their related inflexional forms is not formalized by way of a paradigm, such that any such relationships are purely accidental, as are observed diachronic tendencies towards desyntacticization and concomitant morphologization, which can only be stipulated, but not captured naturally.

1.2.3 Paradigmaticity A considerable amount of the Romance data that we have discussed so far in relation to periphrasis unquestionably point to the need to make reference to a set of paradigmatically related forms, a conclusion which is out of the reach of lexicalrealization theories like DM. A natural consequence of placing periphrasis firmly within the realm of inflexional morphology in this way is that it leads us to expect periphrasis to display many of the recurrent features characteristic of morphology and, in particular, inflexional paradigms. Pace Spencer and Popova (2015:208), this prediction, we contend, is in large part confirmed by the Romance data (cf. also Corbett 2012:180–186), a brief survey of which follows. We begin with Spencer and Popova’s (2015:208) claim that it is difficult to imagine cases of periphrastic defectivity and uninflectability on a par with the equivalents of defective and uninflectable lexemes, since the exponence of periphrasis is through syntax, which, unlike morphology, is held to be less prone

38

adam ledgeway and nigel vincent

to such irregularities (cf. also Kiparsky 2005:§2.2). However, we have already seen in relation to the Salentino progressive, an erstwhile pseudo-coordination structure (cf. Table 1.4), how grammaticalization of the construction over time leads to differing degrees of inflexional attrition and erosion of the auxiliary stare ‘stand’, ultimately resulting in its uninflectability in varieties such as Leccese, where it has been reduced to an invariable preverbal affixal marker of aspect (viz. sta).2⁰ These same varieties simultaneously exemplify defectivity—or ‘underexhaustivity’ in the words of Spencer (2006:291; cf. also Vincent 1987:241; Haspelmath 2000:661; Baerman 2015a:§7.4)—since, as shown in detail by Andriani (2017:Chapter 5), many dialects in the Pugliese–Salentino continuum show defective paratactic paradigms like those exemplified in Table 1.5, where we observe arbitrary feature restrictions on the distribution of the paratactic pseudo-coordinate structure. More specifically, the shaded cells indicate the distribution of the paratactic construction and the unshaded cells the paradigmatic gaps, which are realized instead by a hypotactic construction consisting of fully inflected stare + a ‘to’ + infinitive. Table 1.5 Defective pseudo-coordinate periphrases with stare ‘stand’ in Pugliese (Andriani 2017:ch.5)

1sg 2sg 3sg 1pl 2pl 3pl

(a) Gravina

(b) Rutigliano

(c) Conversano

stóuc’a ssuné sté ssunә sté ssóunә stәm’a ssuné stәt’a ssuné stònn’a ssuné

stɔŋg a ʃˈʃɔ:uk ste ʃˈʃuk ste ʃˈʃɔ:uk stәm a ʃәˈkwe stәtә a ʃәˈkwe stɔnn a ʃәˈkwe

stéc’a ssonә st’a sunә st’a sónә stém’a ssәné stét’a ssәné stànn’a ssònәnә

In addition to illustrating periphrastic defectivity, the paradigms in Table 1.5 also highlight three further typical properties of inflexional morphology within a periphrastic structure. The first is suppletion (cf. Vincent 1987:242; Anderson 2011:§3.2), in that the three paradigms in Table 1.5 variously show gaps in the distribution of the paratactic construction in specific persons which are filled suppletively by the hypotactic structure.21 This blended distribution between 2⁰ Also relevant here are aspectual periphrases of the tornare type discussed by Parry (Chapter 3 in this volume), which in some Romance varieties have also been reduced to invariable aspectual verbal markers (cf. Cos. torna ‘again’, Sal. ntorna ‘again’). In the case of Salentino, we can even find sequences such as (i), in which both invariant preverbal markers may co-occur. (i) Ntorna sta trona! (Matino) again prog thunder.prs.3sg ‘It’s thundering again!’ (Fiorentino 1998:53) 21 For another interesting case of periphrastic suppletion, see the discussion of the distribution of the Italo-Romance passive auxiliaries essere ‘be’ and venire ‘come’ in Vincent (1987:248f.) and Ledgeway (2021:§3.3).

periphrasis and inflexion

39

paratactic and hypotactic periphrastic forms follows arbitrary restrictions on feature realization characteristic of morphological paradigms but not, significantly, of syntactic constructions. The second property evidenced by the paradigms in Table 1.5 regards the extension of the paratactic paradigm according to a morphomic pattern (see also Chapters 3 and 6 in this volume). In particular, we see that in its earliest stages (cf. Gravina) it involves just the second-/third-persons singular, but can subsequently extend to the first-person singular (cf. Rutigliano) to yield a singular–plural contrast within the paradigm—itself a frequent split in inflexional paradigms—from where it can then be further extended to include also the third-person plural (cf. Conversano), a classic example of Maiden’s Npattern (Maiden 2016b:712–716), before finally extending to the whole paradigm (cf. Table 1.4). The third and final significant property of inflexion evinced by the paradigms in Table 1.5 is heteroclisis. Given traditional definitions of heteroclisis (cf. Stump 2006; 2015:§6.4.2.3) such as ‘the property of a lexeme whose inflectional paradigm contains forms built upon stems belonging to two or more distinct inflection classes’ (Stump 2006:279), it is tempting to view the blending of paratactic and hypotactic forms above as part of a heteroclitic distribution. Another frequent property of inflexional paradigms and, in particular, declensional classes, is syncretism (cf. Baerman 2015a:§7.2). If it can therefore be shown to also occur in periphrases, this will provide additional proof of the paradigmatic nature of periphrasis (Vincent 1987:241f.). Above we have already seen several examples, but perhaps the most spectacular involves examples such as (19a–b), repeated here in (22a–b). (22) Ion a fost servit. (Ro.) Ion have.prs.ind.3sg be.ptcp.msg serve.ptcp.msg a. ‘Ion has been served.’ b. ? ‘Ion had served.’ It is striking to note that this Romanian example comes very close to the other frequent example of periphrastic syncretism discussed in the literature (cf. 23), namely the active–passive syncretism found in the progressive of some (obsolete) varieties of English (cf. Anderson 2011:116). (23)

John’s house was building. a. ‘John’s house was built.’ b. ‘John was building his house.’

We turn now to the opposite of syncretism, overabundance (Thornton 2011; 2016; 2019b; Cappellaro 2018; see also Chapter 11 in this volume). In contrast to Ackerman and Stump (2004:126f.), Brown et al. (2012:245) maintain that we must allow for the possibility that periphrasis and synthesis may compete for the realization of a given cell (cf. also Haspelmath 2000:659; Taylor 2011:405). They cite

40

adam ledgeway and nigel vincent

the example of definiteness marking in Scandinavian, where a suffixed form of the noun competes with combinations of a free-standing article + noun, and conditions imposed by the syntax determine which is chosen. Remarkably, this mirrors again what we have already seen for Romance, in this case the alternation between suffixal and free-standing prenominal forms of the definite article in Romanian (§§1.2.1–1.2.1.2) in such alternations as prieteni-i ‘friends.m-def.nom-acc.pl’ vs cei trei prieteni ‘the.nom-acc.mpl three friends.m’. Unlike in the Scandinavian case, however, in some contexts the alternation is not determined exclusively by conditions of the syntax, but also by distinctions of style and register (cf. also Spencer and Popova 2015:224), witness the alternation between standard Ane-i ‘Ana.gen-dat-def.gen-dat.fsg’ and substandard ? lui Ana ‘the.gen-dat.sg Ana.def.nom-acc.fsg’ (for detailed discussion, see also Chapter 10 in this volume). We can also draw a further distinction here between instances like those just considered (e.g., -i vs cei), where there is competition between syncretic and periphrastic expressions in the realization of a given paradigmatic cell, and instances where there arises an alternation between two (or more) periphrastic forms competing for the realization of a given paradigmatic cell. An obvious example here is the Catalan periphrastic go-preterite (cf. Chapter 4 in this volume ), where the function word in the periphrasis, viz. auxiliary anar ‘go’, shows multiple competing forms in accordance with complex stylistic and diatopic variation (cf. Wheeler, Yates, and Dols 1999:301f., 307, and the discussion in §3.3 in this volume) including, among others, vaig/? vàreig menjar ‘go.aux.ind.1sg eat.inf (= I ate)’ and vam/? vàrem/? vem menjar ‘go.aux.ind.1pl eat.inf (= we ate)’. Closely related to overabundance is the phenomenon of overexhaustivity, which obtains whenever ‘a periphrastic expression gives rise to more combinations than might be expected given the ordinary rules of syntax in a language’ (Spencer 2006:291). A possible example here involves once again the Catalan periphrastic go-preterite, which in the subjunctive enters into competition with the synthetic forms of the past subjunctive, e.g., vagi menjar ‘go.aux.sbjv.1sg eat.inf’ vs menjés ‘eat.pst.sbjv.1sg’. While in the indicative alternation between periphrastic and synthetic preterite forms is purely a question of stylistic and diatopic variation (Wheeler, Yates, and Dols 1999:343), inasmuch as their semantic values and functions overlap exactly (24a–b), in the subjunctive there arises a partial aspectual distinction (25a–b; Wheeler, Yates, and Dols 1999:380f.; and discussion in §3.3 in this volume).

(24)

a.

S’ acostà a la porta i self= draw.pfv.3sg to the door and va trocar-hi dos cops. (Cat.) go.aux.ind.3sg knock.inf=there two blows

periphrasis and inflexion

41

b. Es va acostar a la porta i hi self= go.aux.ind.3sg draw.inf to the door and there= trucà dos cops. (Cat.) knock.pfv.3sg two blows ‘He went up to the door and knocked twice.’ (25)

a.

És impossible que ho escrivı´s be.prs.ind.3sg impossible that it= write.pst.sbjv.3sg ‘It’s impossible that she was writing/wrote it.’ b. És impossible que ho vagi be.prs.ind.3sg impossible that it= go.aux.sbjv.3sg ella. (Cat.) she ‘It’s impossible that she wrote it.’

ella. (Cat.) she escreure write.inf

The synthetic past subjunctive form in (25a) is aspectually ambiguous between the imperfective and perfective readings in accordance with the typical Romance pattern, whereas the periphrastic formation in (25b) introduces a novel perfective distinction which is otherwise neutralized in other Romance varieties. Consequently, from a comparative Romance perspective, though not necessarily from a Catalan-internal one, the periphrastic preterite gives rise to more aspectual combinations than would ordinarily be expected given canonical Romance restrictions on the formal realization of aspect in subjunctive paradigms. We conclude this section with a brief remark about the possibility of pure periphrastic morphology. Vincent (2011:434) asks the question whether there could be anything such as ‘pure periphrastic morphology, that is to say situations where a non-finite form occurs in a range of periphrases which are not linked by common features and indeed may even have contradictory feature values’. He hypothesizes that one possible candidate is the Romance perfect participle, which licenses an active reading in the perfect but passive (or, perhaps better, stative) reading elsewhere, but he ultimately dismisses this possibility on the assumption that their respective values can be derived from the effects of the different auxiliaries they combine with (cf. also discussion in §1.2.2.2). While such arguments hold for many standard Romance varieties where there is a robust correlation between auxiliary choice and voice value (cf. 20–21), we have already seen above that, with the exception of the present perfect, the sub-paradigms of the Romanian perfective (cf. Table 1.3) and the substandard pluperfect (cf. 16a–b/19a–b) are genuinely ambiguous, at least when the subject is masculine singular, between active and passive readings. Similarly, the correlation between auxiliary choice and the active/passive interpretation of the participle fails to show a one-to-one mapping in many central–southern dialects of Italy (see Chapters 7 and 8 in this volume), where, among other things, auxiliary choice in the active can also be determined

42

adam ledgeway and nigel vincent

by mood (±realis), tense (±past), person (typically, though not exclusively, 1/2 vs 3), and/or a combination of these (for discussion, see Ledgeway 2019). By the same token, in some Pugliese and south-eastern Basilicatese dialects the auxiliary of the passive can even be have (26), a rather extraordinary development from a cross-linguistic perspective (Loporcaro 1988:291–299; 2012:179–180; Ledgeway 2021:§3.4). (26)

lәlɔčč e g˘g˘әsεppә non a´vәnә paɣeːtә. (Irsina) Raffaele and Giuseppe not have.prs.3pl pay.ptcp ‘Raffaele and Giuseppe are not (being) paid.’ (Loporcaro 2012:179)

Given these observations, there is then an a priori case for taking Vincent’s original supposition seriously to make some allowance for some, albeit limited, cases of pure periphrastic morphology.

1.3 Diachrony Taken together, the case studies in the present volume demonstrate the essential complementarity of synchronic and diachronic perspectives on the description and explanation of linguistic phenomena. And already in this chapter we have seen diachronic and synchronic factors interacting in various ways. In this section, we address the historical side of the question in more detail, focusing on four themes which are especially pertinent to understanding the way periphrasis and inflexion develop and interact over time both in the specific case of Romance and more generally in cross-linguistic terms.

1.3.1 Grammaticalization and directionality The obvious place to start is with the phenomenon of grammaticalization both because of its centrality in debates over the last four decades and because many of the classic examples are drawn from Romance (see also the discussion in §2.4.9 in this volume): the emergence of definite articles from distal deictics, future periphrases incorporating a go verb, progressives built out of the stand verb plus a gerund, locative prepositions assuming a complementizing function, and the like. At the same time, Romance is also home to developments that appear to go against the standard predictions, as with the Cat. anar ‘go’ + infinitive (see Chapters 3 and 4 in this volume) as the expression of simple past exemplified in (24) and (25) or the Lombard and Piedmontese tener ‘hold, keep’ + perfective participle expressing continuative or iterative meaning as in (27) compared to the use of the non-past -ing form in its English translation (see Vincent 2014:14f. for discussion and further references):

periphrasis and inflexion

(27)

Ten spendasciaa. (OMil.) keep.prs.3sg waste.money.ptcp ‘He keeps on wasting money.’

43

(G.P. Lomazzo 1589)

The paradox is that while grammaticalization is a key factor in the development of periphrases such as these and others investigated in this volume, the diachronic dimension does not figure greatly in the literature on this class of constructions. It is, for example, only mentioned in passing in the survey by Spencer and Popova (2015) and not at all in the essays collected in Chumakina and Corbett (2012). We hope that the contributions here will go some way towards building the necessary bridge between the history and structure of these items. To begin with, it is important to be clear about what we mean by the term ‘grammaticalization’ since, beside the extensive literature on the topic, there is a recurrent strand of criticism and even denial of its very existence. Thus, Joseph (2011; 2020), while agreeing that there certainly exists a phenomenon that deserves the name grammaticalization, objects to labelling it a process or a mechanism and prefers instead to call it simply the result of independently motivated mechanisms of change. In other words, his definition is static and synchronic, and hence his suggestion that structures of the kind we have been discussing here should be labelled ‘more grammatical’ and not ‘more grammaticalized’ (2011:196) when compared with their historical source constructions. We will instead suggest that the Romance data strongly confirm the view that there is a mechanism of grammaticalization with its own distinctive properties, and one for which such a family of closely related languages with its rich attestation over a long time span, provides an ideal research environment. We do, on the other hand, agree with Joseph and others that grammaticalization is not a theory in the usual sense of the term but rather a phenomenon in need of appropriate theoretical modelling (Vincent and Bo¨rjars 2010). Other relevant aspects of the questions Joseph raises, in particular those concerning the role of language contact and reconstruction, will be addressed in §1.3.3 and §1.3.4.22 What then are the properties of the grammaticalization process and how do the Romance data help to confirm its validity as a concept? We will start with the issue of the directionality of change, which in turn breaks down into two sub-questions: (a) are the directions unidimensional? And (b) is there evidence of a reverse type of change, so-called degrammaticalization? For both, the history of the Romance languages provides valuable evidence. To deal with the latter first, what is striking is the absence of any convincing instances within Romance. As Willis (2017)

22 One aspect of Joseph’s critique that we will not discuss is his concern that too much attention has been devoted to the topic by comparison with other aspects of language change. Fair or otherwise, this is a charge against the scholarly community and the sociology of the discipline but does not impinge on the nature of the phenomenon in and of itself.

44

adam ledgeway and nigel vincent

shows, there are inherent reasons why a reverse change from grammatical to lexical, if possible at all, is likely to be a rare event but even so, given the richness of the Romance empirical base, it is somewhat surprising that few if any examples to put beside Bulgarian nešto ‘something’ > ‘thing’ (pronoun > noun), Pennsylvania German wotte ‘will’ > ‘want’ (auxiliary > lexical verb), and the like have been highlighted in any of the Romance languages or dialects to date. Negative evidence of this kind cannot be conclusive, but it does tend to confirm the view that degrammaticalization is intrinsically a rare and unlikely diachronic development, and certainly not one which undermines the existence of grammaticalization as a phenomenon in its own right. Let us return to changes which follow Meillet’s classic route from lexical to grammatical or, in Kuryłowicz’s alternative formulation, from lexical to less grammatical to more grammatical. Most of the relevant literature concentrates on recurrent single trajectories defined in semantic terms. Thus, the much cited typological study by Bybee et al. (1994) contrasts futures derived from agent-oriented modalities such as desire and obligation with those derived from movement verbs and notes that both are well attested in languages and language families all around the world. Perhaps not surprisingly, then, Romance is home to examples of all these subtypes. Romance go futures have been much discussed and further treatment is to be found in Chapters 3 and 4 in this volume. A future periphrasis built on a verb meaning ‘desire’ can be seen in Ro. voi pleca ‘I will leave’ with the auxiliary voi < Lat. uolo ‘I want’, and one built on an obligation verb occurs in Srd. des àer cantau ‘you.sg will have sung’ with des < Lat. debes ‘you.sg owe, must’. Bybee et al. (1994:269) further note sub-patterns such as the tendency for futures derived from movement verbs to express immediate or incipient futurity, as with the English imminent It’s going to rain in a minute beside the more generally predictive It will rain. A similar contrast has been noted for the French future periphrasis with aller ‘go’ + infinitive in contrast to the synthetic future (see Mooney 2020 and references there). An alternative approach starts not from a grammatical endpoint like futurity but from a particular lexical concept such as verbs meaning go or come and tracks the range of possible developments, which can be quite varied. For instance, in Figure 1.1 we reproduce the semantic map compiled by Lichtenberk (1991:499) on the basis of a survey of developments of the go verb in Oceanic. In discussing trajectories like this Lichtenberk (1991:476) uses the term ‘heterosemy’, in his words, ‘to refer to cases (within a single language) where two or more meanings or functions that are historically related, in the sense of deriving from the same ultimate source, are borne by reflexes of the common source element that belong in different morphosyntactic categories’. In fact, however, he goes well beyond a single language and demonstrates how the historically

periphrasis and inflexion

45

conditional (protasis) temporal sequencer

temporal distance

persistive

spatial distance

conditional (apodosis)

relative remoteness

conditional (protasis)

‘go’

allative

future tense

past time

Fig. 1.1 Grammaticalization of go forms

connected forms can be found within a whole family of related languages. In similar vein, Craig (1991) uses the term ‘polygrammaticalization’ to characterize the developments of the go verb in the Nicaraguan language Rama. It is instructive to compare the patterns attested in Figure 1.1 with what we find in Romance. In addition to the go pasts and futures already alluded to, there is the Italian progressive or continuative in (28a), which is close in meaning to the Houaïlou example in (28b) (= Lichtenberk’s (34)), which he labels as ‘persistive’: (28)

a.

Gli andavo dicendo che … (It.) 3msg.dat= go.ipfv.1sg say.ger comp ‘I was saying to him that …’ b. Na cuε vi. (Houaïlou) go he sit ‘He goes on sitting (there).’

The difference is that the Italian (28a) refers simply to an ongoing activity while (28b) also implies the beginning and the continuing of the activity, as indeed does the English go on construction. That this distinction is also available in Romance can be seen in the Spanish minimal pair in (29), due originally to Ignacio Bosque and discussed by Bertinetto (2000:577f.): (29)

a.

Juan estuvo colocando libros de 3 a Juan be.pfv.3sg place.ger book.pl from 3 to ‘Juan was placing books (on the shelf) from 3 until 5.’ b. Juan fue colocando libros de 3 a Juan go.pfv.3sg place.ger book.pl from 3 to ‘Juan kept putting books back (on the shelf) from 3 until 5.’

5. (Sp.) 5 5. (Sp.) 5

46

adam ledgeway and nigel vincent

The story is complicated here by the fact that fue in (29b) is the suppletive past of ir ‘go’ as well as being the past of ser ‘be’, but in Bertinetto’s words ‘while (a) simply depicts a durative situation, (b) adds to it the idea that the event be conceived of as a sequence of identical gestures […] To put it differently, (a) describes a static scenario while (b) presents a dynamic one’. Romance not only confirms the patterns identified by Bybee, Craig, Lichtenberk, and others but also adds to them, as with the Italian passives built on go + past participle in (30): (30)

a.

Questo libro va letto. (It.) this book go.prs.3sg read.ptcp ‘This book must be read.’ b. La casa è andata distrutta durante il def house be.prs.3sg go.ptcp destroy.ptcp during def terremoto. (It.) earthquake ‘The house was destroyed during the earthquake.’

Here too there are sub-developments. In (30a) not only does andare ‘go’ serve as the passive auxiliary but a modal component of obligation is introduced. In (30b), on the other hand, andare + past participle expresses change of state but with a semantic restriction to negative or undesirable outcomes and hence the ungrammaticality of **la casa è andata costruita ‘lit. the house has gone built’. The modal component in (30a) is associated with imperfective meaning but is not present in combination with completives as in (30b). Moreover, as Mocciaro (2014) shows, the beginning of these patterns is to be seen in texts from the thirteenth and fourteenth centuries and provides evidence for two different diachronic trajectories. She further shows that, contrary to what is claimed by Giacalone Ramat (2000), these trajectories develop after the go verb had developed as a passive auxiliary. Examples like (30a) with a modal interpretation are not attested before the sixteenth century, while the plain passive use is older as in (31): (31) a. andatagli la elezione confirmata go.ptcp.fsg=3msg.dat def.fsg election.fsg confirm.ptcp.fsg dal papa (OIt.) by.def pope ‘his election having been confirmed by the Pope.’ (Sacchetti Trecentonovelle, 14th c.)

periphrasis and inflexion

47

b. se il detto notaio […] andasse if def.msg said.msg lawyer.msg go.pst.sbjv.3sg electo o vero nominato (OIt.) elect.ptcp.msg or indeed nominate.ptcp.msg ‘if the said lawyer was elected or indeed nominated.’ (Statutes, Siena, 13th c.) c. va più velata che nulla scienza (OIt.) go.prs.3sg more veil.ptcp.fsg comp no science ‘it is more veiled than any other science.’ (Dante, Convivio 1304–1307) We see here the co-occurrence of andare in a passive with a wider range of lexical items than is found in the modern language. Mocciaro (2014:64) interprets the (31a) type as ‘a focus shift’ from ‘result reached through a path’ to ‘movement towards a result which is not yet reached and, for metonymical contiguity, should be reached’, with the modal component now an integral part of the construction in the modern language. The (31b) development, by contrast, is part of a wider cross-linguistic pattern whereby the go verb participates in idiom patterns with seemingly arbitrary semantic restrictions. Compare English, where one can say I told her the news and she went wild/nuts/quiet/silent, but not **she went noisy/angry/happy, etc. Not only do examples like these show how Romance can contribute to the wider typological picture, they also demonstrate how such changes can be tracked over time in a way that is not possible for languages lacking a rich historical documentary tradition. Lichtenberk offers a similar though less diversified profile for come verbs, as illustrated in Figure 1.2.

Agent of passive

locative-distance

standard of comparison

spatial source

ingressive

continuative

‘come’

future time

venitive

relative closeness

Fig. 1.2 Grammaticalization of come forms

48

adam ledgeway and nigel vincent

And here once again Romance both duplicates and complements the patterns he describes. In particular, we find the come verb evolving as a passive auxiliary in both Italian (32a) (Giacalone Ramat and Sansò 2014) and Romanian (34b) (Dragomirescu and Nicolae 2014): (32)

a.

La questione veniva discussa ieri. (It.) def.fsg matter.fsg come.ipfv.3sg discuss.ptcp.fsg yesterday ‘The matter was being discussed yesterday.’ b. Casa aceea vine as¸ezată aici. (Ro.) house.def that.fsg come.prs.3sg place.ptcp.fsg here ‘The house is/will be/should be placed here.’

In Italian, the venire ‘come’ + ptcp construction expresses an event and not a state, in contrast to the be passive in many languages, including Italian, which typically has both values. However, unlike the be passive, it is restricted in its distribution to the inflexional but not the periphrastic forms hence the contrast between (33a) and (33b): (33)

a.

La questione è stata discussa def.fsg matter.fsg be.prs.3sg be.ptcp.fsg discuss.ptcp.fsg ieri. (It.) yesterday ‘The matter was discussed yesterday.’ b. **La questione è venuta discussa def.fsg matter.fsg be.prs.3sg come.ptcp.fsg discuss.ptcp.fsg ieri. (It.) yesterday

The sequence è venuta by itself is perfectly grammatical with the meaning ‘She has come’ so the ungrammaticality of (33b) involves not the status of the component parts but rather their combinability. A different development of the come verb, corresponding to Lichtenberk’s ‘relative closeness’ and defined as ‘an aspectual auxiliary of immediate anteriority’ by Bres and Labeau (2015), is seen in a French example like (34) (= Bres and Labeau’s (1a)): (34)

Mon copain vient de rompre … (Fr.) my boyfriend come.prs.3sg de break.inf ‘My boyfriend has just broken up with me.’

In this case too the combination with the periphrastic perfect fails and hence the ungrammaticality of (35b) in contrast to (35a): (35)

a.

Il vient de pleuvoir. (Fr.) it come.prs.3sg de rain.inf ‘It just rained.’

periphrasis and inflexion

b. **Il est venu it be.prs.3sg come.ptcp c. **Il vint de it come.pfv.3sg de

49

de pleuvoir. (Fr.) de rain.inf pleuvoir. (Fr.) rain.inf

The difference here is that the simple past as in (35c) is also not acceptable, and the explanation for the ungrammaticality of both (35b) and (35c), as Bres and Labeau (2015) demonstrate, does not in this instance concern the combination of forms but rather the combination of meanings. Simple past defines a moment in time rather than an ongoing state and therefore does not permit the aspectual subdomain defined by venir de to intrude. And once again, thanks to the richness of the historical documentation, they can show how this aspectual meaning has developed by stages from the motion sense, which was the only one available prior to the twelfth century, with the aspectual use not becoming predominant until the sixteenth century. Romanian is different again. We have already seen in (32b) that veni ‘come’ can occur as a passive marker, but the multiple glosses there demonstrate that an additional modal meaning component is also a possibility. The details of the various available interpretations are too complex to enter into here, but Dragorimescu and Nicolae (2014:93) provide a neat summary: ‘in contrast with the regular “be”passive, which is static, the veni-passive is dynamic; in contrast with the reflexive passive, the veni-passive contributes a stronger deontic or iterative value […] in the imperfect the passive auxiliary veni yields a habitual, dispositional reading’. What, however, veni shares with the be verb is the ability to act as a copula, as in (36): (36)

Ion îmi vine cumnat. (Ro.) Ion 1sg.dat= come.prs.3sg brother-in-law ‘Ion is my brother-in-law.’

If we ask, how such complexity and variety in the development of a single lexical item comes about, the answer once again is that these various uses are natural outcomes of the mechanisms of semantic change including the passage from concrete to abstract, as detectable in the changing choice over time of accompanying verbs, and the bleaching of the directional component of meaning. To complete the picture for reflexes of Lat. uenire, mention needs to be made of its use with the gerund to express progressive or continuative meaning, in this respect paralleling what we have already seen with go in examples like the Italian (28a) and Spanish (29b). Thus for the modern languages we have: (37)

a.

Gianni veniva scrivendo articoli. (It.) Gianni come.pst.ipfv.3sg write.ger article.pl ‘Gianni kept writing articles.’

50

adam ledgeway and nigel vincent

b. Hace años que viene make.prs.3sg year.pl comp come.prs.3sg mismo. (Sp.) same ‘He’s been saying the same thing for years.’

diciendo lo say.ger def

And once again there is historical evidence to show the emerging pattern—as demonstrated, for instance, for Italian in the recent compilation by Colella (2020), from which the examples in (38) are drawn: (38)

a.

quando il sole si viene when def.msg sun refl= come.prs.3sg tramontando (14th-c. OFlo.) set.ger ‘when the sun is setting.’ (La Metaura di Aristotile) b. sarà venuto augmentando lo calore a be.fut.3sg come.ptcp increase.ger def heat to passo a passo (OIt.) step to step ‘the heat will have been gradually increasing.’ (Restoro d’Arezzo, Composizione, 1282)

(38b) is particularly striking since it involves a periphrastic future perfect of venire combined with a gerund, a combination which is no longer acceptable in the modern language. There are also cross-linguistic generalizations insofar as the periphrases with the come verbs are less frequent than those with the go verbs, co-occur with a narrower range of lexical verbs, and commonly preserve a degree of deictic orientation (Bertinetto and Squartini 2016:949f.). In short, grammaticalization starts earlier and moves faster and further with go than with come. Once again, then, the Romance data both support and extend Lichtenberk’s generalizations and allow them to be documented over close to a millennium of textual attestation. Even more restricted is the third profile proposed by Lichtenberk, namely that for reflexes of the return verb, as illustrated in Figure 1.3. additive repetition

‘return’

reflexive

reditive

Fig. 1.3 Grammaticalization of return forms

Here too there are Romance parallels as demonstrated in detail in the chapter by Parry (Chapter 5 in this volume), where she shows how in a number of northern

periphrasis and inflexion

51

Italian dialects the verb tornare ‘return’ offers synchronically and diachronically documentable evidence confirming all the dimensions and directionality of change identified by Lichtenberk.

1.3.2 Compositionality and univerbation How, one may ask, do all the examples we have just considered relate to the theoretical questions raised in §1.2? The general lesson is that they show how forms and meanings can separate over time. More precisely, they help us to address in more detail the question of (non-)compositionality and its relation to periphrasis (cf. also the discussion in §2.4.6 in this volume). To this end, the principle of compositionality is best taken together with a second principle, which, following Langacker (1987:448), we will call ‘analysability’ and which predicts that complex forms can in general be expected to break down transparently into their constituent parts (Vincent 2015). The possible relations between these two dimensions can be set out as in Table 1.6. Table 1.6 Dimensions of language structure + compositional – compositional

+ analysable regular, ‘transparent’ morphosyntactic structures some compounds and periphrases, idioms, and folk etymologies

– analysable some instances of suppletion arbitrary signs à la Saussure

The combination [+analysable, +compositional] serves as a baseline for defining the way words and morphemes are combined in natural languages given the needs of communication and expression over a potentially infinite range of topics and contexts. Change, however, may disrupt such form–meaning transparency in a variety of ways. At one extreme, both analysability and compositionality are lost with the consequence that a new lexical item is created. This is particularly evident in the case of adverbial particles and prepositions such as Fr. encore ‘still, again’ < Lat. *hinc hac ora ‘hence this hour’ or It. dopo ‘after’ < Lat. de post ‘from after’. Alternatively, analysability may be retained but compositionality lost as we have seen with so-called aoristic drift where perfect periphrases such as Fr. il est parti or (northern) It. è partito have acquired the simple past meaning ‘he left’ even though they incorporate the forms est and è ‘be.prs.3sg’. Note here, however, that the loss of compositionality is not historically immediate but depends crucially on an intermediate stage in which there is still compositionality but involving a lexical head and a grammatical tense/aspect marker rather than between

52

adam ledgeway and nigel vincent

two lexical items. Most grammaticalization is of this latter kind as we have seen with our examples of go, come, and return verbs. A more complex intervening stage is what is labelled ‘Dahl’s paradox’ in Vincent (2015:110–112). This is the case with the surcomposé verb forms such as Fr. il a eu mangé ‘he had eaten’, where the relation between the auxiliary constituents a ‘have.prs.3sg’ and eu ‘have.ptcp’ is non-compositional—they mean ‘had’ not ‘has had’—but where the relation between these two items and the lexical head mangé ‘eat.ptcp’ is compositional: ‘had’ + ‘eaten’ = ‘had eaten’ (see also the discussion in §2.4.2 in this volume). An intriguingly different kind of development is one where analysability is introduced without concomitant suppletion as happens with what is traditionally called ‘folk etymology’ as for example It. battisuocera ‘cornflower’, where the source word Lat. baptisecula has been reanalysed as batti ‘beat’ + suocera ‘motherin-law’. As Maiden (2008) explains, developments like this are not the random curiosities they might at first seem but evidence that analysability of form is in principle independent of compositionality of content. They are thus on a par with the diachronic emergence of morphomic patterns, that is to say structures which unite members of a verbal paradigm in analysable and diachronically reproducible classes but which are not susceptible to morphosyntactic compositionality. That said, the boundaries are not always easy to draw and are the focus of ongoing theoretical debate. Thus, the emergent Romance future and conditional paradigms built from the lexical stem plus endings derived from Lat. habere yields a morphologically productive and analysable pattern, as for example with the future and conditional of correr ‘run’ in Portuguese exemplified in Table 1.7.

Table 1.7 Portuguese future and conditional forms of run 1sg 2sg 3sg 1pl 2pl 3pl

Future correrei correrás correrá correremos correreis correrão

< correre habeo < correre habes < correre habet < correre habemus < correre habetis < correre habent

Conditional correria correrias correria correríamos correríeis correriam

< correre habebam < correre habebas < correre habebat < correre habebamus < correre habebatis < correre habebant

If we subtract the infinitival stem correr- from these forms, we have a set of endings that apply to all verbs in the language whatever their other (ir)regularities might be. And at the outset it is reasonable to suppose that this regular morphological analysability was matched by semantic compositionality: correre habeo ‘I

periphrasis and inflexion

53

have to run’ and correre habebam ‘I had to run’.23 Two developments, however, complicate this picture. On the one hand, the fact that main stress fell originally on the auxiliary for some verbs led to phonetically reduced forms of the infinitival stem being found in both the future and the conditional, as in Pt. direi ‘I will say’ and diria ‘I would say’ beside dizer ‘to say’ and traremos ‘we will bring’ and trarı´amos ‘we would bring’ beside trazer ‘to bring’. The extent and detail of this alternation between the full infinitive and the reduced future/conditional stem varies from language to language but the principle of such an alternation holds across most of the family. Maiden (2018a:Chapter 8) provides rich documentation of the pattern and, following Esher (2014), adopts the label Fuèc for this special stem. Crucially, he goes on to argue that the morphological shape remains even when the semantic link between future and conditional has weakened or disappeared, and that therefore the Fuèc stem is a further instance of morphomic patterning. Alternatively, if one adopts the semantics of the conditional as a combination of [future] and [past] as proposed by von Fintel and Iatridou (2008), the structures both in their original periphrastic version and in their modern univerbated form are still compositional. The jury is still out on this one. A different version of the compositionality question presents itself when we consider the outcomes of the combination of the go verb with the infinitive as in Fr. nous allons danser ‘we will dance (lit. we go.prs.1pl dance.inf)’ beside Cat. vam ballar ‘we danced (lit. go.prs.1pl dance.inf)’. Both languages have structures which are analysable as go + inf and both are compositional but the underlying semantics which define the composition differ. Finally, something which Table 1.6 does not allow for is simple non-availability of forms, or defectiveness. This has always been a puzzle within the domain of inflexion: why, for example, is the expected past participle form **permasto not available for the Italian verb permanere ‘stay, continue’, when the structurally parallel rimasto is in regular use as the participle of the virtually synonymous rimanere ‘stay, remain’, and when the Latin simple verb manere ‘stay’, of which both are compounds, does have the form mansum and indeed old Italian provides evidence for a form permaso? The compositional components are, as it were, ready to go, but the lack of the necessary analysable constituents creates an arbitrary gap. And as we have seen, gaps of this kind exist for periphrases as well, so that while the Spanish compound periphrasis in (39a) is acceptable, its Italian counterpart (39b) is not:

23 For the sake of simplicity, we omit discussion here of the alternative development whereby the conditional forms derive from the past perfect rather than the imperfect of the auxiliary, as with It. correrebbe ‘(s)he would run’ < correre habuit. However, the logic of the argument is the same in both cases.

54 (39)

adam ledgeway and nigel vincent

a.

He estado pensando have.prs.1sg be.ptcp think.ger ‘I’ve been thinking that …’ b. **Sono stato pensando be.prs.1sg be.ptcp think.ger

que … (Sp.) comp que … (It.) comp

However, unlike with inflexion, what we have also seen is that in some cases the absence of the periphrastic forms is not a matter of arbitrary gaps but of semantically motivated failure of compositionality as with the case of the Fr. venir de construction exemplified in (35).

1.3.3 Attestation and reconstruction At various points in our treatment so far we have underscored the need to integrate the investigation of language structure and change, and the inherent complementarity between the two. This theme emerges in a different guise when we come to consider the relation between attestation and reconstruction, a topic which has always been a matter of discussion among Romance linguists. Elcock (1960:33) famously observed that ‘[i]t is the special privilege of Romance philologists that they are not compelled to rely entirely upon reconstruction’. Beside this we can set the following remark from R. A. Hall at the beginning of his Proto-Romance Phonology (1976:1f.): ‘In the Romance languages we find numerous sets of obviously related forms […] Our procedure should always be to reconstruct the Proto-Romance forms first and then to compare the reconstructions with attested Latin material.’ In essence, the techniques of reconstruction apply principles derived from general theory to help open up the lost linguistic past, something which applies as much in the domain of morphosyntax as in phonology (Walkden 2014). In the case of language families with no documentary history they are the only option, but when there is ample documentation, as there clearly is in the case of Latin and in the early stages of Romance even for many of the so-called dialects, there is always the temptation to dispense with the deductive methods of general theory and work inductively on the basis of the surviving texts. An extreme expression of this latter view is Varvaro (2011:300), who asserts that to use reconstruction when texts are available is, in his words, ‘like studying the history of Napoleonic France with the methods that are normal in the domain of prehistory’.2⁴ In fact, however, there is a whole literature on the infectious diseases that were the scourge of some of Napoleon’s armies with the evidence being derived from the analysis of excavated bodies using the same methods as deployed by archaeologists when they study prehistory (Raoult et al. 2006). 2⁴ For more discussion of the context in which Varvaro develops this view and further references, see Vincent (in press).

periphrasis and inflexion

55

Rather than giving priority to one or the other source of evidence it makes more sense to see them as complementary both in the domain of phonology and of morphosyntax. As a case in point, we may consider the phenomenon of differential object marking (DOM): (40)

a.

b.

c.

d.

e.

¡Dejad de atormentar a-l gato! (Sp.) stop.imp comp tease.inf dom-def cat ‘Stop teasing the cat!’ să prinzi pre Iambru Diiac (ORo., 1600) sbjv catch.2sg dom Iambru scribe ‘You should catch Iambru the Scribe.’ (Mardale 2015, ex. 26b) a-j-ju istu a tɔ sureɖɖa. (Crs.) have.prs.1sg see.ptcp dom your sister ‘I have seen your sister.’ (Manzini and Savoia 2005:II:513) si alchuno homini hochirit at alcuno atteru if some man kill.ipfv.sbjv.3sg dom some other homini (14th-c. Srd.) man ‘If a man killed another man.’ (Carta de Logu d’Arborea, 4.1) ad te accareza la pultronaria (15th-c. Nap.) dom you.sg caress.prs.3sg def idleness ‘You’re tempted by idleness.’ (Brancati 108.23)

Within Romance, DOM is widespread and displays many regional variations (see Ledgeway in press for a rich compilation of examples, sources, and variant structures). The common factor, as the term implies, is that in certain contexts the direct object is distinctively marked by a preposition, typically a ‘to’ or whatever other form is used in the variety in question to mark an indirect object, such as reflexes of in medio (ad) ‘in middle (to)’ > ma/me ‘to’ in central Italy (Rohlfs 1969:15; Berizzi 2013) and de+ab ‘from’ > da ‘to’ in Gallo-Italic varieties of Sicily as well as in some Umbrian dialects (Rohlfs 1969:8, 15; 1971:333f.; Manzini and Savoia 2005:II:502, 528). At first sight, this might seem relatable to the fact that in Latin for verbs such as subuenire ‘help’, parcere ‘spare’, inuidere ‘envy’, and suadere ‘persuade’ the object is marked with the dative rather than the more usual accusative case. And, indeed, as Sornicola (1997) notes, there is a degree of correspondence between such items and many of the verbs with which DOM is found in early Italo-Romance and Ibero-Romance texts. However, the Romance phenomena differ in that the presence of DOM depends on information structure and the nature of the object as much if not more than on the governing verb, hence there may be patterns of alternation even when the verb remains the same. There is also the striking formal exception to the general pattern in Romanian where an entirely different preposition, which does not have the function of indirect object marking,

56

adam ledgeway and nigel vincent

has assumed this grammatical function, namely pe (< Lat. super ‘above’). What this means is that, interesting and instructive though the Latin parallels may be, our understanding of the origin of the construction type must depend on comparative reconstruction of the argument marking hierarchies (topics, animacy, pronominals, etc.). Only in a second stage can the outcome of such analyses be compared to and integrated with the data provided by Latin. This is yet more true, as Mardale (2015) demonstrates, when it comes to Romanian, where, as we have seen, the formal marking differs and where the textual attestation is relatively late and varies according to register.2⁵ Even so, the links with particular verb classes are still there in the earliest Romanian texts (Mardale 2015:240), but the issue is complicated by the type and style of the text. Here, as elsewhere, philology and linguistics are companions not opponents.2⁶

1.3.4 Language contact In his critique of grammaticalization, Joseph (2011; 2020) deploys as one of his arguments the history of Spanish adverbs such as claramente ‘clearly’. At first sight this seems neatly to parallel forms such as Fr. clairement and It. chiaramente and to derive via the grammaticalization of the ablative of the Latin noun mens ‘mind’ in an expression such as clara mente ‘clear.fsg.abl mind.fsg.abl (= with a clear mind)’. The problem, as Karlsson (1981:87) notes, is that in that case we would expect the Spanish form of the adverbial suffix to be -miente since in general Lat. stressed [e˘] diphthongizes both in open and closed syllables in Spanish: tiene < tenet ‘(s)he holds > has’, hierba < herbam ‘grass’. And indeed in early texts forms such as de buena miente ‘of good mind, willingly’ are attested. However, according to the data collected by Karlsson, by far the most common form is in fact mientre as in vera mientre ‘truly’, demesurada mientre ‘immoderately’, where the presence 2⁵ Also relevant here is the earliest textual evidence provided by Sardinian, which, in contrast to all other early Romance varieties, shows from as early as the eleventh century a systematic distribution of DOM (i.a–b; cf. also 39d), independently of verb class (Lombardi 2007:143f.). If medieval Sardinian is a verb-initial language, as argued by Virdis (1987) and Lombardi (2007), then the emergence of DOM in old Sardinian could be argued to be a purely endogenous development to ensure unambiguous marking of postverbal animate subjects and objects, often also involving object clitic doubling (Virdis 1987:15f.) as in (i.a). (i). a. mi la furait Petru Tecas a nNastasia de Funtana aue domo me= her= carry.pfv.3sg Pietro Tecas dom Anastasia de Funtana from house dessu thiu (OSrd.) of.the uncle ‘Pietro Tecas abducted Anastasia de Funtana from her uncle’s house.’ (Condaghe di San Pietro di Silki, 15) b. Levait Dorgotori et issos frates a Maria et a Gavini (OSrd.) take.pfv.3sg Dorgotori and these brothers dom Maria and dom Gavini ‘Dorgotori and his brothers took Maria and Gavini.’ (Condaghe di San Pietro di Silki, 68) 2⁶ Further examples of cases where attestation and reconstruction are complementary rather than contradictory are discussed in Vincent (in press:§6).

periphrasis and inflexion

57

of -r- is due to crossing with the Latin adverbial suffix -(i)ter as in celeriter ‘quickly’. What is not attested, however, until the fourteenth century is the suffix -mente, which is now the standard Castilian form. The explanation lies in the fact that Spanish -mente is not a direct continuer of Latin but a literary borrowing, either direct from Latin or more probably from old French or old Occitan (Karlsson 1981:106). That the original structure here was a learnèd form is also the conclusion reached by Hummel (2013; 2014). Examples such as this do not, pace Joseph, counter-exemplify grammaticalization but they do show that diachronic paths can be crossed or blocked as a result of contact. What is special about this case, and about Romance in general, is the way the members of the family provide evidence for these contact effects operating not only between related varieties but also, and unusually, between the daughter languages and their diachronic source. The adverb formation mechanism seen in claramente is a good example of a development from analytic to synthetic without an intermediate periphrastic stage. The original Latin construction had a degree of syntactic freedom, so that examples are attested with both adj + n and n + adj orders (mente sincera ‘with a sincere mind’ Plautus Bacchides 509; turbata mente ‘with a worried mind’ Tacitus Annals 4.22), and with intervening material (mente ferant placida ‘that they may bear with a calm mind’ Ovid Metamorphoses 13.214). The only surviving freedom is that some varieties allow coordinated adjectives as with Sp. sincera y claramente ‘sincerely (lit. sincere) and clearly’ while others require the adverb suffix on both items as with It. sinceramente e chiaramente. At the same time, it is worth noting that the Romance patterns, regardless of the variety of attested patterns, require us to reconstruct the order adj + mente even though the unmarked order in the Romance languages is N + adj. Moreover, this reconstructed form wins out over the sequence adj + modo ‘way’, which Bauer (2006) shows to have originally been the more frequently attested form in Latin. An instance where contact interacts with periphrasis involves complex gerund forms such as Fr. étant parti ‘be.ger leave.ptcp (= having left)’ or Pt. tendo cantado ‘have.ger sing.ptcp (= having sung)’. Although from a synchronic perspective they look simply to belong to the periphrastic part of the verbal paradigm with the choice of auxiliary have/be varying from language to language and verb to verb, these are in fact relatively late innovations due to contact with—and more specifically translation from—Classical Latin. As Menoni (1982) demonstrates, these structures are coined at a time when classical texts were being translated into the vernacular. Thus, to take one of her examples, in Arrigo Simintendi’s fourteenthcentury version of Ovid’s Metamorphoses, the sentence amplexuque dato ait ‘embrace.abl.sg=and give.ptcp say.pfv.3sg’ (Met II.42) is rendered in Italian as e abbiendolo abbracciato li disse ‘and have.ger=msg.acc embrace.ptcp 3sg.dat= say.pfv.3sg (= and having embraced him, he said to him)’. A further consequence of this route of development is that while, as we have seen, structures with simple gerunds are part of the core grammar of all the Romance languages, the compound

58

adam ledgeway and nigel vincent

gerunds are for the most part restricted to higher registers of the spoken, and in particular the written, language. They also always have the straightforward present perfective interpretation while the finite compound forms may develop differently, as for instance in Portuguese, where tendo cantado means ‘having sung’ but tenho cantado means not ‘I have sung’ (which is expressed by the synthetic cantei ‘I sang’ or ‘I have sung’), but the aspectually marked ‘I have been singing/have kept on singing’. Contact via translation is also at work in some instances of the Romanian veni ‘come’ passive construction discussed above. Dragorimescu and Nicolae (2014) cite examples like (41): (41)

a.

Nis¸te corăbii negut¸ătores¸ti […] vin some ship.fpl commercial.fpl come.prs.3pl arse (ORo.) burn.ptcp.fpl ‘Some commercial ships […] are burned.’ (Foletul novel, 1693–1704) b. construzione […] care dă Greci vine construction rel by Greek.pl come.prs.3sg numită sintaxis (ORo.) name.ptcp.fsg syntax ‘construction […] which was/is called syntax by the Greeks.’ (Văcărescu, Gramatica, 1787)

These examples and many similar ones derive either from translations from or imitations of Italian literary models and once again therefore are restricted to older and higher registers in contrast to the more popular usage of the structure discussed earlier. The pattern with compound gerunds also provides an interesting counterpoint to the case of the It. venire ‘come’ passive periphrasis discussed above. There we saw that, although the structural ingredients are available to create compound forms such as **è venuta discussa or **venire discussa, these are considered ungrammatical by native speakers. In the case of the compound gerund, the structural ingredients are in use independently in the texts from the early medieval period but attestation of the combined structure only occurs some half a millennium later and in very specific cultural circumstances. With the venire construction, the early texts shed a different kind of light. Although, as we have said, the modern patterns are considered ungrammatical, examples such as (42) are attested: (42)

disse esserle venuta involata […] say.pst.3sg be=dat.3.fsg come.ptcp.fsg steal.ptcp.fsg una sua collana (16th-c. It.) a her necklace ‘she said a necklace of hers had been stolen.’ (Ascanio de Mori)

periphrasis and inflexion

59

The author of this example is from the Lombard city of Mantua, and it attests to contact between the language of literature and dialects of the region where venire is commonly the unmarked, and sometimes the only, passive auxiliary (Vincent 1987:249; 2014:18f.). Once again, the combination of structural analysis, grammatical change, textual attestation, and regional variation provide a rich and stimulating cocktail for linguists of all persuasions. We have obviously only been able to scratch the surface of the question of language contact and we have concentrated on cases where the Romance data are particularly instructive. Further dimensions are discussed in Finbow and O’Neill in this volume (Chapter 14), where they compare the different mechanisms involved in contact between related and unrelated languages.2⁷ Language adjacency does not, of course, necessarily lead to contact effects and hence the caution about the so-called Balkan influences expressed by Maiden (in press) and other contributors to Gardani, Loporcaro, and Giudici (in press). Indeed, at the other end of the spectrum is the phenomenon we might label ‘anticontact’. A case in point is Guardiol, the variety of Occitan originally spoken in Piedmont and which has survived in the Calabrian village of Guardia Piemontese. The cause of the dislocation from its region of origin was religious persecution and as a result since the sixteenth century the dialect has remained isolated (Kunert 1994). Although it has borrowed vocabulary from the other dialects of the region and from standard Italian, structurally it has resisted change, preserving, for example, a past tense built from go + infinitive, which has died out in other varieties of Occitan (Jacobs and Kunert 2014, and see also Chapters 3 and 4 in this volume).

1.4 Conclusions If there is one general conclusion to emerge from this chapter it is the essential complementarity of synchrony and diachrony. This in turn has important implications for our understanding of broader theoretical issues. Within the literature on (morpho)syntactic change over the last four decades or so, there have been two diametrically opposed stances. On the one hand the Chomskyan tradition in its various current versions (cartography, nanosyntax, universal spine, etc.) argues for the primary importance of defining the properties of UG, in terms of which it is then possible to investigate and model the data of language change. Although the conceptual framework and the theoretical primitives are very different, this is essentially a continuation of the Saussurean dichotomy (or ‘firewall’ in the terms of Kiparsky 2015:87), with priority being given to synchrony and language change 2⁷ For a recent survey, see D’Alessandro (2021).

60

adam ledgeway and nigel vincent

being considered, if at all, in a second phase of enquiry. The alternative view lies behind the question posed by Hopper and Traugott (2003:16): Do we need in our analyses to ‘stop the film’ and fix the grammar of a language as we investigate its structure, or do we need to view ‘grammar’ as a provisional way-station in our search for the more general characteristics of language as a process for organizing cognitive and communicative content?

On this view, the diachronic evidence is as important, maybe even more important, than the synchronic because it reveals the way natural languages emerge and change as the result of a series of intersecting external circumstances rather than being defined and restricted by a set of innate universal principles and mechanisms. It further implies that there is no necessity for there to be a fixed ‘core grammar’ in order to define what counts as a given language. And even if many will find this last conclusion too extreme, the fact remains that there is considerable variation in patterns within any one ‘language/dialect’, and this in turn argues for the desirability of not prioritizing either synchrony or diachrony but rather of treating both as complementary if we wish to achieve a full understanding of any given linguistic situation. Another way of expressing this same conclusion lies in Roger Lass’s words from his inaugural lecture ‘Language and time: a historian’s view’ at the University of Cape Town in 1983: A language never simply ‘is’; it is always looking back over its shoulder, and ahead. It is simultaneously in three states: ‘having been’, ‘being’ and ‘going to be’. Only ‘being’ is the concern of the normal speaker; all three are the linguist’s concern.

In different ways, all three also figure in the chapters that follow.

2 The boundaries of inflexion and periphrasis John Charles Smith

2.1 Introduction In this chapter, I discuss what counts as inflexion and what counts as periphrasis, examining the boundaries of each concept, including the boundary between them.1 Data will be drawn from a variety of languages, but, in keeping with the overall theme of this volume, the main focus will be on Romance. Several criteria have been used to define inflexion and periphrasis, but they do not always yield the same result, and even individual criteria may sometimes prove ambivalent, giving rise to boundaries that are fuzzy. Closer examination reveals that the boundaries of inflexion and periphrasis, like much else in linguistics, are subject to variation, to the extent that the two concepts might themselves be regarded as linguistic variables. As a starting point, we might observe that, structurally, inflexion and periphrasis are distinguished by a simple typological parameter: inflexion is synthetic, whilst periphrasis is analytic. However, although this is a necessary condition, it is by no means a sufficient one. Derivational morphology is synthetic, but it is not inflexional, whilst many syntactic constructions are analytic without being periphrases. A further condition seems necessary: both inflected forms and periphrastic forms are part of a (functional) paradigm. The distinction might then be expressed pretheoretically as follows: inflected forms are single-word members of a paradigm, whereas periphrases are multi-word members of a paradigm. But even this refinement of the definition is unsatisfactory without some indication of what constitutes a paradigm. Clearly, we need to delve more deeply into these basic issues.

1 Thanks first and foremost to Martin Maiden, for four decades of friendship and collaboration. This chapter has benefited from his advice, although he didn’t realize it at the time. It has also been immeasurably improved by the insights of my co-editors, Adam Ledgeway and Nigel Vincent. Further assistance has come from Richard Ashdowne, Ben Bollig, Chiara Cappellaro, Thomas Godard, Françoise Vaissière, Max Wheeler, and Sam Wolfe. Errors and shortcomings are, as always, my own.

John Charles Smith, The boundaries of inflexion and periphrasis. In: Periphrasis and Inflexion in Diachrony . Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © John Charles Smith (2022). DOI: 10.1093/oso/9780198870807.003.0003

62

john charles smith

2.2 The boundaries of inflexion 2.2.1 Preliminary issues The boundaries of inflexion in a given theory will, of course, depend to some extent on where that theory locates the boundaries of morphology more generally and defines its relationship to other components of the grammar. The perspective adopted here recognizes morphology as an autonomous component of grammar and accepts the reality of inflexional paradigms. Other theories have attempted to account for inflexion in terms of syntax or phonology. The syntactic approach is particularly clear in Distributed Morphology (originated by Halle and Marantz 1993; for subsequent surveys, see Harley and Noyer 2003; Matushansky et al. 2013; Siddiqi 2018), which essentially sees no boundary between sentence structure and word structure, and views inflected forms simply as the result of concatenating terminal items (see also the discussion in §1.2.1.1 in this volume). The phonological approach is exemplified by the claim that morphological phenomena which may have a phonological origin, but which have lost any conditioning environment which may have given rise to them, continue to be phonologically determined. This assumption is present in Distributed Morphology (compare the treatment of metaphony in Calabrese 2016), but it also underlies Anderson’s debate with Maiden over the status of the ‘N-pattern’ morphome in Surmiran (see Anderson 2008; 2011; 2013; Maiden 2011b; 2017). For a summary of and rejoinder to the phonological approach, see O’Neill (forthcoming). Although the notion of paradigm is crucial in delineating inflexion from other types of morphology, such as derivation, the issues are not always straightforward. In an overview, Hacken (2014:11) notes: ‘[w]hereas the core opposition between inflection and derivation is fairly obvious, the precise boundary between the two is more difficult to determine’. He identifies two approaches to the problem (Hacken 2014:13): the categorizing tradition, which posits a clear boundary between inflexion and derivation, and the sceptical tradition, which regards the boundary as unclear and/or irrelevant. To some extent, this is a matter of linguistic structure: for instance, the distinction may be less relevant or especially blurred in a polysynthetic language (Korotkova and Lander 2010). Booij (1994; 1996) distinguishes between ‘inherent inflection’ and ‘contextual inflection’. ‘Inherent inflection’ is ‘the kind of inflection that is not required by the syntactic context, although it may have syntactic relevance’, such as number on nouns (when this is not triggered by agreement) and tense on verbs, whilst ‘contextual inflection’ essentially consists of markers of agreement and case (Booij 1996:2). These distinctions are clearly valid, although there is no discussion of noun gender (to which we return below) or of the extent to which sequence of tense might be regarded as agreement, and hence qualify tense-marking in these

the boundaries of inflexion and periphrasis

63

circumstances as contextual. Booij points out that there are some respects in which ‘inherent inflection’ is closer to derivation than ‘contextual inflection’ (for instance, it can feed derivation and compounding—one of Booij’s examples of this process is It. lavapiatti ‘dishwasher, lit. “wash.plates”’). ‘Contextual inflection’ is ‘prototypical inflection’, but both it and ‘inherent inflection’ must be considered as separate from derivation. A paradigm consists of a number of cells. Spencer and Popova (2015:203) suggest that each cell must correspond to some functional or grammatical property or combination of properties, and that the pattern must be broadly productive— that is, it must apply, if not to every relevant item in the language, then to the vast majority. ‘Relevance’ here is to be taken as involving class membership rather than category membership, and excluding marginal exceptions, in which the paradigm may be defective or even non-existent. Thus, for instance, the existence of a large class of Italian adjectives which vary for number but not gender (exemplified by importante/importanti ‘important.sg/pl’ in both masculine and feminine), does not invalidate the existence in a larger class of adjectives of a paradigm which contains four cells (masculine and feminine gender; singular and plural number: compare buono ‘good.msg’, buona ‘good.fsg’, buoni ‘good.mpl’, buone ‘good.fpl’); whilst the small class of completely invariable adjectives (generally ending in a consonant or a stressed vowel), such as chic ‘chic’ and blu ‘blue’, is similarly without consequences for this paradigm. It should be noted that some functional or grammatical properties (such as gender and case) have no necessary extralinguistic correlate, whereas others (such as number and tense) appear to be correlated with independent conceptual notions, such as quantity and time. This distinction is reminiscent of, but not identical to, Booij’s division of inflexion into ‘inherent’ and ‘contextual’. However, the correlation of, say, number with quantity and tense and aspect with time is at best indirect, as demonstrated by metaphorical or quasi-metaphorical usages (compare the discussion in Comrie 1985:18–23), such as pluralia tantum (Corbett 2019), præsens pro futuro (Detges 2020:295f.), the ‘vivid’ or ‘historic’ present (Wilmet 1976:9–40), and the ‘narrative imperfect’ of several Romance languages (Bertinetto 1987); and categories such as number and tense are generally recognized to be grammatical categories in theories of syntax. Issues arise with diminutives and augmentatives. Scalise (1984) viewed these as constituting a separate level of morphology, independent of both inflexion and derivation, which he labelled ‘evaluative morphology’. Although the existence of evaluative morphology as a separate level has been generally contested (see, for instance, Stump 1993), the relationship of diminutives and augmentatives to inflexion and derivation is still a matter of debate, with a general consensus that this relationship may vary from language to language (for a recent survey, see Grandi 2015). Some items which originate as diminutives or augmentatives may become

64

john charles smith

lexicalized and hence clearly non-inflexional. This has been the case in French, where evaluative morphology is no longer productive, and all original diminutives and augmentatives are lexicalized—thus, lunette, originally ‘little moon’ (lune ‘moon’ + -ette ‘dim’), now simply refers to a variety of objects having a circular, semicircular, or crescent shape, most commonly (in the plural), ‘spectacles’; chevalet, originally ‘little horse’ (cheval ‘horse’ + -et ‘dim’), now means ‘easel’. Even in languages where diminutive morphology remains productive, lexicalization may none the less affect certain items—compare It. telefonino, originally ‘small telephone’ (telefono ‘telephone’ + -ino ‘dim’), but now ‘mobile phone’, to the exclusion of all other meanings; It. forchetta, originally ‘small (pitch)fork’ (forca ‘(pitch)fork’ + -etta ‘dim’), but now exclusively ‘(table) fork’ (or, metaphorically, ‘range’). In addition, a diminutive may sometimes have a different gender from its base form, once again implying a degree of lexicalization—for examples, see Merlini Barbaresi (2004:273–275). One criterion which has been used to establish whether evaluative morphology is inflexional or derivational is that of agreement: if evaluative agreement (e.g., a diminutive noun requiring a diminutive adjective) takes place, this indicates that the morphology is inflexional; if there is no agreement, the morphology is derivational (Anderson 1992:80f.). By this criterion, Romance augmentatives and diminutives are derivational. For instance, Spanish can form diminutives of both nouns and adjectives, but this morphology is not implicated in agreement— in other words, diminutive nouns can be found with non-diminutive adjectives and vice versa. Spanish (1) a. Esta mesa es ‘This table is b. Esta mesa es ‘This table is c. Esta mesita es ‘This table.dim is d. Esta mesita es This table.dim is ‘This table is small.’

pequeña. small.’ pequeñita. small.dim.’ pequeña. small.’ pequeñita. small.dim

However, in this connection, it is worth noting a striking construction found in the northern Lengadocian variety of the Gévaudan (spoken in the département of the Lozère), and discussed by Camproux (1958:332f.), who observes that the past participle in the compound past tenses may take a hypocoristic diminutive suffix, and that in this case it must agree in gender with the subject of the verb (even when the auxiliary is have), although agreement between participle and subject when the auxiliary is have is otherwise impossible. Thus, alongside the ‘normal’ sentence (2a), we find (2b); in both cases, the reverse agreement pattern is ungrammatical (2c, 2d):

the boundaries of inflexion and periphrasis

(2)

65

Lengadocian (Gévaudan) a. Ma drouleto l’as ben fach lou my.f girl.dim.f it=have.prs.2sg well do.ptcp.m the souon-souon? sleep-sleep b. Ma drouleto l’as ben facheto lou my.f girl.dim.f it=have.prs.2sg well do.ptcp.dim.f the souon-souon? sleep-sleep c. **Ma drouleto l’as ben facho lou my.f girl.dim.f it=have.prs.2sg well do.ptcp.f the souon-souon? sleep-sleep d. **Ma drouleto l’as ben fachet lou my.f girl.dim.f it=have.prs.2sg well do.ptcp.dim.m the souon-souon? sleep-sleep ‘My darling, have you had a nice sleepy-byes?’

As the grammaticality of (2a) makes clear, we are not dealing here with obligatory diminutive agreement; none the less, the diminutive suffix on the participle triggers gender agreement in a way that a non-diminutive form does not. Moreover, it is extremely hard to argue that the diminutive past participle fachet(o) is lexically distinct from fach(o). In this example, therefore, it seems that the diminutive morphology is at least closer to inflexion than in many others, highlighting the fuzzy and scalar status of evaluative morphology. Another grey area surrounds Aktionsart. If aspect, like tense, is generally assumed to be a grammatical category, Aktionsart is not. The common English equivalent ‘lexical aspect’ recognizes both the similarity of this category to grammatical aspect and the fact that (in many languages, at least) it lies outside the strictly functional realm. Spencer and Popova (2015:198) note that ‘a derivational process may be highly productive with few, if any, defective lexemes and thus, in effect, obligatory. Moreover, it may add content which is highly abstract and hence similar to that of a functional, grammatical element.’ They suggest that the English verbal prefix re- falls into this category, as it ‘behaves to some extent in the way that an iterative Aktionsart marker might behave in other languages’. This is even more true of French, where the corresponding prefix re-, encoding repetition, reversion, or restitution, is almost fully productive—it can appear with practically any verb (compare revouloir ‘to want some more’, ravoir ‘to get back’, Nous resommes en guerre ‘We are at war again’2). 2 Emmanuel Macron, in an interview on the Coronavirus, 9 September 2020—see https:/ /desourcesure.info/les-hospitalisations-augmentent-macron-reagit-nous-resommes-en-guerre/, consulted 1 February 2021.

66

john charles smith

Other aktionsartlich verb forms are more akin semantically to the evaluative morphology of nouns and adjectives, and are also relatively unproductive, making them unlikely candidates for inclusion in a paradigm. We might note, for instance, in French: from taper ‘to tap’, tapoter ‘to tap lightly and repeatedly’; from pleuvoir ‘to rain’, pleuvoter, pleuvasser ‘to rain lightly and intermittently’ (the latter with a more pejorative nuance); from écrire ‘to write’, écrivailler, écrivasser ‘to write (literature or journalism) in haste and carelessly’; from crier ‘to shout’, criailler ‘to shout continually and in a disagreeable way, often to complain about trivial matters’; from travailler ‘to work’, travailloter ‘to work half-heartedly’. All these derived forms contain an element of Aktionsart (they are frequentatives, referring to an action being carried out often—even travailloter), but also an element of manner (and often a pejorative nuance). Further examples can be found in the TLF s.v. -ailler, -asser, -oter (the first two of these suffixes generally add a pejorative nuance; the last is essentially diminutive). Broadly similar arguments can be adduced in respect of derived forms in Italian (for instance, from fumare ‘to smoke’, fumacchiare ‘to smoke slowly and/or absent-mindedly; to be an occasional smoker’; from bruciare ‘to burn’, bruciacchiare ‘to burn superficially in places’), and Spanish (for example, from bailar ‘to dance’, bailotear ‘to dance badly; to dance about’). For further discussion of evaluative morphology and the problems of defining it, see Grandi and Ko¨rtvélyessy (2015).

2.2.2 Case study 1: gender At first sight, the synthetic exponence of gender might seem to be a core instance of inflexion. But the position is more complicated. Adjective agreement in gender is uncontroversially inflexional. For instance, the distinctions between It. buono, Sp. bueno ‘good.msg’; It. buona, Sp. buena ‘good.fsg’; It. buoni, Sp. buenos ‘good.mpl’; and It. buone, Sp. buenas ‘good.fpl’ are regular and have no lexical meaning; they indicate a relationship between an adjective and a noun to which it is linked in some way (for instance, by a copular verb or by being a dependent in a phrase of which the noun is the head). Gender marking on nouns, however, is more complicated. First, whilst in Italian and Spanish singular adjectives, the suffixes -o and -a are always exponents of masculine and feminine gender, respectively, the same is not true of their nominal counterparts—compare It., Sp. mano (f) ‘hand’ and It., Sp. poeta (m) ‘poet’. Second, gender marking on nouns with an identical stem can serve a variety of functions, all of which involve lexical meaning, and are therefore more than simply inflexional—sex (It. ragazzo, Sp. chico ‘boy’ vs It. ragazza, Sp. chica ‘girl’); size (Sp. barco ‘ship’ vs barca ‘(small) boat’ but also Sp. charco ‘puddle’ vs charca ‘pond’); shape (Sp. cesto ‘basket taller than it is wide’ vs cesta ‘basket wider than it is tall’); metaphor vs literal meaning (It. foglio ‘leaf of a manuscript, sheet of paper’ vs foglia ‘leaf of a tree, bush, or plant’); tree vs fruit (It. melo, Sp.

the boundaries of inflexion and periphrasis

67

manzano ‘apple tree’ vs It. mela, Sp. manzana ‘apple’).3 Clearly, the distinction between male and female involves the straightforward mapping of gender on to sex. The semantic distinctions between other masculine/feminine pairs can often be ascribed—originally, at least—to metaphor; and they are not unique to Romance. For instance, in the Manambu language of Papua New Guinea, masculine gender is associated with large size and/or straight shape and feminine gender with small size and/or round shape; moreover, it is normal for a single stem to yield a masculine noun denoting a fruit-bearing tree and a feminine noun denoting its fruit (Aikhenvald 2008:116–124). The gender marking of an agreeing item is inflexional; the gender marking of nouns is in many instances lexical. However, whilst gender agreement of adjectives may be an uncontroversial example of inflexion, not all gender agreement is inflexional. In the French phrases le plus savoureux des fromages ‘the tastiest.msg of cheeses.(m)pl’ and la plus savoureuse des bières ‘the tastiest.fsg of beers.(f)pl’, the adjective savoureux agrees in gender with the noun—an unremarkable example of inflexional morphology. Now consider the phrases (found, for example, in advertisements) le roi des fromages ‘the king of cheeses’ and la reine des bières ‘the queen of beers’, both in the sense of ‘the very best’. Here, too, we appear to have gender agreement, but it is lexical rather than inflexional. Roi ‘king’ and reine ‘queen’ are separate lexical items. Moreover, this type of gender agreement is less strict: ?le roi des bières ‘the king of beers’ is occasionally found, but only generally in translations from other languages, particularly English (where ‘the king of beers’ is the slogan of the American brand Budweiser); French speakers tend to regard this phrase as bizarre and counter-intuitive, but they do not actually rule it out as ungrammatical. Likewise, la reine des fromages ‘the queen of cheeses’ is marginally possible where the name of the cheese is itself feminine (for instance, la tomme, la mozzarella). This indicates that lexical gender agreement is structurally different from inflexional gender agreement—it is possible to say La tomme est la reine des fromages ‘Tomme is the queen of cheeses’, but not **La tomme est la plus savoureuse des fromages ‘Tomme is the tastiest.fsg of cheeses’. There may be an element of metaphor involved in these examples; but they none the less involve gender agreement, although this agreement is lexical rather than inflexional.

2.2.3 Case study 2: number As with gender, the number agreement of adjectives is clearly inflexional. Number marking on nouns may also generally be regarded as inflexional, but there are

3 Apparent pairs where the stem is identical phonologically but not morphologically, such as It. mostro ‘monster’ vs mostra ‘exhibition’ are clearly accidental, and are not relevant here.

68

john charles smith

instances where a distinction of number corresponds to a lexical distinction, although such examples are less widespread than in the case of gender. An example can be drawn from Spanish kinship terminology. If we examine the opposition between terms such as hermano and hermana, generally glossed as ‘brother’ and ‘sister’, respectively, we find that the normal glosses are in fact slightly misleading, in as much as the masculine form, although usually interpreted as ‘brother’, can also have the generic meaning of ‘sibling’. This is particularly frequent in the plural, where mis hermanos can be interpreted as ‘my brothers’, but also as ‘my siblings, my brothers and sisters’. If the context does not make the meaning clear, the interpretation ‘brother’ can be forced by adding varón ‘male’; thus mis hermanos varones ‘my male siblings, i.e., brothers’, un hermano varón ‘a male sibling, i.e., brother’. None of this has any lexical consequences. But if we now consider the Spanish word padre, normally glossed as ‘father’, we find that it behaves like hermano in the plural but not in the singular. Mis padres means ‘my parents’, but un padre cannot mean ‘a parent’, and un padre varón is nothing more than a tautology. In this case, plural marking has lexical consequences. Another instance of plural marking having lexical status concerns the ‘second plurals’ (in -a) of a group of Italian nouns, discussed by Ojeda (1995) and Acquaviva (2002; 2005; 2008:123–161), and exemplified by corno ‘horn’, corni ‘animal horns separated from the animal; musical horns; horns of a dilemma’, corna ‘animal horns still on the animal; horns of the moon; cuckold’s (metaphorical) horns’, and braccio ‘arm’, bracci ‘arms of a cross, lake, etc.’, braccia ‘arms of a person; fathoms (unit of measurement)’. The forms in -a are morphosyntactically plural, but may be construed as more weakly individuated (Acquaviva 2005:259) or ‘undifferentiated’ (Acquaviva 2008:153). For Acquaviva (2008:157–160) ‘[b]raccia “arms” is not the plural of braccio “arm”; it is an inherently plural lexeme derived from the same root as braccio/bracci and provided with a gender value like any other noun. […] The -a of braccia, then, does not carry inflectional information at all, and its status is that of word marker […], a vowel marking the right edge of a word for morphophonological well-formedness.’

2.2.4 Case study 3: vocatives In traditional grammars of many languages, the vocative is treated as a case, and is listed in the paradigm (in Latin, for instance, it appears alongside the nominative, accusative, genitive, dative, and ablative). However, whilst it is (or can be) a form of the noun with a specific ending, it is not obvious that it is a case, or even an inflected form. For Blake (2001:1), ‘[c]ase is a system of marking dependent nouns for the type of relationship they bear to their heads’. By this definition, the vocative is not a case—it is a form of address, which does not encode an intra-sentential relationship (such as argument or adjunct). In fact, as Blake (2001:8) goes on to

the boundaries of inflexion and periphrasis

69

point out, many languages which do not have case systems none the less have a special form of the noun which is used for address, thereby providing support for the view that the vocative is something other than a case. We may also note that, diachronically, the behaviour of the vocative contrasts with that of forms which are uncontroversially cases: for instance, in Slavonic, Slovenian has lost the vocative, but has retained the grammatical case system of old Slavonic, whilst Bulgarian and Macedonian have lost the grammatical case system entirely, but have kept a vocative. Formally, the vocative may resemble a case, but, functionally, it does not behave like one. For Fink (1972:65), the Latin vocative is not a case, but a secondperson form of the noun. Floricic (2011) suggests that the vocative can be analysed as a case, but lies at the periphery of the set of cases—in other words, it is far from prototypical. The vocative may or may not be a case, but, even if it were not, this would not necessarily prevent it from being part of a paradigm (which would list cells containing forms of the noun corresponding to sentential functions).⁴ But the vocative may lie outside the paradigm so defined. For Vairel (1981), the vocative encodes a speech-act relationship, rather than a sentential one. Lyons (1977:217) claims that the ‘use of a common noun with vocative function, whether it is distinguished as such by its form or not, approximates […] to the use of a proper name or a title’. We might therefore see the function of the vocative as being to create a proper name from a common noun, and/or a title from a referential noun (whether common or proper) (see also Ashdowne 2007). A crucial point here is that nouns used as a form of address, whether or not they have a distinct form which indicates this function, may carry semantic restrictions which are lexical in nature. For instance, as Zwicky (1974:788) points out, ‘[t]he most striking general property of vocatives is their extraordinary idiomaticity’. He points to three ways in which this idiomaticity is manifested: • vocatives often have properties distinct from the corresponding referential NPs; • semantically parallel items often have distinct properties; and • morphologically related forms typically have distinct properties. He concludes that ‘the list of vocative NPs in English is largely learned item by item’. As an example, he notes that ‘driver’ can only be used as a form of address to a professional driver, not to a friend or family member who happens to be driving a vehicle, who is none the less a ‘driver’ in the standard lexical sense. Much the same is true of French chauffeur ‘driver’ (although a professional driver is often more likely to be addressed as ‘monsieur’ or ‘madame’). ⁴ As in the Western grammatical tradition and in more recent theoretical work—see, for instance, Stump (2016a:25).

70

john charles smith

Finally, we should note the existence of what Floricic (2011) terms ‘neovocatives’, found in Russian, and also in Romance languages, including (central– southern) Italo-Romance, Sardinian, and Alghero Catalan (see Mazzoleni 1995; 1997; Cabré and Vanrell 2011; Vanrell and Cabré 2011; Sgroi 2012; Iovino and Rossi 2014; D’Alessandro and van Oostendorp 2016). These are derived by a subtractive phonological process of ‘clipping’, which in Romance involves the deletion of all material following the stressed vowel—compare the Italo-Romance forms Pié (Piero), Sa (Sandro), Antó (Antonio), Pa (Paolo). It should be emphasized that this clipping is quite distinct from processes such as back-formation (compare Fr. coût ‘cost’, a noun derived by removing the infinitive ending -er from the verb coûter ‘cost’) and from the use of a bare stem in a marked cell of the paradigm (compare Mil. quaj ‘quails’, plural of quaja), in that the absent material has no morphological status. A similar process of clipping (albeit with different phonological constraints) is involved in the creation of lexical items in a variety of languages— compare Fr. prof (< professeur ‘teacher’).⁵ This fact argues very strongly that this type of vocative is not a part of the nominal paradigm. Of course, the fact that some vocatives may not be inflexional does not necessarily mean than no vocatives are inflexional. It is noteworthy, for instance, that some vocatives may trigger agreement, a criterion, which, as we have seen, has been used to distinguish inflexional evaluative morphology from derivational evaluative morphology. (Examples of such agreement are found in Romanian, although they are generally regarded as archaic and, in the contemporary language, tend to be limited to set phrases, such as iubite cititorule (beloved.voc reader.voc) ‘dear reader’, contrasting with nominative iubit cititor (beloved.nom reader.nom); in general, only the first relevant element in a phrase will exhibit the vocative inflexion (Popescu-Marin 2008:149), which might suggest that the Romanian vocative has become less inflexional.) Indeed, vocative morphology may occupy a similar status to evaluative morphology, being sometimes at the edge of inflexion and sometimes outside it. If at least some vocatives are indeed derivational forms rather than inflected ones, then this fact may also provide an explanation for their greater propensity to be borrowed—as has often been pointed out (see, for instance, Moravcsik 1978b:112; Thomason and Kaufman 1988:74f.; Matras 2007:61), derivational morphology is more likely to be borrowed than inflexional morphology. In Romanian, for example, whilst the inflexional endings of what Maiden (2016a:100) refers to as the adverbal and adnominal cases (traditionally, the ‘nominative–accusative’ and the ‘genitive–dative’, respectively) clearly derive from Latin case forms, the masculine vocative may do so, although it may represent a loan from Slavonic, and the

⁵ The status of prof as a separate lexical item from professeur is clear from the fact that it could be masculine or feminine at a time, before the feminization of job titles at the end of the twentieth century, when professeur could only be masculine and had no feminine equivalent (see TLF, s.v. prof ).

the boundaries of inflexion and periphrasis

71

feminine vocative (for instance, Mario vs adverbal Maria) may also be a borrowing from Slavonic (see Floricic 2011:122 and further references therein).

2.3 The dividing line between inflexion and periphrasis At first sight, the dividing line between inflexion and periphrasis seems clear: in terms of the definition we have adopted, an inflected form is synthetic, whilst a periphrasis is analytic, or an inflected form is a single word, whilst a periphrasis is a multi-word construction. But this generalization comes up against the problem of variation and change. It is likely that most inflexions have arisen from what were once independent items (calling to mind Givón’s celebrated statement that ‘today’s morphology is yesterday’s syntax’—1971:413). Often the multi-word stage is lost in prehistory, but there are more recent instances where we have data demonstrating the change. An example from Romance is the evolution in late Latin of a future tense formed with the auxiliary habere ‘have’ (see §3.2 in this volume). Originally an auxiliary, habere was rapidly reduced to an inflexion (probably passing through a stage of cliticization), a process which, in French at least, is complete by the earliest written record of the language (the Strasbourg Oaths of 842 ce). Despite our lack of detailed textual evidence for the precise processes involved (for discussion of what evidence there is, see Varvaro 2013:32–34), we can make certain ‘uniformitarian’ claims, based on language variation and change which has been observed. First, there will have been no sudden leap from a periphrastic future to an inflected future—there will necessarily have been a period of variation during which both the speech community and individual speakers will have shown variation, using the periphrasis some of the time (especially, one assumes, in formal and/or emphatic contexts) and the inflected form at other times. It may be possible to determine whether an individual token is periphrastic or inflexional, but it will not be so easy to decide whether, at this stage, the habere-future itself is periphrastic or inflexional. One answer is to say that it is both, and that the variation is to be accounted for by postulating that speakers have two different grammars (Kroch’s ‘diglossia’—see Kroch 1989; 2001), permitting each form to be realized. However, such an approach, whilst theoretically plausible, obscures the fact that the periphrastic and inflexional expression of the habere-future are, in a fundamental sense, the same thing. Like Schro¨dinger’s cat, the habere-future exists in two states simultaneously, and it is only when the box (in this case, the speaker’s mouth) is opened that the superposition ceases and we observe one state or the other. Indeed, there may even be intermediate states in which it is impossible to tell. As just noted, in its transition from auxiliary to inflexion, habere may have passed through a clitic phase, and clitics, too, raise issues about where the boundary between periphrasis and inflexion might lie. Anderson (1992) defines clitics as

72

john charles smith

‘phrasal affixes’, and this definition pinpoints a paradox—phrases are (potentially) multi-word entities, whereas affixes (normally) form part of a word. A comprehensive account of this problem lies outside the scope of this survey, but the following may serve as examples of the issues involved. Most Romance varieties exhibit clitic pronouns. In the majority, these are limited to non-subject functions, but in some varieties (such as French, oïl varieties, and some varieties of northern Occitan, Ræto-Romance, northern Italo-Romance, and northern Tuscan), clitic subject pronouns are also found. For a recent survey, with references to earlier work, see Roberts (2019:24–30). In the case of French, these ‘pronouns’ have been variously analysed as lexical subjects which obey certain constraints or as inflexional affixes. An important study is that undertaken by Culbertson (2010), who shows that the analysis of these items as inflexional affixes or something else is dependent on the variety of French involved. Although not a variationist study, it does distinguish between ‘Standard French’ and ‘Colloquial French’, claiming that so-called subject clitics are affixes in the latter, and concluding (Culbertson 2010:126) that ‘[t]he results strongly argue for an approach to the phonological clitic/affix distinction that takes into account evidence from multiple sources and clearly differentiates registers’. In similar vein, Wolfe (forthcoming) claims that the status of these items as subject pronouns or morphological agreement markers, like other aspects of the grammar of subjects in modern French, ‘is characterized by sociolinguistic and diatopic variation’. A particularly thorny problem concerns the placement of (non-subject) clitics in the future and conditional in European Portuguese, old Spanish, and early dialects of northern Italy, which may exhibit mesoclisis, with the clitic intervening between the stem and the ending—contrast Pt. não o farei (not it=do.fut.1sg) ‘I’ll not do it’ with fa´-lo-ei (do=it-fut.1sg) ‘I’ll do it’ (with farei-o (do.fut.1sg=it) also possible in the latter meaning). Once again, it is straightforward to determine the status of individual tokens, but much more problematic to make statements about individual speakers, the speech community, and the language as a whole. Attempts have been made to account for mesoclisis in terms of syntactic differences; however, as Luı´s and Spencer (2004:186) point out, any syntactic approach which ‘is forced to say that the tense/agreement markers and the future/conditional markers are completely different in their structure depending on the presence or absence of mesoclitics […] fails to capture the fact that in other respects the forms are identical in form and in their morphological behaviour’. The alternative would be to assume that the markers always have the same structure in an individual grammar, but that a speaker who systematically uses mesoclisis has a different grammar for the future tense from a speaker who does not use mesoclisis, and therefore that when both speakers use identical future forms in the absence of clitics, the forms are only superficially identical, but underlyingly distinct, because they are produced by different grammars, in which the affix has a different structure or status. Luı´s and Spencer propose a formal extension of Paradigm

the boundaries of inflexion and periphrasis

73

Function Morphology, which treats clitic clusters as affixal in cases of mesoclisis, but has them placed by the syntax in other situations. This is an elegant theoretical solution, but says nothing about variation, change, and the reasons why mesoclisis is limited to the future and conditional. Most Romance verb inflexion derives from Latin verb inflexion—the distribution of the desinences may have been altered by analogy and other processes, but their origin is normally clear. In a discussion of the development of Latin into Romance, the origin of the Latin inflexions themselves is irrelevant. However, the desinences of the future and conditional, as we have seen, originate not in Latin inflexions, but in full forms of the verb habere, which becomes an auxiliary and then a clitic. In late Latin or proto-Romance, therefore, we find verbal clitics and pronominal clitics. To begin with, these clitics respected a Wackernagel constraint, whereby proclisis was impossible in sentence-initial position. In these circumstances, pronominal clitics will have encliticized to the first element of the verbal group (lexical verb + auxiliary)—which forms a prosodic ‘colon’, in the sense of Fraenkel (1932; 1933; 1965)—leading to a sequence of [inflected lexical verb + clitic pronoun] for synthetic forms, [auxiliary + clitic pronoun + lexical verb] for preposed auxiliaries, [lexical verb + clitic pronoun + auxiliary] for postposed auxiliaries, and [lexical verb + clitic pronoun + verbal clitic] in the future and conditional. The reanalysis of the verbal clitic as an inflexion in most Romance languages led to clitic placement with the future and conditional being aligned with that of other inflected forms of the verb. This process took longer in Ibero-Romance, and is still not complete in Portuguese.

2.4 The boundaries of periphrasis 2.4.1 Introduction A starting point for a definition of periphrasis is the view that it ‘expresses a grammatical meaning in a multi-word construction’ (Haspelmath 2000:660). However, as Haspelmath himself points out, such a broad generalization needs to be constrained. It is generally accepted that a definition of periphrasis will not be rigid or monolithic, but, to an extent, ‘fuzzy’; it may therefore be more extensional than intensional. The two major approaches to the problem involve prototypicality and canonicity. As Brown et al. (2012:237) are at pains to point out, these two notions are not the same. Prototypicality involves psychological salience, which is often the result of frequency, whereas canonicity is an analytical tool, which predicts that truly canonical forms are ‘likely to be rare or even non-existent’. They note that ‘a prototype of a phenomenon may actually be one step or two steps down from the [canonical] ideal, because a particular combination of properties is privileged

74

john charles smith

(either because it occurs frequently in languages or because it stands out)’. Prototypicality is an empirical psycholinguistic notion, whilst canonicity is a theoretical idealization. The prototypical approach is represented by Bentein (2011). It ‘takes into account the fact that each construction occupies a position along a scale and is “always on the move”’ (Bentein 2011:22). In a discussion of Ancient Greek, he surveys a number of criteria for periphrasis and concludes that three of them are particularly important. Paradigmaticity refers to the fact that ‘a construction is integrated in the inflectional paradigm and as such is obligatory’ (Bentein 2011:17). Conceptual integration essentially involves the desemanticization of one or more elements of the construction, leading to non-compositionality of the construction as a whole. Finally, he claims (Bentein 2011:14) that ‘[c]ertainly the most prominent syntactic criterion [for periphrasis] is that of “contiguity”’, pointing to ‘the iconic nature of constituent structure’ such that ‘in general, two linguistic elements which are semantically close are syntactically contiguous’. In particular, he notes that ‘in many languages auxiliaries and their complements cannot be separated’. See also Wakker (2006:243). Some scholars who are not advocates of a canonical or prototypical account per se, none the less adopt an approach which might be seen as laying the groundwork for one. For instance, Bertinetto (1990:342), after setting out several criteria for periphrasis, notes that none of them is ‘discriminante’. Ackerman and Stump (2004) propose three criteria, any one of which is by itself sufficient to define a multi-word expression as a periphrasis. First, if it is ‘featurally intersective’ with an inflexional paradigm—i.e., each of the features it expresses (such as tense, aspect, mood, voice, person, case, number, gender) is found elsewhere in the paradigm with inflexional exponence. (This view goes back at least as far as Hockett 1958:212, who claims that periphrases ‘can be recognized only where there is a clear gap in the inflectional patterns, which the phrases serve to fill’.) Second, if its morphosyntactic value is non-compositional. Third, if its morphosyntactic exponence is distributed amongst its parts. Although any one of these criteria is sufficient to define a periphrasis and no single one of them is necessary to do so, it is a natural assumption that an expression will be more periphrastic the more criteria it meets. This assumption is the essence of a canonical approach. Brown et al. (2012) present a fully fledged account of periphrases in terms of canonicity. As already noted, they stress that examples of truly canonical instantiations of periphrasis (as of anything else) may be difficult, or even impossible, to find—the aim of a canonical approach is simply to set up a space within which a category can be defined, to create a yardstick by which to judge exemplars. They regard paradigmatic intersectivity as the only one of Ackerman and Stump’s features which defines a canonical periphrasis, which they view as the sum of canonical inflexion and canonical (functional) syntax, offering the following summary:

the boundaries of inflexion and periphrasis

75

• a periphrastic construction realizes a (canonical) grammatical feature; • a periphrastic construction will occupy a cell in an otherwise inflected paradigm; • a periphrastic construction (like canonical syntax and canonical morphology) will exhibit a transparent relation between form and meaning; and • a periphrastic construction is a canonical functional syntactic construction. Despite such considerations, Haspelmath (2000), amongst others, is prepared to see at least some non-intersective multi-word constructions as periphrases, although of a different kind. He labels intersective periphrases ‘suppletive’ and non-intersective periphrases ‘categorial’. The latter type express ‘the kind of meaning [which] is expressed by monolectic forms elsewhere in the language’ (Haspelmath 2000:660); he gives the examples of Fr. je vais chanter (I go.prs.1sg sing.inf) ‘I am going to sing’ (to which we return below) and Sp. estoy cantando (be.prs.1sg sing.ger) ‘I am singing’. However, as Haspelmath himself recognizes, the notion of ‘the kind of meaning’ is very vague. For instance, is it sufficient for a language to have just one set of inflexional tense forms for any analytic expression of tense to count as a periphrasis? In what follows, we assess various criteria that have been adduced for periphrastic status, with reference to data from the Romance languages.

2.4.2 Intersectivity As we have seen, a multi-word construction exhibits ‘featural intersection’ with an inflexional paradigm if each of the features it expresses—such as (a particular) tense, aspect, mood, voice, person, case, number, gender—is found elsewhere in the paradigm with inflexional exponence. We begin by examining the compound past tense in various Romance languages in the light of this definition. The broad framework is that of Brown et al. (2012), following whom we recognize a feature [perfect]. Brown et al. do not define this feature, and it is not clear whether they consider it to represent a tense, an aspect, or something else. For present purposes, we shall define it as a tense feature indicating that the event precedes the reference point, equating to (E–R) in Hornstein’s recasting of Reichenbach’s formalism of tense (where ‘S’ is the moment of speech, ‘E’ is the event time, and ‘R’ the reference time—see Reichenbach 1947; Comrie 1981; Hornstein 1990) and corresponding to (ST < TT) in the neo-Reichenbachian formalism of Klein 1994 (where ‘UT’ is the utterance time, ‘ST’ the situation time, and ‘TT’ the topic time).⁶ Also following Brown et al., we shall recognize a tense [past], and define this as the event preceding ⁶ In what follows, the labels used will be ‘S’, ‘E’, and ‘R’.

76

john charles smith

the moment of speech—formally, (E–S).⁷ It will also be relevant to consider an aspectual feature [perfective] for a verb form which presents an action as completed or ‘looks at the situation from outside, without necessarily distinguishing any of the internal structure of the situation’ (Comrie 1976:4). In Romanian, the [present, perfect] tense is encoded by the perfect compus, formed with the present tense of have and the past participle—am făcut ‘I have done’. However, this form contrasts with a synthetic pluperfect, which combines the values [past, perfect]. It derives from the Latin pluperfect subjunctive (Lat. fecissem, Ro. făcusem ‘I had done’), despite the fact that it is an indicative mood. Therefore, since the values [present] and [perfect] both receive inflexional exponence elsewhere in Romanian, the analytic present perfect in this language is intersective (see also the discussion in §1.2.2.1 in this volume). (In addition, ‘aoristic drift’ has taken place, with the result that the perfect compus may also have a more general [past, perfective] value, which is not licensed by present relevance; this phenomenon will be discussed later.) In modern (standard) Spanish, the structurally equivalent form (present tense of have + past participle) has only a present perfect value (he hecho ‘I have done’)— no ‘aoristic drift’ has taken place. Moreover, all the other [perfect] verb forms (pluperfect, future perfect, conditional perfect) involve the same auxiliary, have. As there is no inflexional realization of [perfect], the present perfect does not intersect with the morphological paradigm—if it is a periphrasis at all, it is, like estoy cantando, an example of Haspelmath’s ‘categorial’ periphrases. However, in old Spanish, a synthetic pluperfect existed, continuing the Latin synthetic pluperfect indicative (Lat. feceram > OSp. hiciera ‘I had done’).⁸ Once again, therefore, since the values [present] and [perfect] both receive inflexional exponence, the analytic present perfect in old Spanish does intersect with the morphological paradigm. In the history of Spanish, then, the present perfect has become less intersective over the course of time. This development, which at first sight appears to be a sort of partial degrammaticalization, might appear curious; but, on closer examination, it is not. As a general point, if there is a structure which is used for the most frequent or unmarked form in a category, then it is not surprising to find it extended into less frequent and more marked forms in the same category. As a specific point, regarding [perfect] (i.e., E–R) forms, Hornstein (1990) suggests that the reference point (R) is automatically associated with the event time (E), unless specific information to the contrary is provided. The use of auxiliaries, as opposed to synthetic verb forms, is seen as one way of ‘flagging’ the marked relationship between R

⁷ Note that, unlike (E–R), (E–S) is not a basic notion for Hornstein, but one which is derived from the underlying relationship between R and S and the underlying relationship between R and E. ⁸ The cognate form in Spanish, hiciera, is also a pluperfect indicative in old Spanish, but is a past subjunctive in the modern language, alongside hiciese (which derives from the Latin pluperfect subjunctive, and is now largely absent from South American Spanish). The original value of the hiciera form has left a trace in its use as a ‘backgrounding’ device (see Klein-Andreu 1991; Lunn and Cravens 1991); but it can no longer be considered a pluperfect tense.

the boundaries of inflexion and periphrasis

77

and E (i.e., that they are not simultaneous). A similar observation is made, in a typological framework, by Dahl (1985:129). For a related argument situating this issue within the broader context of synthetic vs analytic exponence in Romance, see Coseriu (1971; 1988), and the discussion in §1.2.1 in this volume. Of course, the developments and structures just discussed do not have to apply (compare Romanian). Portuguese has a compound past tense formed with ter ‘have’ (originally ‘hold’), which replaced haver (which can still occasionally be found as an auxiliary in the literary language) in the function of stative ‘have’ (thus, tenho feito, word for word ‘I have done’), but this is not equivalent to an unmarked perfect—it has an imperfective or iterative interpretation (‘I have been doing (continuously or recently)’; ‘I keep doing’) and the standard realization of [present, perfect] is the simple past, fiz, which also encodes ‘I did’ (see Oliveira 2013:517f.). Portuguese has retained a pluperfect indicative which is a reflex of the Latin synthetic pluperfect indicative (Lat. feceram > Pt. fizera ‘I had done’). It has become rarer in speech, but is still common in writing; meanwhile, an analytic pluperfect, using a have-auxiliary (havia feito or tinha feito ‘I had done’), has developed alongside it, and tinha feito is now usual in the spoken language (for discussion, see Oliveira 2013:524f., 530f.). In varieties which exhibit both a synthetic and an analytic pluperfect, the exponence of this tense is ‘overabundant’, in that one cell is occupied by two different forms— see Thornton (2011). In these circumstances, the analytic form intersects with the paradigm by definition, and is therefore periphrastic. In varieties of Portuguese which have lost the synthetic pluperfect, we have the mirror image of the situation in Romanian: because the combination [present, perfect] receives inflexional exponence (and because [past] receives inflexional exponence in, for instance, the imperfect), the analytic pluperfect intersects with the inflexional paradigm, and is periphrastic. Brown et al. (2012:271) discuss the situation of the compound past in French, where they claim that this form ‘genuinely is a periphrasis’. They continue: In French, verbs are morphologically marked for tense, present vs imperfect: mange ‘eats’, mangeait ‘ate, was eating’, a ‘has’, avait ‘had, used to have’. This opposition is maintained in the perfect tense series: a mangé ‘has eaten’, avait mangé ‘had eaten’. So far there is no evidence of (canonical) periphrasis. Now, the present perfect a mangé can also be given a simple past interpretation, the so-called passé composé. This might be taken to illustrate nothing more than vagueness of time reference for the present perfect (and is so analysed by some linguists, e.g. Vet 2007) but this isn’t quite right: there is a semantic opposition between the imperfect mangeait ‘was eating’ and the passé composé, a mangé, in the sense that the imperfect lacks a simple preterite interpretation, and the passé composé lacks the durative, habitual etc. interpretations associated with the imperfect. Interestingly, in the past perfect avait mangé the imperfect form of the auxiliary isn’t interpreted with imperfect semantics. This form is vague between translation

78

john charles smith equivalents ‘had eaten’ and ‘had been eating’. The crucial fact about French is the existence of a passé surcomposé a eu mangé. This is formed by taking the present perfect form of the perfect auxiliary. The interpretation is that of a perfect in the (simple) past, i.e. overlapping with some of the meanings of avait mangé. This means that a mangé really does realize a [Tense:past] feature value but that the [Tense] feature is expressed indirectly […] [T]he French auxiliaries are marked morphologically for non-past tense. Therefore, the simple past (passé composé) interpretation of a mangé is opposed to the morphological present tense form mange. This means there is intersection with a morphologically expressed feature (though of a somewhat complex kind).

In other words, the fact that the auxiliary of the passé surcomposé itself appears in the passé composé with a value which is [past] and not [perfect] (as Brown et al. note, ‘there is no way of semantically interpreting the notion “perfect of a perfect”’) demonstrates that the compound past of French is an intersective periphrasis. Brown et al. offer a somewhat oversimplified characterization of the passé surcomposé (on which, see Cornu 1953; Stéfanini 1954; Jolivet 1984; 1986; Ayres-Bennett and Carruthers 1992; Carruthers 1994; 1996; Apothéloz 2009; 2010). Essentially, three meanings of the passé surcomposé can be distinguished (here exemplified with J’ai eu chanté lit. ‘I have had sung’): (i) immediate anteriority, broadly corresponding to the passé antérieur (Dès que j’eus chanté … (lit. ‘from that I have.pfv sing.ptcp’) ‘No sooner had I sung…’), with the obsolescent preterite form of the auxiliary being replaced by the compound past; (ii) rapid completion or result (En moins de cinq minutes, j’ai eu chanté ‘In less than five minutes, I had sung’); (iii) iterativity or habitualness of an event or activity which took place in a more or less remote past, but no longer does (‘I have sung, but I don’t any more’). It should be noted that none of these values corresponds uncontroversially to the perfective pluperfect meaning claimed by Brown et al. The first two uses are general (although not necessarily common) in French. The last use is often characterized as ‘regional’, and is found predominantly in the originally Occitan- and Francoprovençal-speaking areas of southern France (see Carruthers 1994). Apothéloz (2010) suggests that it in fact corresponds to a ‘existential perfect’, encoding the value ‘It has been the case in the past that X has occurred’, with a reference or ‘validation’ point in a more recent past, rather than the present. Passé surcomposé forms exist in Occitan (Alibert 1935:316f.; Wheeler 1988:265), northern and Balearic dialects of Catalan (Pérez Saldanya 2002:2950; Solà i Pujols 2002:2903; Moll 1975:§151bis), Friulian (Benincà 2014:40f.), the northern Italian varieties of Piedmontese and Lombard (Rohlfs 1969:49), northeastern Italian varieties (Poletto 1992; 2008), Abruzzese (where have combines with be, see D’Alessandro and Ledgeway 2010), and Sardinian (Pisano 2010)—for full discussion of their distribution and meaning, see the works cited, and, for a broader discussion, Vincent (2011:430–432; 2014:12f.; 2015:111f.). Some other

the boundaries of inflexion and periphrasis

79

Italo-Romance varieties, such as old Neapolitan (Ledgeway 1999) and old Venetian (Formentin 2004) also exhibit a passé surcomposé, but this form only occurs in unaccusative verbs, where the auxiliary is be. In addition, some Romance varieties have ‘hypercompound’ or ‘double compound’ tense forms where the auxiliary is in a form other than the compound past (see, for instance, Remacle 1956:70–83, on the eastern Wallon variety of La Gleize, in which the auxiliary may appear in the compound past, the pluperfect, the future perfect, the perfect subjunctive, and the pluperfect subjunctive—but not the past anterior). When Brown et al. refer to the existence of the passé surcomposé as being ‘crucial’, it is unclear whether they mean this in a sufficient or a necessary sense. Is it merely an incontrovertible piece of evidence, or does the absence of a passé surcomposé with a past perfect meaning rule out intersectivity? A careful reading of their argument suggests the latter. If that is the case, then the outcome of ‘aoristic drift’ will be a periphrasis only when the perfect auxiliary can itself appear in the compound past with the value of a perfective but non-perfect past.

2.4.3 Range of forms Haspelmath (2000:661) observes: ‘[i]n an ordinary combination of a finite and a non-finite verb, there are no restrictions on the forms of the finite verb, but in a periphrasis sometimes only a subset of the forms are allowed’, implying that such restrictions may be an indicator of (or a sufficient condition for) periphrasis. Spencer (2003; 2006) makes a similar point, and refers to this phenomenon as ‘underexhaustivity’. Using this criterion, it might be argued that the French go-future is more obviously a periphrasis than both the French ‘immediate perfect’ formed with come and the Spanish and Portuguese go-future. In French, go with temporal value can appear only in the present indicative or the imperfect indicative, and the subjunctive is impossible—**bien qu’il aille chanter (lit. ‘well that he go.sbjv sing.inf’) ‘although he is going to sing’. go in other tense forms followed by the infinitive can only be interpreted as a verb of motion. However, in Spanish and Portuguese, whilst the same tense restrictions apply, the subjunctive is grammatical.⁹ The impossibility of a future auxiliary itself occurring in the future tense, thereby yielding ⁹ The presence of a subjunctive form of the go-future in Spanish and Portuguese but not French may indicate a greater degree of periphrasticity in the latter, but it may also be the result of the different paradigms of the languages. The feature combination [future, subjunctive] has never existed in French, at any stage of its history; so, once the aller ‘go’ + infinitive form is reanalysed as a future, there is no paradigmatic slot that the subjunctive forms can fill. However, the combination [future, subjunctive] is present in both Portuguese (where it is common) and Spanish (where it has been losing ground since the eighteenth century, and is limited in the everyday language to set expressions, such as Sea lo que fuere (lit. ‘be.prs.sbjv.3sg that which be.fut.sbjv.3sg (= Be that as it may)’, although it is still found in formal, especially legal, registers). By contrast, the combination [present, perfect, subjunctive] is found in French (bien qu’il l’ait fait (lit. ‘well that he it= have.prs.sbjv.3sg do.ptcp’) ‘although he has done it’), and so, if venir de ‘come from’ + infinitive is reanalysed as a type of immediate perfect, there is already a slot in the paradigm which can be occupied by the subjunctive.

80

john charles smith

a double future, may be semantically motivated. The French immediate perfect construction venir de faire ‘to have just done’ (lit. ‘come from do.inf’) is also restricted to forms using certain tenses of the auxiliary, which, once again, is largely found in the present (il vient de faire ‘he has just done’) and the imperfect (il venait de faire ‘he had just done’), although the future (il viendra de faire ‘he will just have done’) and conditional (il viendrait de faire ‘he would have done’) are also found, albeit more rarely. We do not find forms corresponding to the past anterior or passé surcomposé—thus il eut fait lit. ‘he have.pfv do.ptcp’ and il a eu fait lit. ‘he have.prs have.ptcp do.ptcp’ (both, broadly, ‘he had done’) are not matched by **il vint de faire lit. ‘he come.pfv from do.inf’ and **il est venu de faire lit. ‘he be.prs come.ptcp from do.inf’ (or **il vient de venir de faire lit. ‘he come.prs from come.inf from do.inf’). However, unlike the French go-future, the venir de construction also allows the subjunctive: bien qu’il vienne de partir ‘although he has just left’. Further away from periphrasis are constructions such as il est en train de partir ‘he is in the process of leaving’, il est sur le point de partir, ‘he is on the point of leaving’, which exhibit a full range of tense and mood possibilities. We therefore encounter a paradox, in that we find a fuller range of tenses in the have-perfect than in the venir de construction, and yet the former is clearly more incorporated into the tense system of French than the latter. Tellingly, for instance, the have-perfect, but not the venir de construction, can be used in questions involving quand? ‘when?’—contrast Quand est-ce qu’il a dit cela? ‘When has he said that?’ (also ‘When did he say that’) with **Quand est-ce qu’il vient de dire cela ‘When has he just said that?’. Bres and Labeau (2015) offer a comprehensive account of these issues (together with a survey of earlier work on the problem), and suggest that venir de is an aspectual marker of (immediate) anteriority, rather than an exponent of tense. The different behaviour and distribution of the haveperfect and the come-anterior derive from this distinction.

2.4.4 Distributed exponence As we have seen, Ackerman and Stump claim that a construction may be regarded as a periphrasis if its morphosyntactic exponence is distributed amongst its parts (see also the discussion in §1.2.2.3 in this volume). Spencer and Popova (2015:215–217) take issue with this criterion of distributed exponence. They discuss compound past tenses formed with have in Macedonian and Bulgarian, which offer a parallel with the comparable tenses of Romance. In the construction from which these forms derive, have was a lexical verb of possession and the past participle modified the direct object of have, and therefore agreed with it. In Macedonian, agreement between the past participle and the direct object has been lost, whereas in Bulgarian it is still found. They conclude that, although Bulgarian arguably exhibits distributed exponence (because the participle agrees

the boundaries of inflexion and periphrasis

81

with the object), whilst Macedonian does not, the construction is in fact more periphrastic in Macedonian, because the absence of agreement demonstrates that it is more grammaticalized. After discussing further examples, they conclude that a more general principle is involved: ‘If the morphosyntax of a construction contravenes the normal patterning for that language, then the construction is (likely to be) a periphrase’. This proposal raises a number of questions. First, the notion of ‘contravenes the normal patterning’ is too vague to be anything other than heuristic. Second, it stands in contradiction to other criteria. If we apply Spencer and Popova’s analysis of Macedonian and Bulgarian to Romance, the compound past tense of Spanish, with no active past participle agreement, is more periphrastic than that of Lengadocian, where past participle agreement with any direct object is the norm. But their subsequent proposal that contravention of normal morphosyntactic patterning is what reveals a construction to be periphrastic invites us to draw exactly the opposite conclusion, for it is wholly exceptional in Romance for verbs to agree with their direct objects, and the compound past tenses formed with an auxiliary and a past participle provide the only examples of this type of agreement. If the agreement is lost, as part of the process of grammaticalization, this is, one assumes, precisely because it contravenes the normal morphosyntactic patterning of the language—but, in that case, Spencer and Popova are offering two contradictory criteria for periphrasticity.

2.4.5 Contiguity As noted above, Bentein (2011:14) claims that ‘[c]ertainly the most prominent syntactic criterion [for periphrasis] is that of “contiguity”’, pointing to ‘the iconic nature of constituent structure’ such that ‘in general, two linguistic elements which are semantically close, are syntactically contiguous’. He notes that ‘in many languages auxiliaries and their complements cannot be separated’. How far do the Romance data conform to this criterion? If we examine the compound past tenses, we find that some Romance languages, such as Spanish, show a tendency to treat the sequence ‘auxiliary + participle’ as an indissoluble unit. In old Spanish this was not the case. The auxiliary and the participle ‘were freely separated in the thirteenth century by any part of the sentence’ (Sturcken 1953:165), but this possibility progressively declined after the medieval period (Company 1983:250– 253). Keniston (1937:452) notes that in sixteenth-century prose ‘the compound tenses are normally treated as an inseparable unit’, although ‘exceptions to the regular practice are not uncommon’. However, intercalation is rare in modern Spanish, and is felt to be literary or affected (Gili y Gaya 1964:§120bis; Carratala´ 1980:141). Native informants tend to reject compound past tense forms in which an item intrudes between auxiliary and participle, and the few examples of such

82

john charles smith

constructions in contemporary Spanish come from literary, or, more rarely, journalistic texts. They are usually limited to the pluperfect, and involve a restricted number of monosyllabic adverbs, chiefly ya ‘already’ and aún ‘still’. Most varieties of Catalan do not separate the auxiliary and the participle, but in northern varieties and the speech of Girona, the monosyllabic items ben ‘well’, pas ‘not’, and ni ‘not even’ may be intercalated (IEC 2016:950). In Romanian, intercalation is limited to the five adverbs cam ‘rather, quite’, mai ‘again, also’, prea ‘too much’, și ‘also, already, indeed, even’, and tot ‘still, also’ (Manea, Pană Dindelegan, and Zafiu 2008:415). By contrast, intercalation, especially of adverbs, is commonplace in French. Appositional phrases may also occur between the auxiliary and the participle, and, especially in the written and non-colloquial spoken language, these may sometimes be quite heavy, as in (3–4) (note that in (3) the intercalated phrase itself begins with a past participle): (3) Malheureusement, M. Robert Galley, alors ministre de l’ équipement, unfortunately M. Robert Galley then minister of the equipment a —encouragé dans un premier temps par have.prs.3sg encourage.ptcp in a first time by the Elysée— préféré imposer par la force le péage dans l’ Élysée prefer.ptcp impose.inf by the force the toll in le Val-de-Marne. (Fr.) the Val de Marne. ‘Unfortunately, M. Robert Galley, the then Minister for Public Works— initially with the encouragement of the Élysée [i.e., the President]—preferred to impose tolls in the Val-de-Marne.’ (Le Monde, 14 October 1976, p. 36) (4) Estce que vous êtes content que le gouvernement be.prs.3sg it that you be.prs.2pl happy that the government ait —malgré tout ce qu’ on a have.prs.sbjv.3sg despite all that which one have.prs.3sg entendu ces derniers jours— finalement décidé de hear.ptcp these last days finally decide.ptcp de confiner? (Fr.) confine.inf ‘Are you happy that the government has—despite everything we’ve heard these last few days—finally decided to impose a lockdown?’ (22 Heures Max, BFMtv, 18 March 2021) In Italian, intercalation is also possible, although grammars of the language have tended to regard it as stylistically marked (see, for instance, Battaglia and Pernicone 1965:410 and Fogarasi 1969:242). By and large, the material intercalated in Italian tends to be less ‘heavy’ than in French. Cinque (1999:45–49) shows that intercalation in Italian is less marked than traditional grammars suggest, and that

the boundaries of inflexion and periphrasis

83

the class of adverb involved is a major factor. For fuller discussion of some of these issues and further references, see Smith (1989; 2021).

2.4.6 Non-compositionality Spencer and Popova (2015:213) suggest that perfect tenses formed with have and past participle are compositional. Harris (1978:135) notes of this type of formation in late Latin that, for encoding a value which ‘combined the meanings of present tense with perfective aspect […], what could be more appropriate […] than a syntagm combining the present tense of an auxiliary verb (habet) with the perfective participle (factvm)?’. It is clear, as Spencer and Popova argue, that a compound past tense which has undergone ‘aoristic drift’ is not compositional (and so, according to this criterion, is more periphrastic) in this sense, because it does not combine present and perfect values. See the discussion in §1.3.2 in this volume. It is unclear to what extent we can speak of compositionality when a construction is the result of a metaphor or a metonymy, as is the case with the go-futures of French, Spanish, and Portuguese, the come-anterior of French, and possibly the go-past of Gascon (see Chapter 4 in this volume). Cruschina (Chapter 6 in this volume) discusses constructions involving the verbs go and come in southern Italo-Romance. Following Cardinaletti and Giusti (2001:386–388), he distinguishes between constructions which present motion and subsequent action as two separate events and andative and venitive aspects, which present them as a single event. The separate-event construction is entirely compositional, and is not a periphrasis. In the aspectual forms, the construction has become monoclausal. Both constructions might be seen as iconic (with a biclausal construction encoding separate events and a monoclausal construction a single event), but the aspectual construction is still compositional, because the verbs which serve to form the andative and venitive remain identifiable, by definition, as go and come; Cardinaletti and Giusti (2001) refer to them in this context as ‘semi-lexical’. The andative may develop further, into a mirative, and here we may be moving closer to non-compositionality, because the verb go no longer encodes motion, and it is not straightforward to define what its precise contribution to the structure is. According to this yardstick, therefore, we might claim that we have a cline of periphrasticity, with the mirative being more periphrastic than the andative, which is in turn more periphrastic than the biclausal structure involving the verb of motion.

2.4.7 Fuzziness In §2.2.1, we discussed the status of re-/ri- as a marker of Aktionsart in Romance, and noted that the claim made by Spencer and Popova (2015) concerning the cognate morpheme of English also applies to its Romance counterpart: despite

84

john charles smith

being derivational, it has much in common with inflexion (it is highly productive and has highly abstract content). If we broaden this discussion to multi-word constructions involving verbs of motion, we may see that the fuzziness at the borders of inflexion is matched by a fuzziness on the boundary of periphrasis. The verbs go and come are widely used as auxiliaries in Romance (see Chapters 3, 4, and 6 in this volume). In addition, some Romance varieties express certain types of iterativity using a multi-word construction involving return (see Chapter 5 in this volume). go, come, and return form a natural semantic class; but, according to a rigid definition of ‘paradigm’, forms involving them represent very different degrees of periphrasis. The go-past of Catalan (discussed by Vincent and Wheeler—see Chapter 3 in this volume) arguably intersects with the inflexional paradigm, and therefore represents a canonical or prototypical periphrasis. The go-future of French (also discussed in Chapter 4 in this volume) may be a ‘categorial’ periphrasis in Haspelmath’s terms, but there is evidence that, in some varieties, at least, it intersects with the inflexional paradigm (we discuss this issue below). Other constructions involving go, such as go + past participle and go + gerund in Italian and middle French, are not intersective, and so are at best categorial. As we have just seen, Cruschina (this volume) argues that, in some southern Italo-Romance varieties, some constructions involving go may become grammaticalized into an ‘andative aspect’, where ‘going to do’ is seen as a single event rather than two distinct events; this ‘andative aspect’ may then be further grammaticalized into a mirative. Finally, multi-word constructions involving return, in as much as they correspond to a derivational distinction rather than an inflexional one, may not be periphrases at all. They may be ascribed quasi-categorial status only to the extent that a quasi-inflexional status may be assigned to the iterative prefix re-/ri.

2.4.8 Structural and sociolinguistic variables Standard approaches accept that periphrasis may be a cline, and that periphrases may be more or less canonical or prototypical; however, this cline is defined according to a set of broad structural criteria: does a particular construction in a language meet certain structural conditions? But what if the construction meets a given criterion some of the time, but not always? Take, for instance, the issue of agreement between past participle and direct object in the active compound past tenses. As we have seen, there is some debate about this type of agreement, which can be seen as lack of grammaticalization, and hence indicative of low periphrastic status (Spencer and Popova 2015:215), or which can be considered as an example of distributed exponence or of normal syntactic patterning being contravened, and would thus be regarded as indicative of periphrasis by Ackerman

the boundaries of inflexion and periphrasis

85

and Stump (2004) and by Spencer and Popova (2015:215), respectively. As already noted, in Spanish, this type of agreement no longer exists, whilst in several varieties of Lengadocian agreement takes place with any direct object. However, many Romance languages fall in between these extremes, with agreement taking place with only certain types of direct object—for instance, only with direct objects which precede the verb (French, at least in its formal prescriptive variety), only with clitic pronouns (standard Italian), only with feminine clitic pronouns (standard Catalan).1⁰ A canonical or prototypical approach might claim that if a language ‘allows’ agreement—i.e., exhibits agreement in any circumstances—then it has distributed exponence or contravenes normal morphosyntactic patterning (because active agreement between a verb and its object is not normal in Romance); but it would be more satisfactory to extend the notion of cline to individual criteria and to argue that the compound past tenses in French, Italian, and standard Catalan were more canonical (or prototypical) than those of Spanish, but less canonical (or prototypical) than those of Lengadocian. Conversely, if active participial agreement is regarded as indicating lower periphrasticity, Spanish and Lengadocian will still be at the two extremes, with French, Italian, and standard Catalan in the middle, and the same argument will hold, mutatis mutandis. Considering now the criterion of contiguity (Bentein 2011:14), we see that it raises the same sort of issue. As we have seen, in French and Italian (although more so in the former) quite substantial material may sometimes be intercalated between the auxiliary and the participle in the compound past tense, whilst in Spanish none may be. In Romanian, intercalation is restricted to five specified lexical items. Some varieties of Catalan pattern with Spanish and others with Romanian. In terms of contiguity, we have a cline of periphrasticity running from Spanish (most prototypical), through Catalan, Romanian, and Italian, to French (least prototypical). However, the Spanish pluperfect is less prototypically periphrastic than the compound past in the same language, whilst being as periphrastic as the Romanian compound past, because it does allow the intercalation of certain lexically specified adverbs. Similarly, if we take non-compositionality as a criterion for periphrasis, this could imply that the compound past was more of a periphrasis when used as a preterite (‘I did’) than as a perfect (‘I have done’), since, as we have seen, the ‘present auxiliary + past participle’ structure could be viewed as compositional as a present perfect, but would not be compositional as a preterite. Such an analysis poses a variety of problems. One possibility is to claim that the compound past reflects two separate structures (we have already seen that Brown et al. 2012:271 incline to this view). But there is no evidence for this violation of Occam’s razor. For

1⁰ For discussion and motivation of these patterns, see Smith (1995; 1996; 1997; 1999; 2001), who claims that the phenomenon may have a functional motivation, related to the (lack of) recoverability of the direct object. See also Loporcaro (1998).

86

john charles smith

instance, Cinque (1999) establishes a link between adverbial modifiers and functional heads, with two of the latter being T(Anterior), required for perfect tenses, and T(Past), required for past tenses. Although the theory is a universal one, it is extensively exemplified from Italian and French. If the compound past had two separate underlying representations, we might expect to find differences between them involving adverb placement (especially with respect to the participle), but in fact there appears to be no distinction between the perfect and preterite readings in this regard (Schifano 2018:59f). The alternative is to say that the compound past is a single tense form with a range of values, and the extent to which it is a periphrasis in a given language is a function of its possible meanings in that language. Thus, the compound past in languages where ‘aoristic drift’ has occurred (French, Romanian, much of northern Italo-Romance) would be more of a periphrasis than in languages with no ‘aoristic drift’ (such as Spanish and Occitan). But, once again, the data are more complex. For instance, the simple past has disappeared from everyday spoken French, but is still found in many written registers, including much literature and some journalism, and can also be present in relatively formal speech (such as lectures, political speeches, and even some news broadcasts), especially if it is a monosyllabic third-person singular form of a frequent verb (fut ‘was’, eut ‘had’, fit ‘did’)—see Yvon (1963), Schogt (1964), van Vliet (1983), Engel (1985; 1996), Monville-Burston and Waugh (1985), Waugh and Monville-Burston (1986). In northern (regional standard) Italian, the simple past forms appear to be less available than in French, but may still be found in very formal contexts. However, in standard Romanian, the simple past is now no more than an archaic literary form. So, in these three varieties in which the compound past has taken over the values of the preterite, the takeover or ‘aoristic drift’ has been more pronounced in Romanian than in French or Italian. It might therefore be argued that the compound past is more periphrastic in Romanian. The facts noted in the preceding paragraphs raise issues of structural granularity: is a periphrasis less canonical or prototypical if the criteria which define it are restricted to particular syntactic contexts (which are independent of the periphrasis itself) or particular lexical items? At a finer level still, they also pose the problem of whether the analysis should apply to types or to tokens: is an occurrence of a candidate construction which does not exhibit, for example, intercalation or past participle agreement (assuming absence of agreement to be an indicator of periphrasis) more periphrastic than one that does? To take a concrete example, when used with a female interlocutor, is It. Ti ho visto (you=have.prs.1sg see.ptcp.msg) ‘I have seen you’, with no participial agreement, more periphrastic than the otherwise synonymous Ti ho vista (you=have.prs.1sg see.ptcp.fsg), with agreement? Finally, they raise the question of sociolinguistic variation. Object–participle agreement in the active compound past tenses of French, for instance, varies at least diaphasically and diastratically. ‘Aoristic drift’ in both French and Italian is a diaphasic, diamesic, and distratic variable—in certain registers and for certain

the boundaries of inflexion and periphrasis

87

individuals, and more commonly in the written language, use of the simple past remains a possibility. How do these facts bear on the definition of periphrasis?

2.4.9 Grammaticalization As Haspelmath (2000:661) notes, ‘we need a comprehensive theory of grammaticalization in order to understand periphrasis. […] Once we have such a theory, the definition of periphrastic construction is easy: The more grammaticalized a construction is, the more it can claim to have periphrastic status.’ (See also the discussion in §1.3.1 in this volume.) But it is not a straightforward matter to define grammaticalization.11 It might best be seen as involving two processes, which are discussed in a general context of (morpho)syntactic change by Langacker (1977), Timberlake (1977), and Harris and Campbell (1995). The first is reanalysis, defined by Harris and Campbell (1995) as ‘a mechanism which changes the underlying structure of a syntactic pattern and which does not involve any immediate or intrinsic modification of its surface manifestation’. Thus, to take the example of the compound past tense, with which we have been most concerned, the verb have acquires the category label Auxiliary, and hence the features, movement, etc. associated with this category (see also Smith 2000). This reanalysis may have actualizations—consequential changes. One of them may be a new reanalysis. For instance, the verb have may ultimately lose the category label, features, movement, etc. associated with main verbs, simply retaining those associated with auxiliaries. When this occurs, a lexical verb may be drafted in to indicate possession, extending or shifting its meaning. This new reanalysis, involving a reflex of Lat. tenere (originally ‘hold’) has taken place in Spanish (Seifert 1930a), Catalan (Seifert 1957), some varieties of southern Italo-Romance (Seifert 1935; Rohlfs 1969:§733), some Sardinian varieties (Seifert 1930b), and the Ambertois Auvergnat dialect of Occitan (Michalias 1906:85). Other actualizations are extensions—‘change in the surface manifestation of a syntactic pattern that does not involve immediate or intrinsic modification of underlying structure’ (Harris and Campbell 1995). In fact, depending on the theoretical stance adopted, it may not always be correct to claim that extension involves no modification of underlying structure (in generative grammar, for example, change in agreement or word order will correspond to some change in movement or feature-checking); but the point is that these changes do not involve relabelling (i.e., change of category), and are purely consequences of the reanalysis. 11 Compare Newmeyer (2001:187) (“‘grammaticalization” is simply a cover term for certain syntactic, semantic, and phonetic changes, all of which can apply independently of each other’) and Campbell (2001:113) (‘grammaticalization is derivative, that is, […] it has no independent status of its own, but rather relies on other processes and mechanisms of linguistic change which are independent of grammaticalization but which provide the explanations for the phenomena involved in grammaticalization’).

88

john charles smith

For instance, in many Romance varieties, the past participle progressively loses agreement with the following noun phrase because this is no longer the direct object of have, modified by the participle, but has become, in the new structure triggered by the relabelling of have as an auxiliary, the direct object of the verb represented by the participle. Within this framework, we may arrive at a better understanding of what periphrasis is and why relationship to a paradigm (whether ‘intersective’ or ‘categorial’) has come to be widely accepted as the most important and relevant criterion for defining this notion. Put simply, reanalysis creates new labels for heads or phrases at an underlying level which is also the locus of the categories which define the cells of the paradigm (tense, aspect, mood, number, etc.), whilst extension does not. A reanalysis which either creates a new functional category or which adds to the membership of an existing functional category is both necessary and sufficient for a construction to be regarded as a periphrasis. Extensions are consequential changes, and are not criterial for periphrasis, even though they may subsequently yield a construction which is more grammaticalized in some surface sense. Seen in this light, Spencer and Popova’s original claim (2015:216f.) that the loss of participial agreement in compound past tenses formed with have represents grammaticalization is almost certainly correct, but has no bearing on periphrasis. There is nothing at all to suggest that, for instance, the compound past of Spanish is more of a periphrasis than the compound past of French simply because the former, but not the latter, exhibits no object–participle agreement, uses a single auxiliary have, and does not allow items to be intercalated between the auxiliary and the participle—all of which are extensions which could be regarded as making it more grammaticalized. What matters is the reanalysis which has taken place in each.

2.4.10 Intersectivity as a variable Whilst these considerations attenuate the problem, they do not dispose of it, because issues of structural and sociolinguistic variation also arise in connection with intersectivity, which, as we have seen, is considered the most important (canonical, prototypical) feature of periphrasis, and which we have suggested is linked to reanalysis rather than extension. We have already noted the claim of Brown et al. (2012) that the existence of a passé surcomposé is ‘[t]he crucial fact about French’, without which, by their definition, the compound past tenses of French would not intersect with the remainder of the paradigm. However, the passé surcomposé is by no means as widely distributed as the other compound past tenses. As Carruthers (1996:185) notes, with further references, a ‘less than whole-hearted acceptance [of the passé surcomposé] by grammarians has been accompanied by a certain hesitation on the part of speakers’. She finds that the passé surcomposé is more common

the boundaries of inflexion and periphrasis

89

in writing (where it is used by 30% of informants) than in speech (where the corresponding figure is 12%) and is more likely to be used by better-educated speakers and by older speakers (Carruthers 1996:193). See also Jolivet (1984). It follows that, in the framework of Brown et al., the degree to which the French compound past tenses intersect with the paradigm is a partly a matter of speaker variation. Similar issues arise in connection with the distribution of the go-future and the synthetic future in Laurentian French. Poplack and Turpin (1999) observe that, in this variety, ‘basically all reference to future states or events is made by [the go-future], which has ousted [the inflected future] from virtually all contexts of productive usage but one’. This one context is negative polarity sentences, where the synthetic future is found with 99% probability. The two exponents of the future are therefore distributed according to polarity, but, crucially, the forms themselves do not encode polarity. In strict paradigmatic terms, we might say that the go-future and the synthetic future occupy the same cell, which therefore manifests overabundance (Thornton 2011), with the choice of form conditioned by factors external to the paradigm, and thus that the go-future axiomatically intersects with the morphological paradigm. For speakers who consistently exhibit this distribution, the go-future is an intersective periphrasis; for speakers who maintain a semantic or stylistic distinction between the two futures, the go-future is a categorial periphrasis. In between lies a range of sociolinguistic variation (discussed in detail by Poplack and Turpin 1999 and by Sankoff and Wagner 2006). Unless we stipulate that the periphrasis is intersective only in the system of speakers who exhibit the polarity distribution 100% of the time, we have to accept that the boundary between intersectivity and categoriality is fuzzy; and, even if we do make the stipulation, we shall have to admit that the boundary is a matter of variation between individual speakers. In sum, it seems clear that if the crucial notion of paradigmatic intersectivity is being defined in terms of a variable which is present only for certain speakers, then what counts as a periphrasis (or the extent to which a given form counts as a periphrasis) will differ from one speaker to the next. In a discussion of verbal inflexion in Acadian French, Smith (2011b) notes that the distribution of the third-person plural inflexion -ont is sensitive to factors such as age, sex, and social network strength. Since one effect of this inflexion (and the associated stem morphology) is to obliterate the ‘N-pattern’ morphome (as defined by Maiden 2011a; 2018a), it follows that the morphome may, in certain circumstances, be a sociolinguistic variable, demonstrating that ‘the concept of (socio)linguistic variability must encompass analyses as well as forms’ and ‘it is not merely (or even essentially) the inflection or the stem which is the variable, but rather more abstract or underlying notions, such as segmentation, paradigm structure, and morphomic patterns’ (Smith 2011b:326). Much the same would appear to be true of the notion of intersectivity and hence of periphrasis.

90

john charles smith

2.5 Conclusion It is clear that the notion of paradigm is of crucial importance in any attempt to define inflexion and periphrasis. Despite this, the evidence from Romance examined in this chapter reinforces the view that monolithic definitions of both inflexion and periphrasis are elusive, and that an adequate approach to the issue must recognize the scalar nature of the phenomenon and involve an element of fuzziness. Gender and number may have both inflexional and lexical values; diminutives and vocatives hover (to different degrees) on the border between inflexion and derivation; French subject ‘pronouns’ may be clitics or affixes, depending on the data surveyed. Romance multi-word constructions with a largely identical structure and (core) function, such as the compound past, may none the less diverge quite widely in the extent to which they are defined as periphrastic according to the various metrics which have been proposed. Moreover, whilst a given multi-word construction may be more (canonically or prototypically) periphrastic in language A than language B according to one criterion, the reverse may be true when some other criterion is applied. These differences, although real, should not blind us to the fundamental similarities between multi-word constructions (such as the compound past) with a common origin and broadly the same (core) value in a variety of Romance languages. We have also seen that, in as much as the data which serve to define several key concepts in both inflexional morphology and periphrasis are subject to sociolinguistic variation, notions such as clitic, affix, morphome, defectiveness, overabundance, and paradigmatic intersectivity, and hence inflexion and periphrasis themselves, are sociolinguistic variables.

PA RT II

PERIPHRASIS

3 Layering and divergence in Romance periphrases Nigel Vincent and Max W. Wheeler

3.1 Introduction Layering (1) and divergence (2) are two of the five key principles of grammaticalization set out in Hopper (1991:22), and, taken together, they characterize the way items may interact with one another on the classic path from lexical to less grammatical to more grammatical (Kuryłowicz 1965). (1)

Layering: Within a broad functional domain, new layers are continually emerging. As this happens, the older layers are not necessarily discarded, but may remain to coexist with and interact with the newer layers.

(2)

Divergence: When a lexical form undergoes grammaticization to a clitic or affix, the original lexical form may remain as an autonomous element and undergo the same changes as ordinary lexical items.

As is clear from the reference to functional domains, the concept of layering explicitly addresses the content side. A frequently cited Romance example concerns the relation between futures expressed by means of inflexions derived from the sequence infinitive + habere ‘have’ and those built from periphrases involving various forms of the go verb plus a following infinitive. What is less obvious is that the definition of divergence, although it refers to ‘a lexical form’, does not refer just to specific word forms but rather to the lexeme as a whole and the set of word forms associated with it. Indeed, one of the examples cited by Hopper (1991:25) concerns precisely the different reflexes of habere as both future inflexion and perfect auxiliary and which at the same time retains its lexical value as a verb of possession in some, though not all, Romance languages.1

1 Many thanks to Adam Ledgeway and J.C. Smith for comments and suggestions on an earlier draft. It goes without saying that all errors remain our own responsibility.

Nigel Vincent and Max W. Wheeler, Layering and divergence in Romance periphrases In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Nigel Vincent and Max W. Wheeler (2022). DOI: 10.1093/oso/9780198870807.003.0004

94

nigel vincent and max w. wheeler

In short, both layering and divergence focus principally on the semanticosyntactic changes that items undergo, and indeed Hopper himself comments: ‘Divergence is perhaps to be understood as a special case of Layering’ (1991:24). It is not surprising then that the term ‘divergence’ has virtually disappeared from the more recent literature; it does not figure at all for example in the index to Narrog and Heine (2011). At the same time, ‘layering’ and ‘layers’ have come to be used to cover both the synchronic patterns and the diachronic mechanisms which generated them regardless of whether they involve a single or multiple etymological sources; compare for example Fischer (2007:150, n.26) and Eckardt (2006:24). While there is a vast body of work that has pursued the syntactic, semantic, and pragmatic dimensions of these questions, as the contributions to Narrog and Heine (2011) testify, there has been much less investigation of the form side. It is taken more or less for granted that there is a cline from independent word to clitic to affix of the kind alluded to in the definition of divergence in (2), with the mechanisms involved being discussed in more detail under the heading ‘bondedness’ by Lehmann (2015:157–167).2 What instead we will consider in the present chapter are changes in form in the context of selected Romance periphrases. Many of these differ from the periphrasis that is the point of departure for the inflexional futures and conditionals in that the grammaticalized or grammaticalizing auxiliary precedes rather than follows the lexical head. Within such a structure the auxiliary tends to resist univerbation and maintains a degree of prosodic and syntactic independence, albeit with differences between languages and constructions in the class of items that can intervene between the individual components of the periphrasis. In the majority of such patterns, the auxiliary element is the locus of person, number, tense, and mood marking and hence retains a full paradigmatic structure. In some instances, the syntactic and semantic developments take place without any changes in the overt forms of the items in question. For example French has both venir in its etymological lexical sense of ‘come’ and as part of the periphrastic structure venir de + infinitive to indicate immediate past as in (3):3 (3)

Pierre vient de me dire quelque chose. (Fr.) Pierre come.prs.3sg de me= say.inf some thing ‘Pierre just told me something.’

However, the morphology of the verb is the same in the periphrastic use as in the lexical use. In Hopper’s terms, there is divergence of content but not of form.

2 The reference here is to the most recent third edition of this classic work but the concept is already to be found in the informally circulated first edition of 1982, and certainly formed part of the background to the thinking of Hopper and others in the work of that era, when grammaticalization, or ‘grammaticization’ as Hopper preferred to call it, was attracting renewed attention following decades of neglect since the time of Gabelentz and Meillet. 3 See also §2.4.3 in this volume, for discussion of the restrictions on the tense/mood contexts in which this periphrasis may occur.

layering and divergence in romance periphrases

95

A more complex case concerns the Latin verb tornare ‘turn on a lathe, round off ’, which has developed both new lexical senses—Sp. tornar ‘give back, go back, change’, Fr. tourner ‘turn (over), ponder’ and more—and in the varieties investigated by Parry in this volume (see Chapter 5) a new aspectual value of repetition, as in the Italian example in (4): (4)

Torno a dire che non ne voglio return.prs.1sg ad say.inf comp neg of.it= want.prs.1sg sapere. (It.) know.inf ‘I say again that I don’t want to know anything about it.’

In the synthetic part of the paradigm, the repetitive use seen here shares the same forms as the lexical sense ‘go/come back’, but in the periphrastic perfect in some dialects there is an auxiliary switch, as evidenced by the Cairese examples in (5) (= Parry’s (2a, b, e)). Cairo Montenotte (5) a. I sun turnòi a piè u liber, 3pl.scl= be.prs.3pl return.ptcp.mpl ad get.inf the book, ma u n’i’èra ciû! but 3msg.scl= neg=there=be.ipfv.3sg more ‘They went back to get the book, but it wasn’t there anymore!’ b. I an turnò a piè u liber, 3pl.scl= have.prs.3pl return.ptcp ad get.inf the book, #ma u n’i’èra ciû! but 3msg.scl= neg=there=be.ipfv.3sg more ‘They took the book again #but it wasn’t there anymore!’ c. A=m sun turnò=me a lavè. I=1sg.refl be.prs.1sg return.ptcp=1sg.refl ad wash.inf ‘I got washed again.’ In (5a), in its lexical sense turnè takes the be auxiliary as is generally the rule for verbs of motion; in (5b) by contrast turnè expresses repetitive aspect and its auxiliary have is determined by the lexical verb piè ‘get, take’, while in (5c) the auxiliary is again be, though now as required by the lexical reflexive lavèse ‘wash (oneself)’ rather than by turnè. A different kind of development concerns the paradigm of the auxiliary itself; we move on therefore in the following sections to the variety of ways in which the form of the item in its auxiliary function has become distinct from its continuing lexical companion. The first possibility we will consider is the one that would seem intuitively the most natural: divergence of form to accompany divergence of function. This is found both across the whole family with the have verb, which we consider in §3.2, and in more restricted fashion with the go verb, which is the

96

nigel vincent and max w. wheeler

focus of §3.3; §3.4 and 3.5 are devoted instead to different and cross-linguistically less well-studied routes leading to formal divergence. In §3.6 we draw some general conclusions.

3.2 Forms of habere ‘have’ Surviving Latin texts and the grammars that are built on them attest to a regular second conjugation verb habere with the principal parts: habeo.prs.1sg, habere.inf, habui.pfv.1sg, habitum.ptcp. At the other extreme, its phonetically reduced form as a future suffix can already be seen in the Strasbourg Oaths of 842 ce much as it is in modern French: with salvarai lit. ‘save.inf.have.prs.1sg (= I shall save)’, prindrai lit. ‘take.inf.have.prs.1sg (= I shall take)’. In what follows we focus on the ways those forms have developed and in so doing demonstrate not only the trajectories of divergence but also the need to integrate attested and reconstructed data in order to achieve a complete account (cf. also the discussion in §2.3 in this volume). In fact, Lat. habere undergoes paradigm split, with three sets of forms in the present tense that need to be reconstructed for the proto-Romance period. Table 3.1 Proto-Romance paradigm variants of habere A 1sg 2sg 3sg 1pl 2pl 3pl 1sg 2sg 3sg 1pl 2pl 3pl 1sg 2sg 3sg 1pl 2pl 3pl

Present indicative ˈabjo ~ ˈajjo ˈaves ˈavet aˈvemos aˈvetes ˈavent Present subjunctive ˈabja ~ ˈajja ˈabjas ~ ˈajjas ˈabjat ~ ˈajja aˈbjamos ~ ajˈjamos aˈbjates ~ ajˈjates ˈabjant ~ ˈajjant Past imperfect indicative aˈvia aˈvea aˈvias aˈveas aˈviat aˈveat aviˈamos aveˈamos aviˈates aveˈates aˈviant aˈveant

B

C

aj ~ o as at aˈvemos aˈvetes ant

aj ~ o as at ˈemos ˈetes ant

ˈia ˈias ˈiat iˈamos iˈates ˈiant

layering and divergence in romance periphrases

97

Table 3.1 shows the forms we believe habere already displayed before Romance orthographies were developed. They are represented in a generalized (Western) proto-Romance phonology. Column A in Table 3.1 displays the forms which are most likely to be the regular reflexes of habere. That said, we need at once to address an issue which is sometimes underemphasized in diachronic inflexional morphology: what exactly are the developments that are to be attributed to regular sound change? The outcome of word-medial /-bj-/ and parallel /-vj-/ is not straightforwardly regular in most Romance languages. Despite what some manuals imply, or assert, the evidence does not show that there is a rule for /-bj-/ and /-vj-/ (with certain exceptions to be explained away). Rather, we see two (or more) possible outcomes, lexically distributed, some more conservative or stronger, possibly retaining labiality, others more advanced or weaker—always non-labial. Examples from different parts of the Romance territory are set out in (6): (6) (old) French: /ʤ/ rubea > roge ‘red’, rabia > rage ‘anger’, cauea > cage ‘cage’, Nouientum > Nogent (toponym) vs /j/ habeas > aies ‘have.2sg. prs.sbjv’, debeo > dei ‘ought.1sg.prs.ind’, *ˈplɔvja > pluie ‘rain’, auiolo> aïeul ‘grandfather’, Nouiomagum > Noyon (toponym) (Nyrop 1935:445f.) Occitan: /bj/ rabia > rabia ‘anger’, cauea > gabia ‘cage’ or /wʤ/ rabia > rauja ‘anger’, cauea > cauja ‘cage’, alleuiat > aleuja ‘lightens’ vs /ʤ/ ~ /j/ debeat > deja ‘ought.3sg.prs.sbjv’, rubeo > roi ‘red’, *ˈplɔvja > plueja ‘rain’ (Grandgent 1905:63) Catalan: /bj/ rabia > ràbia ‘anger’, cauea > gàbia ‘cage’, or /wʤ/alleuiat > alleuja ‘lightens’ vs /ʤ/ ~ /j/ debeat > deja ‘ought.3sg.prs.sbjv’, rubea > roja ‘red’, *ˈplɔvja > pluja ‘rain’ (Recasens 2017:198) old Spanish: rabia > rabia ‘anger’, rubeo > rubio ‘fair’, pluuia > lluvia ‘rain’, vs fouea > foya ‘pit’, rubeo > royo ‘fair’ (Menéndez Pidal 1962:147; Lloyd 1993:421f.) Galician/Portuguese: rabia > ravia ~ raiva ‘anger’, pluuia > chuvia ~ choiva ‘rain’, rubeo > ruivo ~ roiva ‘red-haired’ vs *ˈfɔvjo > fojo Gal. foxo ‘pit, hollow’, *leˈvjarjo > ligeiro ‘light’ (Ferreiro 1999:113, 179) (old) Italian cauea > gabbia ‘cage’, rubeo > robbio ‘red’, debeat > debbia ‘ought.3sg.prs.sbjv’, habeo > abbio ‘have.1sg.prs.ind’ vs debeo > deggio ‘ought.1sg.prs.ind’, *ˈplɔvja > pioggia ‘rain’, habeo > aggio ‘have.1sg.prs.ind’ (Rohlfs 1966:I:386f.; Tekavčic´ 1972:I:262f.). In some cases the lexical split is evidently early, since, for example, rabia has the stronger variant in all the daughter languages, whereas *ˈplɔvja has the weaker variant in both Gallo-Romance (including Catalan) and Italian. The lesson from this is that it is unsafe to assume that the reflexes of the first-person singular

98

nigel vincent and max w. wheeler

indicative form */ajjo/ display analogical reformation, even though irregular phonological reduction is found elsewhere in the paradigm. Paradigm A, then, is what is to be expected as the regular outcome of the present indicative of Lat. habere. Reflexes of it—that is, those parts distinct from paradigm B—were not unusual in old Ibero-Romance, and in Italo-Romance. Or, rather, some parts of the pattern are found, though seemingly no variety maintains the whole ‘strong’ N-pattern, that is to say, the set composed of the three singular persons plus the third-person plural (Maiden 2018a:ch.6). In Iberia, a first-person singular from A is attested only in the Galician-Portuguese form aio seen in (7). (7) de todo herdamento … quanto eu ayo ena of all property how-much I have.prs.ind.1sg in.the villa que dizen Canedo (Glc., 1270) town that call.prs.ind.3pl Canedo ‘of whatever property … I have in the town called Canedo.’ It is the rest of the N-pattern that is present in old Spanish. The 2sg aves, 3sg ave, 3pl aven are quite common, for example in the works of Berceo (before 1264) and the Libro de Alexandre (first third of the thirteenth century), and with some occurrences still in the later fifteenth century. One might suppose that these ‘full’ forms would be found primarily with aver as a lexical verb, or in nuclear stress positions, but this does not seem to be the case, at least in the early period, where they are also found in present perfects,⁴ and in the deontic aver a and aver de constructions (8). (8) a. pésanos de la onta que tú regret.prs.ind.3sg =1pl of the affront that you.sg aves tomada (OSp.) have.prs.ind.2sg suffer.ptcp.fsg ‘we are sorry for the affront you have suffered.’ (Alexandre) b. ellox le rrexpondieron: el llobo çe lo they 3sg.dat= reply.pfv.3pl the wolf refl = it = abe komido (OSp.) have.prs.ind.3sg eat.ptcp ‘they replied: the wolf has eaten it.’ (c.1370 Poema de Yúçuf ) c. y vos, los dichos Condestable, y los Caballeros …, que and you the.pl said.pl Constable and the.pl knight.pl that

⁴ It is perhaps worth emphasizing that all these occurrences are in verse texts: Berceo, Alexandre, Yúçuf.

layering and divergence in romance periphrases

99

con el Rey nuestro Señor aven de ir (OSp.) with the King our lord have.prs.3pl comp go.inf ‘and you, the said Constable and knights …, who are to go with our lord the King.’ (1439 Conde de Haro (Pedro Ferna´ndez de Velasco), El Seguro de Tordesillas) By the sixteenth century, full (A) forms were current, it seems, only in morisco Spanish (that is, the variety of Spanish Muslims), as in (9). (9) a. el rrey al-Hariz abe en voluntad de pelear (OSp.) the king al-Hariz have.prs.3sg in will comp battle.inf ‘King Al-Hariz has in mind to do battle.’ (a. 1600 Libro de las batallas) b. Tres son ke no les aben three be.prs.3pl comp neg 3pl.dat= have.prs.3pl rrepintençia (OSp.) repentance ‘there are three (sins) for which repentance is insufficient.’ (a. 1600 Relatos moriscos) At first sight, Cat. 1sg haig, used in the haver de ‘must/ought’ construction, looks as if it also comes from paradigm A *ajjo, but it is not attested until c.1600. It is surely, therefore, an analogical restitution, in accord with the L-morphome, that is to say, the first-person singular of the present indicative plus all of the present subjunctive, see Maiden (2018a:ch.5). It involves the *ajj- stem of the present subjunctive (haja ~ hagi), and is supported by the parallel forms of go: vaig, vaja, etc. (see §3.3). When we come to the early stages of Italo-Romance, first-person singular and third-person singular forms from paradigm A are found, especially 1sg abbo ~ aggio ~ aço ~ aio and occasionally abbio (Rohlfs 1966:§541); less commonly, it seems, 3sg ave.⁵ A greater proportion of full (A) forms is attested in (old) Sicilian: 1sg aiu, 3sg avi, 3pl avenu ~ avinu. Second-person singular forms from paradigm A are apparently rare in Italo-Romance (Monaci 1955; OVI; Salvi and Renzi 2010), but a full A paradigm is found in modern Neapolitan aggio ~ aio aje ave avimmo avite àveno (Ledgeway 2009:383). As in old Ibero-Romance, the A forms are not restricted to uses with the full lexical meaning ‘have, possess’ but occur in a wide range of syntactic and prosodic contexts. Romanian displays from paradigm A, at most, the 3pl au < *ˈavunt (Conjugation II -ent having been replaced generally by Conjugation III -unt, as largely in Gallo-Romance), and in old Romanian this form au was extended analogically to the third-person singular, on the model of other paradigms that make no number distinction in the third person (Vasiliu and Ionescu-Ruxândoiu 1986:179). True, ⁵ Most cases of ave in OVI and all of the 3pl aveno are past imperfect forms.

100

nigel vincent and max w. wheeler

manuals (for example, Dimitrescu 1978:295) reconstruct a paradigm reflex of A, *ˈaibu, *ˈae, *ˈae, *aˈemu, *aˈeti, au, of which aibu in fact occurs, but only as a first-person singular present subjunctive form (Rosetti 1946:203), alongside aibă < habeam/habeat. But the attested Romanian forms apart from au go back rather to paradigm B, or its analogical developments, as we explain below. Paradigm A, insofar as it differs from paradigm B, thus tends to be defective, at least outside southern Italy. It is paradigm B, or some variant, that predominated, or was exclusive, as a lexical verb, and as a tense/aspect auxiliary, in a variety of prosodic contexts. The characteristic of paradigm B is monosyllabic word forms in the N-pattern (or rather, in the context of all the forms affected, an asyllabic root). Such reduction from the etymological paradigm A has been attributed to the verb’s clitic or pretonic status as a tense/aspect auxiliary.⁶ But this condition is neither necessary nor sufficient to explain the change, which comes to affect all the uses of habere. The essential point, as we see it, is rather that habere joins a small set of verbs that have monosyllabic forms in the N-pattern (singular and third-person plural) of the present indicative (Maiden 2018a:122–140). In Latin there were three such: esse ‘be’: sum es est sunt; dare ‘give’: do das dat dant; and stare ‘stand’: sto stas stat stant. These three verbs also had monosyllabic forms in the N-pattern of the present subjunctive. Three other verbs had monosyllabic forms in the second-person singular and third-person singular indicative only: ire ‘go’: is it; fieri ‘become’: fis fit; uelle ‘want’: uis uult. Though originally these latter verbs will have contributed to supplying models for the general pattern, their own forms largely do not survive. Widely in Romance, a handful of other verbs are attracted to the monosyllabic N-pattern, wholly or in part, adopting etymologically irregular forms: habere ‘have’, facere ‘do’, sapere ‘know’, uadere ‘go’. The development is most complete in Italian. By analogy with etymological dare: do dai dà danno, and stare: sto stai sta stanno, we get avere: ho hai ha hanno; fare: fo, fai fa fanno; sapere: so sai sa sanno; andare: vo, vai va vanno, although in the modern standard language the originally Tuscan forms fo and vo have in turn been replaced by the disyllabic forms faccio and vado. The criteria for joining this group were: being highly frequent, being of the 2nd or 3rd conjugation, and having /a/ as the root vowel. Once again, then, we see a form pattern which develops on its own terms and independently of the content it expresses. Apart from habere itself, to which we return shortly, etymologically irregular shortened forms of uadere are widely found: Italian, as just mentioned, French:

⁶ For a recent reassessment of the relation between stressed and unstressed forms, and the clitic status of the latter, together with rich documentation from early Italo-Romance sources, see Formentin (2020).

101

layering and divergence in romance periphrases

French ai as a avons

ate hanno

avé i l’àn

eve han

avì à

(ha)veis han

vais han

(av)éd ant ~ ont

avez ont

Romanian aux

FrancoProvençal é âs at (av)ens

Romanian lexical verb

Vallader n’ha hast ha vain

Sardinian

Sursilvan hai has ha (ha)vein

Portuguese

Romagnol ò è à avèn

Galician

Piedmontese ai ~ heu has ha oma

Spanish

Ligurian ai ~ on1 ti ài ~ ar a amu

Catalan

1sg 2sg 3sg 1pl 2pl 3pl

aggio hê ha amme

Occitan

2pl 3pl

ho hai ha avemo → abbiamo avete hanno

Acadian French

1sg 2sg 3sg 1pl

Neapolitan

Standard Italian

Table 3.2 Modern Romance present indicative paradigms derived from habere*

ai as a avons avez avont2

ai as a avèm avètz an

he has ha h(av)em3 h(av)eu han

he has ha hemos habéis han

hei has ha habemos habedes han

hei hás há havemos haveis hão

appo as at amus azes an

am ai are avem avet¸i au

am ai a am at¸i au

*

h- in these forms is purely orthographic–etymological. /o/ as in Italian, plus /-n/ analogical with sun < sum ‘I am’. 2 On this paradigm, see Smith (2011b). 3 The forms havem, haveu are standard albeit obsolescent or obsolete, but were normal in old Catalan. 1

vas va; Occitan, Catalan, Spanish: vas va van; Galician: vas van (Portuguese only 3pl vão); Spanish also 1sg voy, parallel to doy ‘give.1sg’ and estoy ‘be.1sg’ (< sto), though the source of -y in all of these is obscure. Shortened forms of facere are somewhat less widespread, and outside Italian, sapere has short forms only of the 1sg. In modern times, Romance languages use versions of paradigm B both as a lexical verb and as the present perfect auxiliary, as set out in Table 3.2. For the modern first-person singular of habere (Table 3.2) four trends can be distinguished. Forms like aggio, appo, and Fr./Occ. ai are inherited from paradigm

102

nigel vincent and max w. wheeler

A *ˈabjo ~ *ˈajjo,⁷ and tend to conform to the L-pattern morphome.⁸ It. ho (with parallel Italo-Romance forms) is modelled on the other monosyllabic present N-stems (see above), specifically, inherited do, sto in line with the rest of the present indicative paradigm.⁹ For Cat./Sp. he and Glc./Pt. hei a form *ai must be reconstructed—as we have done in Table 3.1. (*ajjo would give */aʤ/ in Catalan and */ajo/ in Spanish, Galician, and Portuguese.) This *ai has always been mysterious, but we think worth taking seriously the idea that the model for /-j/, expressing ‘1sg’, after /a/, is to be found in the inflexional endings of the perfective paradigm of the 1st conjugation—*-ai -asti -ˈat/-aut -amos -astes -aront. We mention later in this chapter other cases of elements of auxiliary paradigms being modelled on inflexional suffixes. Fr./Occ. 1sg /-j/ can also be seen in the first-person singular present indicative of esse ‘be’: OFr. sui (→ Fr. suis), OOc. sui ~ soi, which seem to be modelled on 1sg.pfv fui.1⁰ The Fr./Occ. ai may be interpreted in the same way, though a purely etymological source (< *ajjo) is also available here. Romanian 1sg am is likewise problematic. The manuals speak of syncretism with the 1pl am < *aem < *avemos, and while syncretism of first-persons singular and plural present for just this verb is hard to credit, this same syncretism is systematic in the Romanian imperfect indicative, e.g. făceam ‘I/we did’, mergeam ‘I/we went’.11 The modern second-person singular forms all go back to *as; and the third-person singular to *at, except for the Romanian main verb form are. This is said to be derived from the Latin third-person imperfect subjunctive haberet, but there are no parallels for such a shift, which is implausible on general grounds—an unmarked category (present, indicative) being reformed on the model of a marked one (past, subjunctive). We should bear in mind rather, we think, the mutual influence in Romanian between the paradigms of the perfect auxiliary have, and the future auxiliary from a vrea (< *voˈlere) ‘want’. The future auxiliary has modelled its 3sg va on a < *at; reciprocally, are, in turn, will have been modelled on 3sg vare < *voare < *ˈvɔlet.12 The third-person plural forms are from *ant, except Ro. au < *ˈavunt. Fr. ont may be from *ˈavont

⁷ French ai was [aj] in old French, subject to monophthongization of diphthongs > [e] in the Middle French period. ⁸ L-pattern = the set of paradigm cells consisting of the 1sg.prs.ind and all the present subjunctive (Maiden 2018a:ch.5). ⁹ Or < *ao, for those varieties where the 1sg is /ɔ/ rather than /o/, and where the other monosyllabic stems have forms from *fao, *stao*, *vao. 1⁰ Nor should we discount the possibility that this is the source of the mysterious Sp. 1sg -y in soy, doy, estoy, voy. 11 One hypothesis, admittedly speculative, is that am was modelled after sum (1sg.pr.ind of esse ‘be’)—attested su, later replaced by regular syncretism 1sg = 3pl sînt (> sunt/sânt). 12 The lexical verb paradigm of a vrea ‘want’ was later remodelled on the basis of the infinitive form, giving sg vreau vrei vrea. For other examples of contamination of a avea with a vrea, cf. colloquial a vrea forms 3sg o, 3pl or, etc., see Lombard (1974:273), Ledgeway (2012:129 n.57).

layering and divergence in romance periphrases

103

and/or modelled on sont < sunt ‘they are’. Spanish has 3sg.prs.ind hay in the existential copula = ‘there is/there are’, where the -y is a relic of the locative clitic hi (< hic and/or ibi ‘there’) that was used much more widely in old Castilian up to c. 1300 (and much later in Aragonese texts). The complement of the N-pattern in the present indicative of habere, that is to say the first- and second-person plural forms, follows a different trajectory, as is typical of Romance paradigms with stem allomorphy. Whereas the evolution of monosyllabic forms in the N-pattern dates to the proto-Romance period, the corresponding trend in the complement of it, if it takes place at all, does so in the period of recorded history. Many varieties retain here the etymological forms of habere, subject only to modifications such as affect all verbs. Ro. avem avet¸i are not as archaic as they look. Intervocalic -v- is regularly lost. Here a -v- which is anti-hiatic in 1sg.pfv avui < *aui < habui, ptcp avut < *aut < *aˈvuto has been extended analogically (Dimitrescu et al. 1978). In the auxiliary paradigm, however, the /a-/ allomorph of the root is extended from the N-pattern to its complement: am at¸i.13 The same extension can be seen in WCat. ham hau, and also in Neapolitan and Sardinian, though not restricted to the paradigm of the perfect auxiliary. The oddest feature in Table 3.2 is the Spanish paradigm, where, anomalously, the first- and second-persons plural have not followed parallel paths. Here it is possible to trace the evolution in a Spanish corpus (CORDE). Until c. 1500 most cases of hemos ~ emos are examples of paradigm C, that is, the future morph split from the preceding infinitive by a clitic pronoun, as in the following late example: (10) Vamos, ya que habra´n comido, e atarlos go.imp.1pl now comp have.fut.3pl eat.ptcp and tie.inf=them hemos e pagarnos han las gallinas e tortas have.prs.1pl and pay.inf=us have.prs.3pl the hen.pl and cake.pl e bollos que los enviamos (OSp., 1535–1557) and bun.pl comp 3pl.dat=send.pfv.1pl ‘Let us go, now that they will have eaten, and we will tie them up, and they will pay us for the hens, cakes and buns we sent them.’ This pattern of analytic future formation, which is still current in Portuguese, was common in Spanish too until 1500, after which it disappears quite suddenly. In other contexts, that is, as a lexical transitive verb, as a perfect auxiliary, and in the

13 On the divergence of the habere paradigm in Romanian, see Smith (2021:512f.).

104

nigel vincent and max w. wheeler

aver de, aver a (lit. ‘have of, have to’), deontic constructions, avemos (various spellings) was predominant over hemos until 1600, but residual after 1700; see data in Table 3.3. Table 3.3 Spanish hemos / (hemos + habemos) (%)

hemos

1500– 1549 44

1550– 1599 42

1600– 1649 67

1650– 1699 74

1700– 1749 95

1750– 1799 96

1800– 1849 99

Where habemos survives in use in the nineteenth century, it is particularly in the deontic construction habemos de ‘we should’, and in the idiom habemos menester ‘we need’. Whereas Spanish 2pl hedes ~ edes ~ heis was virtually never used outside the analytic future construction, and consequently disappeared with it, hemos appears quite early outside that construction, though at first with low frequency. Before 1250, the CORDE corpus has just three cases, two of hemos as a main verb (11a, b) and one as a perfect auxiliary (11c). (11) a. uendemos a uos Gutier Rodriguez ye a uestra muyer, Beneita sell.prs.1pl to you Gutier Rodriguez and to your wife Beneta Perez, tres terras que hemos en Uillagalegos (OSp.) Perez three land.pl comp have.prs.1pl in Villagalegos ‘we sell to you G. R. and to your wife B. P. three parcels of land that we have in Villagalegos.’ (1243 Doc. León) b. assaz emos razón, materia que digamos (OSp.) sufficiently have.prs.1pl content matter comp say.prs.sbjv.1pl ‘we have plenty of content and matter to tell.’ (1240–1250 Alexandre) c. e pornemos los cuerpos e quant’ and put.fut.1pl the body.pl and how.much emos ganado. (OSp.) have.prs.1pl win.ptcp ‘and we will contribute our persons and whatever we have won.’ (1240–1250 Alexandre) Such examples of hemos reappear only after 1350, as a main verb (12a), as a perfect auxiliary (12b), and in the deontic constructions haber a (12c), and haber de (12d): (12) a. quantos oy dı´a hemos e avremos de how.many today day have.prs.1pl and have.fut.1pl from aquı´ adelante (OSp.) here onwards ‘all those that we have today and shall have henceforth.’ (1358, Los familiares de Mari Llorente perdonan a su marido)

layering and divergence in romance periphrases

105

b. Esto que emos tractado et concertado this comp have.prs.1pl discuss.ptcp and agree.ptcp entre nosotros (OSp.) between us ‘this (matter) that we have discussed and agreed between us’. (1376–1391, Heredia, Gran Crónica de España)1⁴ c. otorgamos e conosçemos que devemos e grant.prs.1pl and acknowledge.prs.1pl comp owe.prs.1pl and hemos a dar e pagar a vos (OSp.) have.prs.1pl to give.inf and pay.inf to you ‘we grant and acknowledge that we owe and must give and pay to you…’ (1475 doc. Mondragón) d. San Agostin en el libro que compuso Saint Augustine in the book comp compose.pfv.3sg del cuidado que hemos de tener of.the care comp have.prs.1pl comp have.inf de los muertos (OSp.) of the.pl dead.pl ‘St Augustine in the book he wrote about the care we ought to give to the dead.’ (1458, Pero Dı´az de Toledo) While we have been able to chart the rise of hemos, we have not identified why hedes ~ heis did not follow it, or indeed, take root at all, outside the split future paradigm. We can note, however, that the first-person plural indicative of aver is considerably more frequent than the second-person plural. For example, in the second half of the thirteenth century, there are in the CORDE corpus 3,553 cases of avemos/emos compared with 957 of avedes/edes. The shift from long forms to short forms in the first- and second-persons plural of the present indicative can also be traced, in part, in the Catalan corpora. In Catalan, outside the split future paradigm (C), the CICA corpus attests only a handful of the short forms hem and heu before 1650, namely eight of hem, from 1450 on, of which seven are perfect auxiliaries, and three certain cases of heu, from 1486 on, of which two are perfect auxiliaries.1⁵ Such a distribution is consistent with the hypothesis that shorter, more grammaticalized, variant forms spread first to more grammaticalized contexts (here, perfect auxiliary) before less grammaticalized contexts.1⁶ The modern Catalan CTILC corpus has texts from 1832 to 2008. In it, hem and heu were the majority forms by the mid nineteenth

⁵2 Many other examples of emos both as main verb and as perfect auxiliary are attested in the Aragonese Heredia. 1⁵ The other is in the idiom heu mester ‘you need’ 1522–1524. 1⁶ We suggest this as likely with the spread of short B forms into prosodically strong, less grammaticalized, contexts that should have originally favoured paradigm A, though such a distribution is no longer evident at the period of the earliest textual data.

106

nigel vincent and max w. wheeler

Table 3.4 Growth of Catalan hem, heu, with respect to havem, haveu, over the last two centuries % hem % heu

1832–1863 58 35

1864–1893 70 75

1894–1923 76 76

1924–1953 92 89

1954–1983 98 97

1984–2008 99.9 99

century, and became virtually exclusive after c.1950, as is clear from the figures in Table 3.4. Catalan has evolved two further paradigms of the present tense of habere. The first has indicative: hec heus heu havem haveu heuen; subjunctive: hegui heguis hegui haguem hagueu heguin; infinitive haver ~ heure.1⁷ These forms may look as if they derive, in part, from paradigm A, but that is not so. Insofar as they differ from the reflexes of paradigm B, they are not attested in the CICA corpus which has texts up to 1650. They appear in the CTILC corpus with texts from 1832. The new forms are analogical, developing in accord with the Catalan G-morphome (the sum of PYTA + ptcp + L-pattern;1⁸ Wheeler 2011), given that early Catalan had already developed stems of aver in -g- in the PYTA + ptcp morphome: e.g., 3sg.pfv hagué, pst.sbjv.3sg hagués, ptcp hagut. Heure thus follows the models of the paradigms of jeure ‘lie’, treure ‘take out’, and so on. The heure paradigm is used primarily in just three constructions: (1) a transitive verb ‘get, obtain’, that is to say, the heir of the telic senses of habere, the atelic senses ‘possess, have’ having been taken over by tenir (< Lat. tenēre ‘hold, keep’); (2) in the idiom heure esment de ‘detect, be aware of ’;1⁹ and (3) in the idiom haver/heure-se-les amb ‘have to do with, deal with’, as in (13). (13) ens les havem amb gent que no refl.1pl= les2⁰= have.prs.1pl with people comp neg toca de peus a terra (Cat.) touch.prs.3sg of foot.pl to earth ‘we are dealing with people who have not got their feet on the ground.’ (1981 CTILC) Though these three constructions are well attested in the modern corpus, they are not particularly frequent. It is perhaps surprising, then, that a new paradigm

1⁷ In Valencian, a variant has /a/ in the first syllable throughout. 1⁸ PYTA = the set of sub-paradigms originally using the Latin perfectum verb-stem allomorph, namely, in Catalan, the preterite, the past subjunctive, and the conditional/past subjunctive derived from the Latin pluperfect (Maiden 2018a:ch.4). 1⁹ Curiously, haver was not used with esment in old Catalan; the construction then was (telic) prendre esment de ‘detect’, or (atelic) tenir esment de ‘be aware of ’. ⁵⁸ The pronoun les is formally 3pl.f.acc, but it is non-referential in this idiom.

layering and divergence in romance periphrases

107

evolved for them. The other new paradigm is a variant of the indicative paradigm B, used as perfect auxiliary, especially in southern Catalonia and Menorca. The forms become proclitic, unstressed, and undergo vowel reduction: [i] ~ [әj] [әs] [ә] [әm] [әw] [әn]. In sandhi after a vowel the [ә] disappears, so all that remains are the person–number affixes.21 The present subjunctive of habere in Catalan had only the expected forms derived from those in Table 3.1 until 1600 at least. By the nineteenth century, however, the dominant forms of the first- and second-persons plural were rhizotonic hàgem ~ hàgim, hàgeu ~ hàgiu, and in the twentieth century the oxytone variants had virtually disappeared, except, that is, in the innovative heure paradigm just mentioned. Rhizotonic first- and second-person plural forms are entirely anomalous in the present tense paradigms of Catalan. How did they get into habere? In fact, columnar stress, with rhizotonic first- and second-person plural forms, is universal in past tense paradigms of Catalan, that is, in the imperfect indicative, the synthetic preterite, the conditional, the past subjunctive in /-s-/, and in the Valencian past subjunctive in -/r-/ derived from the Latin pluperfect indicative. Although columnar stress in the past tense too is an innovation, relative to the proto-Romance paradigms derived from Latin, there, it is very ancient, from before the first vernacular texts. The only plausible link between the present of habere and the past tense paradigms of verbs in general is that habere as an auxiliary verb denotes, in the present perfect, ‘anteriority’ or ‘past with present relevance’. We may say that the present subjunctive of perfect habere adopts the stress pattern of past subjunctives, or of past tense forms in general. (See below for a parallel innovation in the ‘present’ subjunctive of the auxiliary derived from uadere ‘go’.) But observe that this stress pattern is not confined to the subjunctive of habere in the function of a perfect auxiliary but extends to the present subjunctive of habere in other contexts, in particular, in the haver de deontic construction, as in example (14). (14) En el bany: Encara que us hàgiu de in the bathroom although comp= 2pl.refl= have.prs.sbjv.2pl comp rentar o de banyar de pressa la majoria de les vegades, wash.inf or comp bathe.inf of haste the majority of the time.pl procureu seguir, almenys un cop a la setmana els consells try.imp.2pl follow.inf at.least one time at the week the advice.pl segu¨ents … (Cat., 1985) following.pl ‘In the bathroom: although you may have to wash or shower in a hurry on most occasions, try to observe the following advice at least once a week…’

21 There are further cases of morphologically interesting divergence in the present paradigm of habere in perfect auxiliary and in future/deontic functions in the dialects of southern Italy (see, e.g., Schirru 2018). Space does not allow us to explore them further here.

108

nigel vincent and max w. wheeler

Again we observe a morphological innovation justified by a particular context of use being extended beyond the scope of that context. Proto-Romance paradigm C in Table 3.1 is a variant of paradigm B that is used in its present tense forms as a future suffix morpheme, except in Sursilvan,22 many southern Italian dialects such as Lucanian, Sardinian, and Romanian, in which periphrastic futures with verbs other than habere are used. The -av- element is absent not only in the N-pattern as in paradigm B, but also in the first- and secondpersons plural, and throughout the (formally) past imperfective paradigm which constitutes the conditional or future-in-the-past affix. That is, the -av- element is absent when it would have been pretonic (or pre-pretonic). In Italo-Romance, first-person singular future endings deriving from paradigm A (-aggio, -agio, -aio, -abbo) are found both in older texts and in some regional varieties (Loporcaro 1999a) but have not survived in modern standard Italian. The Western Romance infinitive + habere future/conditional is a creation of the proto-Romance dark age. Before the earliest Romance texts, the new future had replaced the Latin synthetic future paradigms,23 and, indeed, the periphrases available in Latin such as that consisting of the future active participle + esse.2⁴ This particular grammaticalization process is hardly observable in the written evidence available to us;2⁵ it appears fully entrenched in the Strasbourg Oaths of 842 ce, as mentioned earlier.

3.3 Forms of the go verb Although, as we have seen, the have verb has a variety of different forms and distributions, they all develop via the mechanisms of linguistic change from the one Latin verb habere. When we come to the go verb, things are rather different. Outside of Daco-Romanian, the go verb in all the Romance languages derives from a combination of one, two, or three Latin lexical ingredients: ire ‘go’ (abbreviated below as i-), uadere ‘advance, go’ (abbreviated below as vad-), and what we can call the a-verb (abbreviated below as all-/an(d)-), an item of contested etymology (Alinei 2009), showing up as Fr. aller, It. andare, Occ./Cat. anar, and the

22 Sursilvan is presumed to have replaced the general Western Romance pattern, seen in other RætoRomance varieties, with the new periphrasis from uenire ‘come’. 23 Except for the suppletive synthetic future of esse ‘be’, whose singular survived in old French and old Occitan alongside forms from *ˈεssere + habere, and whose 2sg form eris is the putative source of Sp. eres ‘be.2sg.prs.ind’. 2⁴ On the loss of the periphrasis facturus esse ‘to be about to/intend to do’, see Vincent and Bentley (2001). 2⁵ Adams (1991:161) concludes ‘while infinitive + habeo was not yet grammaticalised as an exponent of futurity, that order was already by the fifth or sixth centuries rather more closely associated with the idea of futurity than was the reverse order. The relative infrequency of examples of habeo + infinitive which can plausibly be given a future interpretation supports the contention that it was the development of a marked order infinitive + habeo expressing emphatic obligation/necessity which decisively generated the new future.’

layering and divergence in romance periphrases

109

like.2⁶ With the exception of Sardinian, which has generalized the a-verb stem throughout the paradigm, in the remaining languages these verbs come together in a variety of suppletive patterns, representative instances of which are summarized in Table 3.5. Of particular importance for what follows is the distribution in the present tense, which is in accord with the so-called N-pattern (Maiden 2018a:192–204). Table 3.5 Suppletive stems in the go verb across Romance

Galician Spanish French Catalan Italian Sicilian Sardinian

prs N-forms

prs 1/2pl

pst.ipfv

prt

fut

inf

ptcp

vadvadvadvadvadvadand-

ivadallanandiand-

iiallanandiand-

(fu-) (fu-) allanandi—

iiian-iand(ej a + inf) (ai a + inf)

iiallanandiand-

iiallanandiand-

On the content side, in addition to serving as lexical items with the meaning of motion away from a given location, in different ways in different languages these verbs enter into a variety of temporal and aspectual periphrases (see Chapters 4 and 6 in this volume, for further discussion). And in many cases the forms of the verb remain the same across all the uses.2⁷ Thus, in (15) we see three different uses of It. andare with no difference in form for the lexical use (15a), the modal passive use (15b), and the progressive/continuative use (15c). (15) a. Paolo va alla stazione. (It.) Paolo go.prs.3sg to.the station ‘Paolo is going to the station.’ b. Questo libro va letto. (It.) this book go.prs.3sg read.ptcp ‘This book must be read.’ 2⁶ In addition, Spanish, Portuguese, and other Ibero-Romance varieties have a suppletive preterite fui, fuiste, etc. imported from the paradigm of ser ‘be’. For lack of space, we will not address the issues raised by this development in the present discussion. 2⁷ Another pattern of differentiation attested in some southern Italian dialects is for the lexical verb to involve the obligatory presence of reflexive and elative clitics similar to what is found with Fr. s’en aller, It. andarsene ‘go away’ but without the lexical opposition between ‘go’ and ‘go away’ which is found in those languages. Thus in Cosentino we have the contrast between (i) and (ii) (Ledgeway 2016a): (i)

**(Mi nni) vaiu a ru mare. (Cos.) 1sg= elat= go.prs.1sg to the sea ‘I’m going to the seaside.’ (ii) (**Mi nni) vaju fazzu na picca 1sg= elat= go.prs.1sg do.prs.1sg a bit ‘I’m going to do some shopping.’

’i of

spisa. (Cos.) shopping

110

nigel vincent and max w. wheeler

c. Il problema va complicandosi. (It.) the problem go.prs.3sg complicate.ger=3sg.refl ‘The problem is getting increasingly complicated.’ In Italian, this uniformity between lexical and grammatical uses holds for all relevant persons, tenses, and moods. By contrast, in Catalan, for some feature clusters, lexical and auxiliary forms differ. Thus, compare the first-person singulars in (16) and the first-person plurals in (17): (16) a. Vaig a l’ estació. (Cat.) go.prs.1sg to the station ‘I’m going to the station.’ b. Vaig seure. (Cat.) go.prs.1sg sit.inf ‘I sat down.’ (17) a. Anem a l’ estació. (Cat.) go.prs.1pl to the station ‘We’re going to the station.’ b. Vam seure. (Cat.) go.prs.1pl sit.inf ‘We sat down.’ The construction of interest here is the so-called perfet perifràstic formed originally from anar ‘go’ + infinitive and with the meaning of perfective past (Detges 2004; Wheeler 2017), which exists alongside the lexical motion sense. The relevant parts of the paradigms are set out in Table 3.6. Table 3.6 Paradigm split in modern Catalan go Lexical prs.ind

prs.sbjv

1sg 2sg 3sg 1pl 2pl 3pl 1sg 2sg 3sg 1pl 2pl 3pl

vaig vas va anem aneu van vagi vagis vagi anem aneu vagin

Auxiliary A B vaig vàreig vas vares va va vam vàrem vau vàreu van varen vagi vagis vagi vàgim vàgiu vagin

Although in the lexical verb the two suppletive stems /va-/ and /an-/ are distributed according to the N-pattern, there are very rare examples of the /va-/ stem leaking into the first- and second-person plural slots as in (18a–c).

layering and divergence in romance periphrases

111

(18) a. Senyor, nosaltres són hòmens qui vem per lo sire we be.prs.1pl men who go.prs.1pl through the món sercant la nostra fortuna (OCat.) world seek.ger the our fortune ‘Sire, we are men who go through the world seeking our fortune.’ (Xalabin: 87.13 first half of 15th c.) b. El seu pelatge, the 3.poss pelt aixı´ com llurs secretes intencions, thus as 3pl.poss.pl secret.pl intention.pl per evitar que hi haja confusions, for avoid.inf comp loc. have.prs.sbjv.3sg confusion.pl vam a mostrar-vos clarament des del principi go.prs.1pl to show.inf=you clearly since from.the beginning de l’ obra, ja que en= som autors. (Cat.) of the work now comp part be.prs.1pl author.pl ‘Their coats, as well as their secret intentions, to avoid confusion, we are going to show you clearly from the beginning of the work, as we are the authors of it.’ (Rodolf Sirera La pau (retorna a Atenes) 1975)2⁸ c. Per ço que demà vos en puscats entrar en for it comp tomorrow you= abl= can.prs.sbjv.2pl enter.inf into la ciutat de Gerona, vos manam que huy the city of Girona you.pl = instruct.prs.1pl comp today vos en vajats sens alguna falla a 2pl.refl= abl= go.prs.sbjv.2pl without any failure to Medinya, a una legua de Gerona. (OCat.) Medinyà at one league from Girona ‘In order that tomorrow you may be able to enter the city of Girona, we instruct you that you go today without fail to Medinyà, one league from Girona.’ (Ferran I, Epistolari 1416) These attested examples (18) would generally be regarded as ungrammatical. But, in the paradigm of the preterite auxiliary, by contrast, use of the va- stem throughout, without suppletion, is the only pattern that has ever been found. To explain why this is, we need to review the slow process by which anar + infinitive came to be grammaticalized as an analytic perfective construction; a full account is in Wheeler (2017). In the beginning the construction was very restricted, occurring only in indicative main clauses, with positive, declarative predicates of voluntary activity and telic Aktionsart, and, notably, only in the third ⁶⁶ The use of go.prs + inf as a future periphrasis is marginal in Catalan, though go.pst.ipfv + inf ‘was going to’, ‘intended to’ is quite well established.

112

nigel vincent and max w. wheeler

person. While the construction in the third person is attested in the earliest Catalan narrative texts dating from the second half of the thirteenth century, the construction with the first-person singular occurs first after 1350, for example in Ramon Muntaner’s Crònica. For the other persons, the earliest attestations are considerably later: the first plural not before 1500; the second singular not until 1595, and the second plural not at all in the CICA corpus (up to 1650), though it is present in the CTILC corpus which has texts from 1832. Indeed, the first example of the first-person plural of the periphrastic perfect occurs in the prescriptive Regles d’esquivar vocables grossers i pagesı´vols (1492) where it is the periphrastic construction itself that is censured (19a).2⁹ The form vam (Auxiliary paradigm A) is a morphological innovation, inducing a root /va-/ from the N-pattern present of go and/or forms already current of Auxiliary paradigm A, and affixing the regular first-person plural morph /-m/; and likewise for vau (/-w/ 2pl). The two remaining examples of the first-person plural of the periphrastic perfect in the CICA corpus display the form vàrem, from paradigm B in Table 3.6. Example (19b) is one of them. (19) a. [45/49] “vaig anar” e “vaig venir” per “anı´” e “venguı´”, e semblants [48] aquests vocables de “vaig anar” a misser Hierony Pau ne a mi, Pere Miquel Carbonell, no par sı´an bons vocables. Més val dir: “anam”, “venguem”; no: “vam [anar, vam venir]”, etc. (OCat.) ‘these expressions like “vaig anar” appear to Jeroni Pau, and to me, Pere Miquel Carbonell, not to be appropriate. Say rather “anam” [‘we went’], “venguem” [‘we came’], not “vam [anar, vam venir]”, etc.’ b. Aquest segon digué que era this second say.pfv.3sg comp be.pst.ipfv.3sg pare de Amor de Déu, lo qual vàrem trobar father of A. the whom go.1pl encounter.inf per lo camı´, ý vengué-se.n ab nós by the way and come.pfv.3sg=3refl=abl with us e aprés se n’anà. (OCat.) and afterwards 3refl=abl=go.pfv.3sg ‘This second one said that he was father of Amor-de-Déu, whom we encountered on the way, and he came along with us, and afterwards went off.’ (Spill de la Vida Religiosa f10r l13. 1515) That is, by the time the construction came, very gradually, to be extended from the third person to the other persons, the pattern in the third person was so entrenched that the semantic (and formal) link to the present of go had become obscure, and evidently too weak to impose 1pl anam, 2pl anau from the suppletive lexical go paradigm. 2⁹ Jeroni Pau and Pere Miquel Carbonell are generally understood to be the compilers of the whole collection of censured expressions.

layering and divergence in romance periphrases

113

Though van (paradigm A) is by far the predominant third-person plural auxiliary form in the CICA corpus, there are rare cases of varen (paradigm B) (20)—note also the third-person auxiliary van in the same passage: (20) prenı´ una granera en la mà i va tirar take.pfv.3sg a broom in the hand and go.3sg throw.inf la dita granera a la dita Bartomeva i aprés the say.ptcp.f broom to the say.ptcp.f Bartomeva and after es varen acostar la una a l’ altra i 3refl= go.3pl approach.inf the one to the other and van-sa pendra a cabayeras (OCat.) go.3pl=3refl seize.inf at lock.of.hair.pl ‘she picked up a broom and threw the said broom at the said Bartomeva. And afterwards they closed on one another and seized each other by the hair.’ (No serets tots temps batle 1, p. 23, l. 27. 1364–1369)

To explain these forms vàrem, varen, of paradigm B, containing -re-, we need to consider changes taking place contemporaneously in the Catalan synthetic perfective paradigm, particularly of the 1st (-a-) conjugation. (Exactly parallel changes took place in the other conjugations, but are less relevant to the development of the va- auxiliary.) The etymologically inherited forms of the 1st conjugation synthetic perfective are displayed in column A of Table 3.7, where the 3pl -aren derives regularly from -ārunt. After 1300 innovative forms (column B) with -re- start to be attested, at first, only the 1pl -rem. The other (2sg/pl) forms with -re- do not turn up until the last decade or so of the fifteenth century (Wheeler forthcoming:§10.6.3.3). Paradigm B quickly takes over as the norm after 1500.

Table 3.7 Catalan synthetic perfective, 1st conjugation, trobar ‘find’, old (A) and new (B)

1sg 2sg 3sg 1pl 2pl 3pl

Synthetic perfective A B trobé trobí trobest trobares trobà trobà trobam trobàrem trobàs trobàreu trobaren trobaren

114

nigel vincent and max w. wheeler

A form like trobàrem repairs the problem with trobam, namely that it was homonymous with the first-person plural of the present indicative.3⁰ The striking phenomenon affecting the paradigm of the va- auxiliary is that it develops alternative forms (Table 3.6, auxiliary B) modelled on the inflexional suffixes of the 1st conjugation synthetic form, according to the following proportion: trobà: va trobar:: trobaren → varen trobar. We have cited (20) a rare example in the CICA corpus of varen from the fourteenth century, and there is just one other from the mid-fifteenth century. This form becomes frequent only around 1600. In modern Catalan, both auxiliary paradigms of Table 3.6 are current in the CTILC corpus of written texts (1832–2008), which allows us to trace the relative frequency over the last two centuries of the elements of the alternative auxiliary paradigms which have coexisted for nearly 700 years. In this corpus, word forms are tagged, so, for example, auxiliary van can be distinguished from present tense main verb van ‘they go’. 1sg vàreig is constructed mirroring the variation between the remainder of the two paradigms, according to the analogy: vas (vam, vau, van): vares (vàrem, vàreu, varen):: vaig → vàreig, 1sg vàreig is much rarer than the other person–number forms, having no parallel, either in the synthetic perfective paradigm, or anywhere else in the system of verb inflexion. It accounted for up to 5% of the first-person singular forms in the corpus up to 1933, with only negligible numbers thereafter. Until the second third of the twentieth century, vàrem and vàreu were preferred over vam and vau (in the proportions 3:2 and 2:1, respectively), peaking at over 80% in the two decades 1854–1873. Between 1934 and 1993, vàrem declined to 24% and vàreu to 29%. Vares and varen were never predominant. Note that the alternatives vas and van are supported by the presence of these forms in the paradigm of the lexical verb go (unlike vam and vau). Vares accounted for 20% and varen for 41% up to 1933. These proportions declined to 9% and 13% by 1993. Since 1993, all the non-1sg -re- forms have declined to around 5%. They may have come to be perceived as less standard, though it is only vàreig that is excluded from the norm in formal registers (IEC 2016:250). Our impression, however, is that the B paradigm apart from vàreig is considerably more frequent in twenty-first-century spoken and written Catalan than the most recent corpus figures suggest. In the period from 1934, vàrem and vàreu continue to predominate over vam and vau in the Balearic texts in the corpus, and vares and varen are about twice as frequent in Balearic texts as in the corpus as a whole. The Catalan (and old French, old Occitan) periphrastic perfect indicative derived from uadere + infinitive originally had no corresponding subjunctive form. The synthetic past subjunctive, traditionally called ‘imperfect’, though it is aspectually neutral, continued to be used in contexts where a subjunctive was required.

3⁰ The problems with the second persons singular and plural of A are different: theme vowel -e- in -est is anomalous in a 1st conjugation verb, and the second-person singular suffix -st and second-person plural suffix -s are anomalous in the verb paradigm as a whole.

layering and divergence in romance periphrases

115

However, from the late nineteenth century onwards a new TAM category is attested, mostly in the central variety of Catalan (eastern continental Catalan) (see also §1.2.3 in this volume): a perfective subjunctive that uses subjunctive forms of the uadere auxiliary (Table 3.6) together with the infinitive. This usage is not very frequent, but it is acknowledged by descriptive and normative grammars and accepted as standard (IEC 2016:250–251). Since this TAM category is so unusual in Romance, it may be worth giving examples, here (21), one of each PN form, taken from the CTILC, except for the first-person plural item, which is via Google. (21) 1sg Personalment, estic molt content … que essent jo personally be.pr.1sg very pleased comp be.ger I primer Tinent de Batle del mateix Ajuntament, first deputy of mayor of.the same town.council vagi tenir sort de poder dur go.1sg.sbjv have.inf good.fortune comp be able.inf bring.inf a terme la rectificació del carrer de Jaume II (Cat.) to completion the remodelling of.the street of Jaume II ‘Personally I am very pleased … that when I was first deputy mayor of the same town council, I was fortunate to be able to bring about the remodelling of Jaume II street…’ (Miquel Forteza i Pinyà, 1966. Els descendents dels jueus conversos de Mallorca) 2sg allò de no dir al Tribunal que vajas sortir that of neg say.inf to.the court comp go.sbjv.2sg go.out.inf del cuartel ab llicencia meva. Tan mateix tens of.the barracks with permission 1sg.poss at.any.rate have.prs.1sg ben poca memori Jo sı´ que me’n well little memory I yes comp 1sg.refl=3.part= recordaré tota la vida (Cat.) remember.fut.1sg all the life ‘the matter of not telling the court that you left the barracks with my permission. At any rate you have a very poor memory. For my part, I will remember it all my life.’ (Antoni Careta i Vidal, 1905. Narracions estranyes) 3sg Fins i tot encara que aquest recull com a tal fos even though comp this anthology as such be.pst.sbjv.3sg atribuït només secundàriament al rei savi, no hi attribute.ptcp only secondarily to.the king wise neg 3loc= ha cap fonament per a dubtar que Salomó have.prs.3sg any basis for to doubt.inf comp Solomon

116

nigel vincent and max w. wheeler

vagi compondre un gran nombre de poesies go.sbjv.3sg compose.inf a great number of poem.pl didàctico-sapiencials (Cat.) didàctic-wisdom ‘Even if this particular anthology were to be attributed only secondarily to the wise king, there is no reason to doubt that Solomon composed a considerable number of didactic/wisdom poems.’ (Lluı´s Duch i Álvarez, 1973. L’Antic Testament) 1pl No és cert de cap manera que els dos batles neg be.prs.3sg correct of any way comp the two mayor.pl vàgim donar suport a la proposta, quan ens go.sbjv.1pl give.inf support to the proposal when 1pl.acc= go.3sg visitar el promotor de l’empresa. (Cat.) va visit.inf the promoter of the company ‘It is quite incorrect to say that we, the two mayors, gave support to the proposal, when the promoter of the company visited us.’ (2-04-2007 MENORCA) 2pl encara que us vaig entristir amb la carta, though comp 2pl= go.prs.1sg sadden.inf with the letter no em sap gens de greu; i, si abans neg 1sg= taste.pst.ipfv.3sg anything of serious and if before em sabia greu —perquè, de fet, veig 1sg= taste.pst.ipfv.3sg serious because of fact see.prs.ind.1sg que aquella carta, ni que fos per un comp that letter not.even comp be.pst.sbjv.3sg by one moment, us va entristir— ara me n’ moment 2pl= go.prs.3sg sadden.inf now 1sg.refl=3part= alegro, no pas perquè us vàgiu cheer.prs.1sg neg neg because 2pl.refl=go.prs.sbjv.2pl entristir, sinó perquè us vau sadden.inf but because 2pl.refl= go.prs.2pl entristir per penedir-vos. (Cat.) sadden.inf for repent.inf=2pl.refl ‘even if I made you sad with the letter, I am not at all sorry; and if I was sorry before—because, in fact, I see that that letter, did make you sad, even if just for a moment—now I am glad, not because you were sad, but because you became sad in order to repent.’ (Romuald Maria Dı´az I Carbonell, 1961 Epı´stoles de sant Pau) 3pl la pregunta que cal fer no és: the question comp be.necessary.prs.3sg make.inf neg be.prs.3sg per què es van extingir tantes espècies? for what 3refl= go.prs.3pl extinguish.inf so.many species

layering and divergence in romance periphrases

117

sinó: com és que encara vagin quedar but how be.prs.3sg comp yet go.prs.sbjv.3pl remain.inf éssers vius sobre la Terra? (Cat.) being.pl alive.pl on the Earth ‘the question one must ask is not “Why did so many species become extinct?” but “How is it that any living beings survived on Earth?”’ (Francesc Nicolau i Pous, 2007. Panorama actual de la paleontologia) These examples mostly illustrate the typical context in which the construction with the new TAM category appears, namely, in a subordinate clause dependent on a negative polarity matrix clause. The first example is a little different—a clause depending on a mental state verb, estic content … que ‘I am pleased that’. Morphologically the noteworthy feature of this new paradigm, in addition to its divorce from the present tense of the lexical go verb in the first- and second-persons plural that we have already noted in its indicative use, is the rhizotonic pattern of the 1/2pl vàgim/vàgiu, which, as we observed with the subjunctive of the habere auxiliary, is characteristic of synthetic past tense paradigms, but quite anomalous in present paradigms. As far as absence of suppletion in the paradigm of the go auxiliary is concerned, what goes for Catalan also goes for Guardiol—the Occitan variety which has been preserved in the Calabrian village Guardia Piemontese (Jacobs and Kunert 2014:195)—and for (eastern) Aragonese (Martı´nez Cortés et al. 2021). The relevant forms are set out in Table 3.8. Table 3.8 Forms of the present of the go verb in Guardiol and Aragonese

1sg 2sg 3sg 1pl 2pl 3pl

Guardiol Lexical Auxiliary vau vau vas vas vai vai anèm vam anatz vatz van van

Aragonese Lexical Auxiliary voi va vas vas va va imos vam/vamos itz vatz van van

A generalization that seems to hold is that the go verb shows a paradigm split of the kind mentioned here only when it is part of a periphrasis with the infinitive serving to express perfective past; the go + infinitive with future meaning in other Romance languages does not exhibit paradigm split nor does go plus a gerund or participle (14). Thus, in Occitan, where the go past had died out31 but where 31 It is obsolete, or at best obsolescent; according to Sibille (2007) ‘the periphrastic preterite has fallen out of use’ (p. 267), noting that ‘from the end of the sixteenth century its usage was on the decline and

118

nigel vincent and max w. wheeler

there is now a go future (Mooney 2020; Chapter 4 in this volume), the former showed the (non-suppletive) auxiliary forms but the latter has a preference for the (suppletive) lexical forms. Barcélo (2003) cites a text where the both the synthetic future farem petar ‘cause.fut.1pl explode.inf’ and the periphrastic future anam far petar ‘go.prs.1pl cause.inf explode.inf’ are used as equivalents of Fr. nous allons faire péter ‘we are going to explode (a bomb)’. The explanation for these distributions would appear to involve a mix of structural, semantic, and chronological factors. The go future is both more recent than the go past and less structurally integrated, as witness the fact that in some languages such as Spanish the go future construction involves the prepositional complementizer a ‘to’: Sp. voy a estudiar español ‘go.prs.1sg to study.inf (= I’m going to study) Spanish’. In the perfective past, the semantic contribution of ‘go.prs’ has become entirely obscured, while in the later constructions the contribution of ‘go.prs’ meaning is still detectable, or even obvious.

3.4 Reinforcement of the lexical verb In all the cases we have considered up to this point, whatever formal changes have taken place have affected the grammaticalized or grammaticalizing item. The main verb, by contrast, has retained both its form and its meaning(s), subject only to the independently motivated mechanisms of phonetic and lexical semantic change. In our next example we see the working out of the opposite effect whereby the lexical verb undergoes phonetic reinforcement and thereby distinguishes itself from the grammaticalized reflex of the same original item. Thus, compare the forms of the have verb < habere in (22) from the dialect of Ponte di Legno in the province of Brescia (Paoli 2020, her examples (1) and (2)): Ponte di Legno (prov. Brescia, northern Italy) (22) a. garòm … have.fut.1pl ‘we will have (money).’ b. ades el ga fresa now he have.prs.3sg hurry ‘he’s now in a hurry.’ c. go zù la os have.prs.1sg down def voice ‘I’m hoarse.’ today is only occasionally used’ (p. 261). However, it is worth pointing out that the construction persisted in a some Gascon varieties at least until the mid-twentieth century, though lexically restricted to the preterite of go, be, and the perception verbs see, hear, and find, where it is not always semantically interchangeable with the synthetic past (Jacobs and Kunert 2014:187). In Guardiol ‘as in Catalan, there are no semantic restrictions on the complement of the go-past […]; it is the default preterit for all verbs regardless of their Aktionsart’ (Jacobs and Kunert 2014:190f.).

layering and divergence in romance periphrases

119

d. forse el arà mes perhaps he have.fut.3sg put.ptcp ‘perhaps he will have put.’ e. l’a finì she=have.prs.3g finish.ptcp ‘she has finished.’ In examples (22a–c), various subtypes of possessive meaning are associated with the forms garòm, ga, go, while the perfect auxiliary uses in (22d–e) exhibit the forms arà and a. The initial /ɡ/ of the possessive uses is a reflex of the locative clitic ghe < Lat. hic ‘here’ which has become attached to and integrated with the root av-. This reinforcement of the item with independent lexical content had already been noted as a general characteristic of northern Italian dialects (see for further exemplification Rohlfs 1968:274 and the discussion in Smith 2021:514). It is most typically found with finite forms of the verb, but as Paoli (2020) goes on to show, in some dialects this new reinforced form starts to make its way into other contexts as in (23): Treviso (northern Italy) (23) a. a me ga 3sg.scl= me= have.prs.3sg ‘it has stung me.’ b. te a go 2sg= fsg= have.prs.1sg ‘I have given it to you.’

beca´ sting.ptcp data give.ptcp

c. gavendo voia se podessi anche ndar have.ger wish one.scl= can.pst.sbjv.3sg also go.inf ‘if we wanted to, we could also go.’ d. … de gaverte desmentega´ de have.inf=2sg forget.ptcp ‘… to have forgotten you.’ (cf. Paoli 2020:exx. 12c, e; 27b; 17b) In fact, there are two intersecting diachronic parameters involved. The starting point of both is the incorporation of the clitic ghe into finite forms of the have word as in (22). From there, one trajectory leads the reinforced form to in effect regrammaticalize and be used also in auxiliary contexts as in (23a–b) while the other extends the reinforced pattern to non-finite forms of the lexical item as in (23c). The end point is that the two trajectories come together as in (23d) leading by a different route to the same distributional pattern as in standard Italian or French where all forms of the have verb are identical regardless of whether the function is lexical or grammatical. To extend Hopper’s terminology, what we have here is divergence followed by reconvergence.

120

nigel vincent and max w. wheeler

3.5 Loss of inflexion By contrast, grammaticalization of the go verb (and also stare ‘stay’) in some southern Italian dialects leads not to inflexional differentiation but to loss of inflexion altogether (see also the discussion in §1.2.2.5 in this volume). This development is discussed in detail in Ledgeway (2016a), from where all the examples in this section are taken.32 The diachronic starting point is a pair of finite verbs coordinated with a ‘and’ (< Lat. ac), as in (24): (24) a. Va a go.prs.3sg and ‘(S)he goes to call.’ b. Stoc’ a stand.prs.1sg and ‘I stand and do.’

chiama. call.prs.3sg ffazzu. do.prs.1sg

Reflexes of these structures are found in many southern Italian dialects but particularly in those from Apulia, where we can observe a striking contrast in the development of form and content. On the content side, as one might expect on the basis of wider Romance evidence, there are semantic shifts from motion to future, and from location to progressive or imminence. When it comes to form, however, there is a split between northern Pugliese dialects (Foggia, Bari, Taranto) where the inflexion is retained on both verbs, as in the examples in (25), while in the southern (viz. Salentino) part of the region (Brindisi, Lecce) the grammaticalized item has lost all its inflexion as in (26): (25) a. vonә (a) mˈmaɲʤәnә (Martina Franca (Bari)) go.prs.3pl ac eat.prs.3pl ‘they’re going to eat.’ b. u stok a fˈfattsә (Putignano (Bari)) 3.msg= stand.prs.1sg ac do.prs.1sg ‘I’m about to do it.’ c. stok a bˈbeivә (Taranto) stand.prs.1sg ac drink.prs.1sg ‘I’m drinking.’ (26) a. sta ddormu/ ddurmìa/ ddurmimu/ stand sleep.prs.1sg sleep.pst.impf.1sg sleep.prs.1pl ddurmìamu (Lec.) sleep.pst.impf.1pl ‘I/we am/are/was/were sleeping.’ 32 This pattern is also discussed by Cruschina, who proposes an analysis in terms of Distributed Morphology (see Chapter 6 in this volume).

layering and divergence in romance periphrases

121

b. lu va fazzu / facci / fannu (Mesagne (Brindisi)) 3msg= go do.prs.1sg do.prs.2/3sg do.prs.3pl ‘I/you.sg/(s)he/they am/are/is/are going to do it.’ c. lu scia ffacimu / ffaciti (Mesagne (Brindisi)) 3msg= go do.prs.1pl do.prs.2pl ‘We/you.pl are going to do it.’ In effect, the items va, scia, sta have become tense–aspect prefixes. Yet it is notable that even here the morphomic N-pattern makes itself felt. While stare has the same stem in all its inflected forms and hence always reduces to sta (26a), the go verb, as we have seen in §3.3, Table 3.5, is suppletive; hence we find va, the reflex of the Latin uad- stem in the first-, second-, third-singular, and third-plural forms in (26b) but scia, the local reflex of the Latin i- stem (from ire ‘go’), in the firstand second-person plural form in (26c).

3.6 Conclusion By way of conclusion, let us consider some of the implications of the case studies we have investigated for the general principles of change that were our starting point. First, and most obviously, there can be divergence of form as well as divergence of content. In addition, the variant forms can coexist in different ‘layers’. What needs to be emphasized, though, is that there is no overarching requirement that form layers should match with content layers in a one-to-one fashion. Form and meaning can and do go separate and multiple ways; divergence/layering applies to both but not necessarily at the same time or together. It is true that there are pressures that will tend to make them converge. For example, the loss of independent stress that usually accompanies the development of a grammatical as opposed to a lexical function favours the processes of cliticization and univerbation that have long been staples of the grammaticalization literature. What the examples considered here demonstrate, however, is that there are structural pressures which can push back against that tendency. A well-known one is linear order. The emergent Romance order Aux + V preserves the syntactic independence of the grammaticalized item by contrast with the univerbated reflex of the earlier V + Aux order which lies, for example, behind the Romance inflexional futures and conditionals. Even so it is not impossible for later developments to run counter to that tendency as is clear from the case of the prefixed forms va- and sta- discussed in §3.5. By contrast, the data reviewed in §3.4 show a different mechanism of layering, whereby it is the lexical item that is amplified rather than the grammatical item being reduced. In the light of the data considered here we can now add a further factor which stabilizes form even as content changes, namely morphological structure and in

122

nigel vincent and max w. wheeler

particular, the emergence of a morphomic template such as the N-pattern. Developments such as these lead to a unified paradigmatic structure which holds at both the lexical and grammatical layers of an item like habere and in virtue of which they are more resistant to further change. In addition, at the methodological level we have seen how a full picture of the Romance developments, and hence a full understanding of their implications for general theories of linguistic change, requires on the one hand the integration of reconstructed and attested material, and, on the other hand, recourse to the widest possible range of data whether derived from the standard languages or from the less or non-standardized varieties conventionally referred to as ‘dialects’. In short, morphology—and especially the morphology attested in Romance dialects—matters: there could hardly be a better conclusion to a contribution in honour of Martin Maiden!

4 The go-future and go-past periphrases in Gallo-Romance A comparative investigation Sandra Paoli and Sam Wolfe

4.1 Introduction Without a doubt one of the most salient developments in the evolution from Latin to modern Romance has been the emergence of a range of auxiliary verbs and other dedicated functional categories (see in particular Vincent 1997b; 1999; Ledgeway 2016a).1 While the domain of Romance auxiliation is hardly an untrodden path,2 many questions in this area still remain. The relevant developments are significant both from a typological perspective (Schwegler 1990) and in terms of different theoretical approaches to the phenomenon of grammaticalization.3 In this chapter we focus on the development of two periphrases in GalloRomance, namely the French analytic future, the so-called futur proche ‘near future’ or go-future, formed with the verb aller ‘go’ and the infinitive (1), and the Occitan go-past (2), formed via the same means—anar ‘go’ + infinitive—but with a past-time reading: (1) Je vais faire. (Fr.) I go.prs.1sg do.inf ‘I’m going to do/I will do.’

1 It is undoubtedly an oversimplification to think of all functional categories as an innovation, with various periphrases, such as that involving esse ‘be’ + past participle already attested in early Latin texts (Adams 2013:616). Likewise, functional structure is best thought of as having overt exponence in dedicated functional heads in Romance, as opposed to having been absent previously. On evidence for an articulated D-layer in Latin, see Giusti and Iovino (2016). 2 See in particular Harris (1982), Pinkster (1985), Cennamo (2008), and Ledgeway (2016a:767–770). 3 On which, see Hopper and Traugott (2003) and Traugott and Trousdale (2013) for a functional perspective and Roberts and Roussou (2003) and Van Gelderen (2011) for a formal perspective.

Sandra Paoli and Sam Wolfe, The GO-future and GO-past periphrases in Gallo-Romance. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Sandra Paoli and Sam Wolfe (2022). DOI: 10.1093/oso/9780198870807.003.0005

124 (2)

sandra paoli and sam wolfe

e van lo conjurar quel sieu poder, s’ il and go.prs.3pl it= beg.inf that his power if it platz que demostres … (OOcc.) please.prs.3sg that show.pst.sbjv.3sg ‘And they pleaded with him to show his power, if he was willing.’ (Jensen 1990:§684)

Considering (1), note that the grammaticalization path that sees a verb of motion develop into a future marker is well attested cross-linguistically, and movement-verb constructions represent the most frequent and prototypical lexical sources for so-called future grams.⁴ This is no surprise, since motion verbs are known to allow for non-literal motion use: for the verb go this can be understood, for example, as an instance of semantic contiguity between distance from the speaker in terms of space and distance from the speaker in terms of time.⁵ More specifically, the subject of the verb of motion is on a certain path that, if followed, potentially leads to a certain event in the future. Studies taking a cognitive approach to grammaticalization explain the ‘small pool of possible source concepts’ for auxiliaries (Heine 1997:6) on the basis of basic conceptual structures that are lexicalized by individual verbs. Verbs meaning go would under this interpretation be strong candidates for grammaticalization as auxiliaries, since they lexicalize the cognitive schema source–path–goal (cf. Kuteva’s ‘kinesthetic image-schemas’ 1995:379), one of the most fundamental schemas that derive from the idealized cognitive models of our bodily interaction with the physical world as we make sense of it. These schemas have a generally spatial or force-dynamic character, to which motion verbs are perfectly suited. Significantly less frequent is the development of a motion verb construction into an expression of past, such as the Occitan pattern we see in (2).⁶ Within the ⁴ In a sample of 46 dedicated (i.e., not principally aspectual) future forms, Bybee, Pagliuca, and Perkins (1994:252f.) found 20 to be derived from constructions with verbs of movement such as come and go. ⁵ The term ‘contiguity’ is used here and in the next paragraph in its non-technical meaning. However, there is also a technical use of the term, that specifically contrasts it to similarity: semantic contiguity is the basis of metonymy, while semantic similarity is the basis of metaphor (cf. Traugott and Dasher 2001:79). Metaphor and metonymy may involve different axes (i.e., paradigmatic and syntagmatic respectively), but they interact closely and extensively (cf. Goossens’ 1995 coinage of the term ‘metaphtonomy’ to refer to such interaction). Specifically applied to the development of go-futures, the two processes would identify different paths, a spatial goal coming to represent a temporal goal from a metaphor’s perspective, and an action coming to stand for its likely consequence in the case of metonymy. However, since they are not mutually exclusive and interaction between them is highly likely, we do not pursue the specific identification of which of the two may have played a larger role in the process of grammaticalization of the construction. ⁶ Notable exceptions are mentioned in Jacobs (2011:249): a number of unrelated languages spoken in South America and Africa seem to have a similar construction, although its use appears to be limited to specific types of texts or styles, unlike the common, spoken usage of the go-past in Catalan. More productive uses of the periphrasis are recorded by Lichtenberk (1991:498) in Oceanic languages, in which there seems to be an interesting correlation between the presence of a come-future and a go-past construction.

the go-future and go-past periphrases in gallo-romance

125

modern Romance languages, it is only found in Catalan, in some Béarnais varieties of Gascon, and in Guardiolo, the Occitan variety spoken in northern Calabria in southern Italy where it has survived mainly as a residue (Berchem 1968; Jacobs and Kunert 2014:183), but was also found sporadically in older periods of French: (3)

a. Vaig cantar. (Cat.) go.prs.1sg sing.inf ‘I sang.’ b. Ba i. (Gsc.) go.prs.3sg go.inf ‘He/She/It went.’

In contrast to Occitan, the development and establishment of the go-past construction in Catalan has been extensively documented and investigated,⁷ and a variety of analyses have been proposed. Traditional accounts (such as Colón 1978a; 1978b; Badia i Margarit 1994:370f.; Wheeler 2017) have all focused on the use of the historical present in narratives,⁸ that is, a present tense form with past tense value, as one of the key factors in the grammaticalization of the periphrasis as an expression of past. More recent accounts (chiefly Juge 2006; 2008), focus on morphological factors linked to the syncretism between the present and past paradigms of go in the first- and second-person plural forms in old Catalan (i.e., anam and anatz respectively), and the prevalence of past tense forms of go in the early use of the periphrasis. Nagy C. (2010:79) further develops this observation, claiming that the higher incidence of past forms of go over present forms in the early stages also played a fundamental role in the development of the periphrasis as an expression of past, but, more significantly, in ensuring that the construction did not develop into a future. The unusual outcome of go + infinitive into an expression of past is taken by Detges (2004) as an indication that, after all, grammaticalization is not driven by cognitive forces,⁹ rather, by discourse-pragmatic strategies, in this case, the speakers’ desire to give prominence to their past actions. The specific conclusions that he draws are that, as far as grammaticalization of tense markers is concerned, it is not ‘motivated by a desire (or need) to conceptualize time’ (Detges 2004:213). In the same contribution, Detges goes on to formalize the suggestion made by a number of scholars (cf. Diez 1882; Meyer-Lu¨bke 1899; Gougenheim 1929) that the go-past periphrasis, which in the Middle Ages was widespread both in northern ⁷ Jacobs (2011:227) reports that the earliest mentions of the go-past periphrasis go back to the late nineteenth century. ⁸ Within this category it would also be possible to include a number of studies (cf. Berchem 1968) that interpret the periphrasis as a stylistic variant that eventually established itself as a grammatical category. However, since these studies do not address at all the steps involved in this change, they are not considered here. ⁹ According to a cognitive approach, and supported by cross-linguistic variation, expressions marking past develop from a source-oriented schema formed by come from or go from constructions.

126

sandra paoli and sam wolfe

and southern France as well as in northern Spain, developed from an aspectual construction in which go + infinitive conveyed ‘inchoative’ aspect. As will be discussed in §4.4, this label is problematic both in terms of the discrepancies in the ways in which it has been used by these authors, and its actual relevance for the final development of the construction. Against this background, we suggest that the time is ripe for a reappraisal of the evolution of the go-past in Occitan and the go-future in French. The go-future in French is arguably one of the best-studied cases of a periphrastic form competing with, and ultimately replacing, a synthetic paradigm in Romance. Although the Occitan go-past has not been as widely studied, it too has the potential to shed new light on how an analytic form came to take over functions previously associated with a synthetic past in Latin, such as scripsi ‘I wrote, I have written’. Considering them side by side, we propose in this chapter that the two developments into past and future expressions stemmed from a common construction: rather than expressing inchoative aspect, as is commonly assumed, this was a more basic expression of posteriority, closely linked to the semantic contiguity between space and time and the source–path–goal schema associated with go. After setting out the established facts regarding the genesis of the go-future in French in §4.2, we present novel data from old Occitan in §4.3. Considering the data side by side leads to a new comparative analysis and discussion in §4.4 before we conclude in §4.5.

4.2 The go-future in French 4.2.1 From Latin to old French Although traditional manuals of Romance philology and linguistics frequently suggest a simplistic synthetic > analytic evolution to Romance from Latin as regards auxiliation, the picture is rather more nuanced. While Latin did feature two synthetic future forms, amabo ‘I will love’ and regam ‘I will rule’, a periphrastic construction formed from a conjugated form of habere which could either precede or follow the infinitive was in use at least from the second century ce to indicate possibility, obligation, intention, and later, possibly from the fifth century ce (Clackson 2016:9), futurity (for recent critical discussion see Adams 2013:655–666). Eventually, the conjugated form of habere ceased to have independent status and was reanalysed as a functional affix attached to an infinitival stem. It originally had a clear reading of obligation and intention (Pinkster 1985), and there is a degree of debate as to when it lost its original semantics. Recent review of the extant evidence by Adams (2013) suggests that there are very few unambiguous examples of a purely future reading in sub-literary and late Latin texts. The example given in (4) is one often-cited but controversial example

the go-future and go-past periphrases in gallo-romance

127

(cf. Varvaro 2013:33–35) and is also, incidentally, the only example where the habere form does not show independent status:1⁰ (4)

‘Et ille respondebat: “non dabo”. Iustinianus dicebat: and he.nom reply.pst.3sg neg give.fut.1sg Justinian.nom say.pst.3sg “daras” (late Lat.) give.fut.2sg ‘And he replied: “I will not give [them to you]” Justinian said: “You will”.’ (Chronicon Fredegarii II 62, 7th century ce)

It is this habere future which we see used across the earliest French texts, including the Strasbourg Oaths, generally acknowledged to be the very earliest written attestation of Gallo-Romance (Pountain 2011:615–617): (5)

d’ ist di in avant in quant Deus from this day in forward in as.much God savir et podir me dunat si salvarai knowledge and power me=give.prs.3sg si support.fut.1sg eo cist meon fradre Karlo (OFr.) I this my brother Charles ‘From this day forward, insofar as God gives me knowledge and power, I will support my brother Charles.’ (Strasbourg Oaths)

This synthetic future is found across a range of early French texts and appears in all the major descriptive handbooks of old French (Foulet 1919:170f.; Price 1971:202f.; Einhorn 1974:42–44, 115f.; Jensen 1990:350–352). The instances that follow exemplify this and also show that, more rarely as in (7), morphologically present forms could yield a future reading in the earliest French verse and prose (Jensen 1990:343). (6)

a. Si vus plest, a vus parlerai (OFr.) if you= please.prs.3sg to you= speak.fut.1sg ‘If it pleases you, I will speak to you.’ (Desire 191, 585) b. Lors ira la chose autrement (OFr.) thus go.fut.3sg the thing differently ‘Thus the thing will go differently.’ (Roman de la Rose 12, 17062)

(7)

Se uns escapet, morz ies if one escape.prs.3sg dead be.prs.2sg ‘If one escapes, you will be dead and doomed.’

e cunfunduz (OFr.) and doom.ptcp (Roland 97, 288)

1⁰ It is important to note that the ability for the infinitival stem to be separated from the future ending persists with clitic pronouns well into the medieval period for Occitan and other medieval Romance varieties, e.g., servir l’ai dos ans o tres ‘serve.inf cl=have.1sg two years or three (=I will serve her two or three years) (Jensen 1990:273) and persists today in written registers of literary European Portuguese (Duarte and Matos 2000:§4.3.7).

128

sandra paoli and sam wolfe

It is also the case throughout the medieval period that a range of constructions arise where aller ‘go’ appears alongside an infinitive. A search of the Base de Français Médiéval reveals c. 3,000 instances of this construction.11 The consensus in the literature is that during the old French period (ninth–fourteenth centuries) aller yields a purely lexical reading in such constructions, with no sense of intention, obligation, or future temporal reference.12 Our analysis of the examples confirms this, where lexical encoding of motion is clear in examples such as (8). (8)

a. et aprés alerent dormir (OFr.) and after go.pfv.3pl sleep.inf ‘and afterwards they went to sleep.’ (Qgraal_cm 229) b. Mais alons courir sur noz ennemis (OFr.) but go.prs.1pl run.inf on our enemies ‘But let us go and charge at our enemies.’ (Melusine 283)

However, there is also an alternative construction, which is widely employed during the medieval period, where a conjugated form of aller is used alongside an infinitive, allegedly to yield an inchoative reading. Although Gougenheim (1929:92f.) argues that this construction does not fully take hold until the fifteenth century, he cites clauses such as (9) where the core lexical content of aller is not clear from context. Rather, it appears to encode the beginning of an action during a narrative sequence. If we consider (9b), Gougenheim highlights that cooccurrence of aller with ferir ‘strike’ occurs 31 times in the Roland alone, which would be a surprising frequency were it a purely lexical use. However, as we suggest below in § 4.4, although cases such as (9) do suggest an early functional use of aller, that this use is inchoative is not clear. (9)

a. Et al maitin alerent logier devant les and at.the morning go.pfv.3pl camp.inf in.front the portes de la vile (OFr.) gates of the town ‘And in the morning they camped out in front of the town.’ (Villehardouin 1 84) b. Le cheval brochet, vait ferir Oliver (OFr.) the horse spur.ptcp go.prs.3sg strike.inf Oliver ‘The horse spurred, he struck Oliver.’ (Roland 112, 1313)

Note that the use of morphologically present verb forms to encode past events is widespread in old French poetry and prose (10), with Einhorn (1974:116) referring 11 http://txm.ish-lyon.cnrs.fr/bfm. The search was limited to cases where a conjugated form of the verb aller was followed within four constituents by the infinitive. The corpus is not fully lemmatized, so the number may be lower than that indicated. 12 Wilmet (1970:179) suggests there are only two cases of a periphrastic future used in the thirteenth and fourteenth centuries.

the go-future and go-past periphrases in gallo-romance

129

to ‘a striking combination of present and past tenses to describe a sequence of past events’, which frequently entails a past tense usage in the subordinate clause, followed by a narrative present in the main clause. (10)

et quant cil oï qu’ il li crie, si and when he hear.pfv.3sg that he him= cry.prs.3sg si li cort le glaive aloignié (OFr.) him= run.prs.3sg the sword lengthen.ptcp ‘And when he heard what he is shouting to him, he dashes towards him with his sword out.’ (Graal 90.28 from Jensen 1990:344)

To summarize, from the earliest attestations of French through to the fourteenth century, futurity is typically encoded by means of the infinitive + habere construction inherited from Latin, or a morphologically present tense verb. There is no evidence of periphrastic forms with aller encoding futurity, intention, or obligation, though incipient uses of a periphrasis, which has been described in the literature as inchoative, are found. However, we will suggest in §4.4 that caution should be exercised in basing an account of the genesis of the analytic future on this assumption, as the term ‘inchoative’ may have been used in the French literature in a way that does not conform to the modern understanding of that term.

4.2.2 Middle to modern French There is a general acceptance in the literature that the genesis of the aller + infinitive construction with future temporal reference occurs during the fifteenth and sixteenth centuries, often referred to as the middle French period (Wilmet 1970:183–190; Champion 1978:33; Marchello-Nizia 1980:136; Fleischman 1982). We also note, and return to this point in §4.4, that the ‘inchoative’ allerperiphrasis is most widely employed in early middle French (11), typically in portions of text in narrative chronicles, before falling out of use in the majority of French texts in the seventeenth century (Gougenheim 1929:96). (11)

A l’ endemain leur coureur alèrent courir jusques at the morning their horses go.pfv.3pl run.inf as.far.as ès bailles de Bregerach (MidFr.) to.the waters of Bregerach ‘In the morning their horses rode as far as the waters of Bregerach.’ (Froissart, I, §207, III, 46, as cited in Gougenheim 1929:95)

It is also from the fifteenth century onwards that aller, alongside a number of other verbs such as devoir ‘must’ and vouloir ‘want’ (Gougenheim 1929:72–79, 88–92; Wilmet 1970:179–181), which have grammaticalized elsewhere in Romance as future auxiliaries (Iliescu 1972:228; Ledgeway 2011:422; Mensching

130

sandra paoli and sam wolfe

and Remberger 2016:285),13 begins to yield a reading of obligation and intention (12a–b show the latter).1⁴ The example in (12a) is taken from our corpus search of the Base de Français Médiéval and (12b) comes from Wilmet (1970). While the examples in (12) are instructive in demonstrating the emerging intentional reading of the periphrasis, (13) is particularly significant, as the purely andative reading of aller ‘go’ is not directly compatible with the notion of displacing oneself to lose something: (12)

a. Et puis on se dispose d’ aller mourir and thus one refl= prepare.prs.3sg of go.inf die.inf ou vivre avec luy, et pour amour ne l’ or live.inf with him and for love neg him= abandonner point (MidFr.) abandon.inf neg ‘And thus one prepares to go and die or live with him, and due to love never to abandon him.’ (Jouvencel 2, 21) b. Or je vois savoir // au povret now I go.prs.1sg know.inf to.the poor.thing qu’ il me vouldra dire (MidFr.) what=he me= want.fut.3sg say.inf ‘Now I’m going to find out from this poor thing what he means.’ (Pathelin 707, Wilmet 1970:183)

(13)

Tu vas perdre ta conscience // tu t’ en you go.prs.2sg lose.inf your conscience you you= part= vas au dyable server (MidFr.) go.prs.2sg to.the devil serve.inf ‘You are going to lose your conscience// you are going to serve the devil.’ (Moralité de Charité 3, 392, ATF, Gougenheim 1929:98)

However, while the use of devoir ‘must’ and vouloir ‘want’ declines over this period, this is not the case for aller ‘go’. Examples such as (13) show quite clearly that neither obligation nor intention are inherent readings of the construction. Though no contemporary corpus analysis has been completed on the go-future’s frequency, Gougenheim (1929:98f.) shows it to be particularly frequent in theatre and direct speech from the fifteenth century onwards (Champion 1978:34f.) and, furthermore, notes a strong tendency for it to be found with first-person subjects (compare this with the data on the Occitan go-past in § 4.3). The construction’s acceptance as part of the written norm by both Maupas (1607) and Arnauld and Lancelot (1660) suggests it was not exclusively considered a ‘low’ spoken variant at 13 For discussion of the influence of Gascon on early modern regional French, see Gougenheim (1929:89). 1⁴ See § 4.2.1 for similarities with Latin in this regard.

the go-future and go-past periphrases in gallo-romance

131

the time. Furthermore—although contingent on one’s theory of creolization—the fact that future-marking functional heads derived from aller appear in a number of French-lexicalized creoles could be viewed as a piece of evidence for its widespread use in speech at the time (Posner 1997:208; Lefebvre 2006:130f.; Bollée and Maurer 2016:464): (14)

To va allé demain (18th-c. Haitian) you go go tomorrow ‘You will go tomorrow.’

(DeGraff 2008:320)

By the second half of the eighteenth century the main development to note is the ability of aller to occur in this periphrasis in a variety of tense forms and the general decline in the seemingly free alternation between aller + infinitive and s’en aller (lit. ‘to leave, make off ’) + infinitive (Gougenheim 1929:101). Referring to eighteenth-century French, Gougenheim (1929:103), suggests that the construction is used in the same way as in the modern language, in that it encodes temporal proximity between the present and the future event. We will now see, however, that this characterization of the go-future’s use in modern French is far from exhaustive.

4.2.3 Contemporary developments Both the synthetic future and periphrastic go-future persist in French varieties today, though they do so with distinct functions, instantiating distinct points on a grammaticalization pathway. Grevisse and Goosse (2016:§820) suggest that the go-future, in contrast to the synthetic future, indicates a ‘future viewed from the present, often a near-future, sometimes a distant future but one considered inevitable’ (our translation). These factors commonly occur in other descriptive and theoretical treatments of the distinction in modern French. However, there is an increasing body of evidence that the distribution of the go-future relative to the synthetic future is an area of ongoing change in the grammar and one of considerable interspeaker variation (see Tagliamonte, Durham, and Smith 2014; Denis and Tagliamonte 2018 for a parallel with contemporary English, about which the observations in 1 and 2 below also hold to a degree). While not attempting an exhaustive review of the literature (for an overview, see Poplack and Turpin 1999:136–139; Lacross 2018:28–41), the following formal factors have been identified as significant in conditioning the variation observed: 1. Proximity or immediacy of the event. A strong proximity or immediacy between speech time and the future event is said to favour the go-future over its synthetic counterpart (Fleischman 1982:96; Blanche-Benveniste 1990:188; Poplack and Turpin 1999:148; Smith 2016:306).

132

sandra paoli and sam wolfe

2. Certainty that the event will occur. Speakers use the go-future more when the event’s occurrence is thought to be certain (Confais 1995:399; King and Nadasdi 2003:333f.; Rebotier 2015:3). 3. Grammatical person. First-person subjects favour use of the go-future (So¨ll 1983:19f.), while the synthetic future is preferred with the formal secondperson singular form vous, demonstrating the interaction of person with register (Poplack and Turpin 1999:154) 4. Polarity. The synthetic future is more widely attested in negative sentences than affirmative ones (Deshaies and Laforge 1981:27–30; So¨ll 1983:17; Poplack and Turpin 1999:154f.). In addition, varieties of French spoken in Canada have been the subject of several studies into future marking. While Poplack and Turpin (1999) suggested that the synthetic future was heading in the direction of total obsolescence in their Ottawa– Hull corpus, more recent work by Wagner and Sankoff (2011) has revealed an agegrading pattern suggesting such a change may be slowing down, while King and Nadasdi (2003) report that in Prince Edward Island retention rates of the synthetic future are considerably higher than in similar studies for other regions. Overall, there is an increasing body of evidence that the division of labour between the two future constructions is variable across the francophone world today, both within and outside the Metropole. While certain speakers conform to some version of what we might term the ‘conservative’ system outlined in Grevisse and Goosse (2016), this is not the case for other speakers, where the synthetic future has either seen a further narrowing of its function, while remaining part of the core grammar, or is strongly dispreferred.

4.3 Old Occitan Although in the modern Lengadocian variety of Occitan the go + infinitive construction expresses futurity (cf. Vernet 2007:29; Mooney 2020), texts from the thirteenth–fifteenth centuries bear witness to the periphrasis being used with a past interpretation:1⁵ this existed alongside the same construction being used to express a non-past meaning. The source of the Occitan data considered here is a collection of texts in verse, including both stretches of narrative and dialogue, dating between the eleventh and fifteenth centuries, loosely originating from the Languedoc area.1⁶ 1⁵ The first author gratefully acknowledges Balliol College for granting some research funds, and Siarl Ferdinand for being an invaluable research assistant and a fountain of knowledge. 1⁶ The data were partly obtained from the second part of the Concordance de l’occitan médiéval (COM2, Ricketts 2005), and partly from the integral texts. A full list follows, together where necessary with the code by which each text is referred to in this chapter: 11th century: Cançó de Santa Fe; 12th

the go-future and go-past periphrases in gallo-romance

133

In our corpus, the periphrasis go + infinitive first appears in the twelfth century. While in Roland à Saragosse it is almost exclusively found in collocations with the verb ferir ‘hit’ (nine out of a total of 10 occurrences),1⁷ similarly to what we have already seen for French (9b), in GdR and RdJ it also occurs with a variety of verbs used both transitively and intransitively (e.g., penre ‘take’, deslaçar ‘undo’, pausar ‘place, put’, recinglar ‘tighten (the cinch on a horse)’, manjar ‘eat’, cochar ‘lie down, set (of sun)’). There is a single instance of the periphrasis with a verb of motion (prosmar ‘approach’):1⁸ it is only in the fourteenth century that a variety of verbs of motion are found,1⁹ suggesting that not until then had the periphrasis reached a level of grammaticalization that divested go of its lexical meaning. Therefore, a lexicalized (or understood) goal or destination is found in the early examples (15a–b) but becomes less common in the later period: the construction is still compatible with a lexicalized (or understood) goal in the fourteenth century, but it can only yield a literal interpretation implying movement (15c). (15)

part gitar,// una car a. E va s’ en una part and go.prs.3sg refl= part a side throw.inf because totz es las e pesoiantz. (OOcc.) everything be.prs.3sg weary and heavy ‘And he goes over to one side to lie down,// because everything is weary and burdensome’. (RdJ 2822–2823) vers Balenberc b. Lo solel vai cochar vers Balenberc (OOcc.) the sun go.prs.3sg sleep.inf towards Balenberc ‘The sun goes to set towards Balenberc’. (GdR 4633) c. Montatz es mount.ptcp be.prs.3sg va.s metre go.prs.3sg=refl put.inf

tantost sus la immediately on that davant efant efant,// davant l’l’ efant in.front the child

cela// e there and e and

century: Girart de Roussillon (GdR); Roman de Jaufré (RdJ); Roland à Saragosse; 13th century: Canso de la Crotzada; Fierabras; Flamenca; 14th century: Guilhem de la Barra (GdB); La Passion provençale du manuscrit de Didot (MDP1); Le Roman d’Arles (RdA); 15th century: Mystère rouergat de la Passion (MDP2); Lo Jutgamen General (JG); Istoria Petri et Pauli (IPP). 1⁷ Colón (1978b:144) reports that in the old French and old Catalan texts he analysed go + férir (or synonyms) was also the most frequent combination. 1⁸ The relevant example is given in (i): (i)

// prosmar prosmar Mantenen se vai d’ el prosmar/// now refl= go.prs.3sg of him approach.inf lansa sobre man (OOcc.) lance on hand ‘Now he approached him // and carried a lance in his hand.’

e and

portet carry.pfv.3sg (RdJ 1830–1831)

1⁹ We find venir ‘come’, tornar ‘go back’, yssir ‘go out’, intrar ‘go in’, partir ‘leave’, perregir ‘head towards’, layssar ‘leave’, and colarse ‘sneak in’.

134

sandra paoli and sam wolfe

volc cavalgar mais avant (OOcc.) want.pfv.3sg ride.inf more forward ‘He immediately mounted the horse // and he goes and places himself in front of the child, // and was about to ride further’. (GdB 3112–3114) Throughout the corpus, the go verb appears almost exclusively in the present tense,2⁰ and it is inflected for all grammatical persons. On closer inspection, an interesting asymmetry in usage emerges between third person on the one hand, and first and second persons on the other. Third-person subject go + infinitive constructions appear exclusively in narrative stretches of texts. The event they express never occurs in isolation: it is part of a sequence of events of which it is never the first (in terms of temporal succession), but often the second or third, followed by some other(s), and frequently linked to the previous one(s) by the conjunction e ‘and’, the disjunction mas ‘but’, or by pueis ‘then, after’. The surrounding verbs are mostly in the preterite but can also occur in the present (normally understood as a historic present), or, in fact, in a mixture of the two:21 generally, the verb that precedes the go + infinitive construction is in the preterite, and the verb that follows it is very often in the present, as in the examples in (16). (16)

a. E Gilbers quant l’ oit vai se and Gilbers when it=hear.pfv.3sg go.prs.3sg refl= seder. (OOcc.) sit.down.inf ‘And when Gilbers heard it, he went and sat himself down.’ (GdR 4177) b. […] Carles venc apoinent a grant poest.// Charles come.pfv.3sg rushing at great gallop Vai ferir un donzel fran de go.prs.3sg strike.inf a young nobleman of tiest// amont sobre son elme, […]. (OOcc.) German on.top over his helmet ‘Charles came rushing galloping furiously.// He went to strike a young German nobleman// on his helmet, […].’ (GdR 5279–5281) c. E me levet del palafre // senpre e portet and me=take.pfv.3sg of.the palfrey always and carry.pfv.3sg m’ en ab se, // […] Pueis va mi sus un lieg me= part= with self then go.prs.3sg= me on a bed

2⁰ There are a total of 42 instances of the periphrasis with the verb go in the preterite, mostly occurring in the thirteenth and fourteenth centuries, compared with 372 in the present. 21 Tense switching within the same sentence or narrative unit is a typical feature of these medieval texts, which were primarily designed for oral presentation, and it is used as a strategy to mark off different events within the story, and to differentiate between points of view (Fleischman 1990), or to add vividness to the narration.

the go-future and go-past periphrases in gallo-romance

135

gitar, // on me volc forzar et aunir. (OOcc.) throw.inf where me= want.pfv.3sg force.inf and dishonour.inf ‘And he took me from the palfrey// and carried me with him, // […] Then, he threw me onto a bed, // and was about to force me and dishonour me.’ (RdJ 2985–2989) d. E el laisset nos aseguir // loing del castel per and he let.pfv.3sg =us chase.inf far.away of.the castle for nos trair, // e pueis va a ferir lo primier, // us= lead.inf and then go.prs.3sg to strike.inf the first si que l’ abat mort del destrier. (OOcc.) so that him= knock. prs.3sg die.ptcp of.the steed ‘And he let us chase him // to draw us far from the castle, // and then he struck the first (knight), // such that he knocked him dead from the steed.’ (RdJ 891–894) All the go + infinitive constructions above have been translated as past forms, an interpretation that derives from the tense of the surrounding verbs in the narration. First- and second-person go + infinitive constructions, conversely, are restricted to direct speech and do not appear to be connected to other events; however, they are linked to the moment of speech. The second persons, both singular and plural, are clearly imperatives or directives (17a–b), while the first person can either be a prediction or an intention in the singular (17c–d),22 or an exhortation in the plural (17e), and the motion component of the meaning of go is clearly available: (17)

a. ‘Cons, vai parlar o lui, consel li earl go.imp.2sg speak.inf with him advice to.him= quer// e ce qu’ il te dirar fai ask.imp.2sg and that that=he you= say.fut.3sg do.imp.2sg volenter’. (OOcc.) willingly “‘Earl, go to speak to him, ask him for advice // and what he will tell you do without hesitation.”’ (GdR 2978–2979) b. Anatz sezer a una part! (OOcc.) go.imp.2pl sit.inf at a side ‘Go and sit by the side!’ (RdJ 153) c. Ele enbracha Bertran e dist rient:// ‘Eu she embrace.pfv.3sg Bertran and say.pfv.3sg laugh.ptcp I

22 (17c) could also mean ‘I leave in order to get married’, since the following verses see Bertrand inciting the couple to leave quickly: however, the departure is motivated by trying to escape the brideto-be’s parents, who are ‘riding against her’, and not necessarily in order to reach the place where the marriage will take place. It is therefore highly improbable that go is used in its sense of motion.

136

sandra paoli and sam wolfe

m’en vauc a mari tan paubrement, // nen refl=part= go.prs.1sg to wed.inf much poorly neg port od mei aver, aur ne argent, // paile, bring.prs.1sg of my possessions, gold nor silver, silk samit ne porpre ne ornement.’ (OOcc.) satin nor crimson nor jewellery ‘She embraced Bertran and said laughing: // “I am going to get married in such poverty,// I leave behind my belongings, gold and silver, // silk, satin, (GdR 8396–8399) crimson, and jewellery”.’ d. ‘Ieu m’ en vauc negar autresi, // Que ja, I refl=part= go.prs.1sg drown.inf as.well that now per Dieu, sols no.i mora!’. (OOcc.) for God alone neg=loc= die.prs.sbjv.3sg “‘I will drown myself in the same way, // so that, by God, he may not die there alone!”’ (RdJ 8532–8533) e. ‘Cavalier, anem aiudar // a mon seiner lo rei knights go.imp.1pl help.inf to my lord the king Artus! […]’ (OOcc.) Arthur “‘Knights, let’s go to the aid of // my lord King Arthur! […].”’ (RdJ 286–287) In the twelfth century, most instances of the constructions in the first and second persons retain movement as a component of their meaning (although compare (17c) and (17d), in which motion is not the only available interpretation nor the most natural one), and hortatives and imperatives, in the first-person plural and second-person singular and plural, respectively, continue to do so throughout the entire corpus: these are all purposive constructions, in which the periphrasis has the basic meaning of ‘being in motion in order to do something’. This is not the case for the first-person singular expressing a prediction as in (17c), in which go clearly conveys a sense of futurity. By the thirteenth century the first-person singular expressing an intention becomes incompatible with an interpretation including movement, as illustrated by (18) where Fierabras (the Saracen) and Olivier are already standing face to face. (18)

‘[…] vay Sarrazi, si t’ arma: trop mi come.on Saracen then refl= arm too.much me= fas demorar. // si no, vau ti ferir make.prs.2sg wait.inf if not go.prs.1sg you= strike.inf pus senes esperar’. (OOcc.) without anymore wait.inf “‘[…] Come on Saracen, arm yourself: you are making me wait too long. // If not, I am going to strike you without further ado.”’ (Fierabras 935–937)

the go-future and go-past periphrases in gallo-romance

137

Overall, with first and second person, go + infinitive conveys a sense of futurity, either by implying movement towards a goal, or by expressing a prediction or an intention. The temporal point of reference is speech time (hence, the periphrasis works deictically), and the events are interpreted as proximal futures. This person asymmetry is not simply a concomitant of narrative style or text type: the complementary distribution described here is a robust feature found in all the texts throughout the five-century period, hence across genre and register. While it is true that the majority of the narration is in the third person, which may explain the lack of first- and second-person subjects with the go-past construction, there are nevertheless episodes of narration in first or second person, uttered in direct speech, and these never feature the construction it its past meaning. There may in fact be a link between the deixis encoded by first- and second-person pronouns and the anchoring of the construction in the speech time: temporal deixis used to point to the ‘here and now’ of speech time triggered by the presence of speaker and hearer encoding compatible deixis. The lack of deixis compatible with the deixis of speech time in third-person pronouns and referents in general would have made such anchoring unnatural.

4.4 Discussion Focusing on the incipient stages of the use of the go periphrasis affords an insight into the initial usage as well as the initial features of the construction, especially into what its ‘ancestor’, so to speak, could be. In spite of the obvious difference in interpretation between the periphrasis in the first and second person on the one hand, and the third person on the other, we claim that the two are actually different surface interpretations of the same underlying semantics. As to what this underlying semantics may be, there are several proposals that cross-linguistically favour an inchoative construction. For Occitan, Jensen (1986:229, §684), observes that ‘[the anar ‘go’ + infinitive construction] has acquired inchoative value, serving to denote actions that will take place in the immediate future’, and cites a number of Occitan examples that are translated as ‘x is about to do y’, ‘x will do y’, or even, simply, ‘x does y’. For French, Henrichsen (1966:362), following Diez (1882:930) and Meyer-Lu¨bke (1899:§324), claims that the periphrasis originates in the change in meaning, from spatial to inchoative, of the reflexes of Lat. ire ‘go’. Gougenheim mentions the term inchoatif (‘inchoative’) a few times (1929:92, 93, 95) before offering a definition (96); he suggests that ‘Il va dire’ [lit. ‘he go.prs.3sg say.inf’] does not mean ‘he began to say’, but rather ‘he said suddenly’, which he notes in modern French would typically be realized as ‘Il se mit à dire’ [lit. ‘he refl= place.pfv.3sg to say.inf’]. The specific example that he adduces in which the construction has a ‘clear and constant’ inchoative use (Gougenheim 1929:95) uncontroversially shows a lack of any ‘inchoative’

138

sandra paoli and sam wolfe

meaning: there is clearly no possible interpretation such as ‘it starts to happen that …’. In fact, even the assumed meaning of the sudden death of the knight’s spouse is not necessarily supported by the wider context. (19)

Ce chevalier eut sa première femme qu’ il this knight have.3sg.pfv.3sg his first wife who =he ama à merveilles. Si va advenir que la love.pfv.3sg at wonders si go.prs.3sg happen.inf that the mort, qui tout prent, la print, dont le death that all take.prs.3sg her=take.pfv.3sg of.which the chevalier fut si dolent que a peu qu’ il knight be.pfv.3sg so grieved that at little that =he n’ en mourut de dueil et de couroux. (14th-c. Fr.) neg= part= die.pfv.3sg of pain and of sorrow ‘This knight had his first wife whom he loved deeply. It so happened that death, that takes everything, took her too, for which the knight was so grieved that he almost died from pain and sorrow.’ (Le chevalier de la Tour Landry)

As mentioned above, the term ‘inchoative’, as well as being used differently by these authors, also does not correspond to the way it is usually conceived.23 The periphrasis itself is also often reinforced by inchoative verbs: if inchoative aspect really were an intrinsic part of the meaning of the periphrasis, examples such as (20) would be hard to explain. (20)

a. Si se vont mettre à desjuner tous deux. (OFr.) si refl= go.prs.3pl put.inf to lunch.inf all two ‘And both started to eat lunch.’ (Nicolas de Troyes, in Gougenheim 1929:96) b. Et quant il fut là arrivé, il va and when he be.pfv.3sg there arrive.ptcp he go.prs.3sg commencer à prier Dieu et la Vierge Marie. (OFr.) start.inf to pray.inf God and the Virgin Mary ‘And when he arrived there, he started to pray to God and the Virgin Mary.’ (Nicolas de Troyes, in Gougenheim 1929:96) c. Si se bessa de rechef et redressa le roy si refl= bow.pfv.3sg again and raise.pfv.3sg the king d’Espaigne, lequel va commencer à dire […]. (OFr.) of Spain the.which go.prs.3sg start.inf to say.inf ‘The king of Spain bowed and stood up, and began to say […].’ (Jean de Paris, in Gougenheim 1929:96)

23 See, for example, Matthews (2007:187), who defines an inchoative form as ‘indicating the initiation of some process or action’.

the go-future and go-past periphrases in gallo-romance

139

Furthermore, Gougenheim (1929:101) also refers to an entry in a seventeenthcentury grammar (Régnier-Desmarais 1705:417) in which the go + infinitive periphrasis is described as conveying the meaning of ‘being on the verge of / being about to’ do something. A move to expressing what ‘is about to happen’ from an alleged starting point of what ‘starts to happen’ seems cognitively unlikely, but would be plausible if it were instead a development from ‘what occurs subsequent to a given reference time’. This is, in fact, what Gougenheim seems to mean when he uses the term inchoatif, that is, prospective aspect which expresses a relationship between a given moment in time and a (near) future eventuality (Jendraschek 2014:158f.). Careful consideration of all the Occitan data, but especially of the early examples in which the construction is primarily found in collocation with the verb ferir ‘strike’, clearly focusing on the completed action rather than on its initial stages, does not afford any evidence that the go + infinitive periphrasis had, at that point, a genuinely inchoative meaning, nor do they suggest an inchoative construction as its ‘immediate ancestor’ (Detges 2004:213). We therefore wish to pursue an alternative path, rooted in the basic motion meaning of go, which can be assumed for both the go-future and go-past constructions.2⁴ Specifically, building on what may have already been an intuition on Gougenheim’s part, we argue that the spatial component of go is interpreted as temporal, resulting in the event expressed by the infinitive verb being interpreted as posterior to a given reference time. By proposing the same underlying construction for both periphrases, we are able to maintain the idea that grammaticalization in general, but specifically for verbs of motion, draws on universal forms of conceptualization (cf. Kuteva 1995; Heine 1997) that encode certain basic conceptual structures, which are then reanalysed as temporal. More specifically, if we adopt this stance, the typologically robust generalization that the cognitive schema source–path–goal, expressed by structures such as the go + infinitive periphrasis, conceptualizes posteriority, and generally develops into expressions of future, can be maintained. Considering the different path taken by the periphrasis in Catalan, it was mentioned in § 4.1 that the many proposed analyses suggest a number of factors as having played a crucial role in the development of the construction into an expression of past: Nagy C. (2010) in particular attributes pivotal roles to the higher frequency of preterite versus present go forms at the initial stages of the development of the construction, and to the past narrative context within which the periphrasis was interpreted. As noted above, the periphrasis in the third person that we find overwhelmingly used in narrative expresses an action that is part of a sequence of events, of which it is never the first, witness the conjunctions e ‘and’, mais ‘but’, and pueis ‘after, then’ 2⁴ Bres and Labeau (2013b:20) also take a unified approach to the development of the two periphrases, albeit resolving it in favour of the inchoative meaning.

140

sandra paoli and sam wolfe

at the beginning of sentences containing the go + infinitive construction. The order of the clauses tends to reflect the order of events as they unfold in time (i.e., a > b > c). The periphrasis is surrounded by verbs in the preterite or in the present indicative (i.e., historic present): starting from the previous event in the simple past, go + infinitive implies a movement that leads prospectively to the next event. Bres and Labeau (2013b) claim for French that ‘the periphrasis is purely aspectual’, in that it has no temporal content:2⁵ the event expressed by the periphrasis is anchored to the surrounding narration, and its temporal interpretation is established through a link to the sequence of events, all taking place in the past. The periphrasis as a whole, but go specifically, conveys the same prospective orientation towards the action denoted by the infinitive that we see with first and second person, that is, an event that occurs subsequent to a given reference time. The difference between the two, however, lies in this ‘given reference time’: while for the first and second person this is speech time, for third person the time frame is interpreted anaphorically, that is ‘the prospective movement originates in the situation that precedes it rather than in the moment of speech’ (Bres and Labeau 2013a:297). Consequently, the periphrasis in the third person is interpreted as an expression of past. The past and non-past (i.e., prediction/intention, directive and hortative) interpretations of the go + infinitive construction coexist throughout the period between the twelfth and fifteenth centuries, in complementary distribution across third person and first and second person, respectively. From the fourteenth century the first attestations of first and second person used with a past interpretation start to appear (21). (21)

a. Tantost sos vestirs hac trencatz // e totz immediately her clothes have.pfv.3g rip.ptcp and all sos cabels de son cap; // et yeu no m’ o her hairs from her head and I neg= refl= it= tengui a gab: // de la cambra vau tost take.pfv.1sg at joke from the room go.prs.1sg quickly yssir, // e tantost pres mi al leave.inf and immediately take.pfv.1sg I to.the fugir, // […]. (OOcc.) run.away.inf ‘Immediately she had torn off her clothes// and all the hair from her head; // and I didn’t take it as a joke: // from the room I quickly ran out, // and I immediately took flight, // […].’ (GdB 2864–2868)

2⁵ Similarly, Blasco Ferrer (2005:601) sees the periphrasis as a ‘kind of omnitemporal, distinctly unmarked tense, the first value that the periphrastic acquires in the Occitan verbal system’. He concludes that ‘its temporal referent may then become displaced, either to the past or to the future’.

the go-future and go-past periphrases in gallo-romance

141

b. L’ autre jort, vau veyre // Qu’ a ung orb the other day go.prs.1sg see.inf that to a blind.man la veyre // Ly vas rendre tu. (OOcc.) the sight to.him= go.prs.2sg give.back.inf you ‘The other day, I saw // that with a blind man // you restored his sight.’ (IPP 4894–4896) However, as Table 4.1 shows, the past use with first and second person is significantly less frequent than the non-past overall. Furthermore, the past usage is primarily found with first-person singular, only very rarely with second-person singular, and never with first- or second-person plural. Table 4.1 ‘Non-past’ and ‘past’ interpretation with first and second persons Century

Text

Non-past

Tot.

Past

Tot.

14th

GdB RdA MDP1 MDP2 JG IPP

1p (1); 2p (1) 1p (3) 1s (1) 1s (20); 1p (5); 2p (12) 1s (7); 1p (1); 2s (4); 2p (5) 1s (5); 2s (2) Overall total

2 3 1 37 17 7 67

1s (6) 0 2s (1) 1s (4) 0 1s (2); 2s (2)

6 0 1 4 0 4 15

15th

This absence suggests that the ‘prediction’ and ‘intention’ meaning components of the go + infinitive construction found with first-person singular, being at the initial stages of the semantic development of the future gram (Bybee et al. 1991:32), are compatible with a past interpretation (cf. 21a, in which vau tost yssir is part of a sequence of events, and it follows the woman’s actions). The imperative/directive meaning component expressed by second-person singular and plural, however, already having reached an advanced stage in the development of the future gram, could not.2⁶ From a more abstract perspective, this asymmetry could be interpreted as the result of a morphomic developmental distribution, with cells only realized in the third persons to start with, eventually including the first-person singular, but with empty cells for the second-persons and first-person plural. A further point for reflection is also that the past reading of the go + infinitive construction emerges in the first and second person much later than in the third: given that the latter typically originated in written narratives, hence a higher register, the change must have been ‘from above’, only reaching what can reasonably be assumed to be the reflexes of the spoken language at a later stage. However, this was only a temporary state of affairs, since the forces ‘from below’ eventually won 2⁶ Although the stages of development of meaning components proposed in Bybee et al. do not include ‘hortative’, which we see expressed by first-person plural, we assume here for simplicity that it represents a type of directive, and hence can be classified as an imperative alongside second person.

142

sandra paoli and sam wolfe

out and allowed for the complete grammaticalization of the future interpretation of the construction. From the fourteenth century, we find the periphrasis in the third person used for the first time with a possible non-past interpretation: (22)

a. Benezeyt siatz, filh de Dieu, // que va salvar bless.ptcp be.sbjv.2pl son of God who go.prs.3sg save.inf lo poble sieu! (OOcc.) the people his ‘Blessed be you, son of God, // who will save his people!’ (MDP1 328–329) b. Aras s’ en va serquar Raguelh la Vida, now refl=part= go.prs.3sg search.inf Raguelh the life he li digua, sus lo // escadafal: // […]. (OOcc.) and to.her= say.sbjv.3sg on the pedestal ‘Now (the angel) Raguel goes looking for Life, and it says to her, on the // pedestal: // […].’ (JG 1609P.1–2)

In (22a) Jesus is still alive and directly addressed, suggesting a future interpretation; (22b) is an instance of descriptive directions to indicate non-verbal actions to be performed by actors, and the interpretation is directive. It is difficult to come to any firm conclusions on the basis of very few examples. However, it is possible to imagine that as the past interpretation of the go + infinitive construction declines,2⁷ the non-past interpretation extends to the third person, which eventually allowed the anchoring to speech time just as with first and second persons.2⁸ Favouring the grammaticalization of the periphrasis as an expression of futurity rather than past, two main factors are likely to have played a role. First, the almost exclusive use of the present tense form of go in the periphrasis: the higher frequency of preterite forms in medieval Catalan are, according to Nagy C. (2010:79), the reason why the construction did not evolve into a future tense. She reaches this conclusion drawing on an observation by Bybee et al. (1994:268) that, when a future marker develops, the construction includes progressive, present, or imperfective meaning components. This, crucially, was not the case in Catalan. Conversely, the use of the present tense may have favoured the eventual Occitan development into an expression of futurity. Furthermore, the go + infinitive construction coexisted with a go + gerund construction: the latter conveyed a ‘continuous, ongoing action’ (Jensen 1986:229, §685, 255, §756) where go could occur in both the present and preterite (tals obras va pertot fazen lit. ‘such works 2⁷ According to Meyer-Lu¨bke (1925:106), the construction became obsolete in Occitan in the sixteenth century. 2⁸ Bres and Labeau (2013b:20) report that for French the ‘posterior’ use emerged later than the narrative use, appearing sporadically in the thirteenth century, and spreading in the fifteenth century, and that it was especially evident in texts containing stretches of dialogue in which use of first and second person naturally abounds.

the go-future and go-past periphrases in gallo-romance

143

go.prs.3sg everywhere do.ger (= such works he goes everywhere doing)’ (Cançó de Santa Fe 134), ieu trobei, tant anei queren lit. ‘I find.pfv.1sg so.much go.pfv.1sg search.ger (= I found [it], (after) much searching)’ (RdJ 10158)), and it pre-dated the go + infinitive construction, as witness its attestation in the eleventh-century Lo Poema de Boeci. In the early period (twelfth century), it is possible to encounter a go + infinitive construction which clearly is used instead of go + gerund, suggesting that the two overlapped at least partly in the mind of the speakers, most probably on account of their shared imperfective aspect. In (23a) Jaufré is on a quest to find Taulat, and questions a knight as to his whereabouts; since the quest is ongoing, the use of go + infinitive must necessarily be interpreted as equivalent to the use of go + gerund,2⁹ witness also the later reprises by the knight (23b–c): (23)

a. […] mas but cavallier knight

si.m if=me que whom

sabiatz dir lo ver // d’ un know.ipfv.2pl say.inf the truth of a vauc querer, // ves qual part go.prs.1sg seek.inf towards which part

lo poria seguir, // […]. (OOcc.) him= can.cond.1sg find.inf ‘[…] but if you could tell me the truth // about a knight I am seeking, // and where I might find him, // […].’ (RdJ 4757–4759) b. “E per qual ops l’ anatz querentz?” (OOcc.)3⁰ and for what reason him= go.prs.2pl seek.prs.ptcp “‘And for what reason do you go/are you looking for him?”’ (RdJ 4764) c. […]Car vos aquest anatz cercan, […]. (OOcc.) because you this go.prs.2pl seek.ger ‘[…] Because you are looking for him […].’ (RdJ 4792) Second, the grammaticalization of the periphrasis as an expression of futurity rather than past was also favoured by the anchoring of the construction which, as mentioned above, was devoid of temporal features in that it encoded only posteriority: anchoring to the speech time, which led to the development of a future construction, and situational anchoring to the narrative tense, which led to the development to an expression of past. These factors, coupled with the fact that functional polysemy was not tolerated for long, eventually led to the loss of the past interpretation, such that the future reading became conventionalized and finally grammaticalized as an expression of future tense.

2⁹ There is also the possibility, of course, that this may be purely the result of misreading a scribe’s ‘n’ for an ‘r’. 1⁰⁰ Jeanroy (1941:382f.) comments on the frequent substitution of the gerund with forms of the present participle.

144

sandra paoli and sam wolfe

4.5 Concluding remarks In this chapter we have argued that both go-past and go-future periphrases originated from the same construction, in which the basic spatial meaning of the motion verb was reinterpreted as posteriority at the time level. Along with Bres and Labeau (2013a), we understand this construction as being devoid of temporal meaning. In narrative stretches, in which the periphrasis was surrounded by verbs in the preterite form, as well as the historic present, a past interpretation obtained via situational anchoring, while in direct speech, the periphrasis conveyed an exhortation, a command/direction, or a prediction/intention, and derived its eventual future interpretation deictically, in that the prospective movement originates in the moment of speech. The past interpretation never fully extended to productively include the periphrasis with first and second persons in either French or Occitan, and eventually it fell out of use altogether, leaving the road completely free for the periphrasis to fully grammaticalize as an expression of future. Reconciling the two developments of the go + infinitive periphrasis by adopting a unified analysis that focuses on the component of posteriority has the advantage of maintaining the widely attested generalization that structures with verbs of motion develop into future grams. Furthermore, this also allows us to maintain the equally justified cognitive schema source–path–goal.

5 The tornare-periphrasis in Italo-Romance Grammaticalization ‘again’! Mair Parry

5.1 Introduction Abstract grammatical notions may derive from a limited number of typical event schemas that are basic to human experience (Heine 1993:30f.). For the expression of verbal tense and aspect, recourse to the ‘concrete’ concept of motion is prevalent cross-linguistically, with propositions based on the generic verbs go and come being particularly productive. This chapter aims to account for the grammaticalization in several Italo-Romance varieties of an intransitive directional motion verb meaning ‘return’ or ‘go/come back’ (originally deriving from transitive Lat. tornare ‘turn (on a lathe)’) that acquires aspectual auxiliary functions (‘go back to being/doing’, ‘keep doing’, ‘do again’) in periphrastic constructions for the most part with an infinitive. Some of the dialects show a further development into an uninflected adverb expressing repetition, torna (and cognates) ‘again’. The auxiliary development is a classic illustration of Heine’s Verb-to-TAM chain, allowing one to trace the emergence of various semantic values associated with the notions of restitution and repetition, while also exhibiting unusual auxiliary-switch phenomena of particular note for Romance and general syntactic theory. In comparison with go and come verbs, little has been written about the grammaticalization in Romance of verbs meaning ‘return’ (but see SetterbergJørgensen 1950; Yllera 1980; Giacalone Ramat 2001; Rosemeyer 2016), or about the development of expressions for ‘again’.

5.1.1 Grammaticalization Grammaticalization, the development of a lexical (or functional) element into a functional (or more functional) element, is viewed in the functional/typological literature as a complex process that involves pragmatic extension and inferencing,

Mair Parry, The TORNARE-periphrasis in Italo-Romance In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Mair Parry (2022). DOI: 10.1093/oso/9780198870807.003.0006

146

mair parry

semantic change (extension followed by reduction, frequently referred to as bleaching; Hopper and Traugott 1993:87–89); change of grammatical status (reanalysis leading to decategorialization, or transcategorization; Ramat 2001); and often phonological and morphological attrition (Lehmann 1985). One of the typical clines of development is that represented by full verbs with an infinitival complement that are reanalysed as light verbs, then auxiliaries, and may eventually become invariable tense, aspect, or mood markers or reduced to variable affixes on the main verb, as in the case of the Romance future formed from the infinitive plus inflected forms of habere ‘have’ (cf. discussions in §1.3.2 and §3.2 in this volume). Such changes often derive from a subjective perspective on the event, entailing the metaphorical or metonymic use of common lexical items (for the role of subjectification in grammaticalization, see Langacker 1990 and Traugott 1995), a use that if adopted by the community of speakers leads to polysemy, as with It. andare ‘go’, venire ‘come’. A consistent by-product of grammaticalization is the phenomenon of divergence, whereby earlier lexical meanings persist alongside later more grammatical meanings,1 e.g., It. venire ‘come’ retains its original deictic sense of ‘motion towards the speaker’, as well as having auxiliary functions in the aspectual periphrasis (+ gerund: veniva ripetendo ‘come.pst.ipfv.3sg repeat.ger (= (s)he was repeating)’) and in the passive construction (+ past participle: viene ripetuto ‘come.prs.3sg repeat.ptcp (= it is repeated)’) (Giacalone Ramat 1995). Lichtenberk (1991) describes extensive heterosemy deriving from go, come, and return verbs in Oceanic languages. While occasional reference is made to other Romance varieties, this chapter focuses on Italo-Romance and is structured as follows: §5.2 presents nineteenth- and twentieth-century data regarding the auxiliary use of reflexes of tornare; §5.3 analyses the diachronic development; §5.4 concludes the discussion.2

5.2 Italo-Romance verbal periphrases with tornare These occur to a greater or lesser extent in most Italo-Romance varieties, although their use is in some regions, e.g., Piedmont and Liguria, more restricted nowadays than in the past, due to the development of an invariable adverb torna. Modern varieties may have both the auxiliary periphrasis (+ infinitive) and the adverb, but 1 See Hopper and Traugott (1993:116–120, 124–126), and Chapter 3 in this volume. 2 I dedicate this chapter to my long-standing friend Martin, who has made an invaluable contribution to Romance linguistics and dialectology as well as important advances in morphological analysis. I am grateful to the following for dialect data: Sergio Aprosio (Ligurian), Paola Benincà (Paduan), Tiziana Buxton (Ladin), Silvio Cruschina (Sicilian), Rosa D’Agostino (Cairese), Franco Finco (Friulian), Giulio Lepschy (Venetian), Alessandra Lombardi (Calabrian), Iolanda and Giancarlo Mazzucco (Cairese), Fiorenzo Toso (Ligurian) and especially to my informants from the Monferrato, Luciano Origlia (Aglianese) and Tommaso Scaglione (Astigiano). For valuable comments and suggestions during the preparation of this chapter, I wish to thank in particular Paola Benincà, Michela Cennamo, Rodney Sampson, and the editors. All responsibility for remaining errors or misinterpretations is naturally mine.

the tornare-periphrasis in italo-romance

147

usually one of these constructions is dominant (for Piedmontese, see Bertocci and Rossi 2011; for Ligurian, Toso 2004). For the wide range of meanings they express I shall follow Wa¨lchli (2006) by distinguishing between repetitive, continuative, and restitutive ‘again’.3 The notion of ‘doing something again’ concerns the ‘external’ repetition of an event:⁴ this is a complex notion which can include not only the repetition of an action one or more times (i.e., excluding the frequent and the habitual) = repetitive, but also, depending on the Aktionsart of the verb and of the sentence, the semantic interpretation may be rather one of continuation of an interrupted activity as in ‘she went back to sleep (again)’ = continuative, or the restoration of a previous state of affairs, ‘she was able to walk again’. The restoration of a former state may also be achieved through the reversion of an earlier event, as in, ‘she closed her eyes and opened them again’ = restitutive. The semantic changes undergone by tornare derive ultimately from the human propensity for perceiving connections between external phenomena and abstract human experiences in terms of more general features; this leads to the extension in use of a linguistic form denoting one of them to refer to both. The meanings of restoration of a state and continuation of an interrupted activity thus derive from a metaphorical interpretation of human behaviour in terms of redditive motion in space (return to a point of departure), while the idea of repetition of an action requires that the first and the second event be conceived as temporally separate (see §5.3). The frequent link between again expressions and the verb meaning ‘return’ is highlighted in a detailed typological study of a sample of 100 languages selected from all major language groups: Wa¨lchli (2006) distinguishes between two key uses of again: heavy again is used in emphatic contexts, where an explicit again expression is found in all languages, whereas light again represents an inflationary use (often associated with restitutive or continuative meaning) found only in some languages. There is a significant but not necessary correlation on the one hand between heavy again and syntactic–analytical expression (adverbs,

3 Repetitive and restitutive are used by Kamp and Rossdeutscher (1994:190f.) following FabriciusHansen (1980; 2001). Hartmut Haberland (EUROTYP handouts and p.c.) distinguishes between repetition and meanings which do not necessarily imply a ‘second time’: reversal/restitution, and revision (‘do differently’) which may be covered by the same adverbs, verbal affixes, or by both. ⁴ There is a wide range of linguistically relevant distinctions associated with the notion of repetition: verbal plurality may be event-internal (repetitive actions) or event-external (repeated actions) (Cusic 1981), and the latter may range from a simple repetition (‘doing something again’) through several repetitions to frequent and habitual actions. The situation is actually even more complex: ‘[t]he interesting fact is that plurality in the event domain is a much more composite phenomenon than its counterpart in the objectual domain because of the multidimensional nature of events. In fact, events can be plural with respect to time, space, or one of their arguments or with respect to more than one of these variables at the same time’ (Lenci and Bertinetto 2000:273). Numerous terms have been coined in an attempt to distinguish between the different notions (e.g., iterative, reiterative, semeliterative, pluractional, frequentative, continuative, repetitive action, repeated action) but because of the close semantic associations and diachronic links between them, and the frequent confusion between lexical (Aktionsart) and grammatical aspect, terms such as iterative have been used for both and are thus ambiguous.

148

mair parry

particles, auxiliaries) and on the other between light again and morphological expression (i.e., affixes, e.g., Eng. re- and Romance r(e/é/i)-).⁵ Tornare periphrases are used for both heavy and light again and often correspond in the dialects to Italian derivational forms in ri- (riaprire ‘reopen’, rifare ‘redo’, rimandare ‘resend’).

5.2.1 Italo-Romance data (nineteenth–twentieth centuries) 5.2.1.1 Infinitival periphrases In Italian, tornare + a ‘to’ + infinitive is classed by Giacalone Ramat (2001) as an ‘emergent auxiliary’ with an aspectual meaning of iterativity. Similar structures are found in Occitan (Ronjat 1937:III:555, 562), Gascon (Rohlfs 1970:197), Catalan (Coromines 1988:607; Blasco Ferrer 1984:289), Portuguese (Morais Silva 1958:46, 606), and early Spanish (Yllera 1980:198), modern Spanish having replaced tornar with the verb volver, which exhibits a parallel grammaticalization path. It occurs infrequently in a recent corpus of spoken Italian (De Mauro et al. 1993), being restricted mainly to verbs of saying and predicates of perception and mental activity: (1)

Torno a ripetere. (It.) return.prs.1sg to repeat.inf ‘I say it again / I repeat.’

(Giacalone Ramat 2001:127)⁶

In Piedmontese and Ligurian, for example, the construction is frequent in nineteenth-century texts and although still used in conservative dialects, e.g., in Cairese and Aglianese, it has been largely replaced by adverbial torna. Compound tenses have been selected for illustration, so as to exemplify the noteworthy features of auxiliary switch, a phenomenon whereby the temporal (perfect) auxiliary is determined by the infinitive (and no longer by the finite verb which is undergoing grammaticalization), and clitic climbing, whereby clitic arguments of the infinitive are located on the finite verb. These features have been attributed to the phenomenon known as Restructuring, ‘[t]he process of structural simplification (Restructuring) which turns an original biclausal configuration into a monoclausal one, forming a complex verb out of the complement and matrix verbs’ (Cinque 2004:134, but see Cinque 2006).⁷ In Piedmontese and Ligurian the ⁵ The inflationary use of the affix persists, e.g., Telmon (1993:124) cites examples from the Abruzzo bearing no meaning of repetition or return: riportare ‘carry’, rimandare ‘send’. 1⁰⁶ This represents the most frequent collocation of tornare in the corpus; strictly speaking it means ‘I repeat again’, but as the concept of ‘again’ lends itself to inflationary use, torno a ripetere can simply mean ‘I repeat’. ⁷ Cinque (2006) argues that all verbal periphrases involving restructuring verbs are always functional and monoclausal, even in the absence of transparency effects such as clitic climbing and auxiliary change. However, motion verbs and seem may have a lexical and a functional use; the former use allows a complement, in which case a biclausal structure is posited (Cinque 2006:36).

the tornare-periphrasis in italo-romance

149

selection of perfective auxiliary have or be has in the main the same distribution as in Italian: be for unaccusatives, have for unergatives and transitives, resulting in the selection of be for the unaccusative motion verb meaning return (see (2a)). The difference in meaning (lexical ~ aspectual: repetitive/continuative/restitutive) emerges clearly in (2a) and (2b), the latter reading being semantically impossible, since there is only one event. That restructuring applies in (2b) is apparent from the auxiliary switch to have. Note also that, although clitic climbing normally correlates with this, it is not obligatory (2h): (2)

Piedmontese/Ligurian (dialect of Cairo Montenotte, a transitional variety)⁸ a. I sun turnòi a piè u liber, 3pl.scl= be.prs.3pl return.ptcp.mpl to get.inf the book, ma u n’i’èra ciû! but 3msg.scl= neg=there=be.pst.ipfv.3sg more ‘They went back to get the book, but it wasn’t there any more!’ [physical motion; lex. tornare, unacc: be] b. I an turnò a piè u liber, 3pl.scl= have.prs.3pl return.ptcp to get.inf the book, **ma u n’i’èra ciû! but 3msg.scl= neg=there=be.pst.ipfv.3sg more ‘They took the book again #but it wasn’t there any more!’ [repetitive; asp. tornare + trans. inf.: have] c. A i’eu turnò=ie a 1sg.scl=3dat.cl.=have.prs.1sg return.ptcp=3dat.cl to telefunè.⁹ telephone.inf ‘I telephoned him again.’ [repetitive; asp. tornare + unerg. inf.: have] sun a mnì ieri. d. I turnòi 3pl.scl= be.prs.3pl return.ptcp.mpl to come.inf yesterday ‘They came again yesterday.’ [repetitive; asp. tornare + unacc. inf.: be]

1⁰⁸ The turnè periphrasis is also used in the reflexive or si-passive construction in Cairese, but not in compound tenses (for other restrictions on si constructions in Piedmontese, see Parry 1997b): (i) Er cadreghe i=s turnu a mitè=se a pósct. (Cairo Montenotte) the chairs 3pl.scl=refl return.prs.3pl to put.inf.=refl to place ‘The chairs must be put back.’ (deontic use of reflexive) 1⁰⁹ Most Piedmontese varieties (except Canavese) attach complement clitics enclitically to the past participle, whereas conservative varieties (e.g., Cairese) preserve the earlier (intermediary) stage in which the pronoun is both proclitic to the perfective auxiliary and enclitic on the past participle (as in 2e–f) (Parry 1995). (2g) shows clitic climbing, no cliticization to the past participle, but a copy remains in its original position.

150

mair parry

e. A=m sun turnò=me a lavè. I=1sg.refl be.prs.1sg return.ptcp=1sg.refl to wash.inf ‘I went back to washing / I got washed again.’ [continuative/repetitive; asp. tornare + refl inf.: be] f. U=m=a turnò=me a ciapè. 3sg.scl=1sg.ocl=have.prs.3sg return.ptcp=1sg.ocl to catch.inf ‘He caught me again.’ [repetitive: asp. tornare + trans. inf.: have] g. Mé surèla m=a turnò a my sister 1sg.ocl=have.prs.3sg return.ptcp to mandè=me indré. send.inf =1sg.ocl back ‘My sister sent me back (again).’ [restitutive–reversal asp. tornare + trans. inf.: have] h. I an turnò a fè=le 3pl.scl= have.prs.3pl return.ptcp to do.inf =it cum u l’èra prima. as 3sg.scl=3sg.scl=be.pst.ipfv.3sg before ‘They restored it to its former state.’ [restitutive; asp. tornare + trans. inf.: have] Similar paradigms are found in the dialect of Agliano: (3)

Piedmontese (Aglianese) a. A i’oma tornà=ro a fè. 1pl.scl.1pl.scl=have.prs.1pl return.ptcp=it to do.inf ‘We did it again/re-did it.’ [repetitive; asp. tornare + trans. inf.: have] b. I’eve a balè? tornà 2pl.scl=have.prs.2pl return.ptcp to dance.inf ‘Did you go back to dancing / dance again?’ [continuative/repetitive; asp. tornare + unerg. inf.: have] c. La fia a r’è tornà-ia a sven-e. the girl 3sgf.scl=3sg.scl=be.prs.3sg return.ptcp-fsg to faint.inf ‘The girl fainted again.’ [repetitive; asp. tornare + unacc. inf.: be] d. A son tornà=me lavè.1⁰ 1sg.scl be.prs.1sg return.ptcp=1sg.refl wash.inf ‘I got washed again.’ [repetitive; asp. tornare + refl inf.: be]

11⁰ In rapid speech the preposition may be omitted, but my informant always inserted it when asked to repeat the sentence. However, erosion typically results from the grammaticalization process (Heine 1993:106–112).

the tornare-periphrasis in italo-romance

151

Auxiliary switch with reflexes of aspectual tornare appears in Ladin, Friulian, Calabrian, and Sardinian, as well as in Occitan (Wheeler 1988:270f.; Mok 1991:105): (4)

Friulian no=l veve nančhe tornât a viergi neg=3msg.scl have.pst.ipfv.3sg not.even return.ptcp to open.inf i vôj. the eyes ‘He had not even opened his eyes again.’ (Faggin 1985:1498) [restitutive; asp. tornare + trans. inf.: have] (5) Calabrian (Cosentino) Ste littere, l’annu turnate a mannà. these letters, them=have.prs.3pl return.ptcp.fpl to send.inf ‘These letters, they have sent them again.’ [repetitive; asp. tornare + trans. inf.: have] (6) Sardinian a. So torratu a léghere cussu libru. be.prs.1sg return.ptcp to read.inf that book ‘I went back to read that book.’ (Jones 1993:151–153) [lex. tornare: motion: be] b. Appo torratu a léghere cussu libru. have.prs.1sg return.ptcp to read.inf that book ‘I read that book again.’ [repetitive; asp. tornare + trans. inf.: have] A slightly different scenario is found in the varieties of north-east Italy, where auxiliary choice is less clear-cut: in the Ladin of San Vito di Cadore there has been an optional extension of have to unaccusative verbs. The alternation between the two auxiliaries with unaccusative verbs (as in 7b–c), however, involves a semantic distinction: given the Undergoer role of the subject (Van Valin and LaPolla 1997:141), the use of be is the unmarked choice (with agreement of the past participle), and according to my informant it also implies that the subject is more affected. For her, the use of have suggests a more impersonal occurrence, often unexpected, with focus on the event rather than on the end result. (7)

Ladin (San Vito di Cadore) a. Ai tornà a béte=lo in pé, have.prs.1sg return.ptcp to put.inf =it in foot, ma’l é tornà a tomà. but 3msg.scl= be.prs.3sg return.ptcp to fall.inf ‘I put it back up again, but it fell down again.’ [repetitive; asp. tornare + (i) trans. inf: have (ii) + unacc. inf. be]

152

mair parry

b. L’ è tornada a tomà. 3fsg.scl= be.prs.3sg return.fsg to fall.inf ‘She fell again.’ [asp. tornare + unacc. inf.: focus on end result: be] c. L’ ha tornà a tomà. 3fsg.scl= have.prs.3sg return.ptcp to fall.inf ‘She fell again (unexpectedly).’ [asp. tornare + unacc. inf.: focus on event: have] Venetian shows a similar extension of have to unaccusative constructions for the aspectual repetitive sense (see (8d)). The extension of have may be related to the fact that in Ladin and Venetan dialects perfective auxiliary have is used more widely than in Italian, such as in impersonal and reflexive constructions (Marcato and Ursini 1998:250–254). A Venetian informant commented that the use of be could favour the literal meaning of physical motion. (8)

Venetian a. Lo go tornà a metar in pie. it= have.prs.1sg return.ptcp to put.inf on foot ‘I put back it up again.’ [repetitive; asp. tornare + trans. inf.: transitive: have] b. Ma po el xe tornà a cascar. but then 3msg.scl= be.prs.3sg return.ptcp to fall.inf ‘But it fell down again.’ [repetitive; asp. tornare + unacc. inf.: be] c. La se ga ripreso, ma po 3fsg.scl= refl have.prs.3sg regain.ptcp butthen la xe tornada a svignir. 3fsg.scl be.prs.3sg return.ptcp.fsg to faint.inf ‘She regained consciousness, but then fainted again.’ [repetitive/restitutive; asp. tornare + unacc. inf.: be] d. El xe passà e el xe/ga 3msg.scl= be.prs.3sg pass.ptcp and 3msg.scl= be/have.prs.3sg tornà a passer. return.ptcp to pass.inf ‘He passed by and then he passed back / he continued to pass back and forth.’ [repetitive; asp. tornare + unacc. inf.: be/have]

A further development is found in Paduan, where both auxiliaries are seemingly available for tornare periphrases involving all types of infinitive, the variation tending to reflect the distinction between the restitutive and continuative meanings on the one hand (be) and repetitive meaning on the other (have). The oscillation between the two auxiliaries supports the Auxiliary Selection Hierarchy

the tornare-periphrasis in italo-romance

153

proposed by Sorace (2000), since the use of be tends to correlate with telic change, and that of have with a more agentive, less affecting process. More detailed research is required to ascertain the subtle nuances of variation, especially in relation to verbal Aktionsart, although there is a risk that judgements may be influenced by infrequent use and by Italian structures. In the meantime, the following examples represent those of one informant: (9)

Paduan a. El zè/ga tornà cascare. 3msg.scl=be/have.prs.3sg return.ptcp fall.inf ‘He went back to falling’ vs ‘he fell down again.’ [asp. tornare + unacc. inf.: (Accomplishment verb)] b. El zè/ga tornà a parlare in dialeto. 3msg.scl= be/have.prs.3sg return.ptcp to speak.inf in dialect ‘He went back to speaking in dialect’ / ‘he spoke again in dialect.’ [asp. tornare + unerg. inf. (Activity verb)] c. El zè tornà a scrivare la letara. 3msg.scl= be.prs.3sg return.ptcp to write.inf the letter ‘He went back to writing the letter (after an interruption).’ [continuative; asp. tornare + trans. inf. (Active Accomplishment)] d. El gà tornà a scrivare la letara. 3msg.scl= have.prs.3sg return.ptcp to write.inf the letter ‘He rewrote the letter (because it contained errors).’ [repetitive; asp. tornare + trans. inf. (Active Accomplishment)]

In (9a) be conveys the restoration of a previous (more enduring) situation (thus restitutive or continuative senses), e.g., he went back to falling (after being momentarily stopped by a tree, in the case of a long descent, or, after getting better briefly, in the case of a physical weakness and tendency to lose one’s balance). In (9b) be implies either that the person carried on after a pause or went back to speaking in dialect (e.g., after having lost his speech in an accident). In both (9a– b) have represents semelfactive use or a few repetitions (repetitive sense), e.g., he fell again (once or twice), he used dialect again (in a given conversation). However, a referential direct object may induce the repetitive meaning with be also. Thus, the greater the transitivity of the sentence as a whole (in the sense of Hopper and Thompson 1980), the greater the likelihood of auxiliary have. One may conclude that in most Italo-Romance varieties the choice of perfective auxiliary for periphrastic tornare depends directly on the Aktionsart and argumental structure of the dependent infinitive. However, in some north-eastern ones (Venetian, Paduan, Friulian) it is sensitive to semantic and pragmatic variables at the sentence level. The data indeed suggest that the availability of two auxiliaries may be exploited by speakers seeking to distinguish between the different senses of again: restitution, continuation, and repetition.

154

mair parry

5.2.1.2 Paratactic or serial periphrases Less well known are constructions based on the sequence of two finite verbs, the first usually a modal or motion verb (Ascoli 1896). These constructions, which may be asyndetic or coordinated with the reflexes of Lat. et or ac (Benincà 1997:132; Ledgeway 1997; 2016a), are not found in standard Italian, except for the fossilized expression comprising two imperatives: (10) Vattelapesca! ( auxiliary status) comes from the restructuring transparency effects described in §5.2, namely auxiliary switch and clitic climbing (abbreviated here as AS and CC). As a lexical unaccusative verb, tornare takes perfective auxiliary be, but when used aspectually, the restructured complex verb takes have with non-unaccusative verbs, the following being to my knowledge the first (rare) attestation in north-western Italo-Romance texts:1⁵ (32) lo supradicto domino Baptista ha tornato the above.mentioned master Baptista have.prs.3sg return.ptcp a mete a se ipso … dominus Franciscus to put.inf by refl self master Franciscus a lo dicto officio de la cancellaria to the said office of the chancellery ‘the above-mentioned Master Baptista of his own initiative put Master Franciscus back in the said chancellery post.’ (diplomatic letter from Consul Battista Giustiniani at Caffa in 1474, in Vigna 1879:407, 15th-c. Lig.) [restitutive; asp. tornare + trans. inf.: have] Examples abound in 19th-century Piedmontese and Ligurian dialect literature:

1⁵ My thanks to S. Aprosio for this reference. As already noted, auxiliary switch with tornare is unacceptable in modern standard Italian.

the tornare-periphrasis in italo-romance

163

(33) 19th-c. Turinese a. sto mé fio¨l l’era mort, e this my son 3sg.scl=be.pst.ipfv.3sg dead, and l’è tornà a vive 3sg.scl=be.prs.3sg return.ptcp to live.inf ‘my son was dead and he has returned to life.’ (Biondelli 1853:505) [restitutive; asp. tornare + unacc. inf.: be] b. e l’o¨ tornàlo a trové (ib.) CC and him=have.prs.1sg return.ptcp=him to find.inf ‘and I have found him again.’ [restitutive; asp. tornare + trans. inf.: have] c. e l’o¨ tornàlo a vede viv (ib.) CC and him= have.prs.1sg return.ptcp=him to see.inf alive ‘and I have seen him alive again.’ [restitutive; asp. tornare + trans. inf.: have] d. L’ autra, … a l’ha tornaine the other 3scl.3scl have.prs.3sg. return.ptcp=for.her=of.it a veuidè CC to empty.inf ‘The other, however, poured her some more of it.’ (Ferrero 1976:180) [repetitive; asp. tornare + trans. inf.: have] (34) o l’à tornòu à piccâ CC 3sgm.scl =3scl=has return.ptcp to knock.inf ‘he knocked again.’ (Poggi 1992:209, 19th-c. Gen.) [repetitive; asp. tornare + trans. inf.: have] As already noted, in standard Italian auxiliary switch from have to be is common for modals and certain aspectuals as a consequence of restructuring (note also past participle agreement with the subject with auxiliary essere be): (35)

Modern Italian a. Ho voluto andarci. have.prs.1sg wanted.ptcp.msg go.inf=there b. Ci sono voluta andare. there= be.prs.1sg wanted.ptcp.fsg go.inf ‘I (fem.) wanted to go there.’

–CC, –AS ~ +CC, +AS

Normally, however, in the standard language motion verbs do not show a switch from be to have, although this is attested in some dialects, as in the Cosentino example (39). Auxiliary switch in this direction was, in fact, deemed by many linguists to be impossible (Rizzi 1982:19; see also Burzio 1986:333; Kayne 1989:253;

164

mair parry

Andrews and Manning 1999:56; Monachesi 1999:172). Cinque (2004:178) writes: ‘[a]uxiliary change is possible only from avere “have” to essere “be” (not viceversa) and with a subset of the restructuring verbs (volere “want”, potere “be able”, dovere “must”, cominciare, iniziare “begin”, continuare “continue”) for reasons that remain to be understood’. Two apparent cases of auxiliary switch involving be > have in restructuring contexts in old Piedmontese (36) and old French (37) (but not in infinitival predicates) are cited by Benucci (2001:75). (36) queste preere que sum encoy avue fayte en these prayers that be.prs.3pl today have.ptcp do.ptcp.fpl in chesta casa this house ‘these prayers that have been done (= said) today in this house.’ (Le ‘Recomendaciones’ di Saluzzo, l.133, ed. Gasca Queirazza 1965:27, OPie.) (37) Amis, ou est li reis? Molt l’ai alet friend, where be.prs.3sg the king much him=have.prs.1sg go.ptcp querant seek.ger ‘(My) friend, where is the king? I have travelled a long way seeking him.’ (Pélerinage Charlemagne 279, OFr.) However, on closer inspection, these are not convincing counterexamples: the first relates to a passive construction and the past participle of esse ‘be’ in this text is regularly avu (for this suppletive use with the verb esse in early Piedmontese, Lombard, and other Romance texts, see Rohlfs 1969:19 n.1). As for the second example, old French aler can take auxiliary have in certain contexts, especially if there are ‘indications of duration, distance or road travelled, such as tant “so much, so long”, molt “much, a long time”’ (Jensen 1990:291). See also Buridant (2000:373), and Rohlfs (1969:123) for Italian dialects, e.g., Lig. a ndèt ‘has gone’ (AIS Map 522). A transitive use of andare in old Neapolitan, with agreement of the past participle with the direct object referring to the distance covered, is recorded by Cennamo (2002): (38) un viandante, avendo andata multa via a traveller, have.ger go.ptcp.fsg much.fsg way.f ‘a traveller, having travelled a long way.’ (Vita e favole di Esopo, Gentile 1988:161.10, 15th-c. Nap.) The ban on auxiliary switch is attributed by Butt (1997:145f.) and Cinque (2004:172f. n.31) to the fact that these motion verbs are ‘light’ or ‘semi-functional’ verbs, which retain a reduced argument structure (whereas auxiliaries do not). However, Italo-Romance dialect data reveal that there is no inherent incompatibility between the aspectual use of motion verbs and auxiliary switch: in the dialect

the tornare-periphrasis in italo-romance

165

of Cosenza (Calabria), lexical jì ‘go’ requires be (39a), whereas aspectual use shows a switch from be to have (39b), if the infinitive is not unaccusative: (39) a. Signu jutu be.prs.1sg go.ptcp ‘I went to take it.’ b. L’aju jutu it=have.prs.1sg go.ptcp ‘I went and took it.’

a ru piglià. to it= take.inf a piglià. (Cos.) to take.inf

Furthermore, with unaccusative infinitives both auxiliaries are again possible even though both ji ‘go’ and mora ‘die’ on their own would select be: (40) a. Cc’è gghjut’ a there=be.prs.3sg go.ptcp to ‘He went there in order to die.’ b. Cc’ha gghjut’ a there=have.prs3sg go.ptcp to ‘He went and died there.’

mora. die.inf mora. (Cos.) die.inf

There is, however, a semantic, aspectual distinction between the two uses, relating to the degree of telicity and probability of the event: have frequently implies an unexpected, unintentional event as in (40b).1⁶ This recalls the use of have in Ladin (see (7c)) to denote focus on the event rather than on the resultant state of the affected argument.1⁷ That many Italo-Romance dialects manifest auxiliary switch in the case of aspectual tornare confirms the loss of its own argument structure and its auxiliary status. However, clitic climbing and auxiliary switch do not necessarily have semantic consequences (compare the synonymous Italian modal constructions, e.g., ho dovuto andarci have.prs.1sg must.ptcp.msg go.inf=there and ci sono dovuta andare there=be.prs.1sg must.ptcp.fsg go.inf ‘I had to go there’, both uttered by a female speaker, and the examples in (35)). As pointed out by Giacalone Ramat (2001:125), restructuring in Italian does not entail the replacement of the lexical 1⁶ See Carden and Pesetsky (1977:89) for the English ‘unexpected event’ construction. In similar Italian uses only be is allowed: (i) Perché sei andata a mettere questo vaso sul caminetto? (It.) why be.2sg.prs go.ptcp.fsg to put.inf this vase on.the mantelpiece ‘Why did you go and put this vase on the mantelpiece?’ 1⁷ Such variation ties up with the correlation between have and activity semantics with simple verbs that allow the use of both auxiliaries. For example, Sorace (2000:878), referring also to Benincà and Cinque (1992), notes that with weather verbs in Italian be can be used only when a change of location or directional value is implied, have being used for non-motional processes. See also for Neapolitan, Ledgeway (2000:301 n.23) and Cennamo (2002) regarding the use of both auxiliaries with figliare ‘give birth’ in early Neapolitan depending on the way the event is conceived.

166

mair parry

meaning of motion verbs by the new grammatical meaning: both torno a prenderlo return.prs.1sg to take.inf=it and lo torno a prendere it=return.prs.1sg to take.inf can refer to physical motion, while both torno a dirlo return.prs.1sg to say.inf=it and lo torno a dire it=return.prs.1sg to say.inf can mean ‘I say again/repeat’ (Giacalone Ramat 2001:125). Indeed, clitic climbing can occur also in the sixteenth-century Piedmontese lexical use of torné: (41) e de tut ve=tôrnreu dé avis and of all you.pl=return.fut.1sg give.inf notice ‘and I’ll return to tell you about it all.’ (Alione, B285, OPie.) However, the more advanced grammaticalization process undergone by tornare in Italo-Romance dialects that have two auxiliaries (have and be) appears in most varieties to associate the loss of syntactic independence of tornare (after restructuring) with the aspectual function, while the use of have is definitely excluded if the sense is the original lexical (motion) interpretation.1⁸ When asked for grammaticality judgements on the use of be for the aspectual functions if the infinitive is a have-selecting verb, my Piedmontese and Ligurian informants did not totally reject this possibility but, given the fact that they would normally use the adverb torna rather than the tornare+infinitive construction, it is difficult to gauge whether their acceptance is due to Italian influence or the persistence of a situation prior to the grammaticalization of auxiliary switch. The use of be with be-selecting infinitives can, of course, give rise to ambiguity between a lexical and an aspectual interpretation, a situation which may have contributed to the spread of have as the aspectual marker in perfective contexts in north-eastern dialects (see the Paduan examples in (9)).

5.3.3 Cliticization and erosion? As with the modal auxiliaries there are no cases of a reduction from verbal auxiliary to affix, but in rapid speech the preposition a ‘to’ may fall. Most interestingly, semantic, morphological, and syntactic factors have led to the emergence in some dialects of a form torna ‘again’, which has been recategorized as an independent adverb. It is possible that this represents a further stage of development of the sort that Ledgeway (2016a) proposes for the Salentino aspectual marker va, from the motion verb go (see also aspectual markers in African languages (Heine 1993:77) and Welsh dal ‘still’ (Griffiths and Jones 1995:1371)). The change from an auxiliary verb to a positionally restricted aspectual marker and then to an adverb that 1⁸ For example in Sardinian, ‘[w]ith the movement interpretation, torrare is an unaccusative verb (taking éssere as its perfective auxiliary) but with the iterative reading it shows the same transparency with regard to auxiliary choice as modal auxiliaries (e.g., it takes áere if the following infinitive is not unaccusative)’ (Jones 1993:152).

the tornare-periphrasis in italo-romance

167

may be omitted without affecting the syntax of the sentence represents an increase in what Lehmann terms ‘syntagmatic variability’ and may be considered a case of syntactic lexicalization (Willis 2007), the development of lexical forms from earlier functional items. This will be investigated in future research.

5.4 Conclusion Verbal periphrases involving the motion verb tornare ‘go/come back’ (originally deriving from Lat. transitive tornare ‘turn’) evolved through pragmatic extension and inferencing to convey the notion of ‘again’ across the whole Italian dialect spectrum. As a functional verb in a monoclausal structure, tornare often shows transparency effects such as clitic climbing and, in some dialects, a type of auxiliary switch (be > have) with non-unaccusative infinitives that had been deemed impossible. These new data, which trace the emergence of the various aspectual auxiliary functions (restitutive, continuative, and repetitive) of tornare, will also hopefully broaden typological and semantic discussions regarding the linguistic expression of the complex notion of ‘again’.

Textual sources AIS = Karl Jaberg and Jakob Jud (1928–1940). Sprach- und Sachatlas Italiens und der Su¨dschweiz. Bern: Zo¨fingen. Biondelli, Bernardino (1853). Saggio sui dialetti galloitalici. Milan: Bernardoni. (Anastatic reprint, Bologna: Forni, 1970). Bottasso, Enzo (ed. 1953). ‘Comedia de l’homo’, in L’opera piacevole di Gian Giorgio Alione. Bologna: Palmaverde, 1–35. Castellani, Arrigo (1976). I più antichi testi italiani. Bologna: Pàtron. Coromines, Joan (1988). Diccionari etimologic i complementari de la llengua catalana. Barcelona: Curial. Du Cange, C. Du Fresne (1733). Glossarium mediae et infimae latinitatis. Niort: Favre. Faggin, Giorgio (1985). Vocabolario della lingua friulana. Udine: Del Bianco. Faggin, Giorgio (1997). Grammatica friulana. Udine (Campoformido): Ribis. Ferrero, Carlo B. (1976). La bassa Russia. Turin: Viglongo. Gasca Queirazza, Giuliano (ed.). (1965). Documenti di antico volgare in Piemonte, I. Le ‘Recomendaciones’ del Laudario di Saluzzo. Turin: Bottega d’Erasmo. Gasca Queirazza, Giuliano (ed.). (1966). Documenti di antico volgare in Piemonte, II. Gli ordinamenti dei Disciplinati e dei Raccomandati di Dronero. Turin: Bottega d’Erasmo.

168

mair parry

Gentile, Salvatore (ed.) (1988). Giovanni Brancati, Vita e favole di Esopo. Volgarizzamento del secolo XV (Glossario di Rosa Franzese). Naples: Liguori. Griffiths, Bruce and Dafydd Glyn Jones (1995). Geiriadur yr Academi/The Welsh Academy Dictionary, Cardiff: University of Wales Press. Jensen, Frede (1990). Old French and Comparative Gallo-Romance Syntax. Beihefte zur Zeitschrift fu¨r Romanische Philologie 232. Tubingen: Narr. Jones, Michael Allan (1993). Sardinian Syntax. London: Routledge. Leone, Alfonso (1995). Profilo di sintassi siciliana. Palermo: CSFLS. Lizio-Bruno, Letterio (1871). Canti popolari delle Isole Eolie. Messina, n.p. Matalon, Zuan Nazzi (1977). Marilenghe: gramatiche furlane, Udine: Institût di Studis Furlans. Mok, Quirinus I. (1991). ‘Concurrence de tornar + infinitif et re- en occitan’, in Dieter Kremer (ed.), Actes du XVIIIe Congrès International de Linguistique et de Philologie romanes, Université de Trèves (Trier) 1986, III, Tu¨bingen: Niemeyer, 104–111. Moll, Francesc de B. (1969). Diccionari català–valencià–balear. Barcelona: Moll. Morais Silva, António de (1958). Grande diciona´rio da lı´ngua portuguesa. Lisbon: Editorial Confluência. OVI = the Opera del vocabolario italiano textual database. www. vocabolario.org or http://gattoweb.ovi.cnr.it/(S(se3fsd1kvoeagufqio3tu4uv)) /CatForm01.aspx. Palay, Simin (1961). Dictionnaire du béarnais et du gascon modernes. Paris: CNRS. Parodi, Ernesto G. (1898). ‘Studj liguri’. Archivio glottologico italiano 15: 1–110. Poggi, Giuseppe (1992). Ginn-a de Sampedaenn-a (ed. Fiorenzo Toso). Genoa: Microart. Ronjat, Jules (1937). Grammaire istorique [sic] des parlers provençaux modernes, III. Montpellier: Société des Langues Romanes. Scaglione, Massimo (ed. 1982). Teatro in piemontese: antologia di testi con note storiche e di regia. Turin: Piazza. Vigna, Amedeo (1879). ‘Codice diplomatico delle colonie Tauro-Liguri durante la signoria dell’Ufficio di San Giorgio (1453-1475)’. Atti della Società Savonese di Storia Patria 7: 5-1003.

6 Periphrases and irregular paradigms in Italo-Romance Silvio Cruschina

6.1 Introduction In the study of grammatical periphrases many scholars have focused on the intersection between these multi-word expressions and inflexional morphology, as well as on their integration within the inflexional paradigm (Spencer 2001; Ackerman and Stump 2004; Anderson 2011). This has resulted in a number of studies investigating the criteria to define inflexional periphrases in relation to the typical properties that bring them closer to inflexional morphology, such as intersectivity, non-compositionality, and distributed exponence (Sadler and Spencer 2001; Spencer 2001; Ackerman and Stump 2004; cf. also the discussion in §1.2.2 in this volume). Since periphrases typically straddle the border between morphology and syntax, they have also been treated with reference to grammaticalization, acknowledging their gradient nature and the possibility of identifying intermediate stages from fully fledged syntactic combination to full morphologization (Vincent 2011; Brown et al. 2012; Cruschina 2013). Within this line of investigation, most scholars have only focused on inflexional periphrases, namely, periphrastic constructions that realize cells in the inflexional paradigms of lexemes. In order to qualify as an inflexional periphrasis, a syntactic construction must therefore be integrated into the inflexional paradigm of a given lexeme and express a grammatical property that is typically realized inflexionally in the same language. Haspelmath (2000; 2002) calls such periphrases suppletive periphrases, which serve a gap-filling function within the paradigm, as opposed to categorial periphrases, which do not enter into a paradigmatic relation with synthetic forms. An example of suppletive periphrasis in Latin is given in Table 6.1 (Bo¨rjars, Vincent, and Chapman 1997; Haspelmath 2000). Latin verbs inflect for tense, aspect, mood, and voice; combinations of these features are usually realized by synthetic forms except in the forms of the perfectum passive, where

Silvio Cruschina, Periphrases and irregular paradigms in Italo-Romance. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Silvio Cruschina (2022). DOI: 10.1093/oso/9780198870807.003.0007

170

silvio cruschina Table 6.1 3sg indicative forms of Latin capere ‘take’ Present Imperfect past Perfective past Pluperfect Future Future Perfect

Active capit capiebat cepit ceperat capiet ceperit

Passive capitur capiebatur [captum est] [captum erat] capietur [captum erit]

a periphrasis consisting of the perfect participle and copula esse ‘be’ is used. In this inflexional paradigm, therefore, the intersection of grammatical features that are otherwise expressed by a single exponent (aspect and voice) is realized periphrastically. As examples of categorial periphrasis, Haspelmath (2000:660) cites the Fr. aller ‘go’ future (1a) and the Sp. estar ‘be’ progressive (1b), which are both entirely periphrastic. In this case, tense and aspect features are always realized by multi-word expressions—no synthetic form expressing the same grammatical categories exists in the same language (even though in French a degree of ‘competition’ and overlap exists with the synthetic future paradigm, cf. je chanterai ‘I will sing’; for fuller discussion see §2.4.10 in this volume)—and so, unlike suppletive periphrases, they do not fill gaps in an inflexional paradigm: (1)

a. Je vais chanter. (Fr.) I go.1sg sing.inf ‘I am going to sing.’ b. Estoy cantando. (Sp.) stay.1sg sing.ger ‘I am singing.’

In this chapter, building on previous work (Cruschina 2013), I discuss the contribution that the research on periphrastic constructions with a morphomic distribution (‘periphrastic morphomes’) can make to the debate on the integration between periphrases and inflexional paradigms, as well as on the grammaticalization and morphologization of a periphrasis. The constructions I consider here do not meet the criteria proposed for the definition of inflexional periphrases and, according to Haspelmath’s distinction, qualify as categorial periphrases. Nevertheless, they display properties that are typical of highly grammaticalized structures, and show features that are peculiar to the inflexional morphological domain. If periphrases are situated at the interface between morphology and syntax, an expected property that follows from their morphological nature, especially when

periphrases and irregular paradigms in italo-romance

171

grammaticalization places the syntactic construction (increasingly) within the realms of morphology, is ‘the paradigmatic organization of canonical morphology’ (Brown et al. 2012:245). Indeed, periphrases can display many of the recurrent features that are characteristic of inflexional paradigms, including canonical and non-canonical properties. Romance data confirm that these recurrent features can also emerge when the organization of a periphrastic paradigm with irregular forms follows a morphomic pattern (Corbett 2012:180–186; Cruschina 2013; Andriani 2017; Di Caro 2019a; 2019b; cf. also discussions in §1.2.2.1, §1.2.2.5, and §1.2.3 in this volume). Periphrastic morphomes provide us not only with a distinct type within the full typology of the possible manifestations of morphomic patterns (Maiden 2018a:§6.2.6, §12.2), but also with an interesting test bed to investigate the correlation between the degree of grammaticalization of a periphrastic construction and its conformity to relevant inflexional paradigms (Cruschina 2013). In what follows we will look at the parallelisms between inflexional and periphrastic morphomes, focusing on the extension of significant properties of the irregular inflexional paradigms to the periphrastic paradigms. These parallelisms include defectivity and suppletion within the periphrastic paradigms of motion and progressive constructions in Sicilian, Pugliese, and Salentino dialects (§§6.2–6.3). I will also address the question of alignment between inflexional and periphrastic morphomes. Insofar as they display the same distribution as inflexional morphomes, periphrastic morphomes satisfy the most important condition for morphomehood, that is, the irreducibly erratic paradigmatic distribution of irregularities (Maiden 2018a), but also the necessity for abstract generalization: their patterns are not unique, but rather recurrent and systematic. What happens, however, if an inflexional morphome in a specific Romance variety deviates from the systematic distribution? Will the corresponding morphome at the periphrastic level follow this language-specific divergency? This question will be answered with reference to the special distribution of the irregular preterite root in Sicilian and the morphomic attraction of the preterite cells of a modal periphrasis to the same pattern (§6.4).

6.2 Motion and progressive double inflexion constructions In most Italo-Romance dialects motion verb constructions involving an initial verb such as go and come (V1) typically feature a second verb in the infinitive (V2), as shown in (2) from the Pugliese dialect spoken in Bari. In these constructions, V1 and V2 are often linked by the subordinating element a ‘to’ (< Lat. ad); however, this is not obligatory and is in fact absent in some dialects. I will refer to this construction with an infinitival V2 as the infinitival construction:

172

silvio cruschina

(2) Mә vògg’ a ’ccattà u cappìddә névә. (Bar.) me= go.prs.1sg to buy.inf the hat new ‘I am going to buy a new hat.’ (Andriani 2017:231) In southern Italy, the infinitival construction is not the only option for speakers: a number of different combinations of a motion V1 followed by a finite V2 are available. In these constructions the two verbs can be connected by different elements, most commonly by the linker a.1 Following Cruschina (2013), I use the term Doubly Inflected Construction (DIC) for this kind of motion verb construction, where the two verbs share the same inflexional features and express a single event (Cardinaletti and Giusti 2001:386–388) which combines the activity of purchasing with the physical displacement necessary to achieve that goal. Examples of the DIC are provided below, where both V1 and V2 are in the first-person singular (3), the third-person singular (4), and the third-person plural (5) of the present indicative:2 (3) Vaju a pigghiu u pani. (Marsala, Sicily) go.prs.1sg a take.prs.1sg the bread ‘I go to fetch the bread.’ (Cardinaletti and Giusti 2001:373) (4) U veni a piglia dopu. (Mus.) him= come.prs.3sg a collect.prs.3sg later ‘He is coming to pick him up later.’ (Cruschina 2013:266) (5) ˈvonә (a) mˈmaɲʤәnә. (Martina Franca, Puglia) go.prs.3pl a eat.prs.3pl ‘They’re going to eat.’ (Ledgeway 2016a:159) On the basis of a number of syntactic and semantic tests, Cardinaletti and Giusti (2001; 2003) convincingly show that the DIC (the inflected construction, in their terminology) is monoclausal. They argue that the motion verbs involved as V1 are lexical categories merged as functional heads in the extended projection of the V2. More specifically, they define these verbs as ‘semi-lexical verbs’ because, while it is true that they lack or have lost most of their canonical lexical properties 1 For simplicity, in the examples the connecting element a is glossed as a. However, a long tradition of scholars have considered this element as the continuation of the Latin coordinating conjunction ac ‘and’ (Ascoli 1896; 1901; Rohlfs 1969:§710, §761; Leone 1973; Sornicola 1976; Cardinaletti and Giusti 2001; Ledgeway 2016a; Di Caro 2019a). Indeed, in some cognate Calabrian dialects the connective element is the same as the coordinating conjunction e ‘and’ from Latin et (Rohlfs 1969:§759). For this reason, the construction is treated as an instance of pseudo-coordination in several studies (Ledgeway 2016; Di Caro 2019a). At any rate, as argued in Cruschina (2013:271), since this connecting element is now desemanticized and contributes no meaning to the construction, its origin is not relevant to the synchronic analysis of the DIC. 2 The motion verbs that most typically appear in the Sicilian DIC are the local equivalents of go, come, pass, and send. Other verbs may enter the construction as V1 in some dialects (Di Caro 2019a; 2019b). On the special properties of send as V1, which involves both motion and causative semantics, see Todaro and Del Prete (2019) and Del Prete and Todaro (2020).

periphrases and irregular paradigms in italo-romance

173

(for example, with respect to argument and adjunct selection, the expression of an independent event), they still retain their motion semantics (cf. however §6.3).3 On the basis of the fact that V1, as typical of functional or auxiliary elements, does not select an independent argument structure and does not express a separate event, the DIC can be regarded as an aspectual periphrasis where the motion verbs go and come encode andative and venitive aspect, respectively (Cruschina 2013; Ledgeway 2016a; Andriani 2017). In particular, insofar as this aspect is always realized periphrastically in the varieties displaying the DIC, the construction qualifies as a categorial periphrasis that does not enter into a paradigmatic relation with synthetic forms (cf. §6.1). Both in Sicily and in Apulia, DICs with verbs of motion show a great deal of morphosyntactic microvariation. A type of radical impoverishment is found in western and central Sicily, leading to an uninflected and invariant V1 in the whole paradigm (Cardinaletti and Giusti 2001; 2003; 2019; Cruschina 2013).⁴

Table 6.2 The DIC forms for go (V1) and take (V2) in Sicilian 1sg 2sg 3sg 3pl imp.2sg

inflected V1 vaju a pigghiu vai a pigghi va a pigghia vannu a pigghianu va pigghia

invariant V1 va a pigghiu va a pigghi va a pigghia va a pigghianu va pigghia

It is important to note that the impoverished forms in these Sicilian dialects are not limited to the verb go but are also found with other motion verbs (Cardinaletti and Giusti 2019), and that the DIC version with invariant va (cf. Table 6.2) coexists with the version in which go displays regular inflexions. In some areas of eastern Sicily (for example Marina di Ragusa and Acireale) the grammaticalization of V1 appears to have reached the stage of affixation, where V1 has been reduced to a prefixal element attached to V2 realized as vo in (6) and o in (7) (data from Di Caro 2019a:30f.):

3 The monoclausal analysis can account for the different properties of the DIC, including obligatory clitic climbing, single event interpretation, indivisibility, and incompatibility with the arguments and adjuncts typically associated with motion verbs (see Cardinaletti and Giusti 2001; 2003; Manzini and Savoia 2005; Cruschina 2013; and Di Caro 2019a for more details). Cardinaletti and Giusti (2019) propose a refinement of this analysis, according to which V1 is merged in t, a head immediately above T(ense). See also Del Prete and Todaro (2020) for a more detailed semantic analysis of the single event interpretation, and Manzini, Lorusso, and Savoia (2017) and Lorusso (2019) for a different (biclausal) analysis of these constructions. ⁴ As will be discussed below (cf. §6.3), the paradigm of the Sicilian DIC is defective, in that it is restricted to the cells given in Table 6.2.

174

silvio cruschina

(6) a. Voppigghju u pani. go+fetch.1sg the bread b. Voppigghi u pani. c. Vopigghja u pani. d. Voppigghjamu u pani. e. Voppigghjati u pani. c. Voppigghjanu u pani. (7) a. Occattu u giunnali. go+buy.1sg the newspaper b. Occatti u giunnali. c. Occattunu u giunnali.

(Marina di Ragusa, Sicily) 1sg 2sg 3sg 1pl 2pl 3pl (Acireale, Sicily) 1sg 2sg 3pl

The same range of morphosyntactic variation is found in Puglia and in Salento, where the higher verbal form (V1) in such constructions can be an auxiliary/functional verb but can also be a grammaticalized aspectual marker or prefix (Manzini and Savoia 2005; Ledgeway 2016a; Andriani 2017; Calabrese 2019; see also the discussions in §1.2.2.5, §1.2.3, and §3.5 in this volume). Table 6.3 shows the paradigms for the dialects of Putignano (Bari) and Lecce (from Ledgeway 2016a:168). If we compare the Putignano dialect with that of Lecce, we see that a radical reduction of the inflexional endings has occurred in the dialect of Lecce, although there is still preservation of suppletion in the present (see Calabrese 2020a for an analysis of a similar situation in the neighbouring dialect of Campi), while some more specified inflexional forms are still used in the present in the dialect of Putignano. Table 6.3 The DIC forms for go (V1) + V2 in the dialects of Putignano and Lecce

Present

Past

1sg 2sg 3sg 1pl 2pl 3pl 1sg 2sg 3sg 1pl 2pl 3pl

Putignano go + do vok a fˈfattsu vε fˈfaʃә vε fˈfaʃә ʃa faˈʃeimә ʃa faˈʃeitә von a fˈfaʃәnә ʃε ffaˈʃevә ʃε ffaˈʃivә ʃε ffaˈʃevә ʃε ffaˈʃemmә ʃε ffaˈʃivәvә ʃε ffaˈʃevәnә

Lecce go + lose va pˈpεrdu va pˈpεrdi va pˈpεrde ʃa ppεrˈdimu ʃa ppεrˈditi va pˈpεrdεnu ʃa ppεrˈdia ʃa ppεrˈdia ʃa ppεrˈdia ʃa ppεrˈdiamu ʃa ppεrˈdiuvu ʃa ppεrˈdianu

periphrases and irregular paradigms in italo-romance

175

In Pugliese and Salentino dialects, along with the corresponding infinitival construction (8a), double inflection (8b) is also attested with the progressive construction featuring an outcome of Lat. stare (‘stand, stay’) as V1 (Manzini and Savoia 2005; Ledgeway 2016a; Andriani 2017; Lorusso 2019). In this case too, the morphological realization of V1 ranges from functional verb to invariant aspectual marker or prefix. The paradigms in Table 6.4 (from Ledgeway 2016a:168) show the varying degrees of reduction of the inflexional endings in different dialects: in the present, more richly specified inflexional forms are used in the dialects of Putignano and Martina Franca, whereas an invariant form is used in the dialect of Lecce; an invariant form, possibly akin to a prefix, is also used in the past for all three dialects. Table 6.4 stare ‘stand’ paradigms Present

Past

(8)

1sg 2sg 3sg 1pl 2pl 3pl 1sg 2sg 3sg 1pl 2pl 3pl

Putignano do stok a fˈfattsә ste fˈfaʃә ste fˈfaʃә sta ffaˈʃeimә sta ffaˈʃeitә ston a fˈfaʃәnә sta ffaˈʃevә sta ffaˈʃivә sta ffaˈʃevә sta ffaˈʃemmә sta ffaˈʃivәvә sta ffaˈʃevәnә

a. stek a fε u stand.1sg a do.inf the ‘I am making bread’ b. stek a fatsә u stand.1sg a do.1sg the ‘I am making bread.’

Martina Franca call stɔ cˈcεmә stε cˈcәәmә stε cˈcεmә stε ccaˈmεmә stε ccaˈmεtә stɔnәә (a) cˈcamәәnә stε ccaˈmεvә stε ccaˈmәәvә stε ccaˈmεvә stε ccaˈmammә stε ccaˈmavәvә stε ccaˈmavәnә

Lecce lose sta pˈpεrdu sta pˈpεrdi sta pˈpεrdε sta ppεrˈdimu sta ppεrˈditi sta pˈpεrdεnu sta ppεrˈdia sta ppεrˈdia sta ppεrˈdia sta ppεrˈdiamu sta ppεrˈdiuvu sta ppεrˈdianu

pεn. bread (Conversano, Bari) pεn. bread. (Lorusso 2019:204)

Despite the inflexional peculiarities, the structural differences between the DICs and their equivalent infinitival constructions are not so obvious. Cruschina and Calabrese (2021) argue that both types of construction are restructuring configurations where V1 is a functional verb. Semantically, the bleaching of the original lexical meaning is evident in that V1 provides specific aspectual information about the event, under a single-event interpretation. The motion verbs go and come encode andative and venitive aspect, respectively (Cruschina 2013; Ledgeway 2016a), while the function of stare is to mark progressive aspect (but cf. §6.3

176

silvio cruschina

for a discussion of the desemanticization of these constructions). The DIC configurations differ notably from the infinitival counterparts in that they present double inflexion. Double inflexion, however, arises independently of restructuring and can be analysed as a case of agreement within the extended vP which takes place via concord, through a mechanism resembling adjectival agreement (Baker 2010). In Cruschina and Calabrese’s (2021) account, the infinitive is the morphological realization of a [–pronominal] agreement node lacking phi-features, while the DIC arises from an identical structure in which the agreement node is assigned the feature [+pronominal], thus agreeing with V1 in person and number. In other words, the different inflectional properties of V2 in the DIC and in the infinitival construction are not to be attributed to structural differences, but rather to the phi-features encoded in the relevant agreement nodes. Similarly, the different morphological status of V1 as inflected verb, invariable element, or affix depends on the different morphosyntactic realizations of the functional head hosting V1. Crucially, a strong correlation emerges between the inflexional properties of the construction and the availability of morphologically (and phonologically) reduced forms for V1: the reduced forms are only attested in combination with an inflected V2, leading to the generalization that being part of a DIC must be a precondition for the reduction of V1. As just mentioned, the variation in relation to V1, in turn, is the result of different morphosyntactic realizations of the same functional head in the syntactic structure (Cruschina and Calabrese 2021). Depending on the grammatical status of V1, the DIC can be treated as a construction involving two independent verbal forms, but also as a synthetic structure when V1 has undergone inflexional attrition to become an aspectual prefix. At the level of syntax, however, in both cases V1 is a functional element that selects a reduced vP constituent and which is syntactically and semantically integrated in the extended projection of the adjacent lexical verb (V2). If the non-synthetic verbal forms correspond to the realizations of different verbal extended functional projections, the synthetic forms are the result of the cyclic application of head movement that is able to convert the extended functional projections of a verb into a complex head, that is, a single word involving a root and an affix.⁵ The invariant form alternating with the fully inflected variants provides evidence for the decategorialization of V1, as well as for phonological erosion and a tendency towards cliticization or affixation. The grammatical status of V1, therefore, functions as a benchmark for the various stages of the process of grammaticalization that goes from non-synthetic to synthetic, corroborating the observation that ‘the more a periphrastic construction is grammaticalized, the

⁵ See Svenonius (2012; 2016) for the alternative mechanism of ‘spanning’ that could give rise to a similar realization of a sequence of functional heads.

periphrases and irregular paradigms in italo-romance

177

more it can be claimed to have morphological status’ (Cruschina 2013:264). As we shall see in the next section, the (more advanced) morphological status of the periphrasis is also reflected at the level of the paradigm, where double inflexion is a necessary condition for the emergence of a specific morphomic pattern, the N-pattern.

6.3 The N-pattern as a periphrastic morphome In many Sicilian, Pugliese, and Salentino dialects, the DICs discussed in the previous section display irregular paradigms similar to those that have been widely attested at the morphological levels of stem and inflexion. These paradigms follow morphomic patterns. In Sicilian, for example, the motion DIC exhibits a defective paradigm with a distribution that closely resembles the ‘N-pattern’ described in Maiden (2004a; 2005; 2018a): it is only available with all three persons singular and the third-person plural of the present indicative and with the second-person singular of the imperative. This pattern is found in several Sicilian dialects, including Mussomelese, as shown in Table 6.5 (Cruschina 2013; Cruschina and Calabrese 2021), Marsalese (Cardinaletti and Giusti 2001; 2003), and many other dialects (Di Caro 2019a). Table 6.5 The DIC paradigm of (j)ìri ‘go’ + pigliari ‘take’ in Mussomeli, Sicily 1sg 2sg 3sg 1pl 2pl 3pl

Present (indicative) ˈvajʊ a pˈpɪjjʊ ˈva a pˈpɪjjɪ ˈva a pˈpijja **ˈjamʊ a ppijˈjamʊ **ˈjɪtɪ a ppijˈjatɪ ˈvannʊ a pˈpijjanʊ

Imperative va ˈpijja

**ˈjɪtɪ (a p)pijˈjatɪ

The same pattern is found in Apulia. In the provinces of Bari and Brindisi, the motion DIC shows a full paradigm in most dialects (for example, Alberobello and Cisternino), but a defective paradigm is also attested within the same area, as in the dialect of Conversano (Manzini and Savoia 2005; Andriani 2017:220; Lorusso 2019). These full and defective paradigms are compared in Table 6.6. Crucially, the distribution of the DIC in the dialect of Conversano is identical to that found in the Sicilian dialects, where the DIC is sensitive to the N-pattern and is not available with the first- and second-persons plural of the present indicative. In the place of the defective cells, the DIC is replaced by the infinitival construction

178

silvio cruschina Table 6.6 Full and defective paradigms in some Pugliese dialects

1sg 2sg 3sg 1pl 2pl 3pl

Full paradigm Alberobello go + fetch vókә ppìgghiә vè ppìgghiә vè ppìgghiә scì ppәgghiéimә scì ppәgghiéitә vàunә ppìgghiәnә

Cisternino go + play vò ssòne vè ssùәnә vè ssònә scì ssunémә scì ssunétә vònә ssònәnә

Defective paradigm Conversano go + play vek a ssónә ve a ssùnә ve a ssùnә scémˈa ssәné scétˈa ssәné vànnˈa ssònәnә

(cf. the shaded cells in Table 6.6). Andriani (2017:§2.1) reports that the N-pattern is also observed in other central Apulo-Barese dialects, as well as in an isolated case in the Garganic variety of San Marco in Lamis where the infinitival construction otherwise prevails. The same distributional N-pattern characterizes the progressive DIC in Pugliese dialects, as shown in Table 6.7, again taken from the dialect of Conversano (Lorusso 2019:209). Table 6.7 Progressive with maˈnʤε ‘eat’ in the dialect of Conversano 1sg 2sg 3sg 1pl 1pl 3pl

stare + eat stek a manʤә ste a manʤә ste a manʤә stεm a **manʤεmә stεt a **manʤεtә stan a manʤәnә

Both in Sicily and Apulia, other irregular distributions within the paradigm are possible, some of which resemble other morphomic patterns (Andriani 2017; Di Caro 2019a). Consider, for example, the distribution of the progressive DIC in the dialects of Martina Franca in Table 6.4, which corresponds to the Upattern that is widespread in Italo-Romance (see Maiden 2011a; 2016b). These alternative organizations of the paradigm can be viewed either as intermediate stages before the stabilization of the N-pattern or as extensions of the DIC to other non-morphomic cells of the paradigm. Indeed, the process that led to the establishment of the periphrastic morphome appears to have been different in different dialect groups. Cruschina (2013) observes that, historically, more cells of the DIC paradigm were implicated and that eastern Sicilian varieties have fuller paradigms

periphrases and irregular paradigms in italo-romance

179

(see also Manzini and Savoia 2005:696; Di Caro 2019a; 2019b). In contrast, Andriani (2017) adduces robust evidence in favour of the hypothesis that, in Pugliese dialects, the DIC derives from the corresponding infinitival constructions and that the expansion of the inflected V2 forms started with the second- and third-persons singular of the present and then spread through the paradigm in compliance with the N-pattern (see also Ledgeway (in press a) for a comparison with Calabrian dialects). Let us here concentrate on the N-pattern distribution of the periphrasis and on its nature as a periphrastic morphome. Cruschina (2013) analyses the morphomic N-pattern in the Sicilian motion DIC as a case of defectiveness, which is also present in inflexional paradigms (cf. Maiden and O’Neill 2010). Indeed, the periphrastic construction exists only in the N-pattern cells. Other scholars, however, have highlighted the competition between the DIC and the corresponding inflexional periphrasis interpreting the distribution of the DIC in terms of suppletion (see also the discussion in §1.2.3 of this volume): in fact, the cells in which the DIC is not available can be realized with an infinitival V2 (Andriani 2017; Cruschina and Calabrese 2021).⁶ Interestingly, suppletion can be seen as the trigger of the periphrastic N-pattern and survives at the level of V1 in many dialects where the DIC has a full paradigm. The N-pattern cells feature a suppletive root, while the remaining cells (first- and second-persons plural in the present indicative, second-person singular in the imperative, and the rest of the paradigm) display a different root. The suppletive alternation is evident not only at the level of V1, when the N-pattern distribution features inflected forms for V1 (cf. the paradigms of Alberobello and Cisternino in Table 6.6), but also at the level of complex forms when V1 has been reduced to an invariant form and displays no agreement morphology, behaving as a prefix, as shown for the dialects in Tables 6.8–6.9 (from Andriani 2017:228).⁷ In the latter case, double inflection is now lost, but it is entirely reasonable to assume that the DIC was a precondition for this stage. Both when emerging at the level of the complex forms and when limited to V1, the sensitivity of the DIC periphrasis to the N-pattern may be due to the independent tendency of go to develop an N-pattern suppletive distribution in these ⁶ Following Cruschina and Calabrese (2021), the special behaviour of the first- and second-persons plural in the N-pattern can be accounted for by assuming that the regular infinitival morphology that appears in these persons is the result of an impoverishment operation (see also Lorusso 2019). A suppletion analysis of the DIC N-pattern for first- and second-persons plural implies a situation of overabundance for the other cells, where two (periphrastic) forms are available to realize the same cells of the paradigm (Thornton 2011; 2019b). As will be argued at the end of the present section, this suppletive interpretation of a relationship between the DIC and the infinitival construction is not problematic when the DIC is used with a special meaning that is not available to the infinitival construction. On the relationship between suppletion and defectiveness, see Plank and Vincent (2019:§3.2). ⁷ Despite the difficulty of drawing boundaries between invariable clitics and affixes, I assume that in these paradigms the andative morpheme has been reanalysed as a prefix. See Cruschina and Calabrese (2021) for evidence and discussion.

180

silvio cruschina Table 6.8 Present motion DIC in Salentino varieties Present 1sg 2sg 3sg 1pl 2pl 3pl

Mesagne do va fˈfattsu va fˈfaʧi va fˈfaʧi sa/ʃa ffaˈʧimu sa/ʃa ffaˈʧiti va fˈfannu

Lecce lose va pˈpεrdu va pˈpεrdi va pˈpεrde ʃa pperˈdimu ʃa pperˈditi va pˈpεrdenu

Nardò do va fˈfattsu va fˈfaʧi va fˈfaʧe ʃa ffaˈʧimu ʃa ffaˈʧiti va fˈfannu

Table 6.9 Past motion DIC in Salentino varieties Past 1sg 2sg 3sg 1pl 2pl 3pl

Mesagne do sa/ʃa faˈʧia sa/ʃa faˈʧivi sa/ʃa faˈʧia sa/ʃa faˈʧiumu sa/ʃa faˈʧiuvu sa/ʃa faˈʧiunu

Lecce lose ʃa pperˈdia ʃa pperˈdia ʃa pperˈdia ʃa pperˈdiamu ʃa pperˈdiuvu ʃa pperˈdianu

Nardò do ʃa ffaˈʧia ʃa ffaˈʧii ʃa ffaˈʧia ʃa ffaˈʧiamu ʃa ffaˈʧii ʃa ffaˈʧianu

dialects. The same morphomic distribution then survives in the periphrastic configurations, irrespective of the grammatical status of V1. The suppletive nature of the verb go, however, does not explain the defective property of the periphrasis in Sicilian and Apulian dialects, which builds on verb forms in v- (from Lat. uadere ‘go, walk’) with the loss of the cells corresponding to the forms derived from Lat. ire ‘go’.⁸ Crucially, moreover, the same morphomic model has spread to other, non-suppletive motion verbs such as passari ‘pass’ (cf. Table 6.10) and mannari ‘send’, which can enter the DIC as V1 (Cardinaletti and Giusti 2001:381; Cruschina 2013; Cruschina and Calabrese 2021). Table 6.10 Paradigm of passari ‘pass, come by’ + pigliari ‘take’ in Mussomeli, Sicily 1sg 2sg 3sg 1pl 2pl 3pl

Present (indicative) ˈpassʊ a pˈpɪjjʊ ˈpassɪ a pˈpɪjjɪ ˈpassa a pˈpijja **pasˈsamʊ a ppijˈjamʊ **pasˈsatɪ a ppijˈjatɪ ˈpassanʊ a pˈpijjanʊ

Imperative

ˈpassa a pˈpijja **pasˈsatɪ a ppijˈjatɪ

⁸ The Sicilian equivalent of come (viniri) also shows the same distributional type of root alternation and might have contributed to the spread of the N-pattern (Cardinaletti and Giusti 2001; Cruschina 2013; Cruschina and Calabrese 2021).

periphrases and irregular paradigms in italo-romance

181

Even if it is possible to identify the original trigger of the periphrastic Npattern in the suppletive nature of the verb go, these paradigmatic configurations can be considered fully fledged ‘morphomic’ distributions, in the sense that they involve patterns of form distribution within periphrastic paradigms that lack synchronic functional or phonological determinants. In particular, defectiveness and suppletive allomorphy must be seen as the concomitant result of independent changes, whereby the paradigm of the construction under consideration was ‘attracted’ to a recurrent pattern of morphological irregularity, namely, the N-pattern. It is reasonable to identify these independent changes as the effects of grammaticalization, which have (gradually) aligned the periphrastic construction with a more morphological status comprising a single lexical verb. Let us now return to the possibility of treating the relationship between the DIC and the corresponding infinitival construction in terms of suppletion, inasmuch as the periphrastic forms with an infinitival V2 are able to fill the gaps of the DIC defective paradigm. Interestingly, this substitution is only possible when the two constructions are semantically equivalent, but not when the DIC exhibits full desemanticization. Cardinaletti and Giusti (2001; 2003) argue that the motion V1 in the DIC has lost its full lexical properties (e.g., the ability to select for arguments) but, semantically, it still contributes a motion meaning to the construction, which thus entails movement and physical displacement. This evidence has led to the hypothesis that V1 is a ‘semi-lexical verb’ (Cardinaletti and Giusti 2001; 2003) or that it has been grammaticalized as a functional verb encoding andative or venitive aspect (Cruschina 2013). Cruschina (2013), however, shows that in Sicilian the DIC can be used to express surprise and unexpectedness with respect to a past event, as exemplified in (9)–(11):

(9)

(10)

Arrivammu dda, nn’u ristoranti, e mi vannu arrive.pfv.1pl there in.the restaurant and me= go.prs.3pl a dunanu na pizza accussì ladia! (Mus.) a give.prs.3pl a pizza so ugly ‘We arrived there, at the restaurant, and they gave me such a bad pizza!’ (Cruschina 2013:279) Cuannu u vitti ca sunava nna banna, when him= see.pfv.1sg that play.pst.ipfv.3sg in.the band vaju a pruvu na gioia! (Mus.) go.prs.1sg a feel.prs.1sg a joy ‘When I saw him play in the band, I felt such a joy!’ (Cruschina 2013:279)

182

silvio cruschina

(11)

Nni ddri momentu va a scattìa in that moment go.prs.3sg a break.out.prs.3sg nu temporali. (Mus.) a storm ‘At that moment a storm broke out.’ (Cruschina and Bianchi 2021)

This use of the DIC, labelled the mirative DIC in Cruschina and Bianchi (2021), does not necessarily involve physical displacement and is limited to the verb go as V1. The motion DIC and the mirative DIC differ in a number of respects. First, because of its movement meaning, the motion DIC imposes specific requirements on the subject and the types of predicate that can enter into the construction: only intentional agents are allowed, while inanimate subjects or predicates that express an emotion or a feeling are not possible. Since the motion import is lost with the mirative DIC, such requirements do not hold: this is particularly evident in (10) which features an emotive predicate and, more clearly, in (11) with a weather predicate. Second, even if it is morphologically present, as already mentioned, the mirative DIC is used within a narrative context to foreground an unexpected or surprising event in the past. In the examples (9)–(11), we can indeed see that all other verbs are in the past tense (see the discussions in §4.2.1 and §4.3 in this volume). The andative DIC, by contrast, can only be used with a present reference time (or with the imminent-future value that is typical of the present tense). As for the paradigm, the mirative DIC can be used with the same N-pattern distribution as the motion DIC. The only difference consists in the fact that the mirative meaning is not available in the imperative, but we can reasonably assume that this restriction is a direct consequence of its semantics. In its mirative usage, therefore, the DIC shows a defective paradigm. Crucially, however, no suppletive (periphrastic) forms are possible in the cells where the DIC does not exist (firstand second-persons plural): the mirative meaning is not available with an infinitival V2, namely, with the infinitival construction that is in competition with the motion DIC. This evidence shows that in the dialects and in the contexts where the DIC has undergone semantic bleaching, the degree of grammaticalization is higher and the robustness and functional specialization of the N-pattern are stronger. Processes of semantic bleaching have also affected the motion and progressive DICs in Pugliese and Salentino dialects, especially in combination with the morphological reduction of V1 to an invariant form. In this respect, Rohlfs (1969:§33) points out that the meaning of progressive V1 ‘has been so weakened that it can now be stereotypically placed in front of almost each verb form’, and similarly, the motion V1 ‘has almost become a normal verbal expression’ (Loporcaro 1997a:134; Ledgeway 2016a:165).

periphrases and irregular paradigms in italo-romance

183

6.4 The Sicilian modal periphrasis In this section an additional case of morphomic attraction of a periphrastic paradigm is presented. The periphrasis in question is the Sicilian aviri a ‘have to’ + infinitive’ construction (abbreviated as AICo, following Di Caro 2019c). As shown in the examples in (12), AICo is used to express deontic modality, which can in turn give rise to epistemic meaning extensions, as in (13).⁹ (12) a. Peppi mi dissi ca avia a partiri pi Peppi me= say.pfv.3sg that have.pst.ipfv.3sg to leave.inf for Palermu. (Mus.) Palermo ‘Peppe told me that he had to leave for Palermo.’ b. Oji amu a fari u pani. (Mus.) today have.prs.1pl to do.inf the bread ‘Today we have to make bread.’ c. Idda s’ avissi a pigghiari setti jorna di she refl= have.pst.sbjv.3sg to take.inf seven days of ferii. (Mus.) holidays ‘She should take seven days off.’ (13) a. A st’ ura ati a aviri fami. (Mus.) at this hour have.prs.2pl to have.inf hunger ‘By now you must be hungry.’ b. Talè! U vitru ruttu c’è. Appi a essiri u look the glass broken there=is have.pfv.3sg to be.inf the ventu. (Mus.) wind ‘Look! The glass is broken! It must have been the wind.’ c. Dintra un ci sunnu! Si nn’ appiru inside not there= be.prs.3pl refl= from.there= have.pfv.3pl a jiri ’n campagna. (Mus.) to go.inf in countryside ‘They are not at home! They must have gone to the countryside.’

⁹ A future temporal value is often also attributed to this periphrasis. This is probably a secondary reading that arises from the deontic meaning, especially with the imminent-future value typical of the present tense in combination with an obligation or necessity interpretation, as in (13a). See Bentley (1997), Amenta and Strudsholm (2002), Amenta (2006; 2010), Amenta and Paesano (2010), and Brucale and Mocciaro (2019) for a discussion of the interpretations of AICo in old and contemporary Sicilian.

184

silvio cruschina

Di Caro (2019c:221f.) stresses the monoclausal and grammaticalized nature of this periphrastic construction, also reflected by the possibility of using reduced forms in the present tense. In Table 6.11, the possible forms of aviri ‘have’ in the present indicative are shown, distinguishing between its lexical, perfective auxiliary, and AICo uses. Table 6.11 Present forms of aviri ‘have’ in Sicilian (Mussomeli) Present 1sg 2sg 3sg 1pl 2pl 3pl

Lexical aju a(i) avi avemu aviti annu

Auxiliary aju a(i) a amu atu annu

AICo aju, a/e a avi amu ati annu

Reduced forms of AICo are exemplified in (12b) for the first-person plural and in (13a) for the second-person plural. Examples of the reduced first-person singular are given in (14),1⁰ where it appears to have been reduced to a prefix.11 Sicilian (14) a. Êffari/âffari la spisa. have.to=do.inf the shopping ‘I/you have to go to the shops.’ (Di Caro 2019c:222) b. M’âccattari i càvusi novi. rfl=have.to=buy.inf the trousers new ‘I have to buy some new trousers.’ (Amenta and Paesano 2010:13)

1⁰ The example in (14a) is from the dialect of Delia. The reduced forms reported by Di Caro (2019c) for this dialect are very similar to those found in the dialect of Mussomeli listed in Table 6.11. As for the reduced forms of the first-person singular, Di Caro (2019c) points out that, although in principle both a and e are possible, e is largely preferred for the first-person singular, while a is more typically employed to refer to the second-person singular. The same holds true for the dialect of Mussomeli. The only difference between the reduced paradigms of AICo in the two dialects concerns the secondperson plural, which is ati in Mussomeli but atu in Delia. Interestingly, atu is also possible in the variety of Mussomeli but as a specialized reduced form for the perfective auxiliary function (cf. Table 6.11). Compare atu/**ati parlatu assà ‘you (2pl) have talked too much’ vs ati/**atu a jiri dintra ‘you have to go home’. In fast speech, however, this distinction is often lost in the AICo because of the of the elision of final vowel before the element a: ati a > at’a. 11 Note that the circumflex on the reduced forms in the examples in (14) is used as a device to represent in orthography the long vowel resulting from the contraction between the vocalic V1 and the element a, which, despite not being perceptible, leaves its audible trace in the initial consonantal doubling of the following lexical verb (so-called raddoppiamento fonosintattico, see Loporcaro 1997b). This phenomenon is also orthographically represented in Table 6.14, where the initial (semi-)consonant of the lexical verb jiri ‘go’ is lengthened (with concomitant velarization) and realized as a geminate voiced velar plosive: gghiri.

periphrases and irregular paradigms in italo-romance

185

What is more interesting, from a paradigmatic viewpoint, is the root alternation found in the preterite, where no morphological variation according to grammatical status is observed. See Table 6.12.

Table 6.12 Preterite forms of aviri ‘have’ in Sicilian 1sg 2sg 3sg 1pl 2pl 3pl

àppi avìsti àppi àppimu avìstivu àppiru

In the preterite paradigm, we observe the root alternation that is characteristic of Romance irregular verbs that display a distinct root for the preterite (Maiden 2000; 2001a, 2001b; 2004a; 2018a, 2018b; see also Calabrese 2012; 2013; 2015). Some cells of the paradigm are built on the arrhizotonic verb forms in av-, which is found for example in the past imperfective (i.e., Mus. aviva, avivi(tu), aviva, avivamu, avivavu, avivanu), in contrast to the other cells that are derived from the rhizotonic preterite root app-. This type of alternation is very common in Romance, but what is striking about the paradigm in Table 6.12 is the rhizotonic form for the first-person plural. In Italian and most Italo-Romance varieties, the first-person plural builds on the arrhizotonic root, hence participating in a paradigmatic distribution known as the E-pattern (see Maiden 1995; 2000; 2001a; 2001b; 2004a; 2011a; 2018a; 2018b).12 The different distribution of the preterite root allomorphy in Sicilian shows coherence throughout the verbal morphology of the dialect, at least for those verbs that display this irregularity in the preterite paradigm, thus functioning as a consistent morphomic pattern. This is evident if we compare the preterite forms of the Sicilian and Italian forms in Table 6.13, where word stress and open/close vowel quality are indicated respectively with grave/acute accents (not used in standard Italian orthography).13

12 Similar rhizotonic forms for the first-person plural were common in old Italian: for example fécimo ‘we did’, èbbimo ‘we had’; the corresponding modern forms are given in Table 6.13 (Maiden 2000; 2018b). 13 The forms given in Table 6.13 are from the dialect of Mussomeli. The arrhizotonic root von- in the preterite of the verb vulìri ‘want’ is typical of this and some neighbouring dialects. In other varieties the corresponding irregular root is vo- as in 1sg/3sg vòsi. The -n- of von- is the outcome of a general sound change that characterizes central Sicilian, where the lateral approximant /l/ becomes /n/ when followed by a consonant: volsi > vonsi (cf. OIt. volsi) (cf. also antu (< altu(m)) ‘high, tall’ and antru (< alteru(m)) ‘other’).

186

silvio cruschina

Table 6.13 Preterite paradigms in Sicilian and Italian

1sg 2sg 3sg 1pl 2pl 3pl

Sicilian Italian have àppi èbbi avìsti avésti àppi èbbe àppimu avémmo avìstivu avéste àppiru èbbero

Sicilian Italian Sicilian Italian want do vònsi vòlli fìci féci vulìsti volésti facìsti facésti vònsi vòlle fìci féce vònsimu volémmo fìcimu facémmo vulistìvu volèste facistivu facéste vònsiru vòllero fìciru fécero

Sicilian Italian know sàppi sèppi sapìsti sapésti sàppi sèppe sàppimu sapémmo sapìstivu sapéste sàppiru sèppero

This pattern, which only differs from the more general Italo-Romance E-pattern with respect to the first-person plural, is labelled W-pattern in Di Caro (2019a; 2019b; 2019c). The Sicilian W-pattern in the preterite illustrated in Table 6.14 is so pervasive in this language that it acts as an attractive force and an abstract distributional model to novel types of morphological variation. This is what happens in the preterite of AICo, where a defective paradigm follows the W-pattern distribution (Di Caro 2019c:227). Table 6.14 Defective AICo preterite paradigm + go 1sg 2sg 3sg 1pl 2pl 3pl

àppi a gghiri **avìsti a gghiri àppi a gghiri àppimu a gghiri **avìstivu a gghiri àppiru a gghiri

As is typical of morphomic patterns, it is hard to find a semantic or phonological reason for the unavailability of the AICo periphrasis in the second persons of the preterite. A distinctive common feature of these cells of the paradigm is that they involve the addressee; however, this property does not provide a satisfactory explanation for their unavailability. The geographical distribution of this pattern, in both diachrony and synchrony, is not entirely clear, but Di Caro (2019c) reports that today it is found in a large number of dialects, in different Sicilian provinces. Interestingly, no clear alternative or suppletive forms are available to replace the second persons in AICo. In this respect, Di Caro (2019c) points out that the speakers he consulted had to resort to a range of alternative constructions.

periphrases and irregular paradigms in italo-romance

187

It is also important to note that this periphrastic morphome is limited to the preterite and does not affect the other tenses and moods of the paradigm, which use the same regular root, such as the past imperfect indicative (i.e., aviva/avivi(tu)/aviva/avivamu/avivavu/avivanu a gghiri) and the past imperfect subjective (i.e., avissi, avissi(tu), avissi, avissimu, avissivu, avissiru a gghiri).1⁴ In this respect, the W-pattern distribution in the periphrastic construction under consideration is less generalized than the N-pattern periphrastic morphome discussed in §6.3.

6.5 Conclusions The data examined in this chapter provide robust evidence in support of the view that periphrastic morphomes emerge in highly grammaticalized periphrases, that is, in multi-word expressions that have arguably reached a more advanced stage of morphologization. Morphologization, as a core manifestation of grammaticalization, should therefore be included as a key factor in the development and characterization of periphrastic constructions, especially if we want to understand the varying morphological properties within individual periphrases. In turn, the presence of properties that are typical of inflexional paradigms, such as a morphomic distribution of irregular forms, must be interpreted as an index of the growing morphologization of the periphrasis, shifting the balance between syntactic and morphological features. In all the periphrastic constructions discussed in this chapter, the first verb V1 involved in the periphrasis exhibits varying degrees of inflexional attrition and, hence, distinct morphological realizations, ranging from a fully inflected paradigm to a set of specified forms with reduced agreement, to the total loss of inflexion and reduction to an invariable marker or prefix. These patterns of variation found in southern Italian dialects can be analysed as the reflexes of different diachronic stages in the morphologization of the periphrasis. This type of dialectal variation is certainly relevant to understanding the relationship between grammaticalization and the distribution of irregular forms in periphrastic paradigms. It is, however, important to note that grammaticalization is not to be seen as directly responsible for the observed patterns. In fact, grammaticalization, and in particular the morphologization of the periphrasis as a whole, is a necessary condition for the emergence of a morphomic pattern, acting as a conclusive factor in determining whether the construction enters the morphological domain and whether therefore it must be ‘treated’ morphologically. Historically, the motion 1⁴ On the optional enclitic -tu in these forms, see Cruschina and Rinollo (2013).

188

silvio cruschina

and progressive DICs, as well as the Sicilian modal periphrasis, displayed a full paradigm, which is still available in some dialects. The diachronic changes that led to a morphomic paradigm should then be interpreted as the results and effects of the kind of diachronic ‘attraction’ to a recurrent pattern of irregularity that is typical of morphomes and which can evidently also shape periphrastic paradigms. The coherence of the paradigmatic distribution and the attraction from the inflexional to the periphrastic paradigms thus identify clear diachronic diagnostics for the morphomicity of these patterns. ‘Morphologization reduces the analytic construction to a synthetic one’ (Lehmann 2015:16), and speakers apply an abstract morphological distributional template that is robustly present in the domain of inflexion to the realm of morphologized periphrases. In conclusion, the theoretical construct of the morphome, which originated in the study of inflexional morphology, also has great relevance in the domain of periphrasis. Periphrastic morphomes originate as the (accidental) by-product of change but are subsequently acquired, replicated, and preserved in diachrony in a way that suggests that they represent abstract generalizations or templates that are cognitively available to the language users. In the derivational model followed by Cruschina and Calabrese (2021), the distinction between syntax and morphology has to do with the different realizations of the same or similar syntactic structure, further to the application of specific morphosyntactic operations. This view implies a different notion of paradigm, which cannot to be seen as a structured (memorized) set of fully inflected words, but rather as a set of morphologically related words that include a derived stem, which is the same in case of a regular paradigm, and different combinatorial realizations of the same set of inflexional features (Calabrese 2008). The forms realizing the cells of a paradigm, therefore, are derived by establishing the feature sets of the terminal nodes of the morphosyntactic structure and by inserting the corresponding vocabulary items in these terminal nodes. In this approach, the morphomic patterns correspond to abstract templates that specify which feature combinations (i.e., which cells of the paradigm) are subject to the morphological constraints or repair operations that give rise to irregular forms. If we consider the present indicative of the N-pattern, for example, the corresponding abstract template specifies that the cells characterized by the feature combinations that give rise to first and second-person plural (i.e., +participant, +plural) are subject to the specific repair or impoverishment operations that, at the moment of morphological spell-out, yield defectiveness or suppletive forms. As observed by Calabrese (2020b:209), ‘these repairs can insert “ornamental” pieces—structures that are not motivated

periphrases and irregular paradigms in italo-romance

189

syntactically or semantically but only morphologically—to mediate the interface between abstract syntactico-semantic structures and surface PF construction’. Even if from a different perspective, this approach arrives at the same conclusion as Maiden that morphomic phenomena are only morphological in nature and lack any phonological, syntactic, or semantic motivation.

PART III

AUXILIATION

7 Auxiliary selection in Italo-Romance and inflexional classes Xavier Bach and Pavel Štichauer

7.1 Introduction The concept of morphome is for the most part discussed in connection with synthetic morphology, although, as, for example, Maiden (2018a:219, 302) observes, morphomic configurations (i.e., distributional patterns which lack phonological, syntactic, or semantic motivation and are internal to morphology) are not confined to synthetic structures but extend to other phenomena. In the chapter devoted to ‘the substance of morphomic structures’ (2018a:302–306), he gives a putatively full typology of such phenomena, which include, for instance, augments of lexical roots, allomorphs of inflexional endings, defectiveness, but also periphrases and conjugation-class markers. While various types of periphrastic structures are dealt with by other authors in this volume (cf., e.g., Chapters 3– 6 and 9), in this chapter, we address an issue which seems to be missing from Maiden’s account, although it represents an important area of morphology by itself (cf., e.g., Aronoff 1994; Stump 2001; 2016), namely the issue of inflexional classes or inflexion-class membership in periphrasis.1 In particular, we will argue that auxiliary selection in Romance could be considered a matter of inflexion-class distinction. Under such a view, a periphrastic conjugation that involves two auxiliary verbs gives rise to two classes of lexemes which are in many respects similar to canonical inflexion classes, where two or more classes are distinguished on the basis of a different set of inflexional endings. Moreover, in rich systems of inflexional classes, the phenomenon of heteroclisis 1 This chapter is intended as a personal tribute to Martin Maiden to mark his important anniversary. We have both been immensely influenced by his outstanding scholarship and we are proud to be able to elaborate on some of his ground-breaking ideas on the nature of morphology. We are indebted to the editors for their extensive and very helpful feedback on the first draft of this chapter. Part of this work was realized with the support of the European Regional Development FundProject Creativity and Adaptability as Conditions of the Success of Europe in an Interrelated World (No. CZ.02.1.01/0.0/0.0/16_019/0000734).

Xavier Bach and Pavel Štichauer, Auxiliary selection in Italo-Romance and inflexional classes In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Xavier Bach and Pavel Štichauer (2022). DOI: 10.1093/oso/9780198870807.003.0008

194

xavier bach and pavel štichauer

frequently occurs, whereby for a set of lexemes some of the forms are inflected according to one class while the other forms inflect following another class. We shall demonstrate that this phenomenon can also be found in various perfective auxiliation systems in Italo-Romance, in particular in reflexives. This chapter is organized as follows. In §7.2, we lay out the basic theoretical foundations for our approach, which stems from Stump’s (2001; 2016) realizational model of inflexional morphology but extends some of his notions in order to include periphrastic structures. In §7.3, we present an overview of mixed perfective auxiliation systems, i.e., those cases where the two auxiliary verbs, have and be (henceforth h and b), are not used across two different classes of verbs as in standard French or Italian, but within one and the same tense, aspect, and mood paradigm. Such mixed systems can be found in a wide array of Italo-Romance varieties. In §7.4, we narrow down our analysis to those cases where only one coherent class of verbs, such as reflexives, follow this type of mixed strategy, while the other classes (transitives, unergatives, and unaccusatives) exhibit a standard selection of one or the other auxiliary throughout the whole paradigm. We present a wide range of data, drawn not only from Manzini and Savoia (2005:II), where, unfortunately, incomplete paradigms are often given, but also from other sources which report the complete set of forms of interest here (e.g., Paciaroni 2009; Andriani 2017).2 In §7.5, we conclude that the situation we describe is interesting from a theoretical point of view, as heteroclisis is frequently found in inflexion-class systems but is rarely remarked on in periphrastic systems (it is merely alluded to in, e.g., Kaye 2015:22f.; cf. also the discussion in §1.2.3 of this volume). The fact that such systems present heteroclisis and also default marking is a good argument in favour of considering auxiliary alternations a matter of inflexion class. We also discuss some theoretical problems and put forward diachronic considerations which pave the way for future investigation.

7.2 Auxiliary selection and inflexional classes In this section, we lay out the basic assumptions of our approach. We start with a (reduced) example of noun inflexion which will provide us with the basic notions of content paradigm, form paradigm, and realized paradigm. We then move on to verb inflexion, introducing the notion of segregated inflexional classes on which we will base our approach to auxiliary verb constructions.

2 Currently, we are constructing a database of mixed perfective auxiliation systems, the aim of which is to include all data already attested and described in various sources. Such a database will allow for advanced (statistical) analyses where various correlations can be discovered (e.g., between a given pattern, the region, locality, and/or tense, aspect, and mood). The database is not as yet available online, but we plan to publish it once a full range of data has been put together.

italo-romance & inflexional classes

195

7.2.1 Content paradigm, form paradigm, and realizations It is widely acknowledged that, for instance, Latin nouns exhibit rich inflexional morphology, i.e., they inflect for certain features required in the syntax. In Latin, a noun, such as femina ‘woman’ or seruus ‘slave’, presents forms which realize singular and plural, as well as one of the six case values. Traditionally, such a system can be captured by way of a paradigm with a given number of cells, the number of cells corresponding to all combinations of values of all the features. In the case of Latin nouns, we thus have a paradigm of 12 cells, as in Table 7.1. Table 7.1 An abstract paradigm of a noun inflecting for number and case sg

pl

nom acc gen dat abl voc

Intuitively, we expect all nouns to possess a form corresponding to each of the paradigm cells, no matter what these forms might look like. In a canonical paradigm, we would expect to find maximally transparent or maximally discriminating forms, where each of the forms differs from all the other forms (see, e.g., Baerman, Brown, and Corbett 2017:100; Corbett 2009:3f.; Stump 2016:31– 42; Bach 2019:33–37). Moreover, we might also expect that such a system would hold for all nouns in the language. If this were the situation in Latin, we would not have any declensional classes.3 Conversely, we have declensional (or, in general, inflexional) classes where more than one set of forms is needed for a given set of lexemes in order to realize the full paradigm. We can represent this ideal situation by the two paradigms in Table 7.2, where instead of full forms we only give an idealized set of inflexional endings -a, -b, -c, etc. As is well known, we hardly ever find this type of maximally transparent set of forms with a unique set of inflexional endings. Instead, we find a range of deviations from this ideal, such as syncretism, deponency, allomorphy (for an overview, see, e.g., Baerman 2015a). But the point is that there is a crucial difference between the ‘abstract’ paradigm in Table 7.1, where we only define a complete 3 This is a controversial issue largely dependent on various assumptions. In general it is agreed that at least two classes (i.e., two sets of exponents) are needed to recognize inflexional classes. For a critical discussion, see Bach (2019:37–41).

196

xavier bach and pavel štichauer Table 7.2 Two idealized paradigms for two classes of lexemes Class 1

Class 2

sg

pl

sg

pl

nom

-a

-g

nom

-m

-s

acc

-b

-h

acc

-n

-t

gen

-c

-i

gen

-o

-u

dat

-d

-j

dat

-p

-v

abl

-e

-k

abl

-q

-w

voc

-f

-l

voc

-r

-x

morphosyntactic space, and the ‘concrete’ paradigms where two different sets of exponents holding for two distinct groups of lexemes realize exactly the same ‘abstract’ paradigm. Furthermore, we could envisage a situation where, for instance, the set of exponents in Class 1 exhibits one of the deviations mentioned above (say, syncretism in the genitive and dative in the singular, whereby we would have -a, -b, -c, -c, -d, -e, etc.), and, similarly, the set of exponents in Class 2 shows similar syncretism, but elsewhere in the paradigm, say, in the dative and ablative in the plural, i.e., -s, -t, -u, -v, -v, -w). In such instances, we might introduce an intermediate level of paradigmatic analysis, stating the distributional pattern (with the syncretism) in a more abstract way, as represented in Table 7.3, where the grey shading captures the two patterns of syncretism. Table 7.3 Two idealized paradigms for two classes of lexemes with two different distributional patterns of syncretism (grey shading) Class 1 sg nom acc gen dat abl voc

Class 2 pl

sg

pl

nom acc gen dat abl voc

Following Stump (2016:110–115; see also Baerman, Brown, and Corbett 2017:77f.), we adopt a tripartite model of inflexional paradigms. The abstract paradigm, where only the morphosyntactic features and values are given, is a ‘content paradigm’ (represented in Table 7.1). The two paradigms in Table 7.3, where only the patterns of syncretism (or any other distributional patterns such as, for instance, those famously identified by Maiden 2018a) are defined regardless of the concrete forms and concrete exponents, are ‘form paradigms’. And, finally, the

italo-romance & inflexional classes

197

concrete sets of full inflexional forms are ‘realized paradigms’ (corresponding to Table 7.2).

7.2.2 Segregated inflexional classes and compound tenses Stump (2016:90) also introduces another notion which is important for our purposes. From the examples discussed in §7.2.1, it is clear that class membership is established globally in the sense that, for a given lexeme of, say, Class 1, the full set of inflexional forms can be inferred (whether or not we use one, two, or more diagnostic forms to determine the membership).⁴ However, in some cases the inflexion class only determines a subset of the realizations of the whole paradigm for a given lexeme, and in order to have a complete knowledge of the paradigm one needs to look at the other sub-paradigms (see, for details, Stump 2016:90f.; Bach 2019:44–47). For example, the Latin first conjugation (verbs in -āre) only defines the realizations of the infectum, and in order to know the full range of the verb’s inflexions one needs to look separately at the perfectum sub-paradigm. This kind of inflexional class membership is referred to by Stump as involving segregated inflexional classes, where a given lexeme belongs to different systems of classes for different subparts of its paradigm. Stump demonstrates this notion with reference to imperfective versus perfective conjugation in Latin. It applies equally well to a specific subset of the verb inflexion of Romance, namely to all compound tenses realized by way of an inflexional periphrasis (the combination of the auxiliary have or be and the past participle). Indeed, two verbs both belonging, for instance, to the Italian first conjugation in -are (such as mangiare ‘eat’ and arrivare ‘arrive’), may each select a different auxiliary in compound tenses. Hence, class membership limited to just, say, the present indicative, does not predict the realization of the other tenses for one and the same verb. If we adopt this view, we could propose an analysis along the following lines, with reference to the three types of paradigm defined above. Restricting ourselves to the present perfect, we start out from a content paradigm represented in Table 7.4. If we had only one auxiliary, as in Spanish or a range of Italo-Romance varieties, we would only have one class of lexemes that select just one set of periphrastic realizations (be it have or be). If, instead, we had two auxiliaries, as is the case in standard French and standard Italian, we would end up with two classes of lexemes with two different realizations. We would thus obtain two form paradigms and two

⁴ See, e.g., Blevins (2016:85–90) for the status of such ‘principal parts’ and, in general, for the implicational structure of inflexion classes.

198

xavier bach and pavel štichauer Table 7.4 The content paradigm for the present perfect 1sg 2sg 3sg 1pl 2pl 3pl

Present perfect

concrete realized paradigms, as in Tables 7.5 and 7.6, respectively. We exemplify the distinction from standard Italian. We thus suggest that systems of auxiliary selection can be understood as systems of inflexion classes, albeit segregated systems restricted to periphrastic tenses. This yields an abstract content paradigm defined as the present perfect to which correspond two form paradigms where two different auxiliaries are required throughout the paradigm (and so the form paradigm simply states this distributional unity without determining the concrete realizations). And, finally, we have the realized paradigms when any concrete verb is to be used in the present perfect. In adopting this approach to auxiliary selection as inflexional classes defined over two different periphrastic strategies, we share the views of Bonami (2015:97) and Baerman, Brown, and Corbett (2017:28f.). Although such an approach is problematic for a number of reasons (to which we will turn in §7.5.1), we believe that its usefulness can be demonstrated especially on the basis of mixed perfective auxiliation systems where the two auxiliaries are not distributed over two (more or less) motivated classes of verbs (and used systematically throughout the periphrastic paradigms), but intraparadigmatically, giving rise to a wide range of interesting distributions. In particular, systems

Table 7.5 Two form paradigms for the classes of lexemes that select the auxiliary h(ave) and b(e) Class 1 1sg 2sg 3sg 1pl 2pl 3pl

Present perfect H H H H H H

Class 2 1sg 2sg 3sg 1pl 2pl 3pl

Present perfect B B B B B B

italo-romance & inflexional classes

199

Table 7.6 Two realized paradigms for the verbs mangiare ‘eat’ and arrivare ‘arrive’ Class 1

Class 2

Present perfect 1sg 2sg 3sg 1pl 2pl 3pl

ho mangiato have.1sg eat.ptcp hai mangiato have.2sg eat.ptcp ha mangiato have.3sg eat.ptcp abbiamo mangiato have.1pl eat.ptcp avete mangiato have.2pl eat.ptcp hanno mangiato have.3pl eat.ptcp

Present perfect 1sg 2sg 3sg 1pl 2pl 3pl

sono arrivato/a be.1sg arrive.ptcp.m/f sei arrivato/a be.2sg arrive.ptcp.m/f è arrivato/a be.3sg arrive.ptcp.m/f siamo arrivati/e be.1pl arrive.ptcp.m/fpl siete arrivati/e be.2pl arrive.ptcp.m/fpl sono arrivati/e be.3pl arrive.ptcp.m/fpl

of inflexion classes frequently exhibit heteroclisis (cf. Stump 2006; Kaye 2015), where a given lexeme takes its inflexions from two or more classes. Systems of inflexion classes also frequently involve a system of default inflexion, whereby all or most lexemes show the same realization for a given feature–value pairing (Brown and Hippisley 2012; Gisborne and Hippisley 2017): there are indeed dialects, such as that of Bitetto (cf. Table 7.10), where a given cell always appears with the same auxiliary while there is alternation in other cells between different verbs. We now briefly review the basic facts about such mixed perfective auxiliation systems.

7.3 Mixed perfective auxiliation systems in Italo-Romance In a considerable number of Italo-Romance varieties, we find perfective auxiliation systems that are far from following the standard have/be distinction described above, where transitives and unergatives usually require the selection of have, whereas unaccusatives and reflexives trigger the choice of be. In these varieties, in fact, the two auxiliaries alternate inside one and the same tense, aspect, and mood paradigm according to a wide range of person/number-based patterns.⁵ Such mixed perfective auxiliation systems have been widely studied over the past few decades, becoming, as Loporcaro (see Chapter 8 in this volume), points out, ⁵ A wide array of varieties also follow the one-auxiliary system, where only either have or be is required in all compound tenses. There appear to be more varieties with generalized have than with generalized be (cf., e.g., Manzini and Savoia 2005:II:779–809, 759–778). Moreover, even in the varieties with mixed patterns in the present perfect, there is an overwhelming tendency to generalize one or the other auxiliary in the other compound tenses.

200

xavier bach and pavel štichauer

a hotly debated topic (cf., e.g., Bentley and Eythórsson 2001; Loporcaro 2001; 2007; 2014; 2016; Manzini and Savoia 2005:II:Ch. 5; 2011:Ch. 6; Cennamo 2010; D’Alessandro and Roberts 2010; Legendre 2010; Ledgeway 2012:317–327; 2019; Štichauer 2018; 2019). In what follows, we examine one example showing how the data fit into our class-based account. In order to provide a detailed picture of the inflexional behaviour of such mixed systems, we limit our discussion to just one variety for which we have a complete set of forms. Let us consider the examples in Table 7.7, drawn from Andriani (2017:156–164).⁶ Table 7.7 The present perfect of the verbs mangià ‘eat’ and scì ‘go’ in the variety of Bari (Andriani 2017:158–160) Transitives/Unergatives Present perfect 1sg so mmangiàtә be.1sg eat.ptcp 2sg si mmangiàtә be.2sg eat.ptcp 3sg a mmangiàtә have.3sg eat.ptcp 1pl simә mangiàtә be.1pl eat.ptcp 2pl sitә mangiàtә be.2pl eat.ptcp 3pl annә mangiàtә have.3pl eat.ptcp

1sg 2sg 3sg 1pl 2pl 3pl

Unaccusatives Present perfect so ssciùtә be.1sg go.ptcp si ssciùtә be.2sg go.ptcp a sciùtә have.3sg go.ptcp simә sciùtә be.1pl go.ptcp sitә sciùtә be.2pl go.ptcp annә sciùtә have.3pl go.ptcp

In this variety, we find one of the most widespread mixed patterns, namely the distribution in which be (highlighted in grey) is selected in the first and second persons of both numbers, whereas the third persons are realized with have.⁷ While this is only one of the many patterns that can be found, and variation is immense (cf., e.g., Loporcaro 2007:185f.; 2014:55; and Chapter 8 in this volume), what is interesting here is that it involves all verbs, including transitives, unergatives, and unaccusatives. Indeed, the same pattern extends readily into reflexive verbs (regardless of the concrete subclass, such as transitive direct reflexives, antipassives) which follow exactly the same pattern (cf. Andriani 2017:158f.). ⁶ We maintain Andriani’s simplified IPA-based transcription. Hence, instead of, for instance [ʃʃutә], we adopt ssciùtә where the doubled initial consonant signals rafforzamento fonosintattico ‘phonosyntactic strengthening’, and the grave accent indicates the position of the main stress. ⁷ It is important to note, for the sake of completeness, that the pattern BBH–BBH now holds for younger speakers, while older generations also allow for HBH–HHH across all classes (for details, see Andriani 2017:156–158). We view this diachronic tendency as a potential example of morphomic attraction, whereby the common BBH–BBH pattern represents the most widespread distribution serving as a model for analogical changes.

italo-romance & inflexional classes

201

We could conclude that this represents the same one-auxiliary system as the Spanish type, where we only have one (inflexional) class with one suppletive periphrasis involving both the auxiliary and the past participle. Even if that were the case, however, we would still have a different morphological realization of the paradigm. We thus have the same content paradigm (the present perfect), with one particular form paradigm where the grey shading represents the pattern to follow (BBH–BBH), and finally the realized paradigms given in Table 7.7. However, we find, in the same variety, two further patterns as we move to the other compound tenses. There is the pattern HHH–HHH, i.e., the generalized auxiliary H throughout the paradigm for the counterfactual, and the pattern BBB–BBB≈HHH–HHH with complete free variation (henceforth represented by ≈) of the two auxiliaries in the pluperfect (for the full realized paradigms, cf. Andriani 2017:160–166). As Loporcaro (Chapter 8 in this volume) crucially observes, this different distribution of the exponents (that is, the two auxiliaries) is much more a matter of morphology rather than syntax. Although we could go on to discuss other patterns with a different distribution in inflexional classes, such as those where we find different mixed patterns for different classes (transitives versus unaccusatives), we intend to focus on one particular kind of situation which is also frequently found across a wide range of Italo-Romance varieties (cf. in particular Loporcaro 2014). In these dialects, the mixed perfective auxiliation system is limited to reflexive verbs, as illustrated by the examples in Table 7.8.⁸ In this variety, transitives (wash) select have throughout the periphrastic section of the paradigm, and unaccusatives (come) consistently require be (as in standard Italian or French). However, reflexives (wash oneself) are split between the two auxiliaries and thus give rise to a third class of lexemes (reflexives), where we find yet another intraparadigmatic distribution of the two auxiliaries. We use the grey shading to highlight those cells which go with one or the other class in the realization of the present perfect. This situation is interesting from a purely theoretical point of view, as it represents, we argue, a clear case of heteroclisis. Heteroclisis, as mentioned in §7.1, is the phenomenon in which ‘a lexeme inflects according to one inflectional class for one part of its paradigm but according to a distinct inflectional class for the other part’ (Spencer 2013:162f.). But to better understand this particular kind of heteroclitic behaviour, we need to take a closer look at the full range of data at hand.

⁸ The example is reconstructed following Manzini and Savoia (2005:II:652f.), and checked with native speakers of the dialect. We are indebted to Giuseppina Silvestri for providing these revised data.

202

xavier bach and pavel štichauer

Table 7.8 The present perfect of the verbs wash, wash oneself, and come in the variety of Altomonte (Calabria, southern Italy) ˈ

ˈ

ˈ ˈ

ˈ

ˈ ε

ˈ

ˈ

ε

ˈ

ˈ ˈ

ˈ ˈ

ˈ ˈ ˈ

ˈ

ˈ

7.4 Mixed systems in reflexives and heteroclisis In a series of fundamental contributions, Loporcaro (2007; 2014; 2016) has shown how the integration of reflexive verbs into a model of auxiliary selection crucially changes the overall picture. Indeed, reflexive verbs appear to be an important locus of change from which all mixed systems might have originated. As already mentioned, there is a wide range of varieties with the mixed system limited to only reflexive verbs. Although Loporcaro (2007:200; cf. also Benincà, Parry, and Pescarini 2016) observes that this distribution is frequently found in northern Italian dialects, attestations are scattered throughout Italy (Piedmont, Veneto, Friuli, Calabria, Marche). In fact, as Paciaroni (2009:51f.) notes, the variation is also ‘micro-diachronic’, between two generations, and ‘micro-diatopic’, involving small distances between local varieties. In addition, a further level of complexity comes into play in that not all reflexives display the same pattern, and it is thus important to make a crucial distinction between various types of reflexive verb (cf. Loporcaro 2007; 2014; 2016). Ideally, then, we should discuss the full range of data according to each subclass of reflexive verb,⁹ but given the lack of attestations in various sources, such a discussion

⁹ In our database, the classes cover all the following types (with an Italian example for each): retroherent-reflexives = accorgersi ‘notice’; direct transitive reflexives = lavarsi ‘wash oneself ’; indirect transitive reflexives = lavarsi le mani ‘wash one’s hands’; indirect unergative reflexives =

italo-romance & inflexional classes

203

is destined to be incomplete. In the following sections, we present four major systems that emerge from the available data. First, in §7.4.1, we discuss the mixed systems where the reflexives follow the commonest pattern BBH–BBH. In §7.4.2, we examine examples of heteroclisis with default marking. In §7.4.3, we deal with a frequent system where free variation creates what we will dub ‘heteroclisis by overabundance’. In §7.4.4, we briefly comment on some elusive cases which merit further attention.

7.4.1 Reflexives with the BBH-BBH pattern The common pattern BBH–BBH, widely discussed in the literature as being motivated on pragmatic grounds (opposing discourse participants and nonparticipants, cf., e.g., Štichauer 2018:13; Ledgeway 2019:355), is found just with reflexives in northern Italian varieties such as Felizzano, Castellazzo Bormida, Castelletto Merli, Traves, Velo Veronese/Cazzano di Tramigna (cf. Manzini and Savoia 2005:II:650–652), and in Paduan (cf. Benincà, Parry, and Pescarini 2016:204). In Table 7.9, an example from Castelletto Merli illustrates this distribution.1⁰ (In Pellegrini’s 1977 classification this variety belongs to the basso piemontese dialects of Piedmontese, within the broader grouping of Gallo-Italian.) Table 7.9 Heteroclisis in reflexives following the pattern BBH–BBH in the variety of Castelletto Merli (data from Manzini and Savoia 2005:II:650f.) ˈ

ˈ

ˈ

ˈ

ˈ

ˈ

ˈ

ˈ

ˈ ˈ ˈ

ˈ

ε

ˈ

ˈ

ˈ

ˈ

ˈ

ˈ ˈ

ˈ

ε

ˈ

rispondersi (da solo) ‘answer (to) oneself ’; antipassive reflexives = mangiarsi (un panino) ‘eat (a sandwich)’; middle reflexives = svegliarsi ‘wake up’, addormentarsi ‘fall asleep’. 1⁰ We only have data for direct transitive reflexives.

204

xavier bach and pavel štichauer

Some remarks on this particular example are in order. First, although Manzini and Savoia (2005:II:651) take the second-person singular of the unergative drumir (< dormīre ‘sleep’) as displaying the auxiliary have (and thus representing the standard HHH–HHH as opposed to the unaccusative pattern BBB–BBB), it is impossible not to note the formal identity across all three classes.11 If further data were to confirm such a situation, we would have a slightly different morphological realization of the three classes. Similar uncertainty may also arise in the case of other examples drawn from Manzini and Savoia (2005:II:650–652). For instance, for Castellazzo Bormida, the same heteroclitic system can be established (BBB– BBB/BBH–BBH/HHH–HHH), but on closer inspection, the third-person plural shows the same auxiliary form for all three classes. In addition, as already observed, this type of mixed system is limited to the present perfect, displaying generalized have or be in the other tenses (in the above example from Castelletto Merli, the pluperfect requires have throughout the paradigm).12 It is thus essential to note that the distribution of heteroclisis is in fact different when all the periphrastic perfect tenses are taken into account. That means that the discourse-based explanation hinted at above cannot hold— synchronically—for most of these varieties. Rather, under a heteroclitic analysis, any type of distribution is acceptable.

7.4.2 Heteroclisis with default marking Default marking is another characteristic present in a wide number of inflexionclass systems (cf. Gisborne and Hippisley 2017).13 In cases of default marking, each class has an overall distinctive distribution of the two auxiliaries have/be, but there is identical default realization in a subset of cells. This again is a hallmark of a wide range of inflexion-class systems: in multiple classes, a subset of cells—across all such classes—makes use of the same exponents (a banal example would be the first plural ending -iamo for all three Italian conjugation classes in the present indicative as opposed to three different endings in the second plural -ate/-ete/-ite). Let us consider the example of Bitetto, a southern-Italian Apulo-Barese variety (cf. Manzini and Savoia 2005:II:725f.), illustrated in Table 7.10. First of all, this system is very interesting in that it seems to group together reflexives and unergatives in the perfect (again, the pluperfect displays the default have for all classes), and so verbs such as ‘sleep’ follow the same pattern as 11 However, on closer inspection, there is possibly a significant difference, in that the reflexive exhibits the subject clitic a fused with the reflexive clitic (hence we have here at ei laˈva-ti versus t ei druˈmi and t ei mˈni). 12 The data for the other paradigms are not reported in Manzini and Savoia (2005:II:650f.). 13 See also the various instantiations of Carstairs-McCarthy’s ‘No Blur Principle’, largely based on the presence of default marking in inflexion (Carstairs-McCarthy 1994).

italo-romance & inflexional classes

205

Table 7.10 Heteroclisis in reflexives/unergatives in the variety of Bitetto (data from Manzini and Savoia 2005:II:725f.) ˈ

ә ˈ

ә

ˈ

ә

ˈ

ˈ

ә

ˈ

ˈ

ә

ә

ә

ˈ

ә

ә

ˈ

ә

ә ә

ә

ә≈

ә

ә ˈ ә ˈ

ˈ

ә

ә

ә ˈ ә ≈ ourselves= .ptcp

ә

әˈ

ә

әˈ

ә

ә≈

≈

ә әˈ ә

ә әˈ

ә әˈ

ә

ә ә

reflexives; see Manzini and Savoia 2005:II:725 for the full paradigm). This kind of class assignment is relatively rare, and it shows that in the group of verbs that took have as auxiliary (if we consider that both unergatives and transitives originally took have), unergatives might have been the first to show change: we are not aware of any system where transitives, but not unergatives, would pattern like reflexives. Such a grouping also has some theoretical relevance, in that it shows that the original motivation in class assignment (transitives/unergatives versus unaccusatives) can be disrupted and that, eventually, assignment of verbs such as sleep to one or another class becomes more or less arbitrary, lexically stipulated. Default marking, highlighted here by the absence of shading, concerns the first singular and the third plural, where have is selected, and the third singular, where instead be surfaces. However, the third-person singular is also exceptional in that there is a further phonology-driven selectional requirement, dealt with in detail by Loporcaro (see Chapter 8 in this volume). Indeed, be is selected only when a consonant-initial participle follows, otherwise have (in the form avә) is required before any vowel-initial participle (e.g., av arrәvˈɨ:tә ‘he/she has arrived’, av aˈpirtә ‘he/she has opened’). But, crucially for our purposes, this seems to be purely a matter of phonology, and does not change the fact that, given the same phonological factors, all three classes select the same auxiliary. The class of reflexives/unergatives is heteroclitic for the non-default sections of the paradigm, highlighted with grey shading: the second-person singular is aligned with transitives, whereas the first and second persons plural exhibit the same realization as unaccusatives.

206

xavier bach and pavel štichauer

7.4.3 Heteroclisis by overabundance We have already seen that free variation between the two auxiliaries can also occur. In some cases, as alluded to in §7.3, it can affect the entire paradigm; in other cases, it only concerns a subset of cells. In inflexional systems, such situations, where we have two (or more) interchangeable exponents for one and the same morphosyntactic property set, are referred to as ‘overabundance’ (cf. Thornton 2011:359, 362; Stump 2016:61 n.1; Štichauer 2018:15–17; and Chapter 11 in this volume). In a number of mixed systems, the heteroclitic class, instead of taking exponents from two classes for specific cells, alternates for some cells between the possible exponents of the other two classes. Although this looks like a strange situation, we argue that what Loporcaro (2007) defines as triple systems, where we witness free have/be alternation for a given subclass of reflexives, might be the origin of all mixed systems, in that such free variation betrays what we might dub ‘realizational uncertainty’ in the selection of either auxiliary.1⁴ An example that we present only schematically is the system of Barcis (cf. Manzini and Savoia 2005:II:660), where we find the standard distinction between transitives/unergatives displaying have and unaccusatives selecting be, but the mixed system in reflexives BBB≈H– B≈HBB, where the third-person singular and the first-person plural display free variation between the two auxiliaries. A more interesting example, which we shall discuss at greater length, is Popoli (eastern Abruzzese, cf. Manzini and Savoia 2005:II:688f.), where we find an exceptional mixed system in all three classes with free variation between classes only in the third-person singular. As Table 7.11 illustrates, all classes present the be auxiliary in the first and second persons singular, and have in the plural. In the third singular, transitives/unergatives use have, and unaccusatives be. Reflexives show free variation between the two auxiliaries in this cell, but here the choice makes them align with either one or the other class: the selection of be in this paradigm cell would make reflexives align with unaccusatives, whereas the selection of have would yield a realization typical of transitives/unergatives (unlike in the case of Barcis, where the selection of have would necessarily create a separate inflexional class with a different morphological realization).

1⁴ Of course, this claim is in need of further explanation, which we can only sketch here. Following one line of thinking about inflexional systems (Blevins 2016), we share the view that speakers must predict some forms of paradigms on the basis of implicative relations within the paradigm. In this sense, in a one-auxiliary system where only have (or be) is selected, any cell of the paradigm is predictive of the other cells, whereas in two-auxiliary systems, speakers might be uncertain about which auxiliary is to be used. However, once the choice is established, there is again a straightforward predictability throughout the paradigm (as the auxiliary is always the same intraparadigmatically). But in a mixed system, one cell with have (or be) might not be enough for the realization of the whole paradigm and so speakers must deduce the rest of the forms on the basis of a pattern of alternation.

italo-romance & inflexional classes

207

Table 7.11 Heteroclisis in reflexives in the variety of Popoli (data from Manzini and Savoia 2005:II:688f.) sɔ ddur’moit ˈ ә be.1sg sleep.ptcp ʃi ddur’moit ә ˈ be.2sg sleep.ptcp a dur’moit ә ˈ have.3sg sleep.ptcp ˈ ˈ a’vemm ә dur’moit ә have.1pl sleep.ptcp a’ve:t ˈ ә dur’moit ˈ ә have.2pl sleep.ptcp ˈ annә dur’moit ә have.3pl sleep.ptcp

mә so ʃʃak’kwa:t ˈ ә myself= be.1sg wash.ptcp ˈ tә ʃi ʃʃak’kwa:t ә yourself= be.2sg wash.ptcp s a ≈ e ʃʃak’kwa:t ˈ ә himself/herself= have.3sg ≈ be.3sg wash.ptcp ˈ tʃ a’vemm ә ʃʃak’kwa:tә/ʃʃek’kwe:tә ˈ ˈ ourselves= have.1pl wash.ptcp.msg/mpl v a’ve:t ˈ ә ʃʃak’kwa:tә/ʃʃek’kwe:t ˈ ˈ ә ә yourselves= have.2pl wash.ptcp.msg/mpl s annә ˈ ˈ ʃʃak’kwa:tә/ʃʃek’kwe:t ә ә themselves= have.3pl wash.ptcp.msg/mpl

ә sɔ vvә’niut ˈ be.1sg come.ptcp ʃi vvәˈ’niutә be.2sg come.ptcp ɛ vvә’niutә ˈ be.3sg come.ptcp ˈ ˈ a’vemmә vә’niutә have.1pl come.ptcp a’ve:tә vә’niutә ˈ ˈ ' have.2pl come.ptcp ˈ annә vә’niutә have.3pl come.ptcp

As can be seen, the two basic patterns are BBB–HHH for unaccusatives (an interesting ‘balanced’ distribution in which the singular is opposed to the plural, cf. Štichauer 2018:12), and BBH–HHH for transitives/unergatives. And, as already noted, reflexives allow for free variation in exactly that cell of the paradigm which would align them with one or the other class. This situation is intriguing and deserves further comment for at least two reasons. First, it seems to go against the predictions of Loporcaro’s implicational scales (cf. Loporcaro 2014:52; Chapter 8 in this volume), according to which direct transitive reflexives would behave much more like plain unaccusatives. In our heteroclitic class, this expectation is borne out only if the third-person singular selects be, whereas the choice of have would lead to auxiliary selection leapfrogging, as it were, over the intermediate classes of reflexive verbs. However, such a claim would need further discussion on the basis of a complete set of paradigms for each of the reflexive verb classes. Second, it is also interesting, as it addresses, albeit to a limited extent, an issue raised by one of the anonymous reviewers, who aptly points out that under our class-based account nothing prevents systems in which transitives and unergatives select have, reflexives select be, and unaccusatives exhibit a mixed pattern. The question is then why such systems seem to be unattested. In fact, as the example of Popoli shows, we do have a system of classes that comes close to such a situation. Transitives/unergatives follow one pattern, reflexives another,

208

xavier bach and pavel štichauer

and unaccusatives yet another different mixed pattern. Our class-based analysis also attempts to account for the origins of such mixed systems. We thus argue that the change or impetus towards mixed systems spreads from reflexives, where the mixing of the two auxiliaries might be due to the different nature of various reflexive subclasses (in the sense that some subclasses are highly transitive, e.g., odiarsi ‘hate each other’, whereas others are unaccusative, such as pentirsi ‘repent’). It is therefore not surprising that reflexives are the first to show a mixed pattern, while transitives, at one extreme, and unaccusatives, at the other, may resist the incursion of mixing (in §7.5, we discuss an interesting prediction made by this claim). A further example that we adduce for such a scenario is the variety of Macerata (belonging to the central Marchigiano dialects), discussed in detail by Paciaroni (2009). In this variety we find a variation of what we have called ‘heteroclisis by overabundance’, but only in a subset of reflexive verbs. In line with Loporcaro’s implicational scales, mentioned above, middle reflexives and direct transitive reflexives pattern with unaccusatives and exhibit the generalized BBB–BBB pattern, and transitives select auxiliary have (throughout the paradigm), but the class of indirect reflexives (along with antipassives, as expected) stands out as being heteroclitic between these two major patterns. Indeed, indirect transitive reflexives, for instance lavarsi le mani ‘wash one’s hands’, allow for free variation between the two auxiliaries in the third person in both the singular and plural, thus following the BBH≈B–BBH≈B pattern. So this example can again be used in order to show that heteroclisis in reflexives—or in a subset of reflexive verbs—might represent a transitional stage of overlap in the periphrastic realization of the present perfect. In one case, such convergence may affect the whole paradigm; in other cases, it may affect only a subset of cells. In addition, in a range of examples, such a subset of cells does not represent any natural collection of cells being very much close to a morphomic distribution.

7.4.4 Problematic cases of heteroclisis We believe that, in a sense, all the examples discussed so far are to an extent problematic, but they none the less allow a more or less stable set of inflexional classes to be established on the basis of the different intraparadigmatic distribution of the two auxiliaries. In some cases, as already alluded to in the case of the variety of Bitetto discussed in §7.4.2, the patterns are further complicated by phonologydriven requirements. Let us consider the example of Poggio Imperiale (Terranova) in Table 7.12.

italo-romance & inflexional classes

209

Table 7.12 Heteroclisis in unergatives with respect to unaccusatives/reflexives and transitives in the variety of Poggio Imperiale (Terranova) (data from Manzini and Savoia 2005:II:720f.) ˈ

ә

ˈ ә

ә

ˈ

ә

ˈ ә

әˈ

ˈ

ә

ˈ ә

ә≈

ә

≈

ˈ ә≈ ≈

ˈ

ә≈

ә

ˈ ә

ә

ә ә

ˈ

ә

ә

≈

ә

ˈ ә

ˈ ә

ә ә

ә әˈ

ә

әˈ

ә

әˈ ә

ә

әˈ

ә әˈ

ә

Poggio Imperiale can be taken to be a variation on Bitetto,1⁵ in that this time reflexives are grouped with unaccusatives for all forms (with be), but unergatives stand out as being heteroclitic, just as they are in Bitetto (probably because Poggioimperialese, too, is a variety where be progressively extends to certain cells in transitives, and possibly be extends first to unergatives then to transitives). Most importantly, Poggioimperialese also seems to be the only attested variety where unergatives are singled out, and exhibit heteroclisis. Yet there are some problems with these paradigms, and more complete and reliable data are needed. In fact, Manzini and Savoia (2005:II:720f.) report the paradigm only with object clitics, but these seem to have an important impact on auxiliary selection, as they determine the outcome of ‘free variation’: u/a simә caˈmatә lit. ‘him/her= we are called’ versus l εmmә caˈmatә lit. ‘him/her= we have called’, and the same in the thirdperson plural: u sonnә caˈmatә lit. ‘him= they are called’ versus l εnnә caˈmatә lit. ‘him= they have called’. It is not clear whether there is genuine free variation in auxiliary choice which then triggers the choice of the allomorphic form of the object clitic. A comparison with Neapolitan might be instructive.1⁶ In Neapolitan, we witness an apparently similar situation, in that there are two allomorphic forms of the object clitic: ’o and (l)l’. However, the selection of auxiliary be is limited to 1⁵ It is interesting to note that the varieties of Bitetto and Poggio Imperiale, though belonging to two different subgroups within Pugliese dialects (Apulo-Barese and Dauno-Appennine, respectively), crucially differ in the realization of the other compound tense, aspect, and mood paradigms. In the case of Bitetto we find generalized have throughout the paradigm, whereas in Poggio Imperiale (Terranova) be is generalized. This situation shows that ultimately the concrete auxiliary realization can be arbitrary (on a similar situation in other varieties, see Ledgeway 2019:367). 1⁶ We thank Adam Ledgeway for an important discussion of this point.

210

xavier bach and pavel štichauer

cases of object-clitic doubling, as in ’o so’ pigliato ’o bicchiere lit. ‘it= I am taken the glass’, while have surfaces when the object clitic alone is expressed, as in ll’aggiu pigliato lit. ‘it= I have taken’ (cf. Ledgeway 2009:305, 623).

7.5 Conclusions We have presented examples of systems in which reflexives exhibit a mixed auxiliation pattern, i.e., reflexives are, as it were, split between two major classes where the generalized (and expected) have or be is selected. We have argued that this situation is very similar to cases of heteroclisis found in a wide array of inflexional systems. Yet we have so far restricted ourselves to presenting the data rather than discussing its theoretical implications and problems. Moreover, we have only touched upon some diachronic considerations. In conclusion, we address these two issues in turn.

7.5.1 Theoretical problems As far as theoretical problems are concerned, at least two relevant objections could be raised. First, the distinction between content paradigms and form paradigms crucially rests on the featural identity of the content paradigm. In order to have two inflexional classes (i.e., two different form paradigms), the morphosyntactic features to be realized must be the same. In some of the data we have been discussing, this criterion is obviously not met. It is already clearly visible in the case of standard Italian introduced in §7.2.2. (Tables 7.5 and 7.6), where the selection of be triggers agreement with the subject. We could thus say that the paradigm cells (i.e., the morphosyntactic property sets in Stump’s terminology) in Class 1 and Class 2 are actually not the same, in that in Class 2 subject agreement in number and gender is required. We could dismiss such an objection by appealing to the notion of non-canonical inflexional classes (cf. Corbett 2009:3, Criterion 2), but it is also interesting to note that, apart from cases where we do find such a mismatch, as for instance in the case of Altomonte (Table 7.8), we also have examples where past participle agreement is neutralized (due to the phonological evolution of the final vowels) and is, eventually, eliminated (this might be the case of the last example of Poggio Imperiale presented in Table 7.12). In such a case—provided that there really is no trace of the subject agreement—we have an absolutely identical content paradigm. The second objection would probably be the first to be raised. Do we really need an inflexion class-based account of the data at hand? In a canonical perspective (cf. Corbett 2009:5f.), Principle II (independence) states that, synchronically, inflexional classes are unmotivated and that the distribution of lexical items, verbs in

italo-romance & inflexional classes

211

our case, is arbitrary and hence must be lexically stipulated. As Corbett (2009:6) notes, predictability and motivation would allow an account in terms of subcategories, in particular where, for instance, we find just a small number of lexical items within a given group. But in real instances of inflexion classes, it has been shown that most of the inflexion classes show some degree of relationship with phonological, syntactic, or semantic categories that are often remnants of how the classes came into being (cf. Bach 2019). In addition, we have seen that, in some cases, the distribution comes close to being unmotivated in that it separates verbs that would have normally been treated together (e.g., transitives and unergatives). Thus, it might be that speakers, when dealing with, say, call and speak (as, again, in the case of Poggio Imperiale in Table 7.12, or in the example of Bitetto in Table 7.10), must simply list one verb as belonging to one class, and the other to another class with a different realization. In addition, we also note that, although the overall grouping seems to be motivated (clearly, we are also dealing with differences at the level of mood and tense when there are different patterns between, say, the counterfactual and the pluperfect), the internal distribution of the two auxiliaries is simply unmotivated (in cases such as the first-person singular versus third-person plural). Finally, we also believe—moving beyond Italo-Romance—that there are clear instances of inflexion-class behaviour in other Romance languages, such as standard (European) French, many oïl varieties (see e.g., Smith 2016), and Laurentian French (see Sankoff and Thibault 1977; 1980; Canale, Mougeon, and Bélanger 1978; Rea 2020), in all of which, as is well known, we witness varying degrees of extension of have, resulting in a division of labour between have and be which has little synchronic motivation (see e.g., Ledgeway 2019:375f.; for a criticism of this type of class-based account, see Chapter 8 in this volume). Another straightforward example would be generalized be in all kinds of reflexives in standard French or Italian regardless of their different semantic and syntactic nature. Reflexives would thus be an autonomous inflexional class defined only on the basis of their identical morphological and syntactic realization (in that they all take a reflexive clitic). Indeed, as we have seen, and as Loporcaro has also demonstrated, change in the auxiliation patterns found with reflexives is at best marginal. This leads us to a discussion of diachrony.

7.5.2 Diachronic considerations Apart from one view in the literature (cf. Bentley and Eythórsson 2001:67–70), according to which the origin and rise of mixed perfective auxiliation systems might have been due to a homonymic clash between two identical forms of the auxiliary

212

xavier bach and pavel štichauer

have in the second- and third-persons singular,1⁷ there are two competing accounts which appear to be much more in line with the data we have presented. The first account, which we only mention, without dealing with it in detail, is explicitly based on a parameter hierarchy which aims to capture microparametric variation within all mixed auxiliation patterns (cf. Ledgeway 2019; see also Andriani 2018). The other, implicit in Loporcaro’s work (2007; 2014; 2016; and Chapter 8 in this volume), builds on wide-ranging cross-dialectal comparison showing how triple auxiliation systems are originally connected with the versatile behaviour of reflexive verbs. In particular, the data seem to suggest that the locus of diachronic variation is precisely reflexive verbs which diachronically undergo movements between the two major patterns. In his overview of the old Florentine auxiliation systems, Loporcaro (2014:57–61) stresses the shift from an original point of departure where unaccusatives and retroherent reflexives select be to a system where be extends through intermediate classes of reflexives, and he claims that such ‘variation in the in-between categories, with shifting usage (aux H > aux E) over time […], must have resulted in transitional stages with three auxiliation options […] show[ing] that triple-auxiliation systems, far from being an exotic peculiarity of some minor dialect varieties, are a fairly normal product of diachronic change in auxiliary selection rules’ (61). All the examples we have discussed so far represent such a transitional stage of competition between two major classes, thus creating a heteroclitic class of reflexives, realized with—very often—a morphomic mixing of the two auxiliaries within one and the same paradigm. One interesting prediction follows from our discussion of such a change. The mixing of auxiliaries seems to be a transitional stage towards one auxiliary encroaching on more classes, and both auxiliaries seem to proceed in opposite directions: have starting to be used first in some cells in reflexives, and only then in unaccusatives, and be appearing first in reflexives, then in unergatives, and only then in transitives. We should therefore not find systems with mixed auxiliation patterns restricted to unaccusatives unless reflexives all had have, nor systems with mixed auxiliation restricted to transitives, unless all other classes only used be. But we can find cases where both directions of change have proceeded at the same time for different cells and all classes end up showing mixed auxiliation patterns. Of course, more work is needed, and in particular more varieties need to be studied, in order to see whether this claim is falsified or stands.

1⁷ For some problems with this view, see, for instance, Štichauer (2018:14f.; 2019:94–96).

8 The morphological nature of person-driven auxiliation Evidence from shape conditions Michele Loporcaro

8.1 Introduction An interesting and much-debated topic in the study of Romance perfective periphrases is the analysis of person-driven variation and splits in mixed auxiliation systems, intermingling forms of have and be, which occur in particular, though not exclusively, in central–southern Italo-Romance.1 The issue at the core of the controversy concerns the division of labour between morphology and syntax. On the one hand, studies in generative syntax generally strive for a syntactic account of person-driven splits, tracing the selection of different auxiliaries in different persons (as well as free variation in one and the same person) back to differences in syntactic structure which are sometimes compared (e.g., Manzini 1 If morphology is ‘the Poland of linguistics’ (Spencer and Zwicky 1998:1), ‘given its tendency to disappear and reappear cyclically in the history’ of the discipline (Janda and Kathman 1992:153), there are a few scholars who may compete for the role of the Woodrow Wilson of morphology. In Romance linguistics, I can think of no better candidate than Martin Maiden, whose life’s work is a source of inspiration to many fellow linguists, and one of whose teachings—particularly resonating with my own interests—is that Romance linguistics can feed linguistic theory in a much more challenging way if it takes dialect variation seriously. For these reasons, I thank the editors for inviting me to contribute to this volume. I am also indebted to Angela and †Francesco Leone as well as Donatello Navarra (Gravina in Puglia), Dina Ferorelli and Gianni Giannini (Bitetto), Laura Caldarola, Aurelia D’Ingeo, Salvatore Eremita, and Pietro Stragapede (Ruvo di Puglia), who kindly shared with me their native intuitions on their dialects, as well as to Giovanni Manzari, Damiana Santoro, and Salvatore Santoro, who helped with fieldwork. Thanks also go to Anıl Yildiz for the software used to draw Map 8.1. Unreferenced data come from my own fieldnotes. Lastly, I would like to thank Cecilia Poletto and the audience at the Romance Linguistics Colloquium in Frankfurt (November 2020), as well as Federica Breimaier, Greville Corbett, G. Manzari, Anna M. Thornton, and the editors for comments and suggestions. Usual disclaimers apply. In glosses, I have omitted prs in verbs in the present indicative. The two auxiliaries will be referred to generically as have/be, but the small caps do not imply, as sometimes in morphology, that these are cited as lexemes, not only because they are, if anything, the translation equivalents of the Romance (ancillary) lexemes under discussion, but also because, in mixed systems, these are not (inflexionally consistent) lexemes at all, as explained in due course.

Michele Loporcaro, The morphological nature of person-driven auxiliation. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Michele Loporcaro (2022). DOI: 10.1093/oso/9780198870807.003.0009

214

michele loporcaro

and Savoia 1998) with alignment splits in languages such as Dyirbal. On the other, an alternative line of research which I have contributed to shaping since Loporcaro (1999b) claims that syntax goes as far as distributing contrasting auxiliation patterns across classes of constructions (unaccusative vs unergative, etc.), while the distribution across persons within one and the same pattern is morphological, in the same sense as the selection of a specific personal ending contrasting with those from different inflexional classes (e.g., 1pl -amo vs -emo vs -imo in Italo-Romance). In the present chapter, I will bring into the debate a crucial argument in support of the latter view, discussing data from three dialects of central Apulia in which selection of have vs be in one paradigm cell (the third-person singular of the compound perfect) is sensitive to the postlexical phonological context. This is—I will argue—a shared innovation (a) that is unique, to the best of our knowledge, (b) that would not have been possible in non-mixed systems, and (c) whose successful (i.e., contradiction-free) analysis crucially rests on the linguist’s assumptions about the boundary between morphology and syntax. The chapter is organized as follows. §8.2 introduces mixed auxiliation and considers it against the backdrop of current studies in Romance auxiliary selection; §8.3 moves on to consider the data from three Apulian dialects, explaining why they represent an insurmountable problem for a syntactic approach to mixed auxiliation; §8.4 puts forward the solution, arguing that selection of the ‘has’/‘is’ form depends on a shape condition comparable to that accounting for the a/an alternation in the English indefinite article; §8.5 discusses mixed auxiliation with reference to the concepts of overabundance and heteroclisis, drawing on a parallel from (non-periphrastic) verb inflexion; §8.6 addresses what I take to be the limits of a morphological account of auxiliary choice, which has scope over personrelated alternations and variation but whose extension to auxiliary selection tout court—with the ensuing denial of the existence of a syntactic rule—is doomed to failure: the syntactic auxiliation rule is defended against recent claims that auxiliary selection in French should be viewed as a matter of inflexional class. Finally, §8.7 offers a brief conclusion.

8.2 Perfective auxiliation: syntax vs morphology Consider the following data from the dialect of Altamura (province of Bari; cf. Loporcaro 1988; 2007):2 2 Data are reported in a simplified IPA transcription, with CC instead of Cː for geminates. Here and in what follows, I will concentrate on the compound present perfect, since perfective periphrases in which the auxiliary occurs in other tenses/moods display different distributions in this and the other dialects considered. Ledgeway (2019) is an example of a theory of Romance auxiliation which also encompasses tense/mood-related variation.

the morphological nature of person-driven auxiliation (1)

Altamura i. transitive/unergative ii. a. ˈsɔ mmanˈʤεi ̯t/ˈaɟɟә manˈʤεi ̯t (la ˈpast) B/H ‘I have eaten (the pasta).’ b. ˈsɪ mmanˈʤεi ̯t/ˈa manˈʤεi ̯t (la ˈpast) B/H ‘You have eaten (the pasta).’ c. ˈε/ˈa mmanˈʤεi ̯t (la ˈpast) B/H ‘He has eaten (the pasta).’

215

unaccusative ˈsɔ ˈdʤʊu̯t/ˈaɟɟә ˈʃʊu̯t B/H 1sg ‘I have gone.’ ˈsɪ ˈdʤʊu̯t/ˈa ˈʃʊu̯t B/H 2sg ‘You have gone.’ ˈε/**ˈa ˈdʤʊu̯t B/**H 3sg ‘He has gone.’

In several articles over recent decades, I have argued that on the one hand the contrast between free variation in (1c.i) and categorical selection of be in (1c.ii) must be dealt with in the syntax—on a par with the contrast in standard Italian between ha/**è mangiato ‘(s)he has/is eaten’ and è/**ha andato ‘he is/has gone’, see (Table 8.1(a))—while on the other hand the alternation between (1c.ii) (third-person singular) and free variation in the first and second person must be considered a matter of inflexional morphology. This is what I term in (2a) a morpholexical approach to person-driven auxiliary choice, meaning that when person-alternating forms are selected by one and the same lexical predicate, these constitute the paradigm of a suppletive split lexeme (in the sense of Corbett 2012; 2015):3 (2)

Mixed (person-driven) auxiliary selection a. morpholexical approach: Loporcaro (1999b:213; 2001:462, 470; 2007:186, 2014), Bentley and Eythórsson(1999; 2001:70f.), Štichauer (2018; 2019); Bach and Štichauer (Chapter 7 in this volume). b. syntactic approach: Kayne (1993), Cocchi (1994; 1995), Ledgeway (1998; 2019), Manzini and Savoia (1998; 2005; 2007).

In other words, in (1ii) there is no sense in which the syntactic rule for auxiliary selection operates directly on have vs be, a claim that comes in different flavours in the studies in (2b). Rather, auxiliary selection in (1) operates on mixed auxiliary paradigms (i.e., split lexemes) such as those introduced in (14a–b) in §8.4.⁴ This basic idea capitalizes on—and is intertwined with—what is labelled in Loporcaro (2007) a ‘syntactic approach’ to (non-mixed) auxiliary selection ((3a)), to be contrasted with the alternative approaches in (3b–c).

3 Corbett (2012:184f.) includes data comparable to (1a–c). ⁴ Obviously, this is not to deny that, in a trivial sense, syntax is relevant here too, as person inflexion, determined by agreement, pertains to contextual inflexion (Booij 1994; 1996).

216

michele loporcaro

(3) Auxiliary selection a. syntactic approach: Rosen (1982; 1997); La Fauci (1988; 1989; 1992; 2004); La Fauci and Loporcaro (1989); Perlmutter (1989); Loporcaro (1999b; 2001; 2007; 2008; 2011; 2015; 2016); Formentin (2001; 2002); Vecchio (2006); Paciaroni (2009); Loporcaro and Putzu (2013). b. lexical approach: Diez (1843:293), Lois (1990), Manzini and Savoia (2007:151), etc. c. morphological approach: Bonami (2015); Štichauer (2018:9); Bach and Štichauer (Chapter 7 in this volume). The line of research in (3a) is based on the definition of auxiliary in (4): (4) Auxiliary (Definition) ‘Auxiliaries are a lexically designated closed class of verbs whose defining property is that they inherit a 1.’ (1 = subject; Rosen 1997:192) Under this view, auxiliation in Romance perfective periphrases—for example, the occurrence of two auxiliaries in Italian vs one auxiliary in Spanish—is best described as the selection of what Bonami (2015) terms an ‘ancillary lexeme’, contributing finite verb morphology to the periphrasis and lacking an argument grid of its own. In Italian, this selection depends on the syntactic properties of the clause in terms of alignment of grammatical relations, as first recognized in the seminal work of La Fauci (1988), elaborating on Perlmutter’s (1978) Unaccusative Hypothesis (and as shown in Table 8.1). Table 8.1 Auxiliary selection in Italian and Spanish Inactive unaccusative a b

Italian Spanish

Active unergative transitive

be have

This view is incompatible with the many accounts inspired by the idea—as old as Diez (1843:293) at least—that a perfective auxiliary has the same properties as lexical predicates. This is termed the ‘lexical approach’ (3b) in Loporcaro (2007), to which the reader is referred for a discussion of its inadequacy (one example of such an analysis is cited in note 14 in this chapter). Once one agrees that perfective auxiliary selection must be accounted for in the syntax (cf. §8.5 for a refutation of the alternative view that perfective auxiliation per se is a fact of inflexional morphology), one still has to evaluate the merits of competing syntactic approaches, which brings us back to mixed auxiliation. In fact, Kayne (1993:3)—the first reference in (2b)—put forward an influential treatment of the distribution of have/be across Romance, both in mixed and in

the morphological nature of person-driven auxiliation

217

non-mixed systems, assuming that ‘[t]here is no auxiliary selection rule. “Have” is identical to “be” but for the incorporation of an abstract preposition.’ Such incorporation is assumed in the presence of any distribution of have and, conversely, non-incorporation is assumed wherever be is selected, including mixed systems such as the dialect of Trento (Kayne 1993:21, based on Gatti 1990:174), where in reflexive constructions be occurs in the first/second persons and have in the third person. The problem with this approach, as argued in Loporcaro (2007:185f., 193), is that it overgenerates and is hence non-predictive, since it could accommodate an infinity of data sets that, for non-mixed systems, do not seem to exist, such as those shown in Table 8.2.⁵ Table 8.2 Unattested auxiliary systems unacc. a b c

*variety x *variety y *variety z

H B

Inactive reflexive dir. trans. unerg. indir. trans. B H B B H H B H

Active trans./unerg. H B H

The structural reason why these do not occur is provided by the implicational scale first proposed in Loporcaro (1999b:213; see Table 8.3) and further articulated in subsequent work mentioned in (2a).⁶ Table 8.3 Implicational scale of auxiliary systems unacc. retr. a b c d e f

Fr. Log. Pcn. OFlo. Lec. Sp.

B B B B B

Inactive reflexive dir. trans. indir. unerg.

Active trans./unerg. indir. trans. H H H H H H

⁵ The abbreviations in Tables 8.2–8.3 read as follows: unacc(usative), dir(ect) trans(itive) reflexive (Italian Maria si è lavata ‘Maria washed herself ’), indir(ect) trans(itive) (Maria si è lavata le mani ‘Maria washed her hands’), indir(ect) unerg(ative) (Maria si è risposta ‘Maria answered (to) herself ’), retr(oherent) (Maria si è pentita ‘Maria repented’). ⁶ The implications are grounded in the structural representations assumed in Relational Grammar, though most of the representational assumptions involved, starting with the Unaccusative Hypothesis, have since been taken up in most other formal frameworks.

218

michele loporcaro

This scale is the result of inserting the different subtypes of reflexive constructions between the two poles of unaccusatives and unergatives in Table 8.1. In Loporcaro (2001; 2007; 2014; 2016), I showed that systems displaying free variation and/or person-driven alternations can be elegantly plotted onto the same scale, as illustrated in Table 8.4.

Table 8.4 Implicational scale of auxiliary systems unacc. a b c d e f

? Pietrarolo ? ? Bitettese Aquilano

retr.

Inactive reflexive dir. trans. indir. unerg.

mixed paradigm y

indir. trans.

Active trans./unerg.

mixed paradigm x

The scale generalizes over the following situation: (a) there are very many possible combinations of B, H, or free variation of B/H: in Pescarini and Loporcaro’s (forthcoming) sample of 76 dialects with mixed auxiliation, 52 distinct combinations are attested; (b) there are dialects in which just one such combination— whatever the distribution of B, H, or B/H across verb persons—is selected uniformly from unaccusatives to transitives/unergatives (f in Table 8.4) while there are others in which two different combinations occur (b, e in Table 8.4).⁷ In addition, several dialects have been shown to display a three-way choice with (some subsets of) reflexives patterning differently with respect to both unaccusatives and unergatives. This is spectacular evidence in support of the approach in (2a)/(3a). In fact, if mixed auxiliation were to be analysed under (2b)—for example, based on Kayne’s (1993) assumptions—we would not be able to restrict possible options to the limited set in Table 8.4. Rather, as argued in Loporcaro (2007:186), assuming that person-driven auxiliation involves the two auxiliaries H/B and must be accounted for in the syntax in the same way as in Italian (a in Table 8.1), this would force us to treat 729 conceivable options for systems such as (f) in Table 8.4, some half a million different options for those like (b, e) in Table 8.4, and over three billion options for triple auxiliation systems, a combinatorial explosion which is at odds with arrangements of alignment systems known to occur

⁷ Whether the hitherto unattested combinations in (a, c–d) in Table 8.4 occur is an empirical question which will be answered as the database expands.

the morphological nature of person-driven auxiliation

219

cross-linguistically.⁸ By contrast, if person-driven auxiliation is a matter of morphology, then it is just the arrangements of inflexional morphology that vary wildly in these dialects. In other words, inflexion arranges forms originally pertaining to the two proto-Romance ancillary lexemes esse ‘be’ and habere ‘have’ to form one, two, or three mixed auxiliary lexemes whose paradigm is suppletive. This can be expressed in several ways in current morphological theories (cf. §8.5): in Stump’s (2016) terminology, each one of the mixed auxiliary lexemes has a heteroclitic form paradigm, while, following Thornton’s (2018) application of Fradin’s (2003) lexeme vs flexeme distinction to inflexional morphology, it maps onto a heteroclitic flexeme. With these admittedly rather lengthy preliminaries, we have now paved the way to move on to the crucial data demonstrating the inflexional nature of persondriven auxiliation.

8.3 Perfective auxiliation depending on the phonology in three dialects of Apulia The crucial data come from three dialects in central Apulia, those of Ruvo, Gravina, and Bitetto, whose perfective auxiliation is described in Manzini and Savoia’s (2005:II:724–726; III:29f.) monumental monograph. Since I will use their data to prove that their analysis is incorrect, let me first point out how their enormously detailed empirical and theoretical contributions have been instrumental in advancing the scientific community’s understanding of morphosyntactic variation within Italo-Romance.⁹ The three dialects, spoken in the province of Bari (see Map 8.1), display, like Altamurano (see (1)), both person-driven alternations and free variation in some paradigm cells. For Bitettese and Ruvese, this is schematized as shown in (a) in Table 8.5 from Manzini and Savoia (2005:II:728), where H/B stands for free variation of have/be. In schematizing the same data in (b) in Table 8.5, Manzini and Savoia (2007:226) add a further symbol, which occurs in the third-person singular only: H~B, explained as ‘selection of essere [be] and avere [have] neither free nor (entirely) according to verbal class’. This paradigm cell will be our main concern, but I will first sketch the full auxiliation pattern for each type of dialect.

⁸ The figure 729 represents the possible combinations of three values (B, H, or free variation of B/H) over six independent variables (the paradigm cells); for systems with two combinations, this figure must be raised to the power of two, and to the power of three, if there are three combinations. Many of these combinations do occur: extrapolating linearly from Pescarini and Loporcaro’s (forthcoming) findings, in a sample of 729 dialects we should expect to find 499 different combinations. ⁹ In particular, recent comparative papers on mixed auxiliation all capitalize on this invaluable mine of data: cf. Štichauer (2018; 2019), Ledgeway (2019), Pescarini and Loporcaro (forthcoming).

220

michele loporcaro

Table 8.5 Ruvo and Bitetto a

Manzini and Savoia (2005 :II:728, table 79), type (xiv)):

b

Manzini and Savoia (2007:226):

1sg

2sg

3sg

1pl

2pl

3pl

H/B

B

H/B

H

H

H

H/B

B

H~B

H

H

H

First of all, for Bitettese, Manzini and Savoia’s (2005:II:725f.) own data are actually at odds with the schemas in (a–b) in Table 8.5, as this dialect, contrary to Ruvese and Gravinese, still contrasts unaccusatives (5a) with unergatives/transitives (and all reflexives, (5b)) in the second-person singular:1⁰ (5)

a. ˈsi vәˈneu̯tә (Bit.) be.2sg come ‘You have come.’ b. ˈa dәrˈmeu̯tә / t= ˈa laˈvɘːtә / u= have.2sg slept refl.2sg= have.2sg washed him= ˈa caˈmɘːtә (Bit.) have.2sg called ‘You have slept/washed yourself/called him.’

This qualifies Bitettese as a dialect of type (e) in Table 8.4, while Ruvese and Gravinese are of type (f), since they show one and the same mixed auxiliation pattern in all constructions, as exemplified for the second-person singular in (6):11 (6)

a. ˈa vәˈnʊu̯tә / dәrˈmʊu̯tә/ caˈmзːt a ˈffrat=tә // have.2sg come slept called to brother=2sg t= ˈa laˈvзːtә (Grv.) refl.2sg= have.2sg washed ‘You have come/slept/called your brother/washed yourself.’ b. ˈsi vәˈnɪu̯tә / drәmˈmɪu̯tә // be.2sg come slept u= ˈsi caˈmɔːtә // tә= ˈsi laˈvɔːtә (Ruv.) him= be.2sg called refl.2sg= be.2sg washed ‘You have come/slept/called him/washed yourself.’

By contrast, free variation occurs in both dialects in the first-person singular: (7)

a. ˈsɔ / ˈaɟɟә vәˈnʊu̯tә /dәrˈmʊu̯tә /caˈmзːt a ˈffrat=tә // be.1sg have.1sg come slept called to brother=2sg

1⁰ My consultants diverge from Manzini and Savoia’s: see Table 8.8–8.9 and examples (9a–b). 11 Manzini and Savoia’s transcriptions have been revised following conversations with my consultants. Transcriptions diverge especially for stressed vowels, which is partly explained by the fact that vowel systems are quite unstable all over this area (cf. the recent monograph by Manzari 2019).

the morphological nature of person-driven auxiliation

221

mә= ˈsɔ / m= ˈaɟɟә laˈvзːtә (Grv.) refl.1sg= be.1sg refl.1sg= have.1sg washed ‘I have come/slept/called your brother/washed myself.’ b. ˈsɔ vvәˈnɪu̯tә /ddrәmˈmɪu̯tә // ˈaɟɟә vәˈnɪu̯tә /drәmˈmɪu̯tә be.1sg come slept have.1sg come slept u= ˈsɔ ccaˈmɔːtә / ˈaɟɟә caˈmɔːtә // him= be.1sg called have.1sg called mә=ˈsɔ llaˈvɔːtә / m=ˈaɟɟә laˈvɔːtә (Ruv.) refl.1sg=be.1sg washed refl.1sg=have.1sg washed ‘I have come/slept/called him/washed myself.’ Bitettese, on the other hand, according to Manzini and Savoia (2005:II:725f.) has free variation only in the first-person plural (8a) and categorical selection of either B or H in all the remaining persons but the third-person singular, as exemplified with the first-person singular in (8b): (8)

a. ˈsiːmә / ˈaːmә vәˈneu̯tә / dәrˈmeu̯tә // be.1pl have.1pl come slept nә= ˈsiːmә / n= ˈamә laˈvɘːtә (Bit.) refl.1pl= be.1pl refl.1pl= have.1pl washed ‘We have come/slept/washed ourselves.’ b. ˈaɟɟә vәˈneu̯tә /dәrˈmeu̯tә //**ˈsɔ vvәˈneu̯tә / ddәrˈmeu̯tә // have.1sg come slept be.1sg come slept m= ˈaɟɟә laˈvɘːtә / **mә = ˈsɔ llaˈvɘːtә (Bit.) refl.1sg= have.1sg washed refl.1sg= be.1sg washed ‘I have come/slept/washed myself.’

Summing up, in Tables 8.6–8.8 we provide the overall distributions of perfective auxiliaries in the compound perfect in the three dialects, based on Manzini and Savoia’s data (H–B = have/be according to the phonological context, which will be addressed in (10–(12)): Table 8.6 Ruvese (Manzini and Savoia 2005:II:724f.) 1sg 2sg 3sg 1pl 2pl 3pl unaccusatives = unergatives H/B B H–B H H H

Table 8.7 Gravinese (Manzini and Savoia 2005:III:29f.) 1sg 2sg 3sg 1pl 2pl 3pl unaccusatives = unergatives H/B H H–B H H H

222

michele loporcaro Table 8.8 Bitettese (Manzini and Savoia 2005:II:725f.) 1sg 2sg 3sg 1pl 2pl 3pl a Unaccusatives H B H–B H/B B H b Unergatives H H H–B H/B B H

For Gravinese and Ruvese my consultants’ judgements confirm Tables 8.6– 8.7 while for Bitettese I have found two different distributions (divergences from Table 8.8 are in boldface): Table 8.9 Bitettese (consultant 1) 1sg 2sg 3sg 1pl 2pl 3pl a Unaccusatives B/? H B H–B B/? H B B/H b Unergatives B/? H H/B H–B H/B H/B H

Table 8.10 Bitettese (consultant 2) 1sg 2sg 3sg 1pl 2pl 3pl unaccusatives = unergatives B/(H) B/(H) H–B B/H B/H H

Consultant 1 preserves the contrast between unaccusatives and unergatives (plus all remaining constructions), though this contrast is realized differently and extends also to the second-person plural and third-person plural. Consultant 2, by contrast, has one and the same auxiliation scheme throughout, as exemplified for the second-person singular in (9), where be is always the preferred option, with have marginally acceptable.12 (9) a. ˈsi vvәˈneu̯tә / ddәrˈmeu̯tә // tә= be.2sg come slept refl.2sg= llaˈvɘːtә (Bit.) washed ‘You have come/slept/washed yourself.’ b. ? ˈa vәˈneu̯tә / dәrˈmeu̯tә / t= have.2sg come slept refl.2sg= laˈvɘːtә (Bit.) washed ‘You have slept/washed yourself/called him.’

ˈsi be.2sg

ˈa have.2sg

12 The acceptability of have seems to be affected by various factors here, lexical as well as syntactic: for instance, while ? ˈa ˈʃeu̯t ‘you.sg have gone’ is judged as slightly awkward, nan ˈa ˈʃeu̯t ‘you.sg haven’t gone’ and tә n ˈa ˈʃeu̯t ‘you.sg have gone away’ are perfect. Note that one motive behind consultant 2’s eliminating the auxiliation contrast may be the influence of urban Barese, which, as shown by Andriani (2017:158; 2018), has a stable unary auxiliation system of type (f) in Table 8.4 with have selected in third persons and be elsewhere. Among the centres whose dialects are considered here, Bitetto lies next to Bari, only 16 km WSW from its centre, and much less from its outskirts.

the morphological nature of person-driven auxiliation

223

This is a frequently encountered situation in the whole area, where even asking several informants does not guarantee that one is describing anything but a collection of idiolects among many, which may differ crucially in the relevant aspects in this quite shaky area of grammar.13 Be that as it may, my data coincide with those of Manzini and Savoia (2005) on the point which is of crucial concern for us, namely, auxiliary choice in the third-person singular. Here, the three dialects behave uniformly, as exemplified in (10)–(12) (data from Manzini and Savoia 2005). (10) a. ˈε ddәrˈmeutә / vvәˈneutә // s= ˈε llaˈvɘːtә (Bit.) be.3sg slept come refl.3= be.3sg washed b. ˈav arrәˈvɘːtә / aˈpirtә // s= ˈav aˈpirtә (Bit.) have.3sg arrived opened refl.3= have.3sg opened ‘(S)he has slept/come/washed him-/herself/arrived/opened // it opened (up).’ (11) a. ˈε drәmˈmeutә / vәˈneutә // s= ˈε laˈvɔːtә (Ruv.) be.3sg slept come refl.3= be.3sg washed b. ˈɔv aspәtˈtɔːtә / arrәˈvɔːtә // s= ˈɔv asˈsεis̯ ә (Ruv.) have.3sg waited arrived refl.3= have.3sg sat ‘(S)he has slept/come/washed him-/herself/waited/arrived/sat down.’ (12) a. ˈje ddәrˈmʊu̯tә / vvәˈnʊu̯tә / mˈmurtә// s= ˈe be.3sg slept come died refl.3= be.3sg laˈvзːtә (Grv.) washed b. ˈav aˈpirtә la ˈportә // s= ˈav asˈsiːsә (Grv.) have.3sg opened the door refl.3= have.3sg sat ‘(S)he has slept/come/died/washed him-/herself/opened the door/sat down.’ The two auxiliary verb forms ˈε/ˈavә in (10), ˈε/ˈɔvә in (11), and ˈje/ˈavә in (12) are unambiguously identified as the third-person singular of the present indicative of be and have, respectively. In an unexpected way, though, their selection, with predicates of any syntactic kind, depends on the initial segment of the lexical verb: the third-person singular form of be ˈε/ˈje is selected before consonant-initial participles, and the third-person singular form of have ˈavә/ˈɔvә before vowel-initial participles. Describing these facts, the authors do not seem to realize how problematic they are for their own approach, according to which the choice between B/H

13 Manzini and Savoia’s (2005:1.xi) questionnaire was recorded in Bitetto with one informant.

224

michele loporcaro

depends on properties of clause structure.1⁴ Rather, they comment (for Ruvese and Bitettese):

We can analyse this phenomenon in terms of the model already proposed. The fundamental fact from a syntactic point of view is represented by the alternation between essere [be] and avere [have] in the third person singular, which we can interpret as a reflex of the possibility of treating the third person singular as eventive when avere [have] is inserted. This syntactic base had grafted onto it a phonological restriction, which however is independent of it. (Manzini and Savoia 2005:II:744; Gravinese is analysed along the same lines in Manzini and Savoia 2005:III:17)

However, this description is not true to the facts, as the phonological contrast is not ‘grafted’ onto a(n alleged) syntactic condition but clearly has priority over it: in the third-person singular have can be selected if and only if the following word begins with a vowel. Consequently, under their approach, Manzini and Savoia are creating something which does not (and arguably cannot) exist in the languages of the world. Under their account, the data in (10)–(12) violate the phonology-free syntax principle (Zwicky 1969; 1996; Zwicky and Pullum 1983): (13) Phonology-free syntax principle ‘strictly phonological information is never required for the operation of the syntactic component’ (Zwicky 1969:411) In fact, the normal state of affairs in the syntax–phonology relationship is that:

syntax can be sensitive to abstract properties realized in the distribution of phonological features, but not to the specific phonological features. Though the conditions in a syntactic rule can have certain sorts of indirect or ultimate phonological consequences (like the temporal ordering of the parts of an expression), these conditions never seem to distribute phonological properties directly; no language has a

1⁴ Manzini and Savoia (2007:151) follow Kayne (1993) in maintaining a ‘bi-sentential analysis’ supported, in their view, ‘by the fact that participial clauses and auxiliary verb may be found independently of each other, with recognizably the same basic characteristics, so that auxiliary–past participle constructions appear to be an ordinary case of embedding a non-finite sentence beneath a main verb’. Note in passing that this is a clear instance of a type (3b) stance, where auxiliaries are represented, contrary to the definition in (4), as verbs endowed with an argument grid of their own.

the morphological nature of person-driven auxiliation

225

syntactic rule stipulating that some constituent begin with an obstruent, or have no more than two syllables, or contain only unrounded vowels, or have stress on its penultimate syllable. (Zwicky 1996:4477) Elaborating on this list of syntactic impossibilities, imagine a Romance language in which a given verb takes a (prepositionally marked) indirect object if the relevant argument begins with a consonant, but a direct object if the argument begins with a vowel. Such a language would look like (14), invented by varying an existing language, Logudorese Sardinian, whose real data (from the variety of Bonorva, in the province of Sassari) are displayed for comparison in (15) (the initial single asterisks in (14a–d) are meant to indicate that the whole system is impossible):1⁵ (14) a. *ˈappɔ ˈiːðu ** (a bb)aˈindzu (*Bon.) have.1sg seen to Gavino ‘I have seen Gavino.’ b. *ˈappɔ ˈiːðu a ˈkːaːnεzε / **ˈɣaːnεzε (*Bon.) have.1sg seen to dogs dogs ‘I have seen dogs.’ c. *ˈappɔ ˈiːðu (**a) isˈtεvεnε (*Bon.) have.1sg seen to Stefano ‘I have seen Stefano.’ d. *ˈappɔ ˈiːðu (**a) ˈεspεzε (*Bon.) have.1sg seen to wasps ‘I have seen wasps.’ (15) a. ˈappɔ ˈiːðu a isˈtεvεnε / a bbaˈindzu (Bon.) have.1sg seen to Stefano to Gavino ‘I have seen Stefano/Gavino.’ b. ˈappɔ ˈiːðu ˈɣaːnεzε / ˈεspεzε (Bon.) have.1sg seen dogs wasps ‘I have seen dogs/wasps.’ c. lɔz=/**liz= ˈappɔ ˈiːðɔzɔ (sɔs ˈkaːnεzε / a isˈtεvεnε them.acc/dat= have.1sg seen:m.pl the:m.pl dogs to Stefano ε bbaˈindzu) (Bon.) and Gavino ‘I have seen them (the dogs/Stefano and Gavino).’ Back in the real world, Logudorese Sardinian has both a-marked and unmarked direct objects, but these are direct objects in both cases, since this is required by the valency of the predicate regardless of the phonology. That this analysis is correct 1⁵ The name baˈindzu has its initial /b/ deleted when preceded by a final vowel, owing to a general rule, while isˈtεvεnε, stemming from the Greek vocative Στέϕᾰνε (Wagner 1997:294), has undergone i-epenthesis, like all words originally beginning with an s+C cluster in Logudorese.

226

michele loporcaro

is attested to, for instance, by the fact that the objects in (15a–b) are both cliticized by a direct object clitic, and not an indirect object clitic (see (15c)). Furthermore, selection of an object with or without the marker a depends on a syntactic rule (differential direct object marking) which is well known to be sensitive to semantic-referential properties (animacy, definiteness), not to the phonology. Arguably, cases such as (14) do not exist in the languages of the world, as they are ruled out by (13). Thus, observing that ‘has’/‘is’ selection in Bitettese, Gravinese, and Ruvese is sensitive to the initial segment of the following lexical verb is incompatible with maintaining that auxiliary selection is driven by the syntax here. Though the data provided by Manzini and Savoia invariably display the string auxiliary+participle, some tests come to mind which further demonstrate that the phonological condition takes precedence. For instance, one should check what happens when one and the same verb lexeme possesses two variants with C-/V—such doublets occur in Apulian dialects (e.g., Alt. (at)tʃәˈvars ‘eat one’s fill’)—and when another word is interposed between the auxiliary and the participle. If the phonological condition has priority, we expect (a) doublets to behave differently in spite of their semantic/syntactic identity, and (b) the interposition of a C-/V-initial word to affect have/be selection in the third-person singular. Starting with the second test, this is not easy to carry out, since temporal adverbs are the best candidates, but Apulian dialects are more restrictive than standard Italian when it comes to allowing their collocation in the relevant position, as seen in (16)–(17) for Gravinese and Ruvese: (16) a. l= ˈaɟɟә / ʊ= ˈsɔ it= have.1sg it= be.1sg b. **l ˈaɟɟә ʊ ˈsɔ it have.1sg it be.1sg ‘I did it again.’

ˈffatt done arˈrзːtә done

(17) a. ˈkεːra waɲˈɲau̯nә sә= that:f.sg girl(f).sg refl.3= arˈrεːtә (Ruv.) again b. **ˈkεːra waɲˈɲau̯nә sә= that:f.sg girl(f).sg refl.3= ˈʃɪu̯tә (Ruv.) gone ‘That girl has gone away again.’

arˈrзːtә (Grv.) again ˈfattә (Grv.) again

n= ˈε hence= be.3sg

/

ˈɔːvә ˈʃɪu̯tә have.3sg gone

n= ˈε hence= be.3sg

/

ˈɔːvә arˈrεːtә have.3sg again

However, at least the local counterparts of It. già ‘already’, sempre ‘always’, and mai ‘(n)ever’ can be interpolated and, when they separate the auxiliary from the participle, they affect selection of ‘has’/‘is’ as predicted:

the morphological nature of person-driven auxiliation

(18) a. maˈrɪ nan dz= ˈaːvә Maria neg refl.3= have.3sg b. maˈrɪ nan dz= ˈe Maria neg refl.3= be.3sg ˈddo (Grv.) here ‘Maria has never sat here.’ c. pәpˈpɪˈnә s= ˈe Peppino refl.3= be.3sg ˈddε (Grv.) there ‘Peppino has always sat there.’ (19) a. la ˈportә s= def:f.sg door(f).sg refl.3= ˈddʒε (Grv.) already b. la ˈportә s= def:f.sg door(f).sg refl.3= aˈpεrtә (Grv.) opened.f ‘The door has already opened.’

227

/**ˈe asˈsɪːsә ˈmε ˈddo (Grv.) be.3sg sat never here ˈmmε / **ˈaːvә ˈmε; asˈsɪːsә never have.3sg never sat

ˈssemb / **ˈaːvә ˈsemb asˈsɪːsә always have.3sg always sat

ˈaːvә / **ˈe have.3sg be.3sg

ˈe be.3sg

aˈpεrtә opened.f

/**ˈaːvә ˈddʒε have.3sg already

In the Gravinese examples (18a) and (19a) the adverb follows the V-initial participle, which results in selection of ˈaːvә, while in (18b–c) and (19b) ˈmε, and ˈddʒε precede the participle and the auxiliary form is consequently switched. The same situation is observed in Bitettese: (20) a. ˈfraːtәmә ˈεːvә arrәˈvɘːtә (Bit.) brother=1sg have.3sg arrived ‘My brother has arrived.’ b. ˈfraːtәmә ˈε ˈddʒɘ arrәˈvɘːtә (Bit.) brother=1sg be.3sg already arrived ‘My brother has already arrived.’ Note that my Bitettese consultants have the alternative form ˈεːvә in (20), distinct from ˈaːvә recorded by Manzini and Savoia. This form results diachronically from a blend of the relevant forms of the two auxiliaries (ˈaːvә < Lat. habet ‘have.3sg’ vs ˈε < Lat. est ‘be.3sg’), a fact often observed in Italo-Romance dialects with mixed auxiliation (cf. Manzini and Savoia 2005:III:17f.; Cennamo 2010; Loporcaro 2016:814f.), itself evidence for their belonging to one and the same lexeme. However, contrary to, for example, the blended form ˈε ‘be=have.2sg’ reported by Manzini and Savoia (2005:II:770, III:9f.) for the dialects of Cerano and Trecate— which functions both as a copula and verb of possession—Bitettese ˈεːvә belongs

228

michele loporcaro

synchronically to the paradigm of have, as shown by its occurring in free variation with ˈaːvә (21a–b) as the third-person singular of the periphrastic future (also a deontic periphrasis), which is regularly formed with have (cf. the further forms from the same paradigm in (21c–d), Manzini and Savoia 2005:II:725):1⁶ (21) a. ˈjiddә s ˈεːvә a llaˈvɘ lә 3sg refl.3= have.3sg to wash.inf def.f.pl ‘He will/has to wash his hands.’ b. la ˈportә aŋˈɡoːrә l ˈaːv a def.f.pl hand(f).pl still it= have.3sg to ‘The door, (s)he still has to open it.’ c. u ˈaɟɟ /**ˈsɔ a ˈfɘ (Bit.) it have.1sg be.1sg to do.inf ‘I will/have to do it.’ d. u ˈa /**ˈsi da ˈfɘ (Bit.) it have.2sg be.2sg to do.inf ‘You will/have to do it.’ e. ˈjiddә ˈε /**ˈεːvә ˈjirtә (Bit.) 3sg be.3sg have.3sg tall ‘He is tall.’

ˈmɘːnә (Bit.) hand(f).pl aˈprai ̯ (Bit.) open.inf

For comparison, (21e) shows the third-person singular form of the copula, which is distinct from have in all persons.1⁷ In Ruvese, finally, my consultants do not rule out ˈɔːvә ‘has’ remaining even before ˈmε and ˈddʒa.1⁸ (22) a. maˈrεjә s= ˈɔːvә / **s= ˈε Maria refl.3=have.3sg refl.3=be.3sg ? b. maˈrεjә s= ˈε / s= ˈɔːvә Maria refl.3=be.3sg refl.3=have.3sg ‘Has Maria already sat down?’ (23) a. pәpˈpεin̯ ә Peppino asˈsεis̯ ә sat

nan neg ˈmε ever

dz= ˈɔːvә refl.3= have.3sg ˈsɔːp a ˈkkεːra on to that:f.sg

asˈsεis̯ ә sat ˈddʒa already

ˈddʒa (Ruv.) already asˈsεis̯ ә (Ruv.) sat

/**dz= ˈε refl.3= be.3sg ˈsɪз̯ddʒә (Ruv.) chair(f).sg

1⁶ The local outcomes of Lat. habere ‘have’ have been generally ousted by those of tenere ‘hold, keep’ in their functions as verb of possession and support verb across all of this area, e.g., Bit. ˈteŋɡә/**ˈaɟɟә ˈtrend ˈannә ‘I am (lit. “have”) thirty’. 1⁷ This answers the legitimate concern aired by Corbett (2012:185) in commenting on Loporcaro’s (2007) data from one mixed-auxiliation dialect of Abruzzo: ‘we would need to be reassured about the behaviour of the two verbs independently’. 1⁸ My Ruvese consultants share the same judgements reported by Manzini and Savoia in (11, 21), with complementary distribution of ‘has’/‘is’ when the auxiliary immediately precedes the participle, but the (at least marginal) acceptability of preconsonantal ˈɔːvә ‘has’ in (22b) seems to indicate an incipient weakening of the phonological conditioning (thanks to Adam Ledgeway for discussing this point). This auxiliary form is usually ˈaːvә in connected speech, where the velarization that affected stressed /a/ tends to regress, but can also be realized as ˈɔːvә, as recorded by Manzini and Savoia, in slow speech.

the morphological nature of person-driven auxiliation

229

b. pәpˈpεin̯ ә nan dz= ˈε ˈmmε /dz= ˈɔːvә ˈmε Peppino neg refl.3= be.3sg ever refl.3= have.3sg ever asˈsεis̯ ә ˈsɔːp a ˈkkεːra ˈsɪз̯ddʒә (Ruv.) sat on to that:f.sg chair(f).sg ‘Peppino has never sat on that chair.’ The further test with verbal doublets whose lexical shape comes in two variants also gives the same result for the one relevant verb I was able to identify for Ruvese: (24) a. pәpˈpεin̯ ә s= ˈɔːvә Peppino refl.3= have.3sg b. pәpˈpεin̯ ә s= ˈε Peppino refl.3= be.3sg ‘Peppino has eaten his fill.’

attʃәˈvɔːtә filled ttʃәˈvɔːtә filled

ˈbbuːnә (Ruv.) good.m ˈbbuːnә (Ruv.) good.m

8.4 A shape condition on the distribution of ‘has’/‘is’ This undesired, and highly problematic, violation of (13) evaporates under the morpholexical approach to mixed auxiliation advocated here (2a), assuming that split-auxiliary distributions across verb persons are a matter of inflexional morphology, not syntax. Under such an alternative theory, have vs be selection in Tables 8.9–8.10, examples (10)–(12) and in (17)–(24) is best analysed as obeying a shape condition, of the type constraining phrase allomorphy in well-known cases such as French liaison. Such a condition ‘specifies aspects of the phonological shape of i[nflexional]-forms, but “postlexically”—by reference to triggers at least some of which lie outside the syntactic word’ (Zwicky 1986:310). Consider the statement on the derivation of the English prevocalic form of the indefinite article an: It is not part of the lexical entry for the word, because it refers to the following syntactic context. It is not a phonological rule of English, for it applies only to the indefinite article and has no general applicability to phonological domains. It is a condition on shape that overrides the lexical entry for the indefinite article and stipulates that another shape is called for. (Pullum and Zwicky 1988:262) This perfectly suits our case too:1⁹ the complementary distribution of 3sg ˈε ‘is’ and ˈɔːvә/ˈaːvә ‘has’ prevocalically does not follow from any phonological rule of 1⁹ The parallel is quite compelling indeed: compare (15)–(23) with English a closer union/an ever closer union.

230

michele loporcaro

the dialects but is sensitive to the postlexical phonological context, in a way syntactic rules cannot be. Rather, it has to be analysed as a matter of phrase allomorphy, obeying a shape condition, but this is only possible under the assumption that the choice between forms of have and be in a mixed paradigm has nothing to do with the syntax, as maintained in (2a). Conversely, such an unproblematic description of the observed data would be out of reach for theories which derive the specific forms of auxiliaries from syntactic factors such as Kayne (1993) or Manzini and Savoia (2005).

8.5 From complementary distribution via overabundance to heteroclisis and suppletion Since I have argued that the syntactic account of mixed auxiliation makes a typological oddity of the Apulian data analysed here, I owe the reader a demonstration that the present analysis is free from such flaws and explains these highly unusual data in terms of facts that are independently known to occur in the languages of the world. Indeed, under (2a), dialects such as those of Bitetto, Gravina, and Ruvo showing the mixed auxiliation patterns in Tables 8.6–8.8 are one more example of the type of change described by Maiden (2004b:227) in his inquiry into ‘the genesis of suppletion’, according to which ‘distinct lexemes come to acquire the status of synonymous paradigmatic alternants’. have/be entering mixed auxiliation patterns are one more case where ‘lexemes become allomorphs’ comparable to that which Maiden (2004b:242) adduces from some Romanian dialects of Maramures¸ (in north-western Romania) as illustrated in Table 8.11, where the verb go mixes forms across different persons which, in standard Romanian, belong to the two distinct lexemes merge and se duce (25). Table 8.11 The verb go in some Romanian dialects of Maramures¸ (Maiden 2004b:242) 1sg 2sg 3sg 1pl 2pl 3pl a. Funda˘tura mə duk t jə duc səˈduce ˈmɛrem ˈmɛrets sə duk b. Ta˘ureni mə duk ce duʃ ˈmɛre ˈmɛrem ˈmɛrets sə duk

(25) a. Liana merg-e acasă. (Ro.) Liana go-3sg home b. Liana se duc-e acasă. (Ro.) Liana refl.3= bring-3sg home ‘Liana is going home.’

the morphological nature of person-driven auxiliation

231

Our mixed paradigms are also instances of lexeme splits just like that in Table 8.11. Prior to the change which brought mixed auxiliation systems into being, have and be must also have been in syntactically determined complementary distribution in (the diachronic ancestors of) these dialects, as shown by abundant comparative evidence. The change pushed them into the realm of inflexional morphology.2⁰ Following the lines of argument developed so far, we can add that, if in mixed systems have/be are a matter of inflexional morphology, free variation between them within a paradigm cell can be equated to overabundance (i.e., ‘multiple forms realizing the same cell’, Thornton 2011; 2012) while person-driven alternation can be viewed as heteroclisis (viz. ‘the property of a lexeme whose inflectional paradigm involves two or more distinct inflection classes’, Stump 2006:279);21 on overabundance and heteroclisis, see also §1.2.3 and chapter 11 in this volume. Since ‘[h]eteroclitic paradigms are exceptions set against a backdrop of nonheteroclitic paradigms’ (Stump 2016:184), it is essential to keep in mind that, even in dialects showing just one mixed auxiliation pattern in all syntactic constructions (such as Aquilano in (8f)), the whole paradigms of be and have remain as a relevant ‘backdrop’ in their roles as a copula and future/deontic auxiliary respectively, as shown for Bitettese in (21). A comparable case from Italian verb inflexion is that of compiere/compire ‘fulfil’, which displays—as it has ever since the earliest documentation of the language—forms of both classes 2 (-ere) and 3 (-ire) and hence, together with near-synonymous adempiere/adempire, ‘represent the best example of overabundance in all cells’ in Italian (Thornton 2018:319). Class 3 forms were much rarer in old Italian, as documented in Table 8.12(a) for a selection of paradigm cells with the number (in brackets) of occurrences in the OVI corpus. In contemporary standard Italian things have changed, as seen in Table 8.12(b), where, in addition to grammaticality judgements, I add figures from the CoLFIS in brackets (extensive data from the larger La Repubblica corpus are provided in Thornton’s (2018:317–319) study of this and other Italian overabundant verbs): most class 3 forms have fallen out of use, which is not unexpected, given that, for this lexeme, they were rarer from the outset. However, this is not without exceptions. In particular, in the imperfect indicative -iva is favoured over -ieva, and in

2⁰ This change has not completely eliminated all traces of the former system, as argued in Pescarini and Loporcaro (forthcoming) where it is shown that, in mixed systems, uniform selection of be in all persons exclusively occurs in unaccusatives, never in unergatives, while the reverse is true of uniform selection of have. This and other distributional skewings are due to inertia and are to be interpreted as residues of a once productive former stage of syntactically conditioned auxiliary choice. 21 The former terminological point is made by Štichauer (2018:16; 2019:91) in two papers where he mentions Bentley and Eythórsson (2001) as well as my own work on mixed auxiliation, although I could not find in those papers any acknowledgement of the fact that the idea itself, eloquently advocated there, that mixed auxiliation belongs to inflexional morphology and is best analysed as involving one (split) lexeme is put forward in exactly these terms (except for the label ‘split lexeme’, for which, see Corbett 2012; 2015) in Loporcaro (1999b:213); Bentley and Eythórsson (2001:70f.); Loporcaro (2001:462).

232

michele loporcaro Table 8.12 Compiere/compire ‘fulfil’ Italian a. Grammar 1 (attested, OVI corpus, 13th–14th centuries) b. Grammar 2 (my own; CoLFIS)

Class 2 compiere (590) compie (599) compiete (5) compieva (14) compiesse (80) compiuta (639) compiere (115) compie (43) ?? compiete (Ø) ?? compieva (Ø) compiesse (Ø) compiuta (54)

Class 3a (-isc-) compire (4) compisce (4) compite (1) compiva (Ø) compisse (25) compìta (113) ?? compire (Ø) ?? compisce (Ø) compite (Ø) compiva (1) compisse (Ø) **compìta (Ø)

‘fulfil’ infinitive 3sg prs_ind 2pl prs_ind 3sg ipfv_ind 3sg ipfv_sbjv pst_ptcp_f.sg infinitive 3sg prs_ind 2pl prs_ind 3sg ipfv_ind 3sg ipfv_sbjv pst_ptcp_f.sg

the present indicative there is a similar effect for the second-person plural, yielding the heteroclitic paradigm in Table 8.13. Table 8.13 Compiere (present indicative): from overabundance to heteroclisis 1sg 2sg 3sg 1pl 2pl 3pl Class 2 compio compi compie compiamo ?? compiete compiono Class 3 **compisco **compisci **compisce compiamo compite **compiscono

Note that compio, compi, compiono can, in principle, be analysed as inflected according to the subclass of class 3 without augment (compare riempire ‘fill’: 1sg riempio, 2sg riempi, 3pl riempiono), which is not at adds with the fact that compire class 3 forms (also) inflected according to the complementary subclass showing augment in the present (compisce, never **compe; cf. the double inflexion in, e.g., esegue ‘carries out’ alongside the now outdated eseguisce, the only form documented in the OVI corpus of old Italian).22 Moreover, if not including the augment in the N-pattern cells (the singular plus the third-person plural of the present tense, see Maiden 2018a:ch.6), the first- and second-persons singular are non-distinct for class as they display superstable endings (Wurzel 1984:139–142) common to all classes, which is the case also for the first-person plural. In conclusion, once the many paradigm cells hosting syncretic forms are factored out, compiere shows a tendency towards heteroclisis, which arranges weakly

22 Thornton (2018:316) considers all present indicative forms except the second-person plural as non-distinct for class.

the morphological nature of person-driven auxiliation

233

suppletive allomorphs according to the N-pattern (at least if the N-pattern cell forms are interpreted as inflected according to class 2). This helps to explain why in the present indicative class 3 compite, despite being rarer in old Italian, emerged as the winning form while the rest of the present stayed as it was (compie, etc.), be it class 2 or 3: with the sole exception of the participle, it was the whole partition class complementary to the N-cells which won out. This confirms that an autonomous morphological organizing principle such as the ‘N-pattern’ can override sheer token frequency in driving morphological change, which once again emerges, as Maiden (2018a:16) puts it, ‘as a way of diagnosing the fact that some phenomenon already diagnosed as “morphomic” really exists in the minds of speakers’.23 The parallelism with mixed auxiliation is strengthened by the fact that the latter, as observed by Štichauer (2018:17f.; 2019:88–93), displays the N-pattern morphomic distribution in several Italo-Romance dialects.

8.6 The limits of morphology I have not commented on (3c) yet: this is an approach to (non-mixed) auxiliation which, within a theory of periphrasis, claims that auxiliary selection is literally a matter of inflection class: just as different classes of lexemes may trigger the use of distinct rules of synthetic exponence for the expression of the same feature, they may likewise trigger the use of distinct rules of periphrastic exponence. To take a concrete example, let us consider the situation in French. (Bonami 2015:97, followed by Štichauer 2018:91 and Bach and Štichauer Chapter 7 in this volume) This means that selection of avoir/être ‘have/be’ in French is not by syntactic rule. The reason why French is treated this way is that just ‘a few dozen intransitive verbs use être “be”’. Thus, ‘auxiliary selection tends to correlate with lexical semantics, but has to be recognized as partially arbitrary’, which ‘is reminiscent of the status of inflection classes: similar lexemes tend to cluster in the same classes, but there are exceptions’. Note that mention of lexical semantics alone and the references cited, starting with Sorace (2000), make it clear that the author endorses a non-syntactic view of unaccusativity (for discussion, see Loporcaro 2015; 2016:817f.). However, there is an alternative view of unaccusativity (in the wake of 23 Note that the non-productivity of class 2 cannot be invoked as an alternative explanation, since the non-augmenting subclass 3a is just as non-productive: had class productiveness been the driving force in the change, one would have ended up with a consistent class 3 regular lexeme 1sg **compisco, 2sg **compisci, 2pl compite, etc.

234

michele loporcaro

Perlmutter 1978; Rosen 1984) and, elaborating on the latter, there is an available syntactic account of the French auxiliation facts exemplified in (26). (26) a. Le soleil est apparu à 8 heures. (Fr.) def.m.sg sun(m) be.3sg appeared at 8 hour.pl ‘The sun appeared at 8.’ b. Le soleil a disparu à 8 heures. (Fr.) def.m.sg sun(m) have.3sg disappeared at 8 hour.pl ‘The sun disappeared at 8.’ La Fauci (2000) proposed an analysis in which unaccusative advancement, responsible for be selection throughout Romance, occurs in the predicate sector of the auxiliated participle, as in Italian, for French be-selecting unaccusatives such as apparaître ‘appear’, which are known to represent the conservative option (see, for example, Heidinger 2015:282f. and references therein on the gradual replacement of être ‘be’ by avoir ‘have’). By contrast, verbs such as disparaître ‘disappear’ have switched to have but still test as unaccusatives on other diagnostics such as the availability of participial absolutes: (27) Une fois le soleil disparu derrière les arbres, de gros one time the sun disappear.ptcp behind the trees, indef big nuages sombres commencent à se regrouper dans le ciel. (Fr.) clouds dark start.3pl to refl= gather.inf in the sky ‘Once the sun has disappeared behind the trees, large dark clouds begin to gather in the sky.’ (http://exo-terra.com/fr/expeditions/expedition2007_journal.php) For this latter subclass, La Fauci’s analysis proposes that unaccusative advancement has moved to the predicate sector of the verb root. This hypothesis accommodates the (increasing) depletion of the set of be-selecting verbs—which goes even further in Laurentian French and in many oïl dialects (Loporcaro 2016:812f.)— without losing the host of unaccusativity-based generalizations exemplified in (27), which would be left unexplained under the assumption, at the core of the morphological approach (3c), that have- vs be-selecting verbs belong to ‘two [morphologically defined, m.l.] classes of lexemes with two different realizations’ (see §7.2.2 in this volume). With this proviso, the French auxiliary selection rule finds its natural place among the perfective auxiliation rules so far described for the Romance languages (modified from Loporcaro 2011:82; 2 = direct object, P = predicate; the reader is referred to the cited source for details): (28) Perfective auxiliation in Romance (non-mixed systems) The perfective auxiliary is esse ‘be’ iff the final 1: a. is a 2 [Italian] b. is the first 2 in the clause [Logudorese Sardinian] c. is a P-initial 2 [old Romanesco]2⁴ 1⁷3 P-initial is a technical term of Relational Grammar (cf. Davies and Rosen 1988) meaning ‘initial in the predicate sector’, in turn defined as the set of strata in which a given predicate bears the P-relation.

the morphological nature of person-driven auxiliation

235

d. is: i. a 2; ii. not the P-initial 1 of the auxiliated [old Florentine] participle e. is: i. a 2; ii. not multiattached [Engadinian, Leccese] f. is a 2 in the P-sector of the auxiliated participle [French] Otherwise the perfective auxiliary is habere ‘have’. The evidence from mixed auxiliation systems discussed in §§8.2–3 militates in favour of this view of French auxiliation. In fact, the analysis proposed in §8.4 for have/be selection in the third-person singular in Bitettese, Gravinese, and Ruvese in terms of a shape condition, is based crucially on the assumption that such selection is a matter of inflexional morphology. Assuming that the same is true of French, the cautious statement of Bonami’s (2015:97) that French auxiliary selection ‘is reminiscent of the status of inflection classes’, would obscure the difference between non-mixed systems, such as French, Italian, or the remaining examples in (28), and mixed systems. Only in the latter, in fact, do we expect that such cases may occur at all, while the impossibility of shape conditions switching forms of have/be in non-mixed systems is correctly predicted under (3a), the syntactic approach to auxiliation endorsed here. To exemplify, consider the Logudorese Sardinian examples in (29) (from La Fauci and Loporcaro 1993), which show how perfective auxiliary may be affected by sandhi processes in non-mixed auxiliation systems: (29) Bonorvese (Logudorese Sardinian) a. Pedru est partidu [ˈεs b. Ch’est bénnidu Pedru [ˈεl c. B’est annadu Pedru [ˈεst d. **B’at annadu Pedru [ˈað e. Ch’at bénnidu unu pastore [ˈa

palˈtːiðu] ‘Pietro has left.’ B ˈbenniðu] ‘Pietro has come (here).’ B anˈnaːðu] ‘Pietro has gone there.’ B anˈnaːðu] **H ˈbbenniðu] ‘There came a shepherd.’ H

The third-person singular of be, phonologically /ˈεst/ as seen in the spelling on the left-hand side, is realized in different ways depending on the following initial segment: it undergoes /t/-deletion before consonants (29a–b), and its /s/ is realized as [l] before voiced consonants (29b), while its underlying form /ˈεst/ emerges unmodified before a vowel (29c). Note however that this auxiliary form does not change to /ˈat/ (the third-person singular of have) prevocalically, as shown by the ungrammaticality of (29d). The latter may occur with the same unaccusative verb, but only if selected in the appropriate syntactic context, which is the case (given the auxiliation rule proposed in La Fauci and Loporcaro 1993:164) in impersonal presentative clauses whose argument is indefinite, as in (29e). On the contrary, when the argument is definite (as in 29a–c), the clause cannot be impersonal and

236

michele loporcaro

auxiliary be is selected, whatever the initial segment of the word following the auxiliary. The reason why Sardinian patterns in this way is that we are dealing with a well-behaved Romance variety in which auxiliation depends on syntax alone: this language, like most other Romance varieties, does not feature any free variation, mixing, or person-driven alternations in the selection of have/be. The last of these is precisely what happens in the three Apulian dialects we have focused on in this chapter.2⁵ This is a rare phenomenon, which, however, need not be misconceived as an otherwise unheard-of oddity in which the phonology drives a syntactic process, an error from which the morpholexical approach to person-driven auxiliation protects us. By the same token, the syntactic approach to auxiliation in (3a) automatically explains the reason why the phonologically selected have/be alternation seen in the three dialects of central Apulia cannot extend to well-behaved non-mixed systems such as Bonorvese (29), French (26), or Italian (Table 8.1(a)), an undisputable fact that however becomes mysterious under Bonami’s (2015) morphological account of auxiliary selection (3c).2⁶

8.7 Conclusion In conclusion, the bottom line of the present contribution is that, while it is beyond doubt that periphrases fall naturally across the boundary between syntax and morphology (for in-depth discussion, see Chapters 1 and 2 in this volume) and a satisfactory treatment must take both syntax and morphology into account, one must be careful to treat each and every aspect of perfective auxiliary selection with the appropriate tools. In essence, what I have shown is that, in the domain of perfective auxiliation, one has to assign to (inflexional) morphology what belongs to morphology—in our case, person-driven auxiliary selection—and to syntax what belongs to syntax, that is, all the rest.

2⁵ The geographical position of the three dialects (see Map 8.1) suggests that this was a shared innovation, which reduces this to just one isolated exception, the only one attested to date in the hundreds of Romance varieties whose auxiliary selection systems have been investigated. 2⁶ This is a new argument adding to the others already developed to rebut non-syntactic analyses of Romance auxiliary selection in Loporcaro (2015). Not unlike the semantic accounts discussed there, Bonami (2015) is forced to lump together the be-selecting unaccusatives with reflexives, which uniformly take être ‘be’, thus missing the generalization which, more economically, states just one syntactic rule (see (28f)) accounting for be selection in all the contexts where it occurs.

the morphological nature of person-driven auxiliation

Ruvo di Puglia Gravina in Puglia

Bari Bitetto Altamura

Map 8.1. The towns in the province of Bari (Apulia) whose dialects are mentioned in the chapter.

237

PART IV

ANALYSIS VS SYNTHESIS

9 The loss of analyticity in the history of Romanian verbal morphology Adina Dragomirescu, Alexandru Nicolae, and Rodica Zafiu

9.1 Introduction Addressing the general relation between syntheticity and analyticity in the transition from Latin to Romance, this chapter analyses the diachronic development of verbal periphrases in the passage from old to modern Romanian, including its dialectal varieties.1 Only Daco-Romanian and its regional varieties are considered in this study; the sub-Danubian historical dialects have not been taken into account. Our analysis starts from the empirical observation that in the passage from old to modern Romanian, despite the loss of some inherited synthetic formations (see Maiden 2018a:29–43), there is a subset of novel periphrastic formations which disappeared from the standard language, with just some relics preserved dialectally. While traditional scholarship argued that synthetic forms were replaced tout court by analytic formations in the transition from Latin to Romance (cf. Schlegel 1818 and many references thereafter), later work argued against such a radical view (Schwegler 1990; Ledgeway 2012 and, especially, 2017a),2 focusing on the cross-Romance variation in this area and highlighting the particular developments of each Romance variety in turn.3 In this respect, Romanian developed in the 1 Note that we use the term ‘periphrasis’ here and throughout in the sense of ‘categorial periphrasis’ as defined by Haspelmath (2000:656), although, as we shall see, many of the periphrases discussed do compete with inflexional verb paradigms. We take this opportunity to wish Martin La mulți ani!, thanking him for his scholarly work on the Romanian language and for being one of its best foreign ambassadors. We would like to express our gratitude to the editors, not only for reading our chapter and suggesting many revisions, but also for their initiative in putting the present volume together. We also thank our colleague Ionuț Geană who made many helpful comments on an earlier version of the chapter. 2 Cf. also Coseriu’s (1987; 1988) distinction between ‘internal’ and ‘external’ structures, around which he develops the hypothesis that that the Romance languages differ from Latin by an iconic typology whereby relational concepts (external structures) receive relational, ‘syntagmatic’ (i.e., analytic) exponence, and non-relational concepts (= internal structures) receive non-relational, ‘paradigmatic’ (i.e., synthetic) exponence. As Coseriu himself admits, this principle is not without exceptions. See also the discussion in §1.2.1 in this volume. 3 The disappearance of newly coined periphrases is found in other Romance languages as well (see, for example, Squartini 1999:27–29; Buridant 2000:357f.). In the case of standard French, the demise Adina Dragomirescu, Alexandru Nicolae, and Rodica Zafiu, The Loss of analyticity in Romanian verbs. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Adina Dragomirescu, Alexandru Nicolae, and Rodica Zafiu (2022). DOI: 10.1093/oso/9780198870807.003.0010

242

adina dragomirescu, alexandru nicolae, and rodica zafiu

course of its history a number of novel periphrastic formations in the verbal domain which, although continuing in many other Romance varieties (cf. Squartini 1998), did not survive into the modern language; older synthetic formations characterized by greater morphemic simplicity (interpreted here to mean involving a smaller number of morphemes) prevailed diachronically. The loss of this subset of periphrases raises some non-trivial problems concerning the broader Romance typological tendencies and the relation between syntheticity and analyticity/periphrasis. Consequently, the main research question for which our account seeks to provide an answer is why some novel periphrastic formations disappeared from the language, while others have survived to the present day. The first three sections of this chapter present an empirical overview of changes in the verb system from the perspective of the interplay between syntheticity and a subtype of analytic structures (i.e., periphrastic forms), throwing light on the differences between old Romanian (§9.2) and modern standard (§9.3) and dialectal (§9.4) Romanian. These are then addressed from a theoretical perspective (§9.5); our analysis will show that the auxiliaries of these now defunct periphrases share a set of properties concerning the encoding of tense, aspect, and mood, in particular having a richer feature matrix which sets them apart from the auxiliaries of the surviving periphrases, thereby accounting for the demise of the former subset of periphrases. For the concepts ‘analytic(ity)’ and ‘periphrastic/periphrasis’, we adopt the framework established by Ledgeway and Vincent (see Chapter 1 in this volume). Accordingly, ‘periphrasis’ generally refers to a set of analytic forms that realize the cells of a paradigm that are typically realized by inflexional forms elsewhere in the paradigm; note that not all patterns traditionally included under the heading ‘analytic’ may be considered ‘periphrastic’ (see the discussions in §1.2.1 and §2.3 in this volume; see also §9.5.3). In this chapter we focus on periphrases with tense, aspect, and mood auxiliaries (as opposed to auxiliaries whose feature matrix encodes other grammatical categories, such as voice). We adopt a generally accepted definition of auxiliaries such as that given in Anderson (2006),⁴ which is convergent with the general view on auxiliaries in work on grammaticalization (cf. Heine 1993; Kuteva 2001). Although our analysis is empirically oriented, we adopt the spirit of the generative idea (developed by Cinque 1999 and much subsequent literature: see Ledgeway 2012 and Schifano 2018 for cross-Romance overviews, and Nicolae 2015; 2019 for Romanian) that the inflexional domain (namely, the

of many of the French surcomposé (viz. double compound) forms might have been favoured by the fact that they were largely ignored or condemned by French grammars (Ayres-Bennett and Carruthers 1992; Carruthers 1994). ⁴ ‘[A]uxiliaries are not discrete entities per se but rather mono-clausal form–function combinations occupying a non-discrete space on several large form–function continua that include serial verb constructions, clause-chaining, and verb plus complement clause combinations on the one hand and tense-aspect-mood affixes on the other’ (Anderson 2006:4).

loss of analyticity in romanian verbs

243

structural area of the clause hosting tense, aspect, and mood information) is made up of three hierarchically ordered dedicated (fields of) projections: mood > tense > aspect.⁵ Our analysis endorses a compositional view (see Comrie 1985:76 on compositionality in the encoding of tense, aspect, and mood, and Vincent 2015 on compositionality and language change), assuming that there is a division of labour between the components of a periphrasis with respect to the encoding of tense, aspect, and mood information.

9.2 Periphrastic forms in old Romanian Old Romanian is characterized by a larger number of analytic verb forms than modern Romanian,⁶ some of which have been considered calques due to the translation of learnèd texts from Old Church Slavonic or Greek (see Manoliu 1959; Arvinte 1993; but cf. Zafiu 2016:45). However, the fact that they have been preserved in regional varieties of Romanian shows that they are the products of natural and internal linguistic change, and allows us to surmise that they were genuinely present in actual usage in old Romanian.⁷ As we shall see, a striking formal characteristic of old Romanian is the systematic correspondence between constructions with fi ‘be’ plus past participle and fi ‘be’ plus gerund (present participle) (see §§9.2.2–9.2.3).⁸

9.2.1 Periphrases with the lexical verb in the infinitive In the sixteenth century the set of periphrases with the lexical verb occurring as a bare infinitive includes the voi-future (1a) and the aș-conditional (2a), as well as periphrases headed by the auxiliary vrea ‘want’. The direct selection of a bare infinitive generally indicates a high degree of grammaticalization, and hence the fact that they had grammaticalized in an earlier (and unattested) stage of Romanian (Caragiu Marioțeanu 1969:268). This is in contrast to other complex predicates in the sixteenth century, where the lexical verb is introduced by the infinitival marker a ‘to’. A more recent and less grammaticalized type of future (rare and gradually ⁵ We adopt a traditional, widely accepted view of the tense, aspect, and mood categories (see Comrie 1976; 1985; Palmer 2001). ⁶ In agreement with the Romanian philological tradition (see Timotin 2016 and references therein), the period referred to as ‘old Romanian’ stretches roughly from the beginning of the sixteenth century (the period of the earliest attested written documents) to the end of the eighteenth century. The year 1780 is considered the borderline between old and modern Romanian. ⁷ The idea that these analytic structures are productive and cognate with other Latin–Romance patterns—with learnèd Old Church Slavonic or Greek influences functioning as a catalyst at a certain point—was championed by Frâncu (1983–1984; 2012; 2014). ⁸ Traditional grammars describe these forms in -nd as a gerunziu ‘gerund’. As formatives of periphrases, their actual function is that of a present participle (see Nicula 2013).

244

adina dragomirescu, alexandru nicolae, and rodica zafiu

abandoned towards the end of the seventeenth century, see Zamfir 2007:232–241) was based on the a-infinitive (1b). The two types of future developed—through two of the main pathways of grammaticalization (see Bybee, Perkins, and Pagliuca 1994:254–264)—from constructions with vrea ‘want’ (the agent-oriented modality of desire) and avea ‘have’ (the agent-oriented modality of obligation, derived from possession). The former was extremely frequent and clearly temporal, while the latter remained rare and preponderantly modal. In the sixteenth and seventeenth centuries the auxiliary (1a) was not yet differentiated from lexical ‘want’ from a formal point of view.⁹ (1)

a. vedeaveri slava lu Dumnezeu see.inf= want.aux.prs.ind.2sg glory.def gen God ‘you will see the glory of God.’ (CC2 .1581:99) b. de acmu oameni ai a vâna from now people.pl have.aux.prs.ind.2sg to hunt.inf ‘from now on you will hunt people.’ (CC2 .1581:367)

There was significant variation in the inventory of periphrases based on the future which initially had the value of a future-in-the-past, later becoming conditionals. The most frequent analytic conditional, which is preserved in present-day Romanian, involves the auxiliary aș/ai/ar(ă)/am/ați/ar(ă) and the bare infinitive (2a). Despite many etymological controversies (including Weigand 1896; Philippide 1927; Rosetti 1932:104; Ivănescu 1980/2000:163; Elson 1992; see Zafiu 2017 for a review), the most plausible hypothesis is that the auxiliary originates from past forms (imperfective past forms, perhaps contaminated with forms of the synthetic perfect) of the future auxiliary vrea ‘want’. In modern Romanian, the standard analytic form of the conditional (e.g., aș merge ‘aux.cond.1sg go.inf’) only has modal values; presumably, in an earlier period it must have also had a temporal value (as a future-in-the-past),1⁰ similarly to its equivalents in other languages (see also §9.5.2).11 A variant of the construction uses the ‘transparent’ auxiliary vrea ⁹ In the glosses the etymologically transparent auxiliaries have been translated in accordance with the basic meaning of their original lexeme and are in small caps (e.g., future auxiliary veri is glossed as ‘want.aux’), whereas etymologically opaque auxiliaries, whose relation to their lexical source is no longer transparent due to various historical factors (such as phonetic erosion, morphological attrition), are simply given grammatical glosses (e.g., conditional auxiliary aș is glossed as ‘aux.cond’). 1⁰ The existence of a future-in-the-past which developed counterfactual and conditional values is well attested in the Balkan languages, both diachronically (for Greek, see Markopoulos 2009:216–218, and, for Bulgarian, Kuteva 2001:109) and synchronically. The phenomenon is not conditioned by the absence of a sequence of tense rule or of the so-called Double Access Reading: even if the future-inthe-past is not determined by a superordinate tense (cf. D’Hulst, Coene, and Avram 2004:357), it is permitted and attested in texts. 11 Some occurrences of this analytic form can be found in the earliest attested Romanian texts (from the sixteenth century), in contexts where it is impossible to assign them a conditional meaning. Densusianu (1938:231f.) interpreted them as mere translation errors (influenced by their Hungarian and Latin sources) involving uses of analytic conditional forms instead of past tense forms. However, it is also possible to interpret them as future-in-the-past periphrases used within a narrative context.

loss of analyticity in romanian verbs

245

‘want’ in the imperfective past: vrea/vreai/vrea/vream/vreați/vrea + bare infinitive (2b), with a contextually dependent interpretation, either a future-in-the-past or a conditional. (2)

a. eu naș fi derept I not= aux.cond.1sg be.inf right ‘I would not be right.’ b. minciuna amu vrea fi aiavea lie.def then want.aux.3sg be.inf real ‘then the lie would be real.’

(CC2 .1581:93)

(CC2 .1581:186)

A third, even rarer, analytic form was based on the volitional auxiliary in the periphrastic perfect (am vrut/ai vrut/au vrut/am vrut/ați vrut/au vrut ‘have + want.ptcp’) plus the bare infinitive (3).12 (3)

au vrut vrea amu a munci have.aux.prs.ind.3pl want.aux.ptcp want.inf then to torture.inf ‘then they would have wanted to torture.’ (CC2 .1581:414)

With the exception of the voi-future and the aș-conditional, all the formations discussed in this section disappeared in the passage from old to (standard) modern Romanian. The now defunct analytic conditional with the ‘transparent’ imperfective past form of the auxiliary vrea ‘want’ is functionally equivalent to the canonical aș-conditional, whose form of the auxiliary was no longer transparent. Furthermore, the conditional with vrea ‘want’ in the periphrastic perfect followed by the bare infinitive was replaced by the conditional perfect formed with the ‘opaque’ auxiliary aș and the invariable auxiliary fi ‘be’ plus past participle (see §9.3). Other constructions will not be discussed here in detail as they do not satisfy the criteria for being considered auxiliary-based configurations of the same structural type (see §9.5.3). The avea ‘have’ + a-infinitive future was lost as a consequence of the spread of the equivalent construction based on the subjunctive (am să merg ‘have.aux.prs.ind.1sg sbjv go.prs.1sg (= I will go)’), part of the (incomplete) replacement of the infinitive by the subjunctive in Romanian, a Balkan Sprachbund phenomenon (see also note 12). Another construction based on the substitution of the infinitive by the subjunctive is the vrea + infinitive > vrea + subjunctive (the o să-future diachronically derives from this structure after the first component became invariable, namely o < *oa [wa] < va ‘wants’, cf. Ivănescu 1980/2000:418). These constructions featured either double inflexion, both on the auxiliary and on the lexical verb (e.g., am să merg ‘have.aux.prs.ind.1sg sbjv go.prs.1sg’, voi să merg ‘want.aux.prs.ind.1sg sbjv go.prs.1sg’), or simple inflexion on the lexical 12 Note in (3) that, as an auxiliary, vrea ‘want’ selects a bare infinitive (au vrut vrea), while, as a lexical verb, it selects an a-infinitive (vrea a munci). In the latter construction, the infinitive is systematically replaced by the subjunctive in modern Romanian.

246

adina dragomirescu, alexandru nicolae, and rodica zafiu

verb (o să merg ‘prt sbjv go.1sg’). There are no modal (i.e., grammatical) differences, but only sociolinguistic differences, between these future periphrases in modern Romanian. Only one of each of these periphrastic pairs has been preserved: voi merge ‘want.aux.1sg go.inf’ (but not voi să merg ‘want.aux.1sg sbjv go.1sg’) and am să merg ‘have.aux.prs.ind.1sg sbjv go.prs.1sg’ (but not am a merge ‘have.aux.prs.ind.1sg to go.inf’). This split most probably occurred as an effect of the different diachronic patterns of infinitive-by-subjunctive replacement (note that the infinitive has not been replaced by the subjunctive across the board). Of the conditional periphrases, only the form made up of an opaque auxiliary with no obvious tense inflexion and the infinitive (the aș-conditional) has been preserved in standard Romanian; the structure with an auxiliary overtly inflected for tense (vrea/am vrut) was eliminated over time.

9.2.2 Periphrases with the lexical verb in the past participle The periphrastic perfective past formed with the auxiliary avea ‘have’ (am/ai/a(u)/am/ați/au) in combination with the perfective participle (4) was very frequent and fully grammaticalized in the sixteenth century (au mers ‘have.aux.prs.ind.3sg=go.ptcp’ / mers-au ‘go.ptcp=have.aux.prs.ind.3sg/pl’, namely ‘he/they have gone/went’). The only difference with respect to modernday use of the form was its frequency in relation to the synthetic perfective past form inherited from Latin. The latter form is well represented, especially in learnèd texts, and the opposition between it and the periphrastic perfective past still marked a semantic distinction: the synthetic form is a true preterite (i.e., a past punctual perfective form), whereas the periphrastic form functions more similarly to a present perfect in some contexts, although exhibiting a tendency to overlap with the former (Zafiu 2016:33–35). (4)

Iară acest bolnav nau făcut așa, and this ill not= have.aux.prs.ind.3sg do.ptcp like.this ce și pren târg trecu. but also through town pass.ind.pst.pfv.3sg ‘And this ill man did not do so, but even went through the town.’ (CC2 .1581:56)

There is a consistent set of periphrases built with the auxiliary fi ‘be’ (5)–(6) which are considered to have been influenced by Old Church Slavonic (Sandfeld 1930:149). In old Romanian it was possible to form periphrases with the past participle and the auxiliary fi ‘be’ in almost all modal and temporal forms, with the exception of the synthetic perfective past and the imperative. The periphrasis based on be in the present indicative (which occurs mostly with, but is not restricted to, unaccusatives)—interpreted as a be-based periphrastic

loss of analyticity in romanian verbs

247

present perfect (Dragomirescu and Nicolae 2013) or as a copula verb + predicative complement—is very infrequent (Zamfir 2007:74), hence it has not been considered in our analysis. The auxiliary could exhibit synthetic forms: namely, the imperfective past (5a), the pluperfect (5b), the subjunctive (5c), and the synthetic conditional (5d). The periphrasis with the auxiliary in the imperfective past had the value of a pluperfect (Densusianu 1938:224). The periphrasis with the auxiliary in the subjunctive, which had the value of a perfect subjunctive, was attested in the sixteenth century; the auxiliary later became completely invariable in the nineteenth century (Frâncu 1970). Extremely rare was the construction with the auxiliary in the synthetic conditional. This form, which resulted from a mixture of Latin perfect subjunctive and future perfect (like the Ibero-Romance future subjunctive), disappeared after the sixteenth century (Maiden et al. 2021:296). The ephemeral periphrasis with the auxiliary in the gerund (5e) emerged in the seventeenth century (Niculescu 2016:275), and that with the auxiliary in the infinitive (5f) emerged later (in the eighteenth century, most probably replicating the pattern of the perfect subjunctive, Frâncu 2009:321). (5)

a.

b.

c.

d.

e.

f.

au cetitu ce era have.aux.prs.ind.3sg read.ptcp what be.aux.pst.ipfv.3pl ei scris they.nom write.ptcp ‘he read what they had written.’ (CC1 .1567:2v ) rugu-l văzut [pe drac] fusese briar=cl.acc.3msg be.aux.plpf.3sg see.ptcp dom devil ‘the briar had seen him (= the devil).’ (CSVI .1590–602:60r ) să fim noi iubit pre Dumnezeu sbjv be.aux.sbjv.1pl we love.ptcp dom God ‘that we have loved God.’ (CC1 .1567:25r ) se fure faptu păcatu, lăsaseif be.aux.cond.3sg do.ptcp sin forgive.inf= refl= va lui want.aux.3sg him.dat ‘if he (should have) sinned, he will be forgiven.’ (CV.1563–83:57v ) fiind avut cu dânsul și o căruță be.aux.ger have.ptcp with him also a carriage ‘having also had a carriage with him.’ (DC.a.1794:94) vede-să a fi trăit see.prs.ind.3sg=refl to be.aux.inf live.ptcp înaintea lui Lameh before gen Lameh ‘he seems to have lived before Lameh.’ (ACT.1709:249)

248

adina dragomirescu, alexandru nicolae, and rodica zafiu

There were also constructions with the auxiliary fi ‘be’ in an analytic form: the double compound past (6a), the future perfect (6b), the perfect conditional headed by the opaque auxiliary aș (6c) and its less frequent variant with the transparent auxiliary vrea ‘want’ (6d), as well as the double compound subjunctive (6e), attested in old Romanian from the seventeenth century (Zamfir 2005:415; Frâncu 2010:121–123). We find recurring patterns of formation, so that an analytic form can even have three auxiliary verbs: witness the compound perfect conditional (6f), the compound future perfect (6g), and the compound perfect conditional, built with the opaque aș (6h) and transparent vrea (6i) auxiliaries. (6)

a.

mulți amu oameni au fost făcut many then people have.aux.prs.ind.3pl be.aux.ptcp do.ptcp bunătăți good.deeds ‘then many people had done good deeds.’ (CC2 .1581:44)

b. lucrurile ce va fi lucrat things.def which want.aux.3sg be.aux.inf complete.ptcp pre ceastă lume in this world ‘the things he will have completed in this world.’ (CC2 .1581:120) c.

de n-ară fi pipăit, n-ară if not=aux.cond.3sg be.aux.inf touch.ptcp not=aux.cond.3sg fi nici crezut be.aux.inf neither believe.ptcp ‘if he had not touched it, he would not even have believed it.’ (CC2 .1581:141) d. nu vrea fi tăcut de acesta not want.aux.prs.ind.3sg be.aux.inf be.silent.ptcp of this evanghelistul evangelist.def ‘the evangelist would not have kept silent about this.’ (CC2 .1581:267) e. să fie fost trăit sbjv be.aux.sbjv.3sg be.aux.ptcp live.ptcp cu pace (NL.~1750–66:3) with peace ‘they should have lived there in peace.’

într-însul in-him

f. Pătru au vrut fi zis Peter have.aux.prs.ind.3sg want.aux.ptcp be.aux.inf say.ptcp ‘Peter would have said.’ (CC2 .1581:151)

loss of analyticity in romanian verbs

249

g.

dă se va fi făcut sărutare, if refl.pass= want.aux.3sg be.aux.inf do.ptcp kiss ori dă nu să va fi fost or if not refl.pass= want.aux.3sg be.aux.inf be.aux.ptcp făcut do.ptcp ‘if one had kissed or not.’ (Prav.1652:176) h. de-ară fi fost înțelegut if=aux.cond.3sg be.aux.inf be.aux.ptcp understand.ptcp ‘if he would have understood.’ (CPr.1566:287) i. de vrea fi fost iubit if want.aux.prs.ind.3sg be.aux.inf be.aux.ptcp love.ptcp acest tânăr pre vecinul său this young dom neighbour.def his ‘if this young man should have loved his neighbour.’ (CC2 .1581:323)

The future perfect was also used with a modal and evidential value as a perfect of the ‘epistemic future’, expressing suppositions about past events. The periphrasis with fi ‘be’ in the imperfective past (i.e., the analytic pluperfect) was frequent in the sixteenth century (Pamfil 1973; Zamfir 2007:196–200). The periphrasis with the auxiliary in the compound periphrastic perfective past (am fost ‘have.aux.prs.ind.1sg be.aux(ptcp)’ + participle), frequent until the eighteenth century (Pamfil 1973; Zamfir 2007:37–62) and preserved in the dialects to the present day (see §9.4.2), has been considered a second analytic pluperfect by some scholars (Densusianu 1938:224), but its uses were not always equivalent to those of an anaphoric ‘past in the past’. It seems to have had a mainly aspectual value, emphasizing the perfectivity, as did the compound auxiliary forms (with auxiliary be) in old Neapolitan described by Ledgeway (2009:596–598),13 corresponding therefore to a ‘perfect in the past’, according to the distinction in Squartini (1999). The periphrases with the auxiliary in the conditional (the synthetic fure (5d), the analytic aș fi (6c)/vrea fi (6d), or the double compound am vrut fi (6f)) function as variants of the perfect conditional. The set of forms which contain three auxiliary verbs includes: the future perfect voi fi fost + participle (6g; Zamfir 2007:258f.), the principal perfect conditional aș fi fost + participle (6h; Zamfir 2007:336, 339; Frâncu 2009:124) and also the type vrea fi fost + participle (6i). Many of these formations (6a, e, g–i) contain fi ‘be’ in the past participle (i.e., fost). Given that these periphrases disappeared (with just one exception, preserved dialectally, see (24)), it is not always easy to determine the value contributed by fost: in some cases 13 The same compound auxiliary pattern, but with the auxiliary have, is attested in Friulan, Occitan, French, Catalan, Ræto-Romance, and northern Spanish varieties (Ledgeway 2009; Melchior 2012; Saussure and Sthioul 2012). The French surcomposé forms, which are still in use, have varying values in informal registers and different regional varieties (see Carruthers 1994; Apothéloz 2010; Melchior 2012).

250

adina dragomirescu, alexandru nicolae, and rodica zafiu

it might be taken to express anteriority, similarly to other Romance compound auxiliary periphrases (cf. Poletto 2009).1⁴ Only the analytic forms with the auxiliary in the imperfective past, the periphrastic perfective past, the subjunctive, and the conditional were frequent in early texts. And only the perfect subjunctive (i.c, ii.c), the future perfect (i.a, ii.a), the conditional perfect (i.b, ii.b), and the perfect infinitive (i.c, ii.c) were retained (§9.3). In modern Romanian, these forms contain the invariable auxiliary fi ‘be’, which has two diachronic sources. In the case of most forms, invariable fi ‘be’ continues the bare infinitival form of ‘be’. In the perfect subjunctive, it continues an inflected form of ‘be’ which, although originally displaying inflexional variation in person and number (i.d), later came to lose these inflexional endings in the nineteenth century such that the auxiliary became invariable (ii.d) on a par with other forms (e.g., conditional perfect, future perfect; see also example (34) and discussion in §9.5.2). This change created a structural correspondence between the perfect subjunctive and the other analytic be perfects, even if subjunctive să is an invariable particle, rather than an auxiliary. (i)

a. b. c. d.

old Romanian voi/ve(r)i/va/vom (v(r)em)/v(r)eți/vor aș/ai/ar(ă)/am/ați/ar(ă) a să

(ii)

a. b. c. d.

modern Romanian voi/vei/va/vom/veți/vor aș/ai/ar/am/ați/ar a să

fi fi fi fi

fi fi fi fiu / fii / fie / fim / fiți / fie

participle participle participle participle

participle participle participle participle

The non-passive periphrases made up of be plus past participle also underwent an important change: although in the sixteenth century this pattern frequently shows agreement, in the following centuries it appears without agreement (Dragomirescu 2014; 2016:263). Statistically, if we exclude masculine singular subjects (since it is impossible to say whether the participle shows referential or default agreement), and just examine masculine plural and feminine subjects, then in the sixteenth century approximatively 50% of the periphrases made up of be and the past participle display agreement (Dragomirescu 2016:264, 270). Notably, this form of gender and number participial agreement controlled by the subject does not depend on the class of lexical verb. Rather, what is crucial is the selection of the 1⁴ This raises important questions for future work about the degree to which compositionality plays a part in diachronic change (see Vincent 2015).

loss of analyticity in romanian verbs

251

auxiliary, in that agreement only obtains in the be-periphrases. Therefore, in contrast to most Romance varieties, transitive (7), unaccusative (8), and unergative (9) verbs show past participial agreement with the subject in be-periphrases: (7)

a.

Și ceia ce vor hi făcuți and those what want.aux.prs.ind.3pl be.aux.inf do.ptcp.mpl aceasta this.f.sg ‘and those who would have done that.’ (Prav.1581:258r ) b. și ceia ce vor fi and those who want.aux.prs.ind.3pl be.aux.inf botezați finul (Prav.1581:242r ) baptize.ptcp.mpl godson.def.acc ‘and those who will have christened their godson.’

(8)

a.

(9)

a.

au fost trecuți ai have.aux.prs.ind.3pl be.aux.ptcp pass.ptcp.mpl years de la Adamu până acmu 7105 from Adam until now 7105 ‘7105 years have passed from Adam until now.’ (DÎ.1597:XV) b. păscarii era ieșiți dentr-însele, fishermen.def be.aux.pst.ipfv.3pl go.out.ptcp.mpl from-those spăla mreaja wash.prs.ind.3pl nets.def ‘the fishermen had left there, they were cleaning the nets.’ (CC1 .1567: 110v ) și încă foarte departe n-ară fi and yet very far not=aux.cond.3pl be.aux.inf merși go.ptcp.mpl ‘and they would have not gone too far.’ (PO.1582:155) b. Că acești oameni orbi era that these people.mpl blind.mpl be.aux.pst.ipfv.3pl auziți de vestea hear.ptcp.mpl of news.def.acc ‘these blind people had heard the news.’ (CC1 .1567:55v ) c. după ce era lăcuiți 10 ani after what be.aux.pst.ipfv.3pl live.ptcp.mpl ten years în pământul Canaanului in land.def Canaan.gen ‘after they lived for ten years in the land of Canaan.’ (PO.1582:52)

252

adina dragomirescu, alexandru nicolae, and rodica zafiu

9.2.3 Periphrases with the lexical verb in the gerund/present participle Old Romanian has a series of periphrases formed with auxiliary fi ‘be’ and what is traditionally called the gerund, in parallel with the periphrases formed with the past participle reviewed in §9.2.2. There are even isolated forms, whose participlebased counterpart is not attested, with the auxiliary in the synthetic perfect (10c) (see also Zamfir 2007:75f.) and the imperative (10g). The original value of this periphrasis must have been progressive, even if the available textual evidence does not allow us to identify a clear aspectual contrast. The auxiliary could exhibit synthetic forms: namely, the imperfective past auxiliary exhibited a synthetic form in the present indicative (10a) (see also Zamfir 2007:74f.), the imperfective past (10b), the perfective past (10c), the pluperfect (10d), the subjunctive (10e)—attested since the seventeenth century (Zamfir 2005:415f.; Niculescu 2013:158)—and the extremely rare synthetic conditional (10f) or the imperative (10g). The construction in (10h), with the invariable auxiliary fi ‘be’, is a late innovation (nineteenth century), whose emergence is probably due to analogy with the past participle periphrasis (see §9.2.2; cf. Niculescu 2013:166f.). (10)

a.

vor chema numele lui Emmanuil, want.aux.prs.ind.3pl call.inf name.def gen Emmanuel carele iaste tălmăcindu-se „Cu which.def be.aux.prs.ind.3sg interpret.ger=cl.refl.pass with noi Dumnezeu” us.acc God ‘they shall call his name Emmanuel, which means “God with us”.’ (BB.1688:751)

b. În vreamea aceaia era Isusu învățându in time.def that be.aux.pst.ipfv.3sg Jesus.nom teach.ger în besearecă întru o sâmbătă in church in a Saturday/Sabbath ‘Jesus was then teaching in a church on the Sabbath.’ (CC2 .1581:406) c. fum veselindună be.aux.pst.pfv.1pl be.glad.ger= cl.acc.1pl ‘(we) were glad.’

(PS.1573−8:273)

d. patr-înși-l fusease purtând four.people=cl.acc.3msg be.aux.plpf.3pl carry.ger ‘four people had carried him.’ (CC2 .1581:54) e. dă if

vei vedea pe vreun păcătos […] want.aux.prs.ind.2sg see.inf dom any sinner

loss of analyticity in romanian verbs

să fie petrecând bine sbjv be.aux.prs.sbjv.3sg feast.ger well ‘if you see a sinner having a good time.’ f. să fure întru if be.aux.cond.3sg in ‘if he lived in you.’

253

(Prav.1652:613)

voi lăcuind you.pl live.ger (CPr.1566:194)

g.

fii tocmindu-te1⁵ cu pârâșul tău […]! be.aux.imp.2sg negotiate.ger=cl.refl.2sg with plaintiff your ‘negotiate with your plaintiff.’ (BB.1688:753) h. pare a nu mai fi existând nici o seem.prs.ind.3sg to not more be.aux.inf exist.ger not one idee serioasă idea serious.fsg ‘it seems that there is no longer any serious idea.’ (EOXIII.1882:229) Constructions are attested in which the auxiliary itself is found in an analytic (viz. compound) form, occurring in the periphrastic perfect (11a), the future (11b), the aș-conditional (11c), the vrea-conditional (11d), and the perfect subjunctive (11e). The most frequent was the analytic form with the auxiliary in the compound past (Zamfir 2007:62–74). Structures with three auxiliaries are also found, where the first auxiliary is itself an analytic form; these include a vrea ‘want’ conditional form (11f), a future perfect (11g; cf. Zamfir 2007:258), and a conditional perfect (11h; cf. Zamfir 2007:345). (11)

a.

câți draci au fost având how.many devil.pl have.aux.prs.ind.3sg be.aux.ptcp have.ger întru el inside him ‘how many devils he had inside himself.’ (CC2 .1581:422)

b. de vor fi și nescari nevoi if want.aux.prs.ind.3pl be.aux.inf also some troubles țiind pre noi keep.ger dom us ‘if any troubles will befall us.’ (CC2 .1581:306) c. de n-ați fi având lucrure bune if not=aux.cond.2pl be.aux.inf have.ger thing.pl good.pl ‘if you didn’t have good things.’ (CC2 .1581:526) 1⁹⁰ This structure is the unambiguous result of a loan translation from the Greek original, where the sequence ἴσθι εὐνoῶν is made up of the imperative of the verb be and a present participle. Cf. https:// biblehub.com/interlinear/matthew/5-25.htm [accessed 21 December 2020].

254

adina dragomirescu, alexandru nicolae, and rodica zafiu

d. fi-vreați știind be.aux.inf=want.aux.2pl know.ger ‘you would know.’ (CC2 .1581:526) e. varvarii […] să fie fost țiind Barbarians.def sbjv be.aux.prs.sbjv.3pl be.aux.ptcp hold.ger ‘Barbarians should have held.’ (CH.1717–23:342) f. am vrut fi fiind have.aux.prs.ind.1pl want.aux.ptcp be.aux.inf be.ger ‘we would be.’ (CT.1560−1:51r ) g. va fi fost gătind arme want.aux.3sg be.aux.inf be.aux.ptcp prepare.ger weapons ‘he will have prepared weapons.’ (Prav.1652:98) h. ară fi amu fost fiind aux.cond.3sg be.aux.inf then be.aux.ptcp be.ger ‘he would have been.’ (CPr.1566:553) The functional value (mood and tense) of the progressive periphrases generally was that supplied by their auxiliary, the periphrasis being equivalent to the present, the imperfect, the future, the conditional, etc. They gradually fell into disuse in the passage from old to modern Romanian. Contemporary Romanian preserves only the analytic forms which are the counterpart of the fi-perfects, i.e., the future (11b), the subjunctive (10e), the conditional (11c), and the infinitive (10h). These forms gradually specialized for specific epistemic uses of the future, the subjunctive, the conditional, the infinitive, but only the future periphrasis has a significant frequency and a certain degree of stability as a modal form (part of the so-called ‘presumptive mood’—Zafiu 2013:53). Various hypotheses have been formulated regarding the disappearance of certain periphrases, including competition between functionally equivalent forms and their bookish (learnèd) nature. For example, Frâncu (1983–1984; 2014) suggests that the jettisoning of some earlier periphrases was determined by deliberate attempts to modernize the standard language in the nineteenth century (which is also reflected by their elimination from some influential grammars). In §9.5, by contrast, we formulate an analysis which privileges structural factors as an explanation for their demise. On the whole, the following features distinguish old Romanian analytic structures from those of modern Romanian: • they are more numerous in previous stages of the language, but they have different frequencies reflecting different degrees of stability; • those with the fi ‘be’ auxiliary are built on a pattern which allows for the recursive extension of the structure; • the auxiliary fi ‘be’ may inflect for tense (and aspect);

loss of analyticity in romanian verbs

255

• the auxiliary fi ‘be’ is not restricted to irrealis forms, but also occurs with different tenses of the indicative.

9.3 Analytic forms in modern Romanian The tense, aspect, and mood system of modern standard Romanian has six synthetic forms: the present indicative, the synthetic perfective past, the pluperfect, the imperfective past, the present subjunctive, and the imperative. The analytic forms include the periphrastic perfective past, three competing paradigms with future value, the future perfect, the present and perfect conditional, the perfect subjunctive, the perfect infinitive, and three marginal periphrases (the presumptive periphrases) specialized for the epistemic values of the future, the subjunctive, and the conditional (see Zafiu 2013). Depending on the non-finite or finite form of the main verb, standard Romanian shows four types of analytic forms, which use the infinitive, the past participle and the gerund (present participle), and the subjunctive, respectively. The periphrases which use the infinitive are the voi-future (dubbed ‘the literary future’ in the Romanian grammatical tradition), which is also the oldest type of future, and the conditional. The auxiliary of the future (12a) preserves older forms (1sg voi, 3sg va, 3pl vor) and simplified forms (2sg vei, 1pl vom, 2pl veți) of the present indicative of vrea ‘want’, while the paradigm of the lexical verb has been analogically regularized on the model of the vr- root of the first- and secondpersons plural in vrem and vreți, respectively, witness 1sg vreau, 2sg vrei, 3sg vrea (Maiden et al. 2021:362f.). The conditional auxiliary (12b) has opaque forms which have generated etymological controversies: they present a certain similarity to forms of avea ‘have’, but almost certainly (according to an old explanation, revived in Zafiu 2017) derive from past forms of the same auxiliary want (see also §9.2.1). (12)

a.

voi/vei/va/vom/veți/vor merge want.aux.1sg/2sg/3sg/1pl/2pl/3pl go.inf b. aș/ai/ar/am/ați/ar merge aux.cond.1sg/2sg/3sg/1pl/2pl/3pl go.inf

The periphrases involving the past participle of the lexical verb are heterogeneous. The only pattern formed with the auxiliary avea ‘have’ is the periphrastic perfective past (13a). The other set of forms with a participial lexical verb are formed with the invariable auxiliary fi ‘be’, interpreted as an irrealis auxiliary by Avram and Hill (2007) (but see Ledgeway 2014 for a finer-grained distinction). In one set of constructions, namely the perfect subjunctive (13b) and the perfect infinitive (13c), fi ‘be’ is preceded by invariable particles (să and a). In another set of forms, the first element in the cluster is a proper (inflected) auxiliary, the future auxiliary

256

adina dragomirescu, alexandru nicolae, and rodica zafiu

vrea in the future perfect (13d) and the conditional auxiliary aș in the conditional perfect (13e). (13)

a.

am/ai/a/am/ați/au have.aux.1sg/2sg/3sg/1pl/2pl/3pl b. să sbjv c. a to d. voi/vei/va/vom/veți/vor want.aux.1sg/2sg/3sg/1pl/2pl/3pl e. aș/ai/ar/am/ați/ar aux.cond.1sg/2sg/3sg/1pl/2pl/3pl

mers go.ptcp fi mers be.aux go.ptcp fi mers be.aux.inf go.ptcp fi mers be.aux.inf go.ptcp fi mers be.aux.inf go.ptcp

In fact, the aspectual difference between future and future perfect, and between present conditional and perfect conditional is reanalysed in contemporary Romanian as a difference between the present infinitive and the perfect infinitive (made up of the invariable auxiliary fi ‘be’ + past participle):1⁶ (14)

a. b.

voi [merge] vs voi [fi mers] want.aux.1sg go.inf want.aux.1sg be.aux.inf go.ptcp aș [merge] vs aș [fi mers] aux.cond.1sg go.inf aux.cond.1sg be.aux.inf go.ptcp

The subjunctive and the infinitive can be analysed along similar, yet not fully identical lines; although from an inflexional perspective, the opposition between the present subjunctive and the perfect subjunctive is not of the same type as that between the present infinitive and the perfect infinitive (the subjunctive displays an opposition between a variable/finite form in the present and an invariable/nonfinite sequence in the perfect), from a distributional point of view the present subjunctive and the perfect subjunctive make up a minimal pair. (15)

a.

a [merge] vs a [fi mers] to go.inf to be.aux.inf go.ptcp b. să [merg] vs să [fi mers] sbjv go.prs.1sg sbjv be.aux.inf go.ptcp

Periphrases with fi ‘be’ + gerund (present participle) are the precise counterpart of the be + past participle constructions. They are rare in present-day Romanian; only the construction with the future of the auxiliary (16a) is more frequent. It is specialized for the epistemic value of the future, partially grammaticalized as a form in the presumptive paradigm. The others, learnèd and extremely rare (almost extinct), 1⁶ The interpretation of the conditional perfect and the future perfect as consisting of auxiliary (aș or voi) plus perfect infinitive is adopted, for example, in Avram (1999:40), D’Hulst, Coene, and Avram (2004) and Nicolae (2015:82f.).

loss of analyticity in romanian verbs

257

are variants of the present subjunctive (16b), the infinitive (16c), and the present conditional (16d), specialized for expressing epistemic uses of the subjunctive, the infinitive, and the conditional, respectively.1⁷ The original progressive value of these periphrases (§9.2.3) has been lost, but there is an acknowledged diachronic relation between the progressive and epistemic values (Heine 1995; Anthonissen, De Wit, and Mortelmans 2016; see Bertinetto 2000 for a particular view on Romance) which eventually paved the way for the reinterpretation of these periphrases as epistemic forms. (16)

a.

(v)oi want.aux.2sg b. să sbjv c. a to d. aș aux.cond.1sg

fi be.aux.inf fi be.aux fi be.aux.inf fi be.aux.inf

mergând go.ger mergând go.ger mergând go.ger mergând go.ger

This pattern can be interpreted as having the same segmentation (formal analysis) as the perfect series: (17)

a.

(v)oi [fi mergând] vs voi want.aux.1sg be.aux.inf go.ger want.aux.1sg [fi mers] be.aux.inf go.ptcp b. să [fi mergând] vs să [fi sbjv be.aux.inf go.ger sbjv be.aux.inf c. a [fi mergând] vs a [fi to be.aux.inf go.ger to be.aux.inf d. aș [fi mergând] vs aș aux.cond.1sg be.aux.inf go.ger aux.cond.1sg [fi mers] be.aux.inf go.ptcp

mers] go.ptcp mers] go.ptcp

Periphrases with the subjunctive of the lexical verb are more recent and reflect the phenomenon of replacement of the infinitive by the subjunctive typical of the Balkan Sprachbund (cf. Sandfeld 1930). There are alternative paradigms for the future, only partially accepted by the standard language: one of them (18a) reflects 1⁷ The label presumptive mood is inconsistently used in the literature, sometimes for the partly specialized Romanian epistemic future, which has three paradigms—present oi fi ‘want.aux.1sg be.aux.inf’ and (v)oi fi fiind ‘want.aux.1sg be.aux.inf be.ger’, perfect (v)oi fi fost ‘want.aux.1sg be.aux.inf be.ptcp’ (Zafiu 2013:53f.)—and at other times for all the gerund periphrases which have been preserved and specialized with an epistemic value (Niculescu 2013), and in some cases for specific combinations of these categories.

258

adina dragomirescu, alexandru nicolae, and rodica zafiu

a common Balkan pattern (invariable particle o (< *oa [wa] < va ‘wants’) + subjunctive clause). The type in (18b) is made up of the auxiliary have + verb in the subjunctive. There is an important difference between the future in (12a) and the future in (18a): inflexion (in number and person) is marked on the auxiliary in the first case, but on the main verb in the second; (18b) features inflexion on both components. (18)

a.

o fut

să merg/mergi/meargă/mergem/mergeți/meargă sbjv go.prs.1sg/2sg/sbjv.3/1pl/2pl/sbjv.3 (3pl also: or să meargă) să b. am/ai/are/avem/aveți/au have.prs.ind.1sg/2sg/3sg/1pl/2pl/3pl sbjv merg/mergi/meargă/mergem/mergeți/meargă go.prs.1sg/2sg/sbjv.3/1pl/2pl/sbjv.3

The auxiliary in (18b) has the same forms as lexical have; occurrence in actual usage is uneven, in that the first- and second-persons plural and the third-person singular are infrequent in comparison to the frequent use of the other persons (Berea-Găgeanu 1972). Modern Romanian uses a learnèd future-in-the-past formed from the same auxiliary inflected in the imperfective past (19). (19)

aveam/aveai/avea/aveam/aveați/aveau să have.pst.ipfv.1sg/2sg/3sg/1pl/2pl/3pl sbjv merg/mergi/meargă/mergem/mergeți/meargă go.prs.1sg/2sg/sbjv.3/1pl/2pl/sbjv.3

The analytic forms in (18) and (19) are not of the same structural type as the other (auxiliary-based) analytic structures introduced in this section, as the lexical verb is an inflected (not a non-finite) form; §9.5.3 takes up this issue in more detail. For a classification of modern Romanian analytic forms, see Maiden et al. (2021:369–382), and for a summary of the differences between Romanian and western Romance, see D’Hulst, Coene, and Avram (2004:362–365).

9.4 Analytic forms in (Daco-)Romanian dialectal varieties In addition to the forms of standard (modern) Romanian, dialect data reveal that certain other periphrastic verb forms have been preserved.

9.4.1 Periphrases with the lexical verb in the infinitive The analytic conditional with a ‘transparent’ auxiliary is preserved in a few isolated areas. The pattern vrea ‘want’ in the past imperfective followed by the infinitive

loss of analyticity in romanian verbs

259

is found in the areas of Banat (20a) (Neagoe 1984:264), Crișana, and Maramureș (20b) (Marin 2000:202). In the south of Crișana, an area adjacent to Banat, there is an analytic conditional consisting of an invariable third-person singular form of the auxiliary and the infinitive (20c) (Urițescu 1984:310). (20)

a.

reaș merge want.aux.1sg go.inf ‘I would go.’ b. Urâtu satului ve vorbi ugly.def village.def.gen want.aux.3sg speak.inf Mândra cu satului with beautiful.def village.def.gen ‘the village ugly would speak with the village beauty.’ c. vrę mę noi aclo want.aux.1pl go.inf we there ‘we would go there.’

The pattern with the periphrastic perfective past of the auxiliary vrea ‘want’, rarely attested in old Romanian, is occasionally found in Crișana, Sălaj, and Maramureș (Marin 2000:202; Marin et al. 2017:47; cf. Urițescu 1984:310): (21)

a.

teai vu supăra refl.2sg= have.aux.2sg want.aux.ptcp get.upset.inf ‘you would get upset.’ b. dacă am vrut ști, plecam if have.aux.1sg want.aux.ptcp know.inf leave.pst.ipfv.1sg ‘had I known, I would have left.’

The future consisting of the auxiliary avea ‘have’ with the same inflexion as the corresponding lexical verb and followed by an a-infinitive has been preserved in Bucovina and Maramureș (Marin, Mărgărit, and Neagoe 1998:111). (22)

a.

naveți a mântui repede not= have.aux.2pl to finish.inf fast ‘you won’t finish fast with me.’ b. unde avem a durmi? where have.aux.1pl to sleep.inf ‘where are we going to sleep?’

cu with

mine me.acc

9.4.2 Periphrases with the lexical verb in the past participle forms The periphrasis with fi ‘be’ in the imperfective past and a past participle showing agreement (of the same type as that found in old Romanian, cf. (8b), (9c)) survives

260

adina dragomirescu, alexandru nicolae, and rodica zafiu

in regional varieties in the west, the north, and also the south (Marin 1985:463; 2005–2007:117–119): (23)

nu era mutați acasă la ei not be.aux.pst.ipfv.3pl move.ptcp.mpl home at them ‘they had not moved to their place.’

The periphrasis with fi ‘be’ in the periphrastic perfective past is better preserved in non-standard Romanian, in all areas, even in the south (an area which shows the least degree of analyticity in the verbal domain; Lăzărescu 1984:225; Urițescu 1984:309; Vulpe 1984:337; Marin 1985; 2005–2007:119f.; Neagoe 1992a:83; 1992b:163; Mărgărit and Neagoe 1997; Marin, Mărgărit, and Neagoe 1998). Its temporal value depends on context, in that it can be interpreted either as a pluperfect, or as a perfective past ((24); cf. Neagoe 1992a:83). This latter reading confirms Ledgeway’s (2009:598) observation that the Romance compound auxiliary forms are conditioned and optional variants of the corresponding non-compound auxiliary forms. (24)

a.

apă una am fost dato then one have.aux.1sg be.aux.ptcp give.ptcp= cl.acc.f.3sg la muiere at wife ‘then I gave one to my wife.’ b. n-o fo avut zăce ani not=have.aux.3sg be.aux.ptcp have.ptcp ten years ‘he hadn’t turned ten years old yet.’

Even the rare periphrases with fi ‘be’ in the pluperfect have some regional descendants in the south (Marin 1985:466; 2005–2007:121): (25)

Să duce unde fusese Marin refl= go.prs.ind.3sg where be.aux.plpf.3sg Marin tras. pull.ptcp ‘He is going where Marin had spent the night.’

The conditional perfect vreaș + fi ‘be’+ past participle is still attested in the Valea Almăjului area (an isolated rural region in Banat) and in the north-western part of Oltenia (Marin, Mărgărit, and Neagoe 1998:112): (26)

parcă mai vreaș fi șezut apparently more want.aux.1sg be.aux.inf stay.ptcp ‘I would have kind of preferred to stay there longer.’

acolo there

loss of analyticity in romanian verbs

261

The periphrastic form with fi ‘be’ in the subjunctive followed by the (agreeing) past participle was preserved in the north-western part of Oltenia,1⁸ the west of Crișana, and the Apuseni Mountains (Marin, Mărgărit, and Neagoe 1998:112): (27)

să fie fostă moda sbjv be.aux.sbjv.3sg be.ptcp(.fsg) fashion ‘if that were the fashion.’

9.4.3 Periphrases with the lexical verb in the gerund/present participle The periphrasis with auxiliary fi ‘be’ in the imperfective past followed by the gerund of the lexical verb is rarely attested in the south (Marin 1985:459; 2005– 2007:115; 2012:20): (28)

[…]

erau trecând printr-o be.aux.pst.ipfv.3pl cross.ger through-a ‘[…] they were crossing a forest.’

pădure forest

The periphrasis with auxiliary fi ‘be’ in the periphrastic perfective past is preserved with a low frequency in Wallachia, Dobrodja, Oltenia, Banat, and several areas of Transylvania (Neagoe 1984:263; 1992a:83; 1992b:163; Urițescu 1984:308; Vulpe 1984:336; Marin 1985; 2005–2007:112–114; 2012; 2014): (29)

ăst unchiaș a fost având un fecior this old.man have.aux.prs.ind.3sg be.aux.ptcp have.ger a son ‘this old man had a son.’

An atypical construction involving the invariable auxiliary o and the gerund (30), which has either imperfective or perfective value, is attested in Banat (Neagoe 1984:263). The form of this construction can be explained by the loss of the past participle fost namely o fost dând > o fost dând. Most probably, it was preserved by analogy with the past participle periphrases o (fost) dat (cf. (24)). (30)

o dând have.aux give.ger ‘he was giving.’

The periphrases specialized for expressing the epistemic (presumptive) value are attested with the auxiliary in the future (31a) or in the conditional (31b), but, in contrast to standard Romanian, the gerund can have a ‘feminine’ form (Marin, 1⁸ It is not clear whether this is a genuine case of participle agreement or whether it preserves the archaic invariable past participle ending in -ă, syncretic with the feminine singular form of the past participle (Urițescu 2007), a form also attested in earlier stages of the language in periphrastic forms with auxiliary be (Dragomirescu 2016: 269f.), only very rarely with have (Densusianu 1961:II:143).

262

adina dragomirescu, alexandru nicolae, and rodica zafiu

Mărgărit, and Neagoe 1998:113; see Maiden et al. 2021:340f. for details regarding this apparently ‘feminine’ form of the gerund). (31)

a.

nu știu cum s-o fi făcândă not know.prs.ind.1sg how refl.pass=fut be.aux do.ger.fsg ‘I don’t know how it would be made.’ b. zicea că ar fi avândă say.pst.ipfv.3sg that aux.cond.3sg be.aux.inf have.ger.fsg ‘he was saying that he would be having.’

9.5 An account of the partial loss of analyticity in Romanian verbal morphology The aim of this section is to argue that the group of now defunct auxiliary-based forms share a set of properties which distinguishes them from the auxiliary-based forms which have been diachronically preserved. On the one hand, we see that, on the surface, change is driven by a principle of inflexional simplification, which is, in turn, the effect of structural factors which have to do with the division of labour between the auxiliary verb and the lexical verb in the marking of tense, aspect, and mood categories. In particular, we argue that the feature matrix of auxiliaries in the now defunct periphrases is richer and includes tense features along with mood features, while the auxiliaries of the forms which have been diachronically preserved have an impoverished feature matrix, devoid of tense features. In the latter periphrases, assuming a compositional analysis of tense, aspect, and mood encoding (cf. Comrie 1985:76), the tense specification is provided by the morphology of the lexical verb; Romanian verbal participles have been analysed as encoding perfective aspect and past tense (Stati 1965:195; D’Hulst, Coene, and Avram 2004:360, 364; cf. also Comrie 1985:65 on the featural make-up of (verbal) participles). The idea that modern Romanian auxiliaries do not inflect for tense has been previously advanced by Dobrovie-Sorin (1994), D’Hulst, Coene, and Avram (2004), and Nicolae (2015). On the other, we see that other periphrastic forms with a different internal structure are not affected by the changes which act upon auxiliary-based structures, hence they prove more stable diachronically, since they show distinct properties and a different division of labour in the marking of tense, aspect, and mood (§9.5.1).

9.5.1 Demise of a subset of periphrastic forms When comparing the now defunct forms with the forms which have been diachronically preserved, we observe that the former are more complex than the latter from an inflexional point of view. The inflexionally complex forms fall into several types according to the structure of the first auxiliary:

loss of analyticity in romanian verbs

263

(a) forms whose be or want auxiliary displays mood and tense inflexion (32): be occurs in the present indicative (32a), imperfective past (32b), synthetic perfective past (32c), pluperfect (32d), conditional/future (32e), present subjunctive (32f), and the imperative (32g), or as a non-finite form (a gerund) (32h); want occurs in the imperfective past indicative (32i); (b) forms whose be or want auxiliary occurs in an analytic form (33), hence the periphrastic form is itself a compound periphrasis: be in the periphrastic perfective past in (33a), and want in the periphrastic perfective past in (33b). (32)

a. b.1 b.2 c. d.1 d.2 e.1 e.2 f.1 f.2 f.3 g. h. i.1 i.2 i.3 i.4

iaste tălmăcindu (= 10a) be.aux.prs.ind.3sg interpret.ger era scris (= 5a) be.aux.pst.ipfv.3pl write.ptcp era învățându (= 10b) be.aux.pst.ipfv.3sg teach.ger fum veselindu (= 10c) be.aux.pst.pfv.1pl be.glad.ger fusese văzut (= 5b) be.aux.plpf.3sg see.ptcp fusease purtând (= 10d) be.aux.plpf.3pl carry.ger fure faptu (= 5d) be.aux.cond.3sg do.ptcp fure lăcuind (= 10f) be.aux.cond.3sg live.ger să fim iubit (= 5c) sbjv be.aux.sbjv.1pl love.ptcp să fie petrecând (= 10e) sbjv be.aux.sbjv.3sg feast.ger să fie fost țiind (= 11e) sbjv be.aux.sbjv.3sg be.aux.ptcp hold.ger fii tocmindu (= 10g) be.aux.imp.2sg negociate.ger fiind avut (= 5e) have.ptcp be.aux.ger vrea fi (= 2b) want.aux.pst.ipfv.3sg be.inf vrea fi tăcut (= 6d) want.aux.pst.ipfv.3sg be.aux.inf be.silent.ptcp vrea fi fost iubit (= 6i) want.aux.pst.ipfv.3sg be.aux.inf be.aux.ptcp love.ptcp fi-vreați știind (= 11d) be.aux.inf=want.aux.pst.ipfv.2pl know.ger

264

adina dragomirescu, alexandru nicolae, and rodica zafiu

(33)

a. b.1 b.2 b.3

au have.aux.prs.3pl au have.aux.prs.3pl

fost be.aux.ptcp vrut want.aux.ptcp

au have.aux.prs.3pl am have.aux.prs.1pl

vrut want.aux vrut want.aux

făcut make.ptcp vrea want.inf

fi be.aux.inf fi be.aux.inf

zis say.ptcp fiind be.ger

(= 6a) (= 3) (= 6f) (= 11f)

Note that the want auxiliary, whether inflected for mood and tense or occurring as an analytic form, may be accompanied by auxiliary be in its bare infinitival form or in its past participial form, thereby giving rise to another type of double (32i2 ), (32i4 ) and even triple (32i3 ), (33b2 ), (33b3 ) compound periphrasis. In contrast to these now defunct forms, all the periphrastic forms which have been preserved are characterized by auxiliaries with a simpler inflexional structure: notably, what is absent from the feature composition of these auxiliaries is a tense specification. Thus, as proposed in Nicolae (2015; 2019) on the basis of earlier remarks by Dobrovie-Sorin (1994), modern Romanian periphrastic structures appear to be oriented around the category of mood, encoded explicitly by the auxiliary verb, whereas tense inflexion is not present on the auxiliary. Emergent structures whose auxiliaries have a more complex structure encoding also tense (cf. (32)–(33)) are not diachronically preserved in the standard language. This claim finds support in the compound auxiliary forms. With the double auxiliary forms preserved in modern Romanian—the conditional perfect and the future perfect (as well as their infrequently used gerundial counterparts, specialized for epistemic values)—we observe that the second auxiliary surfaces as the invariable fi ‘be’. The preservation (and vitality) of these forms is highly relevant for the diachronic process discussed here. First, it confirms the hypothesis that the elimination of a subset of auxiliary-based forms is not driven by reasons which have to do with auxiliary selection (as one might be tempted to believe, given that many of the defunct forms are based on fi ‘be’), since be is preserved with these forms (as well as with the perfect subjunctive and with the perfect infinitive, see §9.5.3). Nor does it have to do with the disappearance of compound auxiliary forms, since these periphrases contain two auxiliaries. Second, this form of be—interpreted as an irrealis auxiliary in these modern Romanian forms, see §9.3—occurs in an uninflected form, a form which in and of itself is unspecified for tense. This gives rise to yet another extremely relevant difference between old and modern Romanian. Alongside these invariable be forms (where the auxiliary is syncretic with the bare infinitive), we find in old Romanian forms in which the auxiliary be occurs as a past participle (fost) (cf. examples (6a), (6e), (6h), (6i), (11a), (11e), (11g), (11h)). These past participle be forms also disappeared, we believe, for a similar reason: as a past participle, be is inflexionally complex and bears

loss of analyticity in romanian verbs

265

tense features (i.e., [past]), hence the periphrasis disappeared because it contained a tense-bearing auxiliary, even if be is not the first auxiliary in the cluster. Below, we briefly address the division of labour in the marking of tense, aspect, and mood categories, and we show that in modern Romanian compound auxiliary structures, both auxiliaries encode mood values. However, two more facts should be mentioned before discussing this in more detail. First, from what has been said so far, it follows that some of the now defunct periphrases were subjected to a bidirectional pressure on their auxiliaries: with forms such as (6i) (= (32i3 )), two of the three auxiliaries in the cluster (the highest, want in the imperfective past, and the lowest, be in past participial form) explicitly marked tense, hence their loss was motivated on two accounts. Second, we see inflexional simplification at work from at least two other perspectives as well. On the one hand, participle agreement in non-passive structures (see the discussion of examples (7)–(9) in §9.2.2) disappears from the standard language across the board (Dragomirescu 2014), and is only preserved occasionally in dialectal Romanian (see (23), (27)): it disappears both from structures where the be auxiliary displays inflexion (as a synthetic form (9b), or as an analytic form (8a)), and from structures where be occurs as an invariable auxiliary such as (7b) and (9b). On the other hand, we see forms such as the perfect or gerundial subjunctive, which in modern Romanian consists of the subjunctive particle să, and the invariable auxiliary be plus the past participle/gerund, evolving from structures in which be is originally inflected for the subjunctive, but gradually undergoes inflexional simplification losing the subjunctive marking and turning into an invariable auxiliary (cf. (34)). (34)

a.

să fim noi iubit pre sbjv be.aux.sbjv.1pl we love.ptcp dom ‘should we have loved God.’

Dumnezeu God (CC1 .1567:25r )

aˈ. să fi iubit (ModRo.) sbjv be.aux love.ptcp ‘that we (I/you/(s)he/we/they) have loved.’ b. pentru ca să fie judecând la tot norodul for that sbjv be.aux.sbjv.3 judge.ger to all people.def ‘so they might be judging all the people.’ (BB.1688:340) bˈ. să fi judecând (ModRo.) sbjv be.aux judge.ger It is not clear if the disappearance of participle agreement and subjunctive inflexion is directly related to the type of inflexional simplification which we have taken to explain the disappearance of a subset of auxiliary-based forms, but it essentially reflects a tendency towards inflexional simplicity operating at a pace with other similar processes.

266

adina dragomirescu, alexandru nicolae, and rodica zafiu

Returning to the idea that the now defunct periphrases involved auxiliaries whose feature matrix was more complex than that of auxiliaries found in periphrastic forms which have survived, we briefly bring to the fore some supplementary evidence in support of the hypothesis that modern Romanian auxiliaries are mood-oriented.1⁹ We also comment on the division of labour between the auxiliary and the lexical verb in the marking of tense, aspect, and mood categories. First, the contrast between the periphrastic paradigms with auxiliary have in (standard) French (35) and (standard) Romanian (36) shows that in French, but not in Romanian, the auxiliary overtly inflects for both tense and mood. By contrast, the Romanian periphrasis with the have auxiliary is unambiguously indicative (and may have the functional value of a past punctual or a present perfect); the tense (and aspect) specification is provided by the morphology of the lexical verb (D’Hulst, Coene, and Avram 2004:364): French j’ai I=have.aux.ind.prs.1sg b. j’avais I=have.aux.ind.pst.impf.1sg c. j’aurai I=have.aux.fut.1sg d. j’aurais I=have.aux.cond.1sg e. j’aie I=have.aux.prs.sbjv.1sg

(35)

a.

mangé eat.ptcp mangé eat.ptcp mangé eat.ptcp mangé eat.ptcp mangé eat.ptcp

(36)

eu am mâncat (Ro.) I have.aux.prs.1sg eat.ptcp

Second, the mood-oriented nature of Romanian auxiliaries can also be illustrated by the ‘opacification’ of the conditional auxiliary, a fact that could be related to the complete loss of its initial temporal value. The standard Romance synthetic conditional, considered as a separate mood or as a tense of the indicative (Quer 2016), has its origins in a periphrastic future-in-the-past (Ledgeway 2012:136), and has preserved some of its original temporal values and uses in addition to novel modal and evidential ones. The Romanian conditional auxiliary most plausibly originates from a past form (an imperfective past form, probably contaminated by the synthetic perfective past, Zafiu 2017) of the same auxiliary as the future (want), and probably went through the same process, moving from a future-in-the-past 1⁹ Cross-linguistically, it has been shown that auxiliaries may encode different tense, aspect, and mood categories (Anderson 2006). Therefore, the feature matrix of a given auxiliary needs to be identified for each language/structure in turn.

loss of analyticity in romanian verbs

267

value to the development of modal values. Relics of some temporal values of the aș-conditional can still be found in the sixteenth century, but rapidly disappear (§9.2.1). The present-day Romanian conditional-optative has two important features: an opaque auxiliary, whose relation to its etymon and with the future auxiliary is no longer transparent,2⁰ and a value which is no longer temporal, but only modal (Vlad 2012; Zafiu 2013:50–53). Third, minimal pairs like the following indicate that the entire periphrastic system of modern Romanian (except for the particle + subjunctive periphrases, see §9.5.3) is organized along these lines, with the auxiliary/auxiliaries primarily encoding mood values and lexical verbs encoding tense. In (37) the difference between the future and the future perfect is ensured by the distinction between the infinitive (37a) and the ‘perfect infinitive’ (37b), itself made up of irrealis be plus a past participle (see also note 16), not by the auxiliary voi. (37)

a.

Voi cânta. (future) want.aux.1sg sing.inf ‘I will sing.’ b. Voi fi cântat. (future perfect) want.aux.1sg be.aux.inf sing.ptcp ‘I will have sung.’

This becomes more evident with presumptive and epistemic conditional periphrases: the difference between a past and a present/future orientation is solely conveyed by the difference between the gerund (present participle) (38a)/(39a) and past participle (38b)/(39b) encoded by the inflexional make up the lexical verb; the functional sequence which precedes the lexical verb is identical. (38)

a.

O fi venind azi / mâine / **ieri? fut be.aux.inf come.ger today tomorrow yesterday ‘Could/Should (s)he be coming today/tomorrow/yesterday?’ b. O fi venit **mâine / ieri? fut be.aux.inf come.ptcp tomorrow yesterday ‘Could/Should (s)he have come / Is it possible that (s)he came / Did (s)he come tomorrow/yesterday?’

(39)

a.

Ar fi venind (non-past) aux.cond.3sg be.aux.inf come.ger ‘(S)he would come / be coming.’ b. Ar fi venit (perfect) aux.cond.3sg be.aux.inf come.ptcp ‘(S)he would have come.’

2⁰ The opaque nature of this auxiliary is reflected in the etymological debates which linked the modern forms to another auxiliary, avea ‘have’ (cf. Rosetti 1932:104; Elson 1992).

268

adina dragomirescu, alexandru nicolae, and rodica zafiu

Finally, it is important to highlight that the presumptive—a set of periphrastic forms which, together with the conditional and the subjunctive, make up the class of irrealis moods—partially grammaticalized in Romanian.21 It cannot be coincidental that a language with mood-oriented auxiliaries grammaticalized a periphrastic form with a primarily mood-oriented usage. To conclude, the auxiliary system of modern Romanian appears to be moodoriented, while the extinct auxiliary structures explicitly marked both mood and tense, therefore not conforming to the mood-oriented periphrastic pattern which diachronically came to prevail in modern Romanian.22

9.5.2 Diachronically stable analytic formations Of the forms introduced in §9.3, there is a set of diachronically stable analytic forms which might appear problematic for the mood-based explanation proposed in the previous section, but which are not, as their internal structure is different from that of auxiliary-based forms. The colloquial future consisting of the particle o + subjunctive, the colloquial future consisting of fully inflecting have + subjunctive (40a), and the future consisting of the imperfective past of have + subjunctive (40b) are not auxiliary-based forms of the same type as the periphrastic formations discussed above. In contrast to the typical auxiliary-based forms, which are characterized by obligatory clitic climbing and a single negator, with these subjunctive-based periphrases pronominal object clitics remain within the domain of the subjunctive verb (40) and negation may also marginally adjoin to the subjunctive (41); note that DobrovieSorin (1994) and Avram (1999) argue that the have + subjunctive configuration (40a) instantiates a biclausal structure. (40)

a.

o / am săl ascult. fut have.prs.ind.1sg sbjv =acc.3msg listen.prs.1sg ‘I will listen to him.’ b. aveam săl ascult. have.pst.ipfv.1sg sbjv =acc.3msg listen.sbjv.prs.1sg ‘I was going to listen to him.’

21 While the presumptive is traditionally claimed to be a so-called Balkan feature (see Friedman’s 1986 discussion of the Bulgarian, Macedonian, and Albanian presumptive), the particular forms discussed here, the so-called Romanian presumptive, are more similar to the Romance epistemic future, both from the point of view of their values, and from the point of view of their origin (see Squartini 2005). 22 Note that, according to a long line of generative scholarship (see, for example, Cornilescu 2000; Schifano 2018), finite synthetic verb forms systematically raise to the mood head in modern Romanian. Nicolae (2020) develops this hypothesis in more detail.

loss of analyticity in romanian verbs

(41)

269

nu o / am / aveam să not fut have.prs.1sg have.pst.ipfv.1sg sbjv (?nu) ascult. not listen.prs.1sg ‘I will not listen / I wasn’t going to listen.’

The perfect subjunctive and perfect infinitive (42) are not problematic either on account of the cluster-internal distribution of functional elements. In contrast to auxiliaries, the formatives a and să occur to the left of clitics and the clausal negator and have been analysed as occurring in the complementizer domain (see Dobrovie-Sorin 1994; Alboiu and Motapanyane 2000; Gheorghe 2013; Nicolae 2015:§III). Hence, for these forms, the division of labour in the marking of tense, aspect, and mood is shared between the complementizer domain and the inflexional domain; besides the lexical verb, the sole other component internal to the inflexional domain, is a mood marker—the invariable auxiliary be, the exponent of irrealis (just as in the case of the conditional perfect, perfect future, and gerundial epistemic periphrases)—a fact which ensures the diachronic stability of the periphrases. (42)

să / a nu îl fi văzut sbjv to not acc.3msg= be.aux.inf see.ptcp ‘(s)he should/to not have seen him.’

By contrast, periphrases such as the double compound perfect subjunctive become diachronically extinct for the same reason as the group auxiliary-based forms discussed in the previous section: one of their auxiliaries is the past participle form of be, an overt bearer of tense features. (43)

Află-să această țară să fie discover.prs.ind.3sg=refl.pass this country sbjv be.aux.sbjv.3sg fostu lăcuit și alții într-însa be.aux.ptcp live.ptcp also others in=it mai nainte de noi (ULM.~1725:3v ) before of us ‘one may discover that in this country others had lived before us.’

9.5.3 The relevance of analytic formations which were preserved dialectally The dialectal data presented in §9.4 appear to challenge our analysis. However, if we focus on the usage of these verbal periphrases, we note that most of them are very rarely attested or are attested in very isolated areas. Our analysis has focused on the standard language, therefore the dialect data do not challenge the analysis,

270

adina dragomirescu, alexandru nicolae, and rodica zafiu

as they are not part of the same variety. The only non-standard periphrasis which continues in many areas and has a relatively high frequency is that illustrated in (24), the am fost dat type (44a). The contrast in (44) shows that in this particular periphrasis (44a), the auxiliary avea ‘have’ still exclusively expresses mood, the tense-aspect value being expressed by the second part of the structure (fost + past participle). (44)

a.

am [fost dat] vs have.aux.1sg be.aux.ptcp give.ptcp b. am [dat] have.aux.1sg give.ptcp

9.6 Conclusions: the loss of analyticity in a wider perspective On the basis of the empirical data analysed in this chapter, it has been observed that Romanian privileges mood: auxiliaries systematically grammaticalize as exponents of mood, and this has consequences for their internal structure and for the periphrases they are part of. Periphrases with auxiliaries whose feature matrix is richer and also includes tense features fail to enter the periphrastic system of standard Romanian (and, in some cases, give way to functionally equivalent synthetic formations). Anna Giacalone Ramat’s (2000:125) remark that ‘members of the category “auxiliary” exhibit differences in their degree of grammaticalization and are located at different points along the “Verb-to-TAM” chain’ is highly relevant in this respect. Another consequence of our analysis is that, besides the fact that some synthetic formations prevail over functionally equivalent periphrastic formations (whose internal structure does not observe the mood-oriented option) on account of the mood-oriented nature of the Romanian system, there are values which simply fail to find a morphological exponent. This is the case with the progressive, gerund-based periphrases, whose sole relics in modern Romanian are the gerundial presumptive and the very limited subjunctive and conditional gerundial periphrases.

Corpus ACT.1709 Antim Ivireanul, Chipurile Vechiului și Noului Testament. Ed.: Antim Ivireanul, Opere, ed. G. Ștrempel, Bucharest: Minerva, 1972, 240–321.

loss of analyticity in romanian verbs

271

BB.1688 Biblia. Ed.: Biblia adecă Dumnezeiasca Scriptură a Vechiului și Noului Testament, tipărită întâia oară la 1688 în timpul lui Șerban Vodă Cantacuzino, Domnul Țării Românești, Bucharest: Editura Institutului Biblic, 1977. CC1 .1567 Coresi, Tâlcul Evangheliilor. Ed.: Coresi, Tâlcul evangheliilor și molitvenic românesc, ed. V. Drimba, Bucharest: Editura Academiei Române, 1998, 31–187. 2 CC .1581 Coresi, Evanghelie cu învățătură. Ed. S. Pușcariu, Al. Procopovici: Diaconul Coresi, Carte cu învățătură (1581), vol. I, Textul, Bucharest: Socec, 1914. CH.1717–1723 Dimitrie Cantemir, Hronicul vechimei a romano moldo-vlahilor, ed. S. Toma, Bucharest: Minerva, 1999−2000, 1–274 (vol. I), 5–223 (vol. II). CPr.1566 Coresi, Apostol. Ed. I. Bianu, Texte de limbă din secolul XVI, IV, Lucrul apostolesc tipărit de diaconul Coresi la 1563, Bucharest: Cultura Națională, 1930. CS Codex Sturdzanus. Ed. Gh. Chivu, Bucharest: Editura Academiei Române, 1993, 237–300. CSVI .1590–602—Legenda lui Sisinie. CT.1560–1561 Coresi, Tetraevanghel. Ed.: Tetraevanghelul tipărit de Coresi. Brașov 1560–1561, comparat cu Evangheliarul lui Radu de la Mănicești. 1574, ed. F. Dimitrescu, Bucharest: Editura Academiei, 1963. CV.1563–1583 Codicele Voronețean. Ed. M. Costinescu, Bucharest: Editura Academiei Române, 1981, 229–400. DC.a.1794 Departamentul de cremenalion. Din activitatea unei instanțe penale muntene (1794–1795). Ed. L. Livadă-Cadeschi, L. Vlad, Bucharest: Nemira, 2002 (Anaforaua postelnicului Gheorghe Berindei […], a. 1794, 92–95] DÎ Documente și însemnări românești din secolul al XVI-lea, text stabilit și indice de Gh. Chivu, M. Georgescu, M. Ioniță, Al. Mareș, Al. Roman-Moraru, Bucharest: Editura Academiei, 1979. EOXIII.1882 M. Eminescu, Opere, XIII. Publicistică (1882–1883, 1888–1889). Ed. A. Oprea, Bucharest: Editura Academiei, 1985. NL.~1750–1766 Ion Neculce, Letopisețul. Ed.: Ion Neculce, Letopisețul Țării Moldovei și O samă de cuvinte, ed. I. Iordan, Bucharest: Editura de Stat pentru Literatură și Artă, ed. a II-a, 1959, 31–388. PO.1582 Palia de la Orăștie. Ed. V. Pamfil, Bucharest: Editura Academiei, 1968. Prav.1581 Pravila ritorului Lucaci. Ed. I. Rizescu, Bucharest: Editura Academiei, 1971, 161–83. Prav.1652 Îndreptarea legii. 1652, ed. Colectivul pentru vechiul drept românesc condus de acad. A. Rădulescu, Bucharest: Editura Academiei, 1962 (Adunarea izvoarelor vechiului drept românesc scris, 7), 33–631. ULM.~1725 Grigore Ureche, Letopisețul Țării Moldovei. Ed. P.P. Panaitescu, Bucharest: Editura de Stat pentru Literatură și Artă, 1955, 57–210.

10 The relation between inflexional and analytic marking of obliques in Romanian Gabriela Pană Dindelegan and Oana Uță Bărbulescu

10.1 Introduction: marking of oblique functions in modern standard Romanian The strategies used for what is traditionally labelled synthetic marking are different in the inflexion of nouns (and adjectives), on the one hand, and in pronouns on the other. In the inflexion of nouns (and adjectives), morphological case is regularly marked on the accompanying determiner head, as the following examples demonstrate (observe that in Romanian, definiteness is marked inflexionally through a suffix).1 (1)

modern Romanian a. cartea book.def.adv.fsg ‘the boy’s book.’ b. buna good.def.adv.fsg ‘My good friend.’

băiatului boy.def.adn.msg mea prietenă my.adv.fsg friend.adv.fsg

In the definite declension, all nouns distinguish adverbal and adnominal syncretic forms, both in the singular and in the plural (if countable), as illustrated in Table 10.1. Only in the feminine singular declension are oblique functions overtly marked by discrete inflexional endings distinct from those of the adverbal series, but

1 Following Maiden (2016a:100), we employ the glosses adv(erbal) and adn(ominal) to denote the nominative–accusative and genitive–dative inflexional forms, respectively.

Gabriela Pana˘ Dindelegan and Oana Ut, a˘ Ba˘rbulescu, The relation between inflexional and analytic marking of obliques in Romanian In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Gabriela Pana˘ Dindelegan and Oana Ut¸a˘ Ba˘rbulescu (2022). DOI: 10.1093/oso/9780198870807.003.0011

inflexional and analytic marking of obliques in romanian

273

Table 10.1 Definite declension of modern Romanian feminine and masculine nouns

sg pl

Feminine Adverbal fata girl.def.adv.fsg fetele girls.def.adv.fpl

Adnominal fetei girl.def.adn.fsg fetelor girls.def.adn.fpl

Masculine Adverbal băiatul boy.def.adv.msg băieții boys.def.adv.mpl

Adnominal băiatului boy.def.adn.msg băieților boys.def.adn.mpl

identical—for the overwhelming majority of feminine nouns—with the plural endings: (2)

modern Romanian a. cartea unei fete book.def.adv.fsg a.adn.fsg girl.adn.fsg ‘the book of a girl.’ b. câteva fete few.adv.fpl girls.adv.fpl ‘a few girls.’ c. câtorva fete few.adn.fpl girls.adn.fpl ‘of/to a few girls.’

The binary adverbal vs adnominal distinction present in the definite nominal declension also characterizes the declension of demonstrative determiners and (personal, relative, indefinite, and some negative) pronouns. The unambiguous adnominal morphemes (msg -ui, fsg -ei, pl -or) are inherited from Latin.2 In the pronominal declension, case inflexion is better preserved, not only for tonic pronouns, but also for clitics. In contrast to the declension of nouns, some case oppositions in pronouns are marked suppletively, suppletion being sometimes inherited from Latin and sometimes phonologically motivated. 2 These are the simple adnominal inflexions found in the paradigms of pronouns, determiners, and quantifiers (e.g., acest-ui ‘this.adn.msg’, acest-ei ‘this.and.fsg’, acest-or ‘this.adn.pl’). However, in the definite inflexional paradigms of nouns (and adjectives in prenominal position) suffixation of the definite article has led to a merger with the nominal inflexional ending, such that original sixteenthand early seventeenth-century forms such as case-ei ‘house.adn.f-def.adn.fsg’, lumi-ei ‘world.adn.fdef.adn.fsg’ (where one can transparently identify singular (and plural) feminine adnominal endings -e and -i to which has been added the feminine singular definite adnominal suffix -ei) have today been replaced by the forms casei and lumii, which display the singular (and plural) feminine adnominal endings -e and -i and the reduced feminine singular definite adnominal suffix -i. Note, furthermore, that in the feminine nominal declension the adnominal inflexional ending, whether -e or -i, is the same for the singular and the plural (and also overlaps with the plural adverbal cell of the paradigm, cf. cas-e ‘house.adv.fpl’ and lum-i ‘world.adv.fpl’). Their origin remains controversial.

274

gabriela pană dindelegan and oana uță bărbulescu

The inventory of elements involved in what is traditionally described as analytic marking includes the prepositions a (< ad ‘to(wards)’), de (< de ‘(down/away) from’), la (< illac ad ‘(over) there towards’), and către (< contra ‘against’), all of which have been in use since old Romanian. Besides these prepositions, Romanian also uses two other means: a non-prepositional marker lui, a free-standing proclitic form of the definite article (< ille ‘that’), and the so-called prepositional determiner al (< (ad) ille ‘(to/at) that’; cf. Ledgeway 2012:116), a functional proclitic used in possessor constructions displaying agreement in number and gender with the possessee (viz. msg al, mpl ai, fsg a, fpl ale). (3)

modern Romanian a. cartea lui Ion book.def.adv.fsg lui Ion ‘Ion’s book.’ b. cartea plicticoasă a băiatului book.def.adv.fsg boring.adv.fsg al.fsg boy.def.adn.msg ‘the boy’s boring book.’

In standard Romanian each of these analytic markers displays a distinct distribution (cf. also discussions in §§1.2.1–1.2.1.2 in this volume). For example, the preposition a is specialized for possessor phrases realized as quantified nominals or relative clauses introduced by an invariable wh-pronoun: (4)

modern Romanian a. Cartea a doi book.def.adv.fsg a two ‘the books of two boys.’ b. Este împotriva be.prs.ind.3sg against ‘It is against what I believe.’

băieți boys.adv.mpl a ce cred. a what believe.prs.ind.1sg

By contrast, the free-standing proclitic definite article lui occurs with (masculine and indeclinable feminine) proper names: (5) Cartea lui Lili și a lui Martin (Ro.) book.def.adv.fsg lui Lili.f and al.fsg lui Martin.m ‘Lili and Martin’s book.’ Furthermore, the distribution of functional al must obey the condition of nonadjacency with the definite determiner, hence its presence in (3b) but its absence in the corresponding string without the adnominal adjective where the possessor băiatului now immediately follows the definite noun cartea (viz. cartea (**a) băiatului). Inflexional and analytic marking are in complementary distribution, such that if inflexional marking is available (cf. (6)), then proclitic lui or prepositional markers such as a or la are excluded:

inflexional and analytic marking of obliques in romanian

275

(6) Cartea (**lui) Tomei (Ro.) book.def.adv.fsg lui Toma.adn.msg ‘Toma’s book.’ On the other hand, functional al may co-occur with inflexional marking since, as already noted (cf. (3b)), its distribution is determined by definiteness marking: (7) cartea băiatului și a fetei (Ro.) book.def.adv.fsg boy.def.adn.msg and al.fsg girl.def.adn.fsg ‘the boy’s and the girl’s book.’ These distributional restrictions found in the standard language do not necessarily hold in the same way in the spoken variety or the dialects. For instance, in Aromanian a and inflexional marking can co-occur (8). (8) a kasәλej (Aro.) a house.def.adn.fsg ‘of the house’ The focus of this chapter is the competition between inflexional and prepositional marking of oblique functions; the free-standing proclitic definite article lui and the so-called prepositional determiner al will not therefore be taken further into account. In what follows, we will provide a brief outline of the most important data from old Romanian, after which we will take a closer look at the contexts in which the inflexional forms contrast and/or compete with analytic prepositional markers, or in which a remarkable diversity of overlapping prepositional expressions occurs in old Romanian. We will also examine the competition between the inflexional forms and the prepositional markers in modern Romanian, since this will allow us to draw a more nuanced comparison between old Romanian and modern standard and non-standard varieties of Romanian. The discussion of the data falls into three parts. In the first (§10.2) we consider texts between the sixteenth and eighteenth centuries, investigating different analytic constructions in order to establish to what extent their patterns of use fit with the development attested in modern Romanian (§§10.2.1–10.2.4). The discussion in §10.2.4 focuses on quantifiers, since they represent an area in which there is most variation, apparently free in some cases, between inflexional and prepositional marking. In the second part (§10.3) we discuss the situation in modern Romanian, taking into consideration not only the standard language, but also non-standard registers. In the final part (§10.4) we draw some conclusions about the evolution of the marking of oblique functions in the history of Romanian and its theoretical implications.

276

gabriela pană dindelegan and oana uță bărbulescu

10.2 Old Romanian From the first extant Romanian texts (early sixteenth century), analytic prepositional expressions are in use alongside synthetic inflexional realizations of oblique functions.

10.2.1 Distribution of inflexional and analytic markers: an overview The inventory of prepositional markers available in old Romanian to convey the oblique functions includes: (a) in the nominal domain, the prepositions a (9a–b), de (10), and la (11): (9)

old Romanian a. Întru știrea [a toate limbile] și for knowledge.def a all.adv.fpl peoples.def.adv.fpl and [a toată lumea] a all.adv.fsg humanity.def.adv.fsg ‘to the knowledge of all nations and humanity.’ (CPrav.1560–2:11v ) b. slăbiciunea lui Dumnezău mai puternică-i weakness.def lui God more strong.fsg=be.prs.ind.3sg decât [a oameni] than a people.adv.mpl ‘the weakness of God is stronger than that of men.’ (NT.1648:225r )

(10) deasupra [de hălășteu] (ORo.) above de lake.adv.msg ‘above the lake.’

(DÎ.1576:V)

(11) numele [la doi greci viteazi] (ORo.) name.def la two Greek.adv.mpl brave.adv.mpl ‘the name of two courageous Greek warriors.’ (CDicț.1691–1697:12v ) (b) in the verbal domain, the prepositions a (12a–b), la (13), and cătră/către (14) old Romanian (12) a. cuvine-se [a bârbat înțelept] ought=refl.3.acc a man.adv.msg wise.adv.msg (CT.1560–1561:13v ) ‘a wise man ought to […].’ b. nu se cade [a doi frați] not refl.3.acc= ought a two brothers.adv.mpl să se împreune cu doao veare să.sbjv refl.3.acc= unite.sbjv.3 with two.f cousins primari primary ‘two brothers ought not to unite with two primary cousins.’ (Prav.1581:263v )

inflexional and analytic marking of obliques in romanian

277

(13) Scriem dumitale birău de write.prs.ind.1pl you.2sg.adn mayor de Bistrițî și [la toți giurații Bistrița.adv and la all.adv.mpl jurymen.def.adv.mpl den oraș] from city ‘We are writing to you, the mayor of Bistrița and to all the jurymen of the city.’ (ORo., SB.XVII:63) (14) celora ce aduseră răslăbitulu cel.pl.adn who bring.pfv.pst.3pl weak.def.msg [cătră bunulu vraciu și cătră good.def.adv.msg doctor.adv.msg and mântuitoriului nostru] saviour.def.adn.msgour.msg] ‘to those who brought the paralytic to our merciful doctor and our saviour.’ (ORo., CC2 .1581:243) It is worth noting that constituents introduced by a and la are used respectively as equivalents of inflexional case markers introducing Possessors and Recipients, while de introduces possessive phrases and cătră/către indirect objects. The starting point for the de-construction is late Latin, where ‘the de-expression moves closer to a proper possessive/partitive-genitive equivalent’ (Adams 2013:271) in contexts in which the idea of separation and removal was lost. As for the competition between ad and the dative of the indirect object, the ‘analysis of the use of the preposition reveals that until the medieval period it was not equivalent to the dative at all’ (Adams 2013:260).3 None the less, ‘phonetic changes causing a loss of distinctiveness of dative and other endings might have been one factor at a late period motivating the fading of the inflectional category, and from that time ad, given its long-established overlap with the dative (particularly with verbs of saying), presumably widened its functions’ (Adams 2013:294). That explains the shared function across multiple Romance languages (Salvi 2011:335),⁴ and its grammaticalized status in old Romanian (Stan 2016:317f.). La and către began the process of grammaticalization in Romanian at a later stage and developed from locative/allative lexical prepositions to case markers (and to some extent they replicate the development of a). Old Romanian texts show competition between inflexional and prepositional marking. This competition between synthetic and analytic realizations is instantiated by: (i) the possibility of double marking within the same noun phrase of 3 See too Pinkster (1990:202); Baños Baños (2000). For a different view, see Ciorănescu (2002:15). ⁴ See also Adams and de Melo (2016:87f.); for old French, see Buridant (2000:474f.) and Va¨a¨na¨nen (1956:15f.); for Italian, see Maiden (1995:100), for old Neapolitan, see Ledgeway (2009:125f.).

278

gabriela pană dindelegan and oana uță bărbulescu

oblique relations: prepositional marking in combination with the quantifier, but inflexional marking on the nominal head (15a–b); (ii) the possibility of mixed marking in coordinated structures where one or more of the coordinated elements bears synthetic marking, whereas other coordinated constituents are marked analytically (16a–b); (iii) alternation, within the same text, of de-expressions (as in the (a) examples) and inflexional genitives (as in the (b) examples) in conjunction with the same selecting nominal head ((17)–(18)).

(15)

(16)

(17)

old Romanian a. muma [a toți viilor] mother.def a all.adv.mpl alive.def.adn.pl ‘the mother of all the living.’ (PO.1582:21) b. sântem slugi credincioase [a toată be.ind.prs.1pl servants loyal a all.adv.fsg creștinătăței] Christendom.def.adn.fsg ‘we are loyal servants of all Christendom.’ (DÎ.1600:XLIV) old Romanian a. scriu închinăciune și multă sănătate write.prs.ind.1sg prayer.adv and much health.adv [priiatiniloru nos¸tri], […] și [a friends.def.adn.mpl our.mpl and a doisprădzeace pârgari] twelve administrative.heads.adv.mpl ‘I’m writing a prayer of health to our friends […] and to twelve of your administrative heads.’ (DÎ.1595:CIII) b. grăiaște [cătră ai săi ucenini] și say.prs.ind.3sg cătră al.mpl his.mpl disciples.adv.mpl and [apostolilor] apostles.def.adn.mpl ‘talks to his disciples and apostles.’ (CC2 .1581:88) old Romanian a. robul [de Domnulu] slave.def de God.def.adv.msg ‘God’s slave.’ (CC2 .1581:508) b. iară alalți toți era robi și slugile and others all be.ipfv.pst.3pl slaves and servants.def.fpl [Domnului] God.def.adn ‘and all the others were God’s slaves and servants.’ (CC2 .1581:524)

inflexional and analytic marking of obliques in romanian

(18)

279

old Romanian a. eu tremeț voi ca oile în mijloc I send.prs.ind.1sg you like sheep.def.fpl in middle [de lupi] de wolves.adv.mpl ‘I am sending you like sheep into the middle of a pack of wolves.’ (CC1 .1567–1568:51v ) b. eu tremeț voi ca oile pre I send. prs.ind.1sg you like sheep.def.fpl in mijlocul [lupilor] middle.def wolves.def.adn.mpl ‘I am sending you like sheep into the middle of a pack of wolves.’ (CC1 .1567–1568:230v )

Competition between inflexional and prepositional realizations has been observable since the earliest texts. None the less, synthetic realizations are quantitatively predominant across all texts, as summarized in Table 10.2. Table 10.2 Distribution of inflexional and prepositional markers of oblique functions in old Romanian Texts1 DÎ.16th c. (whole text) PO.1582 (pp. 50–70)

Inflexional genitive

Inflexional dative

Prepositional genitive

Prepositional dative

80

44

8

9

Inflexional vs analytic (%) 88 vs 12

128

148

6

6

95.8 vs 4.2

1 For old Romanian, DÎ consists of non-translated documents (private letters and administrative papers), and PO.1582 is a translated religious text.

10.2.2 Analytic markers: distribution and competition As shown in Table 10.2, in the non-translated texts of the sixteenth century analytic a-constructions equivalent to adnominal inflexional marking are rarely attested (see also Iliescu 2008:65). It is only in translations that we find contexts where the grammaticalized marker a introduces the indirect object, which may be expressed as a bare noun (19a), a modified noun with a generic reading (19b), or a definite DP (19c).⁵ ⁵ In these contexts, prepositional constructions cannot be explained as a simple imitation of similar patterns found in the source texts, since the Slavonic texts translated into Romanian use an inflexional dative case, and not prepositional expressions. They might therefore be interpreted as the result of a deliberate ‘Romanianization’ on the part of the translators, since the Slavonic inflexional dative is

280

gabriela pană dindelegan and oana uță bărbulescu

old Romanian (19) a. o deade [a lucrători] 3fsg.acc= give.pfv.pst.3sg a workers.adv.mpl ‘he leased it to workers.’ (CT.1560–1561:96r ) b. nu fromos iaste [a bărbat ovreescu] not beautiful be.prs.ind.3sg a man.adv.msg Jewish.adv.msg ‘it is unlawful for a Jewish man.’ (CB. 1559–1560:111) c. slujiră [a dumnezeii păgânești] serve.pfv.pst.3pl a gods.def.adv.mpl pagan.adv.mpl ‘they served the pagan gods.’ (SVI.~1670:224v ) The analytic construction finds itself in competition with the adnominal inflexional forms, sometimes even in the same text, as illustrated by the contrast between (19a) and (20): (20) o deade el [lucrătorilor] (ORo.) 3fsg.acc= give.pfv.pst.3sg he workers.def.adn.mpl ‘he leased it to the workers.’ (CT.1560–1561:166r ) In the sixteenth century these analytic realizations prove to be very rare even in translated texts. Possessive constructions marked by the preposition a where the head noun is a bare noun or a modified noun are even rarer than nominal indirect objects marked by the preposition a. In particular, old Romanian only offers a few examples, variously involving a bare noun (21a), a modified noun (21b), and a modified noun with generic reading (21c). old Romanian (21) a. trestie [a cărtulariu] reed a scholar.adv.msg ‘pen of a scribe.’ (PS.1573–1578:140) b. răssipele [a mare case] loss.def.fpl a big.adv.fpl houses.adv.fpl ‘the fall of the illustrious families.’ (CLM.1700–1750:281r ) c. să vor prinde de poala [a refl.3.acc= aux.fut.3pl hang.inf of lap.def a bărbat iudeu] man.adv.msg Jewish.adv.msg ‘they will take hold of the robe of a Jew.’ (DPar.1683:I.7v )

not only translated in old Romanian with a in combination with nominal forms incapable of hosting inflexional case (viz. adnominal) forms (cf. (19a)), but also in combination with definite nominals which are otherwise capable of marking adnominal inflexional distinctions (cf. (19c)).

inflexional and analytic marking of obliques in romanian

281

Their distribution by region is not uniform: possessive functions marked by a are attested especially in northern, more conservative texts (see also Gheție 1997:150), whereas prepositional marking of indirect objects is attested both in northern and southern texts. The distribution of de shows different usages, some of which are emergent in (late) Latin (see, among others, Densusianu 1938:143f.; Adams 2013:268). In texts from the sixteenth century de is already completely grammaticalized as a case marker. In the nominal domain, de-constructions expressing possession, destination, or definition are found in several patterns (for more examples, see Densusianu 1938:143f.; Stan 2013b:75), where the prepositional marker is followed by a definite noun (22a), a bare noun (22b), and a definite noun modified by a possessive (22c). old Romanian (22) a. casa [de Domnul] house. def de Lord.def.adv.msg ‘the house of the Lord.’ b. calea [de cetate] road.def de city.adv.fsg ‘the road of the city / the city road.’ c. în dzua [de rreulu mieu] in day.def de evil.def.adv.msg my.msg ‘in the day of my calamity.’

(PS.1573–1578:424)

(PS.1573–1578:356)

(PH.1500–1510:12v )

In sixteenth-century texts we also find a construction where de is followed by a toponym (23), a construction considered to be a calque from Slavonic (Densusianu 1938:144). (23) lui Budachi Gașpar, birăului [de Bistriță] (ORo.) lui.dat Budachi Gașpar mayor.def.adn.msg de Bistrița.adn ‘to Budachi Gașpar, the mayor of Bistrița.’ (DÎ.1592:LXXXII) The structures in (22a–c) are attested in translations where they compete with inflexionally marked possessives: old Romanian (24) a. casa [Domnului] house.def Lord.def.adn.msg ‘the house of the Lord.’ b. calea [cetației] road.def city.def.adn.fsg ‘the road of the city/the city road.’

(PH.1500–1510:249v )

(PH.1500–1510:92v )

After the second half of the seventeenth century they occur very rarely:

282

gabriela pană dindelegan and oana uță bărbulescu

old Romanian (25) a. era om [de casa be.ipfv.pst.3sg man de house.def.adv.fsg împăratului] emperor.def.adn.msg ‘he was one of the emperor’s court members.’ b. beiul [de Moldova] bey.def de Moldova.adv.f ‘the governor of Moldova.’

(NL.~1750–66:201)

(NL.~1750–66:209)

A better attested structure in the same period is one in which the de-expression occurs after complex prepositions (such as în/pre/pren mijloc ‘in the middle of ’). Prepositional de could combine with a bare noun (26a), a (personal) pronoun (26b), and a quantifier (26c): old Romanian (26) a. în mijloc [de besearecă] in middle de church.adv.fsg ‘in the middle of the church.’ b. pre mijloc [de ea] on middle de it.fsg.adv ‘through the midst of it (= sea).’ c. în mijloc [de mulți] in middle de many.adv.pl ‘in the midst of many.’

(PH.1500–1510:17r )

(PH.1500–1510:116r )

(PH.1500–1510:97r )

As can be seen in the examples in (27a–b), the selection of an inflexional adnominal case form (when available) depends on the definite or indefinite form of the nominal in the spatial expression or complex preposition: old Romanian (27) a. în mijlocul [caseei meale] in middle.def house.def.adn.fsg my.adn.fsg ‘in the middle of my house.’ (PH.1500–1510:82v ) b. în mijloc [de ucenicii săi] in middle de apostoles.def.adn.mpl his.adn.mpl ‘in the midst of his apostles.’ (CC2 .1581:136) In old Romanian, especially following the second half of the seventeenth century, de-phrases begin to appear as complements of deverbal adjectives (28a) and long infinitives (28b): old Romanian (28) a. [de doao limbi] înțelegători de two languages.adn.fpl who.knows.msg ‘the one who understands two languages.’ (CDicț.1691–1697:37r )

inflexional and analytic marking of obliques in romanian

b. isprăvire [de cei doi tâlhari] end de cel.mpl two thieves.adn.mpl ‘the execution of the two thieves.’

283

(LC.1650:424r )

Deverbal adjectives can be contextually substantivized and are also followed by a de-phrase. (29) împăratul și făcătoriul [de leage] al emperor.def and maker.def de law.adv.fsg al.msg spartaneanilor (ORo.) Spartans.def.adn.mpl ‘the ruler and the creator of the law of Sparta (= the legislator).’ (CDicț.1691–1697:179r ) Since old Romanian, there has not only been competition between synthetic (viz. inflexional) and analytic (viz. prepositional) marking, but also competition among various prepositional markers. In the nominal domain, different possessive constructions variously show competition between marking by a (30a) and de (30b) and between de (31a) and la (31b). Similarly, in the verbal domain marking of indirect objects shows competition between la (32a) and cătră (32b), or, as a result of their competition in the nominal domain (cf. (30a), (31b)), la and a may even co-occur (33). old Romanian (30) a. vrednicia [a zeace boiari] diligence.def a ten boyars.adv.mpl ‘the council of ten boyar(s).’ (= decemvirate) (CDicț, 1691–1697:77v ) b. vrednicie [de cinci persone] diligence de five persons.adv.fpl ‘council of five persons.’ (CDicț, 1691–1697:276r ) old Romanian (31) a. iubeaște să se învârtească împrejur loves să.sbjv refl.3.acc= spin.sbjv.3 around [de oameni] de people.adv.mpl ‘he loves being around people.’ (FD.1592–1604:576r ) b. groapă împrejur [la cetăți și la orașuri] grave around la fortresses.adv.fpl and la cities.adv.fpl ‘a grave around fortresses and cities.’ (CDicț.1691–1697:118v ) old Romanian (32) a. Zise șarpele [la muiare] say.pfv.pst.3sg serpent.def la woman.adv.fsg ‘The serpent said to the woman.’

(PO.1582:18)

284

gabriela pană dindelegan and oana uță bărbulescu

b. Și dzise Domnul Domnezeu [cătră muiare] and say.pfv.pst.3sg lord.def God cătră woman.adv.fsg ‘God said to the woman.’ (PO.1582:19) (33) de-aciia se-au arătat împreună from=here refl.3.acc=aux.pfv.3pl show.ptcp together [la a cinci sute de frați] (ORo.) la a five hundred de brothers.adv.mpl ‘Since then they have been shown together to five hundred brothers.’ (CC2 .1581:133/21) With verba dicendi, both inflexional (34) and analytic (35a–b) marking are found. (34) grăiaște și [noao] Hristos, [slăbiților de talk.prs.ind.3sg also us.adn Christ weak.def.adn.mpl of suflet] (ORo.) soul ‘Christ talks to us, the helpless.’ (CC2 .1581:146) old Romanian (35) a. tu grăiaște [a tot nărodul] you talk.imp.2sg a all.adv.msg people.def.adv.msg ‘Talk to all the people!’ (CM.1567:234r ) b. Grăiaște amu Dumnezeu [cătră Isaia proroc] talk.prs.ind.3sg now God cătrăIsaiah.adv.mprophet ‘And then God talks to Isaiah the prophet.’ (CC2 .1581:51) c. cela ce [la voi] grăiaște ca cel.msg who la you.adv talk.prs.ind.3sg like [la feciori] la sons.adv.mpl ‘that one who talks to you as if you were his sons.’ (CPr. 1566–7:570) Since old Romanian, the four prepositional case markers have displayed different degrees of grammaticalization. A and de are grammaticalized as case markers. Among other things, they have undergone semantic bleaching, and they display paradigmatic variability alternating with inflexional forms. The more recent markers la and către, although showing signs of grammaticalization, are far from having completed a full grammaticalization cline. For la and către, signs of grammaticalization are obvious from: (i) doubling (realized either as clitic left-dislocation or clitic right-dislocation) of the prepositional constructions with a dative pronominal clitic (36a–b); (ii) coordination of the prepositional construction with adnominal inflexional forms (37a) or equivalence with an apposition displaying agreement

inflexional and analytic marking of obliques in romanian

285

and inflexional realization (37b); and (iii) the extension of la from the end of the seventeenth century as a marker of possession by analogy with the dative-genitive syncretism of Romanian inflexional forms (38a–b). In the nominal domain the grammaticalization of la is advanced, inasmuch as there are no signs of any connection to verbs of giving or saying or to verbs involving a Recipient which would explain the original prepositional construction. old Romanian (36) a. cuvente bune și frumoasece [ne]-ai words good and beautiful thatus.dat=aux.pfv.2sg scris dumneata [la noi] write.ptcp you.adv.2sg la us.adv ‘good and sweet words that you’ve written us.’ (SB.XVII:85) b. [cătră aceștea] și sfânta evanghelie cătră these.adv.mpl also holy.def.adn.fsg book.adn.fsg grăiaște-[le] say.prs.ind.3sg=3pl.dat ‘the holy Bible talks to them.’ (CC2 .1581:17) old Romanian (37) a. ei au fost mers și they aux.pfv.3pl be.ptcp go.ptcp and fost [la au grăit și aux.pfv.3pl be.ptcp talk.ptcp also la Măria sa] și [lui Piri] highness.adv.fsg his.adv.fsg and lui Piri ‘they had gone and had talked to His Highness and Piri as well.’ (ISBD.1593:XX) b. dzise [cătră muiarea sa], say.pfv.pst.3sg cătră wife.def.adv.fsg his.adv.fsg [Sarăei] Sara.adn.fsg ‘he said to his wife, Sara.’ (PO.1582:44) old Romanian (38) a. nume [la câțiva înțelepți] name la few.adv.mpl wise.men.adv.mpl ‘the name of a few philosophers.’ b. numele [la trei oameni] name.def la three men.adv.mpl ‘the name of three men.’

(CDicț.1691−1697:37r )

(CDicț.1691−1697:343v )

286

gabriela pană dindelegan and oana uță bărbulescu

In old Romanian these uses remain marginal, as la and cătră occur mainly as fully fledged prepositions introducing spatial/locative adjuncts (39a). Sometimes the two uses may occur within the same text and generate ambiguous readings (39b). old Romanian (39) a. cum grăiaște [la cartea] lu proroc David how say.prs.ind.3sg in book.def.adv lui prophet David ‘how it is said in the book of David the Prophet.’ (CC1 .1567:37v ) b. cum grăiaște Pavel apostol [la Rimleani] how say.prs.ind.3sg Paul apostle la Romans.adv.mpl ‘how Paul the Apostle talks in (the Epistle to the) Romans’ or ‘how Paul the Apostle talks to Romans.’ (CC1 .1567:134v )

10.2.3 Mixed structures Mixed patterns of inflexion and periphrasis are attested in old Romanian. One pattern involves a preposition in combination with a synthetically inflected noun/pronoun, although a proves rare in such instances. Most probably, the overlap between both types of periphrastic and inflexional marking was made possible precisely because of their equivalence. old Romanian (40) a. cine poate sluji [a oamenilor] who can.prs.ind.3sg serve.inf a people.def.adn.mpl mai mult more much ‘whoever can serve more people.’ (CC1 .1567:199r ) b. totu se poate [a credinciosului] all refl.3.acc= can.prs.ind.3sg a believer.def.adn.msg ‘all things are possible for one who believes.’ (CC2 .1581:80) In one example from the Floarea darurilor there is a mixed structure where the preposition is followed by the adnominal inflexion of the relative pronoun cine: (41) oamenii [a cui] trebuiia asini (ORo.) men.def.adv.mpl a who.adn need.ipfv.pst.3pl donkeys.adv.mpl ‘people who needed donkeys.’ (FD.1592–1604:557v ) The preposition a is followed by the adnominal form of the pronoun cine ‘who’ (viz cui), which is used instead of the relative pronoun care ‘which’ (oamenii cărora ‘men to whom’), a phenomenon preserved until the present day in southern Banat

inflexional and analytic marking of obliques in romanian

287

(cf. ALR, sn, VI, h. 1685). As a relative pronoun, cine introduces headless relatives with [−animate] reference,⁶ whereas care usually introduces headed relatives that restrictively or non-restrictively modify one of the constituents of the matrix clause. The relative pronoun cine thus marks in (41) formal case restrictions imposed by the embedded clause (cuiadn trebuiia). Another pattern involves a prepositionally marked indirect object doubled by a dative clitic, as in (42a–c): old Romanian (42) a. [a mulți] nu [le] se arată a many.adv.mpl not 3pl.dat= refl.3.acc= show.prs.ind.3pl dulci sweet.pl ‘to many they do not seem tasteful.’ (CC2 .1581: 263) b. pare-[le]-se [a mulți] răotate seem.prs.ind.3sg=3pl.dat=refl.3.acc= a many.adv.mpl evil.fsg a fi a.inf be.inf ‘it seems to many to be an evil act.’ (CC2 .1581: 402) c. [a mulți] [lă] e a many.adv.mpl 3pl.dat= be.prs.ind.3sg întrebarea ce vor question.def.adv.fsg what want.prs.ind.3pl ‘many wonder what they want.’ (FD.1592–1604:467r ) Clitic doubling, realized either as clitic left-dislocation or as clitic right-dislocation, confirms the ‘dative’ status of these analytic constructions. Clitic doubling is also possible in constructions with numerical quantifiers (cf. (46b)).

10.2.4 Quantifiers In many cases prepositional marking is triggered by the impossibility of inflexional case marking on the leftmost constituent of the DP, giving rise to analytic structures which are attested throughout the old Romanian period and which continue into the modern language. In the sixteenth century most attestations of analytic marking are in conjunction with the singular universal quantifier tot ‘all’, the indefinite quantifiers mult/mulți ‘much/many’ and puțin/puțini ‘little/few’, and with cardinal numerals. In the plural, the universal quantifier and the indefinite quantifiers behave differently in relation to inflexional case marking, inasmuch as plural toți ‘all’ has an inflexional case form tuturor/tutulor, whereas for puțini ‘few’ ⁶ In old Romanian cine can introduce headless relatives with [−animate] reference.

288

gabriela pană dindelegan and oana uță bărbulescu

there are no attestations of the adnominal inflexional form puținor, and for multor ‘many’ there are only a few attestations dating from the earliest old Romanian period. In non-translated texts—whether official documents or documents written in the princely chancelleries or private correspondence—and in most translations, the use of analytic expressions for indefinite quantifiers and numerals is well established. old Romanian (43) a. denaintea episcupului și [a mulți in.front.of bishop.def.adn.msg and a many.adv.mpl boiari] boyars.adv.mpl ‘in front of the bishop and many boyar(s).’ (DÎ.1594:X) b. dennaintea [a mulți oameni buni] in.front.of a many.adv.mpl people.adv.mpl good.adv.mpl ‘in front of many respectable people.’ (DÎ. c.1597–1600:XVI) c. cu ¸stirea [a mulți b[oi]ari] with knowledge.def a many.adv.mpl boyars.adv.mpl ‘with the knowledge of many boyar(s).’ (DÎ.1600:XXXIX) Among the group of indefinite quantifiers that lack adnominal case inflexions we can also include o samă ‘some’, întune(a)rec’ 10,000; many’,⁷ ceva ‘something’, and feal de feal ‘many and various’, which occur in the early period in both original and translated texts. old Romanian (44) a. împotriva [a o samă] de domni against a a.adv.fsg number.adv.fsg of kings ‘against some kings.’ (ULM. ~1725:69r ) b. zborul [a întunearece] de îngeri gathering.def a ten.thousand.adv.pl of angels ‘the gathering of many angels.’ (Ev1642:459) c. spre clădirea [a ceva] for building.def a something.adv ‘to build something.’ (CDicț.1691–1697:185v ) d. probăluirea [a feali de feali de lucruri] proof.def a sort.adv.msg of sort.adv.msg of things.pl ‘understanding of many and various things.’ (CDicț.1691–1697:108r ) The indefinite determiner nește/niște ‘some’ also occurs in prepositional constructions: ⁷ In old Romanian întunearec has the meaning ‘ten thousand’ calqued on Slv. developed the meaning ‘many’.

, from which it

inflexional and analytic marking of obliques in romanian

289

(45) ziseră [a nește bărbați răi] (ORo.) say.pfv.pst.3pl a some.adv.mpl men.adv.mpl bad.adv.mpl ‘they said to some wicked men.’ (CPr.1566:79) Structures containing numerals also usually involve analytic exponents: old Romanian (46) a. Nimea nu poate [a doi domni] lucra nobody not can.prs.ind.3sg a two kings.adv.mpl work.inf ‘No one can serve two masters.’ (CT.1560–1561:11r ) b. [a doisprăzeace apostoli] numele lă a twelve apostles.adv.mpl name.def 3pl.dat= sânt be.prs.ind.3pl ‘the names of the twelve apostles are […].’ (CT. 1560–1561:18v ) However, synthetic marking is also attested, especially with quantifiers such as amânduror(a) ‘both’. old Romanian (47) a. iaste sminteală [amânduror țărilor] be.prs.ind.3sg harm both.adn countries.def.adn.fpl ‘it harms both countries.’ (DÎ.1592:LXXXII) b. [amândurora] le iaste să se both.adn 3pl.dat= be.prs.ind.3sg să.sbjv refl.3.acc postească fast.sbjv.3 ‘both of them must fast.’ (CC2 .1581:83) An asymmetry, however, occurs in the context of universal quantifiers. In the singular the universal quantifier usually appears in analytic formations: old Romanian (48) a. spre binele [a toată creștinătatea] for good.def a all.adv.fsg Christendom.def.adv.fsg ‘for the good of the whole Christian world.’ (DÎ.XXXVI:1600) b. va sluji împăratului și aux.fut.3sg serve.inf emperor.def.adn.msg and [a toată creștinătatea] a all.adv.fsg Christendom.def.adv.fsg ‘he will serve the emperor and the whole Christian world.’ (DÎ. XXXIII:1600)

290

gabriela pană dindelegan and oana uță bărbulescu

In non-translated documents the oblique functions of the universal quantifier in the plural are almost always expressed synthetically whether used pronominally (49a) or adnominally (49b): old Romanian (49) a. Facem șcire [tuturoru] make.prs.ind.1pl news.adv.fsg all.adn.mpl ‘we inform all.’ b. Și cu știrea [tuturor and with knowledge.def.adv.fsg all.adn.mpl vecinilor] neighbours.def.adn.mpl ‘with the knowledge of all the neighbours.’

(DÎ.1593:CXIII)

(DÎ.1582: VII)

In the case of the universal quantifier when used in the plural, the synthetic forms are clearly preferred over the analytic variants: in non-translated documents tuturor registers ten occurrences in both adjectival and pronominal uses, in contrast to just one occurrence with an analytic expression a toți. Translations, irrespective of their regional origin, exhibit a preference for the synthetic realization. For instance, in CT.1560–1561 tuturor registers 20 occurrences in both adnominal and pronominal uses, in contrast to just three adnominal occurrences (there are no pronominal uses) in an analytic formation a toți, a toate. Although a translation, the Floarea darurilor in the version dating back to the end of the sixteenth century presents us with a situation similar to that attested in the non-translated documents, in that only inflexional marking—whether in adnominal (50a) or pronominal (50b) uses—is found with the universal quantifier in the plural. old Romanian (50) a. elu e împărat [tuturor pasărilor] he be.prs.ind.3sg emperor all.adn.fpl birds.def.adn.fpl ‘he is the king of all birds.’ (FD.1592–1604:602v ) b. merge dendărăptul [tutora] go.prs.ind.3sg in.back.def.adv.msg all.adn.pl ‘he leads the funeral procession.’ (FD.1592–1604:483v )

10.2.4.1 Homogeneous constructions vs heterogeneous constructions The examples in (47a), (49b), and (50a) are homogeneous constructions in which both the quantifiers and the nouns bear the adnominal case inflexion. The examples in (43a–c), (45), and (48a–b) are also homogeneous constructions, but both the quantifiers and the nouns are used with adverbal forms. The constructions in (48a–b) are extremely stable throughout the history of Romanian, but towards

inflexional and analytic marking of obliques in romanian

291

the end of the seventeenth century there occur analogical adnominal inflexional forms for the singular universal quantifier, namely toatei, most likely by analogy with pronouns which display adnominal inflexional forms in -ei in the feminine singular. This tendency to use the feminine singular form of the universal quantifier in homogeneous constructions with synthetic marking continues to occur until the mid-nineteenth century. The fact that the adnominal inflexional form is -ei, not -ii, which would have produced affrication of the stem-final consonant (cf. -t- > -ț- [ts]), as happens in the earliest old Romanian period until 1640, proves that these forms were created later, after the inflexion -ii had been replaced by -ei (for discussion, see Uță Bărbulescu 2018). The form toatei may occur in constructions which only show inflexional marking (51a) or in mixed constructions where a co-occurs with an inflexionally marked form (51b): old Romanian (51) a. puind capete și otcârmuitori romani put.ger heads.adv.pl and rulers.adv.mpl Roman.mpl [toatei Dachii] all.adn.fsg Dacia.adn.fsg ‘appointing magistrates and governors to the entire province of Dacia.’ (CIst.1700–1750) b. domn și stăpân [a toatei lumi] king and ruler a all.adn.fsg world.adn.fsg ‘ruler and master of the entire Universe.’ (AD.1722–1725:79r ) In the masculine singular, a corresponding inflexional form **totui is not attested. However, another definite form totului ‘all.def.adn.msg’ is attested, which in old Romanian only occurs in fixed expressions such as cu totului tot lit. ‘with all.def.adn.msg all (= completely, entirely)’ (CD.1698:IXv ). The plural form tutor, attested in the Psaltirea Hurmuzachi (52), is most likely a scribal error, and not an archaism (for a slightly different opinion, see Densusianu 1938:192). This form is only attested in the manuscript of the Psaltirea Hurmuzachi, which was not wholly accurately copied, leading us to believe that we may be dealing with a genuine error. Tuturor is the reflex of Lat. totorum ‘all.gen.pl’ + illorum ‘that.gen.m-npl’ (Rosetti 1986:136), and the compound was created very early on, as shown by its oxytonic accentuation. (52) se afle mără ta [tutor să.sbjv find.sbjv.3 hand.fsg your.fsg all.adn.mpl vrajmașilor] tăi (ORo.) enemies.adn.mpl your.adn.mpl ‘your hand will find out all your enemies.’ (PH.1500–1510:16r )

292

gabriela pană dindelegan and oana uță bărbulescu

In contrast to tuturor(a), the adnominal forms multor(a) ‘many and puținor(a) ‘few’ show a different behaviour in relation to their first emergence and frequency in early texts. Multor(a) is not attested until the mid-seventeenth century, occurring only sporadically in texts up to 1780 (its frequency increases afterwards): old Romanian (53) a. [multor sfinți] au arătat Dumnezeu many.adn.mpl saints.mpl aux.pfv.3sg show.ptcp God ‘God revealed to many saints.’ (CÎ.1678:5v ) b. mică țară [multora] au părut small.fsg country many.adn.mpl aux.pfv.3sg seem.ptcp că iaste that be.prs.ind.3sg ‘to many people, it seemed to be a small country.’ (CIst.1700–1750:3r ) In different versions of the same biblical verse, one can also note that the analytic formation with la was already fixed in the language (54), being preferred to inflexional forms (55): old Romanian (54) a. [la mai mulți] îl b. [la mai mulți] îl c. [la mai mulți] îl la more many.adv.mpl msg.acc= ‘I commended him to many.’

adevăram (BB.1688:726) cunoșteam (MS 45:872) cunoșteam (MS 4389:910) know.ipfv.pst.1sg

(55) [multora] îl închinam (ORo.) many.adn.mpl msg.acc= worship.ipfv.pst.1sg ‘I recommended him to many.’ (BV5 .1760–1761:222v ) Puținor(a) ‘few’ is not attested in the earliest period and occupies an entirely marginal position in modern Romanian. It is, to be sure, a late, analogical creation. From the second half of the old Romanian period, inflected numerals such as doaor ‘two’ and trior ‘three’ prove most rare (see Pană Dindelegan 2013a:165f.). old Romanian (56) a. numele [doaor orașuri] name.def two.adn towns.pl ‘the name of two towns.’ b. numele [trior ostroave] name.def three.adn islands.pl ‘the name of three islands.’

(CDicț.1691–1697:184v )

(CDicț.1691–1697:215r )

Most probably, these forms emerge by analogy with the adnominal case form amânduror(a) ‘both.adn’, but at least for Corbea (and later, for writers of the Transylvanian School) a possible influence from the Latin form duorum ‘two.gen.m-n’

inflexional and analytic marking of obliques in romanian

293

cannot be excluded. Trior must be an entirely analogical creation (cf. Lat. trium ‘three.gen’).

10.2.4.2 Mixed constructions Although homogeneous constructions are robustly attested in old Romanian, these are not the only attested structures in both non-translated and translated texts. In sixteenth-century documents, two examples show a mixed marking where the universal quantifier occurs with an analytic marker followed by a case inflected DP. old Romanian (57) a. slugi credincioase [a toată servants.fpl faithful.fpl a all.adv.fsg creștinătăței] Christendom.def.adn.fsg ‘faithful servants of the whole Christian world.’ (DÎ.1600:XLIV) b. cu știre [a toț with knowledge a all.adv.mpl megiiașilor] și [a tuturor neighbours.def.adn.mpl and al.fsg all.adn.pl fraților] de ocină brothers.def.adn.mpl of property ‘with the knowledge of all the neighbours and of all the brothers endowed with property.’ (DÎ.1592:VIIIb) In Palia de la Orăs¸tie, a text translated at the end of the sixteenth century (1582), we find more frequently than in any other old Romanian text heterogeneous constructions where the plural universal quantifier agrees in gender and number with its associated noun, but appears with an analytic marker whereas its associated noun bears an adnominal inflexion. old Romanian (58) a. tatțl [a toți ficiorilor] lui Eber father.def a all.adv.mpl sons.def.adn.mpl lui Eber ‘the father of all the children of Eber.’ (PO.1582:39) b. Și dziseră Moisi și Aron [a and say.pfv.pst.3pl Moses and Aaron a toți fiilor] lu Israil all.adv.mpl sons.def.adn.mpl lui Israel ‘and Moses and Aaron said to all the sons of Israel.’ (PO.1582:231) To these we may add mixed constructions with the quantifier in the singular: old Romanian (59) a. sfârșitul [a tot trupului] end.def a all.adv.msg body.def.adn.msg ‘the end of all flesh.’

(PO.1582:28)

294

gabriela pană dindelegan and oana uță bărbulescu

b. Grăiți [a toată adunăriei] speak.imp.2pl a all.adv.fsg gathering.def.adn.fsg feciorilor lu Izdrail sons.def.adn.mpl lui Israel ‘Tell all the congregation of Israel!’ (PO.1582:214) In heterogeneous constructions, the quantifier—whether in the singular, where it lacks an inflexional case form for the oblique functions, or the plural, where it shows inflexional marking for the oblique functions—assumes the adverbal form, whereas the nominal—whether countable or not—presents adnominal inflexional marking. Tuturor(a) is also attested in PO and remains predominant in comparison to analytic expressions such as a toți, a toate (56 occurrences of the former as opposed to just 11 occurrences of the latter). From the examples in (57)–(59) we can note that heterogeneous constructions are only possible in conjunction with the universal quantifier, since it is followed by a nominal which bears the inflexional definite article. Such heterogeneous constructions are not attested for existential quantifiers, which are followed by nominals that do not bear the suffixal definite article, thus ruling out the possibility of adnominal inflexional forms. The heterogeneous constructions are attested throughout the entire old Romanian period, but begin to decline from the end of the eighteenth century and appear only rarely in the nineteenth century.

10.3 Modern Romanian Prepositional marking of oblique functions was more frequent in old Romanian than in the modern language. It thus follows that inflexional marking of oblique functions is still predominant today, as illustrated in Table 10.3. Table 10.3 Distribution of inflexional and prepositional markers of oblique functions in modern Romanian Texts1

Inflexional genitive

Inflexional dative

HSGD.2014 Liiceanu.2019 (20pp.)

13 140

17 21

Analytical– Analytical– Inflexional prepositional prepositional vs analytigenitive dative cal quantity ratio (%) 1 5 84% vs 16% Ø 2 98.8% vs 1.2%

1

For the contemporary language, HSGD.2014 is a collection of regional texts, and Liiceanu.2019 a philosophical text.

In the passage from old to contemporary Romanian, there have been some changes in the distribution and use of the four prepositional markers. For instance,

inflexional and analytic marking of obliques in romanian

295

the marker a has seen a considerable reduction in its distribution. As a marker of possession, a is only used to mark certain structures within the noun phrase. In particular, it is obligatory if the first element of the noun phrase, most typically an invariable quantifier, cannot be inflected for case (60a–b). modern Romanian (60) a. Plecarea [a cinci copii] departure.def a five children.adv.mpl ‘The departure of five children.’ b. Plecarea copii] / [a numeroși [a departure.def a many.adv.mpl children.adv.mpl a fel de fel de copii] / [a foarte type of type of children a very mulți copii] many.adv.mpl children.mpl ‘The departure of many children / the departure of various children / the departure of so many children.’ As a result of the process of restriction and ultimately of exclusion of this prepositional marker, a is no longer used to mark indirect objects. However, it still occurs after those prepositions that take an oblique complement (61a–b). modern Romanian (61) a. În fața [a cinci copii] in front.def a five children.adv.mpl ‘In front of five children.’ b. Grație [a cinci copii] thanks.to a five children.adv.mpl ‘Thanks to five children.’ There are a number of older patterns, such as the quantifier tot ‘all, entire’, which continue to occur in analytic constructions in the singular. In informal, colloquial registers there is also a strong tendency to use the analytic construction in the plural too, as illustrated by the contrast between (62a–b). modern Romanian (62) a. Plecarea [tuturor copiilor] departure.def all.adn.mpl children.def.adn.mpl b. Plecarea [a toți copiii] departure.def a all.adv.mpl children.def.adv.mpl ‘The departure of all children.’ In informal registers, competition between the inflexional forms and analytic prepositional markers is extended to other quantifiers in the plural, witness the contrast in (63a–b).

296

gabriela pană dindelegan and oana uță bărbulescu

modern Romanian (63) a. Plecarea [câtorva copii] departure.def few.adn.mpl children.mpl b. Plecarea [a câțiva copii] departure.def a few.adv.mpl children.mpl ‘The departure of a few children.’ Mixed constructions are still attested, but they occur (frequently) only in spoken non-standard registers (64), whereas heterogeneous constructions have been lost. (64)

?

Părerile [a multora] dintre ei (non-standard ModRo.) opinions.def a many.adn.mpl of them.acc ‘the opinions of many of them.’

Turning now to de, this marker has considerably restricted its uses, and is now in complementary distribution with the inflected form (65a–b) and has values other than those associated with a case marker. It functions as a marker of nominal complements (66a), as a marker of modifiers of the head noun (66b), and as a predicate marker to indicate property-denoting values (66c). While inflected possessives form part of a phrase that obligatorily takes a determiner (65b), the other values of de occur in phrases without a determiner such as (65a) and (66a–c). modern Romanian (65) a. Furnizare [de energie] provision de energy.adv.fsg ‘Energy supply.’ b. Furnizarea [energiei] provision.def energy.def.adn.fsg ‘The energy supply.’ modern Romanian (66) a. Iubitorii [de dulciuri] lovers.def.mpl de sweets.adv.pl ‘The sweet-toothed.’ b. Casă [de țară] / casă [de cărămidă] / cei [de house de country house de brick cel.mpl de acolo] there ‘Country house / brick house / those over there.’ c. Copil [de preot] / picior [de balerină] (Niculescu (2008)) child de priest leg de ballerina ‘Son of a priest / ballerina’s leg.’

inflexional and analytic marking of obliques in romanian

297

In standard contemporary Romanian, de is no longer used to express possession. However, remnants of this archaic use of de as a possessive marker persist in regional Romanian, but only in conjunction with noun phrases introduced by an indefinite determiner (67a–b). regional modern Romanian (67) a. o coadă [de un topor] a.f fang de a.adv.m axe.adv.msg ‘a handle of an axe.’ (TDR:372) b. c-ar fi pierdut urma because=aux.cond.3sg be.inf lose.ptcp track.def [d-o căprioară] de=a.adv.f deer.adv.fsg ‘because he might have lost the track of a deer.’

(TDR:258)

As for indirect objects, these can be expressed by the preposition la, preferably in conjunction with animate noun phrases, regardless of definiteness, witness the following indefinite (68) and definite (69) examples. Such constructions are tolerated in the standard language. (68) Trimite [la copii] / dă send.prs.ind.3sg la children.adv.mpl give.prs.ind.3sg [la copii] / le spune la children.adv.mpl dat.3pl= say.prs.ind.3sg [la studenți] (ModRo.) la students.adv.mpl ‘(S)he sends to children / (s)he gives to children / (s)he’s telling the students.’ (69)

Trimite [la copiii străzii] / send.prs.ind.3sg la children.def.adv.mpl street.def.adn.fsg dă [la cei din orfelinate] / give.prs.ind.3sg la cel.mpl from orphanages le spune [la cei din 3pl.dat= say.prs.ind.3sg la cel.mpl from primul rând] (ModRo.) first.def line ‘(S)he sends to homeless children / (s)he gives to those who live in orphanages / (s)he’s telling the ones in the first row.’

There are other manifestations of the overlap of la with adnominal inflexional marking, but usually these are attributed specifically to non-standard registers. For example, in the non-standard examples (70) la replaces long-established uses of the adnominal inflexional forms and is used regardless of the lexical features of the noun, animate or inanimate, and its (in)definiteness (see also Stan 2013a:268–271; Maiden 2016a:101).

298

gabriela pană dindelegan and oana uță bărbulescu

(70)

?

Urmările [la vaccin] / tivul consequences.def la vaccine.adv.msg hem.def [la rochie] / culoarea [la mașina asta] / la dress.adv.fsg colour.def la car.def.adv.fsg this din cauza [la doctor] (non-standard Ro.) from cause la doctor.adv.msg ‘Consequences of the vaccine / the hem of the dress / the colour of this car / because of the doctor.’

The shift towards its grammaticalized use as a marker of Recipients emerged in the context of verbs of sending (a trimite la ‘send to’), verbs of accusing (a pârî la ‘denounce (lit. to))’, and speech-act verbs (a porunci la ‘command (lit. to)’). Constructions such as those in (71a–b), in which la marks argumental Experiencers, are not accepted in the standard language. This also explains the difficulty in constructions such as (71c) of distinguishing between a lexical locative la and a functional la. In non-standard usage, where constructions such as (71a–b) are possible, the degree of grammaticalization is more advanced than in the standard language. Furthermore, in non-standard usage the la-construction functions as a genuine equivalent to the possessive uses of the adnominal inflexional forms (cf. (70)) without any restrictions on the noun, a state of affairs which confirms that the preposition is a fully grammaticalized substitute for inflexional marking in non-standard usage. (71) a. ? Nu le convine / not 3pl.dat= be.convenient.prs.ind.3sg Nu le place not 3pl.dat= like.prs.ind.3sg [la copiii mei]. (non-standard ModRo.) la children.def.adv.mpl my.mpl ‘It does suit my children / My children don’t like it.’ b. ? [La copacii ăștia] le cad la trees.def.adv.mpl these.mpl 3pl.dat= fall.prs.ind.3pl frunzele. (non-standard ModRo.) leaves.def ‘Leaves are falling from these trees.’ c. A trimite [la copii] / a.inf send.inf la children.adv.mpl a da [la animale]. (ModRo.) a.inf give.inf la animals.adv.fpl ‘To send to children / To give to animals.’

inflexional and analytic marking of obliques in romanian

299

The marker către (and variant cătă) is restricted to regional varieties. Dialectally, the marker către retains its lexical status as well (namely ‘towards’), as shown by the impossibility of examples like (72a–b) where the noun phrase introduced by către encodes an Experiencer, a relation usually associated with the dative, witness the third-person singular dative clitic îi. modern Romanian (72) a.** Îi place către Ion. 3sg.dat= like.prs.ind.3sg către Ion.adv ‘Ion likes it.’ b.** Îi pasă către Ion. 3sg.dat= matter.prs.ind.3sg către Ion.adv ‘Ion cares about it.’ It cannot be argued that the spread of prepositional marking resolves the ambiguities of the genitive–dative syncretism of adnominal inflexional marking, since prepositional expressions tend themselves to be as multifunctional as the inflexional case forms they replace. In Romanian, the prepositional constructions are ambiguous as a result of confusion with homonymous forms with similar values (cf. a) and overlap between la-phrases which indicate direction and la-phrases which encode Recipients. For instance, in Romanian there are different strategies for expressing the possessive relation: the prepositional expression with a which is used as an invariant marker (73a) and the prepositional determiner al (73b). As noted above, in standard Romanian the selection of the prepositional determiner obeys the adjacency constraint: the article is licensed whenever the possessor phrase is not immediately adjacent to a definite nominal or adjectival form of the preceding possessee phrase (cf. (73b)). Furthermore, the prepositional determiner al behaves as an inflexional element and displays agreement in number and gender with the possessee, such that in (73b), for instance, it occurs in the feminine singular form a in conjunction with the feminine singular head noun jucărie ‘toy’, but exhibits the feminine plural form ale in conjunction with the feminine plural head noun jucării ‘toys’. By contrast, the prepositional marker a in (73a) shows no adjacency condition and does not agree with the head noun, be it singular jucărie/jucăria ‘toy/toy.def.adv.fsg’ or plural jucăriile ‘toys.def.adv.fpl’. modern Romanian (73) a. Această jucărie / jucăria / jucăriile [a câțiva this.fsg toy.fsg toy.def.fsg toys.def.fpl a few.adv.mpl copii] children.mpl ‘this toy / the toy / the toys of a few children.’

300

gabriela pană dindelegan and oana uță bărbulescu

b. Această jucărie [a câtorva copii] / this.fsg toy.fsg al.fsg few.adn.mpl children.mpl aceste jucării [ale câtorva copii] these.fpl toys.fpl al.fpl few.adn.mpl children.mpl ‘this toy of a few children / these toys of a few children.’ The preposition la functions in present-day Romanian as a marker of Recipients and Possessors, but continues to function as a locative lexical marker selected by verbs of both literal and metaphorical movement such as those in (74a). In conjunction with certain verbs, the two values are clearly differentiated (cf. (74a)), but there are contexts such as (74b) in which it is difficult to distinguish between the two values. modern Romanian (74) a. A merge la / a alerga la / a to go.inf to to run.inf to to se gândi la / a se uita la (lexical la) refl.3.acc= think to to refl.3.acc= look.inf to ‘Go to / run to / think about / look at.’ b. A trimite la / a oferi la / a împărți to send.inf to to offer.inf to to share.inf la (?lexical/functional la) to ‘Send to somebody / offer to somebody / share with somebody.’ The ambiguity of de surfaces in examples such as (65a) and (66a–c), where it marks multiple values according to context (e.g., verbal vs non-verbal nature of the head, semantic value of the subordinate phrase).

10.4 Conclusion The two patterns of inflexional and prepositional marking have been shown to coexist throughout the history of Romanian, with a prevalence of the synthetic pattern regardless of period and register. The development of analytic marking for the oblique functions shows the cyclic repetition of some grammaticalization processes, witness the ‘new’ markers către and, especially, la, which repeat the grammaticalization process previously undergone by the prepositional marker a. In old Romanian, de can encode possession in contexts in which, usually, the Possessor is instantiated by an indefinite nominal. Although structures in which the Possessor is instantiated by a definite nominal are attested in old Romanian, they are not in genuine competition with their synthetic variants insofar as the sequence de + DP only occurs in translated texts where, even there, they have a

inflexional and analytic marking of obliques in romanian

301

very low frequency. In modern Romanian, de can be found in limited and specialized constructions; in particular, de-marked constituents cannot take the definite determiner and are not referential, such that they have been analysed as kind/property-denoting genitives and non-anchoring genitives. In modern Romanian, probably under the influence of the other Romance languages, especially French, de-expressions are preferred in contexts with deverbal nouns and adjectives, on condition that the complement is a property-denoting NP (see also Mardale 2009). Dialectally, de-constructions can encode possession, but only in contexts in which the Possessor is preceded by the indefinite determiner. In non-standard registers, la-constructions encoding possession can be used with definite nominals and function as genuine equivalents to the inflexional genitive. Since old Romanian, the analytical marker a has tended to specialize as an equivalent to the adnominal inflexional markers in contexts in which the latter are not available in the leftmost constituent of the DP (e.g., cardinal numerals, universal quantifiers), a pattern subsequently extended by analogy to almost all neological quantitative modifiers. As an equivalent to the inflexional marking of Recipients, by the end of the old Romanian period the prepositional a marker in conjunction with numerical or indefinite quantifiers begins to find itself increasingly in competition with la. For some constructions current today in literary registers where a is embedded within another PP governing adnominal case (e.g., grație ‘thanks to’), we cannot wholly rule out French influence (cf. grâce à). The analytic marker la is attested in old Romanian to mark both Possessors and Recipients but only rarely, and in modern Romanian shows different degrees of frequency according to stylistic register. In all registers, however, the prepositional construction for the indirect object relation is triggered by the impossibility of case inflexion on the leftmost constituent of the DP: in this respect the distribution of la is regulated by the same constraint as a in old Romanian. In literary registers, la is obligatory in contexts in which the relevant abstract nominal is derived from (or related to) an intransitive verb that takes an argument in the adnominal form. Some other configurations also have good literary credentials, especially those in which neological verbs assign the thematic Goal/Recipient role and the PP introduced by the preposition la takes a complement which bears the feature [−animate] or, through a metonymic process, comes to acquire this feature. In non-standard registers, la also occurs in configurations in which the Goal/Recipient, Experiencer, or Possessor (regardless of the type of possession, alienable or inalienable) is realized as a DP with the feature [+human]. In old and pre-modern Romanian, there has been a tendency to analogically create adnominal inflexional forms for quantifiers such as puțin ‘little’, mult ‘much’, not only in the plural, but also in the singular, and these compete with the analytic structures. Modern standard Romanian shows a tendency towards specialization: (i) possessives marked inflexionally have an entity type, referential denotation,

302

gabriela pană dindelegan and oana uță bărbulescu

whereas complements marked by de denote kinds or properties; (ii) the prepositional markers a and la usually occur whenever an inflexional form is not available. Contrary to old Romanian and the standard language, non-standard registers prefer the analytic constructions even when the inflexional forms are available, generalizing the preposition la as marker for oblique functions. The distribution of these two formally distinct grammatical devices varies throughout the history of the language and the transition from synthetic to analytic marking and the use of prepositional expressions should be seen as a feature of the language in general, rather than a feature restricted to the substandard register. In broader theoretical terms, the data we have discussed here show not only that grammaticalization paths may repeat themselves within a single functional domain, but also that with a group of closely related dialects there may none the less be subtly different outcomes at different periods, in different regions, and in different registers. Theories of grammaticalization therefore need to be sufficiently nuanced to accommodate such variety rather than relying on a set of single unidirectional trajectories.

Textual sources AD. 1722–1725 Antim Ivireanul, Didahii, ed. G. Ștrempel, Antim Ivireanul, Opere. Bucharest: Minerva, 1972, 1–238. ALR, s.n., VI Atlasul lingvistic român, serie nouă, vol. VI. Bucharest: Editura Academiei, 1969. ALRII Atlasul lingvistic român (serie nouă). Bucharest: Editura Academiei, 1956– 1972. BB.1688 Biblia. Ed.: Biblia adecă Dumnezeiasca Scriptură a Vechiului și Noului Testament, tipărită întâia oară la 1688 în timpul lui Șerban Vodă Cantacuzino, Domnul Țării Românes¸ti, Bucharest: Editura Institutului Biblic, 1977. BV.1760–1761 Biblia Vulgata, Blaj, 1760–1761. CB.1559–1560 Codicele popii Bratul. Ediție de Alexandru Gafton: http://media.lit. uaic.ro/gafton. 1 CC .1567 Coresi, Tâlcul Evangheliilor, ed. V. Drimba, Tâlcul evangheliilor și molitvenic românesc. Bucharest: Editura Academiei Române, 1998, 31–187. CC2 .1581 Coresi, Evanghelie cu învățătură, ed. S. Pus¸cariu, Al. Procopovici, Diaconul Coresi, Carte cu învățătură (1581), vol. I, Textul. Bucharest: Socec, 1914. CD.1698 Dimitrie Cantemir, Divanul, ed. V. Cândea, D. Cantemir, Opere complete, I, Divanul. Bucharest: Editura Academiei, 1974.

inflexional and analytic marking of obliques in romanian

303

CDicț.1691–1697 Theodor Corbea, Dictiones Latinæ cum Valachica interpretatione, ed. A.-M. Gherman, vol. I. Cluj-Napoca: Clusium, 2001. CÎ.1678 Cheaia înțelesului. ed. R. Popescu, Ioannykij Haleatovskyi, Cheia înțelesului. Bucharest: Libra, 2000. CIst.1700–1750 Constantin Cantacuzino, Istoria Țării Românești, ed. O. Dragomir, Istoria Țărâi Rumânești atribuită stolnicului Constantin Cantacuzino. Bucharest: Editura Academiei Române, 2006. CLM.1700–1750 Miron Costin, Letopisețul Țărâi Moldovei, ed. P. P. Panaitescu, M. Costin, Opere. Bucharest: Editura de Stat pentru Literatură și Artă, 1958, 41–201. CM.1567 Coresi, Molitvenic, ed. V. Drimba, Tâlcul evangheliilor și molitvenic românesc. Bucharest: Editura Academiei, 1998, 189–211. CPr.1566 Coresi, Apostol, ed. I. Bianu, Texte de limbă din secolul XVI, IV, Lucrul apostolesc tipărit de diaconul Coresi la 1563. Bucharest: Cultura Națională, 1930. CPrav.1560–1562 Coresi, Pravila, ed. Gh. Chivu, in I. Gheție (coord.), Texte românești din secolul al XVI-lea. Bucharest: Editura Academiei Române, 1982, 218–231. CT.1560–1561 Coresi, Tetraevanghel, ed. F. Dimitrescu, Tetraevanghelul tipărit de Coresi. Brașov 1560–1561, comparat cu Evangheliarul lui Radu de la Mănicești. 1574. Bucharest: Editura Academiei, 1963. CV.1563–1583 Codicele Voronețean, ed. M. Costinescu. Bucharest: Minerva, 1981. DÎ Documente și însemnări românești din secolul al XVI-lea, text stabilit și indice de Gh. Chivu, M. Georgescu, M. Ioniță, Al. Mareș, Al. Roman-Moraru. Bucharest: Editura Academiei Române, 1979. Ev.1642 Evanghelie învățătoare, ed. A.-M. Gherman. Bucharest: Editura Academiei Române, 2011. FD.1592−1604 Floarea darurilor, ed. Alexandra Roman Moraru. Bucharest: Minerva, 1996 (Cele mai vechi cărți populare în literatura română, 1), 119–182. HSGD.2014 M. Marin, M. Tiugan, ‘Harta sonoră’ a graiurilor și dialectelor limbii române. Bucharest: Editura Academiei, 2014. ISBD Nicolae Iorga, Scrisori de boieri. Scrisori de domni, 2nd ed. Vălenii-deMunte: Tipografia ‘Datina românească’, 1925. LC~1650 Lemnul crucii, ed. E. Timotin. Bucharest: Fundația Națională pentru Știință și Artă, 2001 (Cele mai vechi cărți populare în literatura română, 5), 193–199. Liiceanu, G. (2019). Caiet de ricoșat gânduri. Bucharest: Humanitas. MS 45—Vechiul Testament Septuaginta. Versiunea lui Nicolae Spătarul Milescu (MS 45 de la Biblioteca Filialei din Cluj a Academiei Române), ed. E. Munteanu. Iași: Editura Universității Alexandru Ioan Cuza. MS 4389 Manuscript held by Biblioteca Academiei Române.

304

gabriela pană dindelegan and oana uță bărbulescu

NL.~1750–1766 Ion Neculce, Letopisețul, ed. I. Iordan, Ion Neculce, Letopisețul Țării Moldovei și O samă de cuvinte. Bucharest: Editura de Stat pentru Literatură și Artă, 2nd ed., 1959, 31–388. NT.1648 Noul Testament, ed. Alba Iulia: Reîntregirea, 1998. PH.1500–1510 Psaltirea Hurmuzaki, ed. I. Gheție and M. Teodorescu. Bucharest: Editura Academiei Române, 2005. PO.1582 Palia de la Orăștie, ed. V. Pamfil. Bucharest: Editura Academiei, 1968. Prav.1581 Pravila ritorului Lucaci, Ed. I. Rizescu, Bucharest: Editura Academiei, 1971. SB Al. Rosetti, Scrisori românești din arhivele Bistriței (1592–1638). Bucharest: Casa Școalelor, 1944, 34–75. SVI.~1670 Varlaam și Ioasaf, ed. M. Stanciu Istrate, Reflexe ale medievalității europene în cultura română veche: Varlaam și Ioasaf în cea mai veche versiune a traducerii lui Udriște Năsturel. Bucharest: Editura Muzeului Național al Literaturii Române, 2013. TDR Valeriu Rusu (coord.), Tratat de dialectologie română. Craiova: Editura Scrisul Românesc, 1984. ULM.~1725 Grigore Ureche, Letopisețul Țării Moldovei, ed. P. P. Panaitescu. Bucharest: Editura de Stat pentru Literatură și Artă, 1955, 57–210.

11 A diachronic perspective on polymorphism, overabundance, and polyfunctionalism Rosanna Sornicola

11.1 Polymorphism and polyfunctionalism The relation between form and function involves many issues for the theory of morphology and its relation to syntax and semantics. In the present chapter I will explore these through a case study based on documents from Italy from the ninth century ce, considering in particular the place of the modern concept of overabundance in the context of earlier theoretical discussions.1 The definitions of form and function and the relation between the two have long been a matter of debate, particularly within structural linguistics, as the writings of Saussure, Bloomfield, and Hjelmslev among others amply attest. It will be as well therefore to be clear from the outset about the definitions that will be assumed in the present study. Form as conceived here has two ingredients: first a sequence of phonological units that can be realized phonetically or graphically, and second any more abstract grouping of these units into morphological components such as stem, suffix, prefix, and the like. In other words, form is doubly defined, being characterized in both phonological and grammatical terms (compare Hjelmslev 1928:112–116 and the definition in Matthews 2014:144). Function refers to the content of a linguistic form and also has two dimensions: a paradigmatic one defined in terms of contrast and opposition and a syntagmatic one defined in terms of combinability.2 Form–function relations may then be one-to-one (or biunique— see §11.3.1) or many-to-one (non-biunique). The latter is the point of departure for 1 I dedicate this chapter to my dear friend Martin Maiden, whose writings on morphology have always been a source of great inspiration. I would like, too, to thank the editors of this volume for their very helpful comments and suggestions on earlier drafts. This study is part of the Italian research project PRIN 2017 (2017WLBK3Z, ‘Writing expertise as a dynamic sociolinguistic force: the emergence and development of Italian communities of discourse in Late Antiquity and the Middle Ages and their impact on languages and societies’). 2 Here I differ from Hjelmslev (1928:123) and follow instead Saussure (Komatsu and Wolf 1997: 50, 142). For further discussion of the history of the concept ‘function’ see Sornicola (2014).

Rosanna Sornicola, A diachronic perspective on polymorphism, overabundance, and polyfunctionalism. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Rosanna Sornicola (2022). DOI: 10.1093/oso/9780198870807.003.0012

306

rosanna sornicola

the concepts of polymorphism, the mapping of many forms to one function, and polyfunctionalism, the mapping of many functions to one form. In what follows we understand polymorphism in the broad sense as covering both heteroclitic patterns in which different morphologically definable forms realize a given function and those where the formal variation is due to phonetic or phonosyntactic factors. This in turn will allow us to develop a general theory within which the synchronic and diachronic behaviour of different subtypes can be investigated. Our first case study concerns the variant case forms of some Latin nouns. These are the result of synchronic analogical attraction from one inflexional schema to another. This is very evident with many fourth-declension nouns, which admit alternant forms that follow the pattern of other declensions, and in particular the second, as can be seen from the forms of domus ‘house’ as set out in the Table 11.1.3 Table 11.1 Polymorphism of Latin domus ‘house’ nominative accusative genitive dative ablative

Singular domus domum domūs, domī, domuis, domōs, domuī, domō domō, domū

Plural domūs domōs, domūs domōrum, domūs domibus domibus

Alternatively, the co-presence of more than one form in a given paradigmatic cell may be due to contact between different regional varieties, as is well exemplified by the different pronominal and adjectival forms in the alpine Piedmontese dialects (Terracini 1937) and in Francoprovençal (Diémoz and Kristol 2014; Kristol 2018). Segmental and suprasegmental variation in spontaneous speech is also a fertile terrain for the development of polymorphism and once again Romance dialects are a rich source of relevant data. A case in point concerns the cliticization of the first-person singular pronoun io to the verb form o/go ‘I have’ in the interrogative as exemplified in two different varieties from the Veneto region in (1):⁴ (1)

a. San Donà del Piave (Venice) ojo [‘ɔjo], ogno [‘ɔɲo] ‘have I?’ b. Quarto di Altino (Treviso) gojo [‘gɔjo], gogno [‘gɔɲo] ‘have I?’

The palatal nasal has emerged independently in the two dialects as an optional sandhi effect leading to a pattern of free variation and hence polymorphism. 3 Space does not permit me to consider the interesting questions concerning the historical and textual distribution of these forms, a topic to which I hope to return on a future occasion, but in the meantime, see Ernout (1945:107f.), Leumann, Hofmann, and Szantyr (1972:441–443). ⁴ Examples courtesy of Pierluigi Cuzzolin, a native speaker of both varieties.

polymorphism and polyfunctionalism in diachrony

307

Free variation here is crucial. If the variants have particular geographical, cultural, social, or generational connotations, they are not strictly speaking free and it is then open to question whether there is true equifunctionality. This in turn implies that to establish genuine polymorphism requires detailed textual and historical analysis. A good case in point is ModIt. braccio ‘arm’ with its two plural variants bracci and braccia. In some synchronic accounts these forms would not be considered polymorphic since they are not equifunctional: braccia ‘arms (human limbs)’ and bracci ‘parts of machines, stretches of water, etc.’, and indeed Acquaviva (2008) treats bracci and braccia as plurals of two independent lexemes (see also the discussion in §2.2.3 in this volume). However, analysis of the different senses associated with these forms in diachrony as well as different current regional and diastratic norms suggests that they have belonged, and may still belong, to polymorphic paradigms. For polyfunctionalism, on the other hand, the most typical instances involve syncretism, a topic to which we return in §11.3.3. Although our principal concern in what follows will be with the theoretical and historical implications of polymorphism, we will continue to bear polyfunctionalism in mind since a comparison of the two will prove to be instructive. Both these non-biunique relations can also fruitfully be compared with the lexical concepts of homonymy (one word form with more than one meaning) and synonymy (one meaning associated with more than one lexeme). The latter in particular is usually taken to be a key component of linguistic creativity. Naturally, any such comparison requires us to consider the differences between lexical meaning and morphological function. The latter manifests itself in the inflexional markers of grammatical categories (tense, aspect, mood, person, case, number, gender) and/or through stem alternations. Inflexional markers are the exponents of grammatical morphemes and as such determine functional alternations or syncretisms within morphological paradigms, while the semantic relations between lexemes are defined with respect to one or more semantic fields. A different situation obtains with polythematism, in which variant stem forms alternate within a lexical paradigm. Such a situation, which in some respects falls within the more general category of polymorphism, at the same time also has affinities with the lexical relation of synonymy. It can show up in different ways. There is, for example, the alternation of the stem according to the morphosyntactic category in the differing case and number forms of Latin imparisyllabic nouns such as homo/homin‘man’, patēr/patr- ‘father’, flōs/flor- ‘flower’.⁵ In these cases there is a degree of shared phonological identity across the stems. By contrast in so-called suppletion, the stems are completely distinct as with Breton den ‘man.sg’ vs tud ‘man.pl’ or Greek ὁρα´ω ‘see.prs.1sg’/oἶδα ‘see.prf.1sg’, both meaning ‘I see’ but

⁵ For details of the intricate historical processes which gave rise to these patterns, see Weiss (2020:Ch 25).

308

rosanna sornicola

where the latter has what was once a perfect stem now marking the present, beside ὄπωπα ‘see.pfv.1sg’ with a perfect stem displaying Attic reduplication. As a kind of polymorphism, suppletion has a closer affinity to lexical synonymy, although it is distinguished from it by the functional conditions which determine the distribution of its alternants. More interesting would seem to be the problem discussed by Plank (1991) of differences in parts of speech when it comes to inflexional homonymy and synonymy. It has been argued that nominal paradigms tend to have more homonymic irregularities when compared to verbs, but, as Plank (1991:35) observes, this hypothesis does not as yet have extensive empirical support.

11.2 Polymorphism and overabundance: on the history of the terms The historical and empirical side of polymorphism has long been a topic of relevance to philologists, for whom a careful study of orthographic (and hence phonetic) and morphological variation in manuscripts has been key to the preparation of critical editions. Within Romance linguistics it has been central to work both on late medieval scriptae and on modern dialects (for details and references, see Sornicola 2016b; 2018). Thus, Allières (1954:70) understands polymorphism as ‘the coexistence in a speaker’s language of two or more phonetic or morphological variants of the same word, used concurrently to express the same concept, the choice of one or the other being independent of articulatory conditioning (tempo, etc.) or any expressive effect’.⁶ By contrast, for Kristol (2018) some of the instances of polymorphism found in Francoprovençal dialects such as [alˈa]/[alˈɑ]/[alˈɔ] ‘go.inf’ are systemic and not individual; they co-occur in a single region and across all registers. A key question, then, has been the historical interpretation of polymorphic patterns and the possible implications for their theoretical modelling as paradigms with multiply filled cells. By historical interpretation I mean both the descriptions proposed and the external factors—time, place, cultural level, movement, and contact—which have been adduced by way of explanation. These issues of necessity require us to take into consideration the multiple traditions and norms that exist in a given community, which are in turn due to the interaction of different social groups who may, consciously or unconsciously, use different variants and thus give rise to situations where different forms have the same function, and who may feel a ‘sense of attachment’ to one form rather than another (Terracini ⁶ The original French reads as follows: ‘la coexistence, dans le langage d’un sujet parlant, de deux ou plusieurs variantes phonétiques ou morphologiques d’un même mot, utilisées concurremment pour exprimer le même concept, le choix de l’une ou de l’autre apparaissant comme indépendant du conditionnement articulatoire (tempo, etc.) ou d’une recherche quelconque d’expressivité’.

polymorphism and polyfunctionalism in diachrony

309

1935–1936). Such a historical interpretation concerns a group of patterns which elsewhere I have labelled ‘exogenous’ polymorphism, where the co-presence of different forms is due to contact and where the speakers will have a clear awareness of their different connotations. There is also ‘endogenous’ polymorphism, where the coexisting forms result from internal change within the linguistic system, as with allophonic alternations (what Baudouin de Courtenay calls ‘divergent or alternating neophonetic’ forms)⁷ or with prosodic morphology and the development of ‘paragogic elements’ (Sornicola 2006; 2020). In the case of both exogenous and endogenous polymorphism the different variants may not coexist in the linguistic competence or usage of all members of the speech community. This in turn raises the question of the unity or diversity of the community or communities, an issue which is relevant both synchronically and diachronically. Although most theoretical studies of polymorphism have been synchronic, the phenomenon also poses interesting diachronic questions. It may serve as an indicator of ongoing change while at the same time complicating the analysis because of the need to separate out ‘cyclic’ shifts from those without an obvious directionality, both of which are not without theoretical difficulty, as evidenced in the patterns of diphthongization explored in Sornicola (2003). From the Romanist’s historical–philological perspective, we can use the intuitive metaphor of ‘debris’: the multiple occupants of the paradigmatic cells can be seen as the debris brought there by the mechanisms of change.⁸ In addition to such implications for the theory of language change, there remain to be addressed a number of open questions about the very nature of multiple variation which require an integrated vision of synchrony, diachrony, and history. To represent polymorphism as the continued presence of older, more opaque and often fossilized, forms superimposed on more recent forms within the same paradigmatic cell is to attribute to the speakers’ competence memorized forms heard and reproduced without a ‘sense’ of their syntagmatic and paradigmatic properties and therefore without a ‘sense’ of the system to which they once belonged.⁹ However, this way of representing polymorphism can also correspond to an awareness on the part of the speakers of the diverse functional connotations and social characteristics of each variant or, in some instances, a ‘sense’ of complete denotational and connotational equivalence of the forms in question. If particular variants are associated with different social levels, for whatever reason, discrimination on the part of the speakers can lead to one variant being ⁷ For Baudouin de Courtenay (1972:162), writing in 1894, ‘divergents are always of internal origin, for their difference is the result of the varying pronunciation of the members belonging to the same speech community’. ⁸ I prefer the term ‘debris’ to Lass’s (1990) ‘junk’, which seems to me to imply a different conceptualization of the historical underpinnings of the changes involved. ⁹ This state of affairs is reminiscent of what has been called morphological Erstarrung, with forms that have been overextended in use and border on fixed formulae. The remarks of Kruszewski ([1883] 1995:124) are still instructive in this connection.

310

rosanna sornicola

favoured and the other disappearing. If, on the other hand, any such variation is non-existent or below the threshold of speakers’ awareness, then the variants can coexist for long periods.

11.2.1 Overabundance Within current morphological theory, there are many different models, methods, and goals to be seen in the way phonetic and morphological variation of a single unit is handled. However, whatever other differences there may be, all these approaches have in common a prioritization of synchrony (or panchrony), even if there are occasional dips into the waters of diachrony. The term ‘overabundance’, borrowed from studies of the history of the Italian language, has established itself in the theoretical morphological literature in the sense of one kind of ‘deviation from canonicity in inflectional paradigms’. In the words of Thornton (2011:359): ‘This deviation occurs when a cell in a paradigm is filled by two or more synonymous forms which realize the same set of morphosyntactic properties, as in the case of the two English past tense forms burnt and burned.’ The corresponding Italian term sovrabbondanza had been used by Serianni (1988) to describe verbs which belong to two inflexional classes such as the verbs starnutare/starnutire ‘sneeze’ or empiere/empire ‘fill’ (see also the discussion in §8.5 in this volume) or the case discussed above of braccio ‘arm’ with its two plurals bracci/braccia. Thornton goes on to say (2011:360) that ‘[t]he two or more forms that compete to realize the same cell could be called “cell-mates”’.1⁰ As we will see better in due course, this approach is eminently synchronic (panchronic). The two terms overabundance and polymorphism reflect different theoretical stances even if they can and must be reconciled. Both address the problem of how and why there can be equifunctional variants. In some accounts the preservation of overabundance in a given cell correlates with low frequency and salience of the items in question, as for example in the treatment of Italian personal pronouns in Cappellaro (2018:218f.). Both polymorphism and overabundance can be integrated within a theoretical framework in which there can be paradigms which admit both multiple forms corresponding to a single function or multiple functions corresponding to a single form.11 From my point of view, the concept of multiplicity is preferable to overabundance since the former is neutral when it comes to the assumption of

1⁰ On the definition of overabundance, see also Cappellaro (2018) and Thornton (2019a, 2019b). Both have in mind a concept which is distinct from what Plank (1991) calls paradigmatic ‘abundance’. 11 On the need to recognize these two dimensions compare Carstairs (1987:14): ‘Any deviation from consistent one-to-one pairing of exponents and properties must fall into one of two categories: […] either a many-to-one or a one-to-many relationship’.

polymorphism and polyfunctionalism in diachrony

311

a principle of biunique correspondence between form and function, something which the latter instead necessarily takes as given.

11.3 On representing the relations between form and function When it comes to the relations between form and function, the different analyses raise a number of fundamental problems which it is important to address albeit briefly, since they serve to define the current theoretical horizon.

11.3.1 The relation between form and function A first fundamental question concerns the nature of the linguistic sign with its intrinsic link between form and function. Several theories, despite their other differences, take for granted the core principle of a biunique relation between the two, labelled by Venneman ‘Humboldt’s Universal’. For Vennemann (1972:183), who discusses the problem from the perspective of phonology, ‘in a conceptually ideal human language, the linguistic sign […] is uniform: a single concept is symbolized by a constant sound image, and the derivation of complex concepts is reflected in a corresponding derivation in sound’, although as he goes on to say ‘a conceptually ideal language need not be phonologically ideal’ (Vennemann 1972:183f.). The same idea recurs in the work of the German proponents of so-called natural morphology Mayerthaler and Wurzel. Thus, Wurzel (1989:12) provides the following translation of Mayerthaler (1981:34f.): ‘The symbolization/encoding of a paradigm Pi is uniform if Pi is organized according to the principle of “one function-one form”, otherwise it is more or less non-uniform … A paradigm Pi is transparent if it is constituted by monofunctional operations or has only monofunctional flexives/derivatives’ (italics in the original).12 Both scholars emphasize here the importance of perceptual factors; biunique relations are more easily perceived than those which are one-to-many or many-to-one (Wurzel 1989:11f.). The question has also been considered by Carstairs (1987:12f.), who describes the natural morphological approach ‘as a kind of ideal state towards which morphological systems allegedly tend, unless perturbed by phonological and other interference’. For him instead the biuniqueness principle is simply the point of departure for an investigation of other possible inflexional schemata, a topic which is central to his own thinking. As he goes on to say, ‘morphological principles 12 Mayerthaler’s original German reads: ‘Die Symbolisierung/Enkodierung eines Paradigmas Pi ist uniform, wenn Pi gema¨ss “one function-one form” organisiert ist, andernfalls mehr oder minder nicht-uniform … Ein Paradigma Pi ist transparent, wenn es sich durch monofunktionale Operationen konstituiert bzw. nur monofunktionale Flexive/Derivative aufweist’.

312

rosanna sornicola

which operate to promote compliance with Humboldt’s Universal may well exist alongside principles which limit the nature and extent of deviation from it’ (Carstairs 1987:13). What is more important, in his view, is to see whether there is an underlying logic to the various inflexional patterns that may occur in languages with the aim of showing that some logically possible options are never attested, since if all conceivable deviations were possible there would be nothing more than a heap of idiosyncratic factors of little interest to linguistic theory (Carstairs 1987:12). His aim is to search for universal factors, a goal which he sees as different from but not necessarily incompatible with that of the natural morphologists. The principle of an essential one-to-one relation between form and function inside paradigm cells also lies at the heart of the canonical approach developed by Corbett (2005; 2007a, 2007b). Within such a framework, as Thornton (2011:360f.) notes, ‘with overabundance there are two (and sometimes more) forms in the same cell, one of which may (but need not) be regular, while the other one(s) will be in some way irregular / non-canonical’. Overabundance therefore requires a model in which ‘canonical cell-mates are defined as a set of two or more forms that realize the same cell (i.e., the same set of morpho-syntactic features) in a lexeme paradigm and can be used interchangeably, with the choice of one or the other form subject to no condition’. Such unconditioned interchangeability of coexisting form raises fundamental theoretical issues which deserve consideration in the light of earlier work in general linguistics, as regards both the phoneme–allophone and the morpheme–allomorph relations.

11.3.2 Polymorphism and free variation It is important to recall in this connection the controversial status of allophones in free variation and the secondary role assigned to them within structuralist ‘representational phonology’ (Darstellungsphonologie). Trubetzkoy was a particularly clear exponent of this approach with his distinction between stylistically significant and non-significant features. The former express differences due to emotional, contextual, or social factors, while the latter have no special value and are freely interchangeable. A classic example is Kabardian, in which the palatal stops have as allophones velar stops and palato-alveolar affricates, for which the distribution is not determined either by context or changes in expressive or connotative value. Such variants can then be considered optional since ‘from the point of view of representational phonology the “variant” is a purely negative concept: a relation of variance exists between two sounds if they cannot be used to differentiate lexical meaning’ (Trubetzkoy 1969:48, italics mine). Since only some of the articulatory properties of each sound are phonologically distinctive, optional phonetic variants exist to the extent that they are marginal and not relevant in representational terms. Such a theoretical prejudice precludes the study of most of the interesting

polymorphism and polyfunctionalism in diachrony

313

synchronic and diachronic aspects of polymorphism and helps to explain why there have not generally been positive outcomes from the few encounters between dialectologists and phonologists that have taken place. However, more recently possible parallelisms between polyfunctionalism, understood as morphological homonymy within inflexional paradigms, and phonological neutralization have been explored, with further theoretical groundwork being required, as Carstairs (1987:87–90), to my mind convincingly, argues. Thornton (2011:262) defines canonical overabundance as that in which the multiple forms of the same cell in a paradigm are not subject to any restrictions and she develops a scalar model of canonicity represented as in (2): (2)

no condition >

conditions (where > = ‘more canonical than’)

The criterion of canonicity has already been applied in typological work also to other domains such as agreement and suppletion (Corbett 2006:26; 2007a:27). Such an approach is founded on the premise that canonical instances of a given phenomenon are rare or possibly even non-existent. An empirical–statistical implication of this way of doing things would then be that canonical overabundant forms are more or less equally frequent. In fact, equiprobability of occurrence, even if considered as the statistical correlate of interchangeability of forms, does not seem to be a satisfactory prerequisite property of coexisting variants. Indeed Thornton would seem to join sides with the critics of free variation when she observes ‘two completely interchangeable forms of equal frequency, whose use is not subject to any (speaker-related) diaphasic, diastratic, diamesic, diatopic, or diachronic conditions, do not exist’ (2011:362). At the same time, she goes on to argue that ‘breaking up the criterion relating to conditions into […] several facets allows us to establish canonicity clines, classifying specific sets of forms as more or less canonical according to certain sub-criterion, and to observe what sort of factors, if any, tend to influence the appearance of a specific form from a set of cell-mates’. She then develops a typology of overabundance in relation to:

i. alternation of forms, some of which do not have a diachronically restricted distribution while others do, as for example It. vedo ‘I see’ and chiedo ‘I ask’ beside the older but functionally equivalent forms veggo and chieggo; ii. alternation of forms some of which are not linked to diatopic or diaphasic conditions while others are as with It. vado ‘I go’ and faccio ‘I do’ besides the Tuscan and literary forms vo and fo; iii. alternation of forms with no constraints on the distribution of the individual variants as with It. siedo/seggo ‘I sit’ and possiedo/posseggo ‘I own’; iv. alternation examined on the basis of corpus data for journalistic writing as compared to the spoken language as with It. debbo/devo ‘I must’.

314

rosanna sornicola

According to Thornton, the patterns in (iii) and (iv) are good examples of the kind of phenomena that the canonical model would predict to be ‘almost non-existent’, an opinion with which I find myself entirely in agreement since in some varieties of modern Italian, including my own, these forms are indeed freely interchangeable. The overabundance of siedo/seggo and possiedo/posseggo is more canonical than that of the diachronically restricted chiedo/cheggo. This is a case where detailed examination of the historical record would have been valuable but was outside the scope of this investigation. The conclusions that Thornton (2011:380f.) draws are nonetheless interesting: (a) that overabundance is a phenomenon that belongs to autonomous morphology; and (b) that the stem alternants are distributed morphomically, and in particular that ‘overabundant cells tend to occur in the same partitions in which morphomic alternants occur, and to cluster mainly in those partitions in which the phonologically most diverse stems appear’ (Thornton 2011:381). By contrast, the conclusion that there is no necessary relation between overabundance and morphomic distribution of irregular stems suggests that these two conclusions are a posteriori descriptions of static properties of the system rather than factors which explain the diachronic mechanisms at work here.

11.3.3 Suppletion and syncretism A third set of problems concerns the nature of morphological paradigms and their intrinsic properties such as suppletion, syncretism, stability, economy, and the like. Of these, the first two are of particular importance when it comes to understanding the dynamics leading to non-biuniqueness. Under the heading of suppletion have been studied instances of one-to-many relations between morphosyntactic features and their realizations. In some theoretical approaches, suppletion properly speaking is taken to involve patterns in which the allomorphs do not derive from the same etymological source and therefore require more abstract modes of representation than simple item-and-arrangement. For instance, the English plural is marked not only by the suffix -s, but also by -en (oxen, children) and by zero or apophony (feet, geese). In other models, even the phonologically conditioned variants such as the allomorphs [-ız, -z, -s] of plural -s also fall under the heading of suppletion. Thus, Vennemann (1972:184) observes: ‘Paradigmatic variation, even though it may be regular phonologically, is suppletion conceptually. Suppletion is the antithesis of uniform linguistic symbolization. Suppletion is undesirable, uniformity of linguistic symbolization is desirable: Both roots and grammatical markers should be unique and constant.’13 In fact, however, as is well known, the label suppletion traditionally refers to paradigms in which the cells exhibit patterns of stem alternation such as the Italian first-person singular vado 13 See Carstairs (1987:18–21) for further discussion of the different theoretical approaches involved.

polymorphism and polyfunctionalism in diachrony

315

‘I go’ beside the first-person plural andiamo ‘we go’, or tense alternation such as the French present je vais ‘I go’ beside the imperfect j’allais ‘I went’ and Lat. fero ‘I carry’ beside tuli ‘I carried’, and many more such cases (Corbett 2007a). Whatever the theoretical interest of synchronic models that unite such diverse cases, the fact remains that from a diachronic point of view they are the result of very different kinds of processes. While phonologically conditioned allomorphy may be due to natural phonetic processes, the patterns with stem alternations are often the product of historical mechanisms not easily observable in the surviving texts. We may hypothesize that such dynamics are due to pragmatic factors of language use. These are likely also to lie behind defectiveness, when individual relics of earlier patterns may survive in the memory of speakers as fixed items linked to particular semantico-syntactic contexts, or they may be innovations which for various reasons have not extended to the full range of grammatical contexts. In short, both suppletion and defectiveness involve historically idiosyncratic and irregular circumstances and therefore stand in direct contrast to those patterns which are the consequence of regular phonological processes.1⁴ In any case, not all instances of allomorphy involving multiple stems are derivable in this way, since individual items may have well-motivated historical explanations. For example the stem alternation of It. uomo ‘man.sg’ vs uomini ‘man.pl’ is the outcome of a diachronic development whereby the singular is the reflex of the Latin nominative homo while the plural derives from the Latin stem homin- which is found in all inflected forms other than the nominative singular. The inverse relation, in which one form corresponds to more than one function, is found with syncretism. Thus, Latin, like many other Indo-European languages, has nominal paradigms in which the same form may realize two or more cells. Some representative instances are set out in Tables 11.2 and 11.3.

Table 11.2 Latin noun declensions: singular II -o-

II -ro-

II neuter

III -C-

III -i-

III neuter

III neuter

Nom rosă

I

dominus

puer

templum

rex

turris

caput

animal

Acc

rosām

dominum

puerum

templum

regem

turrim

caput

animal

Gen

rosae

dominī

puerī

templī

regis

turris

capitis

animālis

Dat

rosae

dominō

puerō

templō

regī

turrī

capitī

animāli

Abl

rosā

dominō

puerō

templō

regē

turrī/turrē

capitē

animāli

Voc

rosă

domine

puer

templum

rex

turris

caput

animal

1⁴ For further discussion of these issues, see now the contributions to Vincent and Plank (2019).

316

rosanna sornicola

Table 11.3 Latin noun declensions: plural II -o-

II -ro-

II neuter

III -C-

III -i-

III neuter

III neuter

Nom rosae

I

dominī

puerī

templa

regēs

turrēs

capita

animālia

Acc

rosās

dominōs

puerōs

templa

regēs

turrīs/turrēs

capita

animālia

Gen

rosārum dominōrum

puerōrum templōrum regum

turrium

capitum

animālium

Dat

rosīs

dominīs

puerīs

templīs

regibus

turribus

capitibus

animālibus

Abl

rosīs

dominīs

puerīs

templīs

regibus

turribus

capitibus

animālibus

Voc

rosae

dominī

puerī

templa

regēs

turrēs

capita

animālia

As is evident, the genitive and dative singular of first-declension nouns have the single exponent -ae, the dative and ablative plural of both first- and seconddeclension items have the form in -īs, the nominative and accusative plural of third-declension consonant-stem nouns share the form -ēs, and the dative and ablative plural of all subtypes of the third declension have the inflexion -ibus. Indeed, in some instances syncretism is cross-declensional as with the nominative and vocative, where only second-declension nouns in -us show distinct forms.1⁵ Likewise, when it comes to neuter nouns, regardless of the declension class and the form of the ending, in both singulars and plurals the nominative and the accusative have the same inflexion. The diachronic motivations for these patterns are various. Sometimes, the syncretisms are the surface outcome of sound change, as for example the genitive singular -ī (< -iī) and the nominative plural -ī (< -oi) of second-declension nouns, or analogy as with the etymologically distinct genitive singular -ae (< -āī < ās) and nominative plural -ae (< -āi) of nouns of the first declension. Elsewhere there may be a deeper semantic explanation.1⁶

11.3.4 Polymorphism and polyfunctionalism: paradigmatic and syntagmatic relations Of particular interest is the problem of how to represent polymorphism and polyfunctionalism in relation to the paradigmatic and syntagmatic dimensions of linguistic structure. It raises a number of issues concerning the way the properties of the distributional context justify the rise of variant forms of a given categorial unit. In this connection, we follow Carstairs (1987:14–18) who advances an interesting quadripartite model of the way structures can deviate from the

1⁵ For the debate about whether the vocative is truly a case, see the discussion in §2.2.4 in this volume. 1⁶ On the general principles at work here, see Baerman (2009). For the specifics of the individual case forms in their historical context, see Kuryłowicz (1964:190f.) and the relevant sections of Weiss (2020).

polymorphism and polyfunctionalism in diachrony

317

ideal one-to-one pattern as traditionally conceived, and one which takes both paradigmaticity and syntagmaticity into account. • Deviation I: One property to many exponents syntagmatically, as with a German past participle like ge-hab-t ‘had’ which involves both a prefix ge- and a suffix -t. • Deviation II: One to many paradigmatically, as with the different mechanisms for plural formation in English (-(e)s, -en, apophony, and zero) or the Latin second-person singular which is marked by a suffixal -s in all finite tenses and moods except the perfect indicative which has the ending -isti. • Deviation III: Many to one syntagmatically, as for example in the contrast between Lat. dominum ‘master.acc.sg’ and dominōs ‘master.acc.pl’, where the suffixes -um and -ōs cumulatively express case and number but are not further segmentable. • Deviation IV: Many to one paradigmatically, evident in syncretisms as with Latin nouns, all of which have a single form to express both dative and ablative plural as can be seen from Table 11.3.

11.3.5 Polymorphism and polyfunctionalism in diachrony Within theoretical morphology, polymorphism and polyfunctionalism have been mainly discussed from the synchronic angle. As far as diachrony is concerned, most attention has been paid to potential changes in the organization of a paradigm, whether due to internal factors, especially the effects of sound change, or to extra-paradigmatic pressures. Vennemann (1972:184) reminds us of the widely accepted ‘Sturtevant’s paradox’, according to the first half of which sound change is regular but causes irregularity and thus disrupts the ideal ‘one form–one function’ relation. In his words, ‘[n]atural phonological changes lead to preferred phonetic structures, but may destroy the uniformity of the linguistic sign by introducing paradigmatic variation’. In similar vein, according to Wurzel (1989:12), the two principles of uniformity and transparency also influence morphological change. Even though languages favour iconicity together with uniformity and transparency, there are no languages which do not exhibit some degree of morphological markedness due to the continuous effects of phonology, although this will be less in strictly agglutinating languages. In fact, however, the problems to which the diachronic dimension and its relation to synchrony give rise are manifold, and some insight into them can be gained through the analysis of polymorphism and polyfunctionalism in different text types of older languages. It is essential, however, that any such investigation should bear in mind the different morphological properties of different periods

318

rosanna sornicola

with a view to inferring possible diachronic paths. It is to a case study of this kind that we now turn our attention.

11.4 Polymorphism and polyfunctionalism in late-medieval documents from Italy In many late-medieval Latin documents from Italy from the eighth to the tenth centuries the inflexional morphology of the noun phrase is characterized by a notable degree of polymorphism. Of particular interest in this connection are the documents collected and published in the Codice Diplomatico Longobardo, of central-northern provenance, and those of the Abbazia della Santissima Trinità in Cava dei Tirreni in what is now Campania, some of which have been published in the Chartae Latinae Antiquiores (Galante 1997; 1998; Magistrale 1998). The latter in particular, which go back to the ninth century, allow us to develop hypotheses about polymorphism and polyfunctionalism from a diachronic perspective. The writing centres of Lombard Italy were home to scribes and notaries whose linguistic competences were complex and interestingly differentiated. In particular those documents which come from the writing centres of southern Italy exhibit a spectrum of more or less grammatically correct Latin, a broad and stratified version of latinum circa romançum (‘Latin which comes close to Romance’). It is no coincidence that it is from this part of Italy that we encounter the first attempts to record the vernacular in writing.1⁷ The numerous morphosyntactic variants of a given word that are found in these texts are not easily integrated into paradigms, even multiple ones, in the way that is possible for the forms of a more systematically codified language. The difficulty lies first in deciding how to project the micro-tendencies observable in the documents written by each notary into individual micro-systems, and second how to incorporate all such micro-paradigms into a single overarching system. A further complication lies in the fact that the phenomena in question derive from written texts, with the consequent difficulty of projecting the micro-systems reconstructible from the written language onto those appropriate for the colloquial spoken language. We may wonder whether the inflexional patterns we find in these texts have an inherent ‘systematicity’, albeit only tendential, or whether they are merely the relics of a system which is already dying out. We may further ask to what extent they are the harbingers of the inflexional systems of the emergent Romance vernaculars. Some of the patterns have an evident link both in meaning and form to ‘classical’ Latin.1⁸ Others share only the form, which has become atrophied, but have 1⁷ For detailed discussion of these issues, see Sabatini (1965; 1968), Sornicola (2017a), and Proietti (2019). 1⁸ By ‘classical’ here we do not mean the Latin of Cicero and other writers of the so-called Golden Age (otherwise referred to in this volume as classical Latin) but rather that Latin whose structural

polymorphism and polyfunctionalism in diachrony

319

assumed different functions. Yet others have diverged from the classical language in both form and function. In any case, comparison with classical Latin is never more than a heuristic allowing us to identify divergences that can be traced back to different chronological strata. Here we focus in particular on nominal morphology. At the outset it needs to be emphasized that any coincidence of form and function with classical usage is by itself no guarantee of continuity. A key consideration is how long certain important abstract properties of the system lasted, in particular the conflation of case and number whereby every form gives expression to both, so that for example -um is accusative singular and -ōs is accusative plural for nouns of the second declension. As is well known, by the time of the earliest Italo-Romance texts this property has been lost so that the forms realize only number and not case. In this connection the documents from Cava dei Tirreni represent a grey area in which, for various reasons, for some forms the case value has been atrophied or lost and the number value is only deducible from external factors such as verbal or adjectival agreement or the semantic context, while others have preserved a full case value even if that is not always expressed by the form that would have had that value in the classical language. Polymorphism in the classical Latin noun is constrained by patterns of alternation determined by gender and declension.1⁹ With some lexical exceptions, both the polymorphism and polyfunctionalism attested in the documents examined here are in large measure different from the classical patterns and, to a lesser degree, from those to be seen in late Latin literary texts and high-register documents. There is a further problem when it comes to the analysability of the form– function relation in our texts, since the structural patterns based on the traditional cases (nominative, genitive, and the like) are not found here. We will therefore adopt instead the relational categories subject, object, specifier, and indirect object together with the syntactically defined category prepositional phrase (PP). This last is required both for adverbial structures and for the specifier or possessive when it is realized by de ‘of ’ + NP and the indirect object realized by ad ‘to(wards)’ + NP.

11.5 Inflexional systems, polyfunctionalism, and polymorphism The results of our investigations allow us to reconstruct inflexional microsystems and to derive some implications, both synchronic and diachronic, therefrom. At the same time, problems arise due to individual variation between features are consistent across different periods, regions, and registers, and which in consequence can be considered ‘perennial’ (I owe this term to Pierluigi Cuzzolin). For further discussion, see Rosén (1999), Cuzzolin (2010; 2013), and Vincent (2016b). 1⁹ There are, however, no absolute dependencies between gender and inflexional class. See Aronoff (1994:61–87) for a particularly interesting theoretical discussion of this question.

320

rosanna sornicola

notaries and cultural variation between writing centres. Our general conclusions can be summarized as follows, with more detailed reflections set out in §§11.6–11.15. A. One relevant methodological principle appears to involve the need to consider the initial morphological conditions (stem and ending) of the units under investigation. Changes in the inflexional forms seem to be relatively more homogeneous for nouns of class I and II, and more differentiated and fragmented for class III nouns, which had a greater variety of stem structure and inflexional allomorphy.2⁰ B. Despite significant phonetic and morphological changes, the Latin inflexional classes maintain a clearly recognizable identity, not least because of the persistence of the stem patterns. C. Inflexional allomorphy across the cells of the paradigm tends to be reduced by comparison with classical Latin, in both the singular and the plural of all classes, although the details differ from class to class. Moreover, even within those paradigm cells which remain active the classes differ in the number of available variants and the conditions which determine their distribution. Class I polymorphism is determined both by sound changes affecting the final syllable and by the linguistic competence of the individual notaries, while for class II and particularly for class III the structural reorganization of the paradigm is more important. Reduction in allomorphy assumes various guises, all ultimately connectible to the neutralization of case contrasts, with those between the nominative and the accusative being of particular interest, as is the way these developments play out differently in the singular and the plural. The structures of the inflexional paradigms of the first two noun classes are summarized in Table 11.4.21 Such synchronic representations of the paradigms reconstructible for late Latin from our documents pose the question as to whether it is feasible, diachronically speaking, to consider the reductions when compared to classical, grammatically correct, Latin as loss of cells, where cell is understood as a way of modelling the relation between morphological form and syntacticosemantic function. It is no coincidence, for example, that a corresponding table to 11.4 for class III nouns is not feasible given that the structures vary from lexeme to lexeme as we will see in §11.11. We return to the general question in §11.16 but in the meantime we will use the cells in Table 11.4 as the framework for the analysis of form–function relations.

2⁰ Henceforth I will use ‘class’ to refer to data derived from the documents analysed here—class I nouns have stems in -a-, class II in -o-, and class III in consonants or -i- — and ‘declension’ to refer to the paradigms of traditional Latin grammar. 21 The inflexional variants and the minority patterns are placed in parentheses and in order of frequency. Cells marked with ‘–’ are those where our corpus does not contain relevant data. Here and in what follows we use the term ‘specifier’ for the various relations that obtain between a genitive or de-phrase (‘of-phrase’) and its head noun.

polymorphism and polyfunctionalism in diachrony

321

Table 11.4 Inflexional paradigms of classes I and II I -aSingular Subject -a Specifier de + X_-a Indirect object (-ae) Object -a/(-am) PP P + X_-a

Plural -e/(-as)/(-es)1 de + X_-e/(-aru)2 – -e/-as3 -e/(-as)/(-is)/(-abus)4

II -oSingular -us/(-os) -i/-de + X_-o (-i) – (P + X_-um)

Plural -i/(-is) – – -os P + X_-i/ (P + X_-is)/ (P + X_-os)5

1

-es occurs only once in filies. Only in the form nuzzaru in the formula in die nuzzaru (= nuptiarum ‘marriage.gen.pl’) ‘on the day of the marriage’. 3 Forms in -as make up roughly one-third of the forms used to mark plural objects. 4 The form -abus is only found with the lexeme filia ‘daughter’; the form filiabus had been long attested in juridical and legal Latin presumably in order to distinguish the female descendant from the male filius/filiis ‘son.nom.sg/dat-abl.pl’. 5 The less frequently attested endings often do not respect the rules of classical Latin when it comes to cases governed by prepositions.

2

D. There are some infrequent but not negligible patterns which we can call ‘case/number polyfunctionalism’, and in which the same form recurs in different contexts either as singular or as plural and sometimes with the same syntactic function (see §11.14). E. Although, in the synchronic patterns deducible from our documents, some syntactic contexts seem to be associated with significant differences compared to classical norms (PPs, and equative and appositional constructions), we can nonetheless say that the impact of syntactic factors is secondary. More directly relevant to understanding the breakdown of the classical system and the development of diverse, lexically determined, microparadigms are the purely morphological nominal class patterns (Sornicola 2016a; 2017b), as predicted by theories of autonomous morphology (Maiden 2018a). F. The study of form–function variants allows us to understand the mechanisms relevant to our theoretical interpretation of the synchronic and diachronic processes at work here and thus reveal the logic underlying paradigm shift. Some outcomes are the consequence of natural phonological processes such as the loss of final nasals, while others result from neutralizations, as for example case and/or number syncretisms, or from states of inertia which have as a consequence so-called atrophization. We discuss these in turn in the sections that follow.

322

rosanna sornicola

11.6 Influence of phonetic factors Phonetic factors have different effects according to the class of the nouns. They are particularly evident in respect of -a- and -o- stem nouns. In the singular of -a-stem nouns, the high percentage of forms lacking a final nasal—a phenomenon which in its own right is very old—22 means that inflexion is levelled to -a, reinforced by the overlap with the old ablative which had by this time lost its distinguishing vowel length. In sum, the levelling of case alternations is the outcome of phonetic and prosodic changes affecting the final syllable of the word. By contrast, in the singular of class III masculine and feminine consonant and -i-stem nouns, the structure of the final syllable, including the consonants -s and -m, is relatively well preserved.23 One can therefore establish the following hierarchy of the degree to which phonetic factors impinge on noun classes: I class < II class < III class. As anticipated in §11.4, this has repercussions on the internal polymorphism of paradigm cells, which will differ from class to class.

11.7 Atrophization By atrophization we mean the failure of certain form–function pairings to participate in paradigmatic alternations. Such reduced forms are typically restricted to fixed syntactic contexts, and are particularly characteristic of the legal formulae which abound in these texts They must have been memorized by the notaries as part of their training, are not productive and are not necessarily further analysable into their constituent parts. There are various subtypes: (a) the form is unchanged and appears in a syntactic context appropriate to its original function but only in fixed expressions as for instance the genitive Heredum (Eredum) and the crystallization of forms in -ibus; (b) the form has undergone sound change but still appears in contexts consistent with its original function as for example the genitive plural of nouns in -a- in formulaic expressions (nuzzaru = nuptiarum, «in die nuzzaru» ‘on the day of the wedding’, 855, Salerno, L, 34, 3); (c) the form is unchanged but appears in contexts where it has lost its original function; for example Latin nominative singulars which appear as direct objects or inside PPs. This particularly applies to class II nouns such as presbyter ‘priest’, uir ‘man’, and some class III items such as lex ‘law’, iudex ‘judge’, mulier ‘woman’, pars ‘part’. 22 It is frequently attested in the inscriptions from Pompeii in words belonging to various morphological classes (Va¨a¨na¨nen 1966:71–77). 23 Both these conclusions are confirmed by the analysis of documents from adjacent regions—see for example the evidence of documents from Amalfi and Naples in Sornicola (2007; 2012).

polymorphism and polyfunctionalism in diachrony

323

11.8 Overextension of the stem Many third-declension imparisyllabic nouns in the singular exhibit oblique stems such as capite ‘head’, fratre ‘brother’, genetrice ‘mother, female parent’, muliere ‘woman’, which have extended their function to include the subject marking that would classically have required the nominative (caput, frater, genetrix, mulier). This phenomenon, well documented in historical grammars, is standardly seen as a core part of the morphological shift from Latin to Romance. Indeed, for many originally third-declension lexical items the Romance languages display forms derived from the oblique stem, which was by far the most common since it occurred in all case and number forms except the nominative singular. Here too, therefore, the original distribution has determined the historical remodelling of the paradigm but in this instance, unlike with the -a-stems, the factors are purely morphological and analogical rather than phonetic. Viewed diachronically, these developments demonstrate the effect of analogy in reducing polymorphism.

11.9 Overextension of inflexion The overextension of inflexion can be observed in the plural of class I and II nouns. Thus, -a-stem nouns tend to generalize the inflexion -e to all functional contexts, with a notable frequency of agreement in -e with a class I adjectival modifier or a participle: abente fine una de ipse pezze de binea ‘having as boundary one of the parts of the vineyard’ (837, Salerno, L, 14, 7). Allomorphy involving -e and -as (with -as clearly in the minority) is consistent with the system of marking which contrasts the subject with the direct object or the object of a preposition and is only found in documents drawn up by the more learned notaries. The same general tendencies can be seen with nouns in -o, with the difference that the accusative plural in -os marking objects and in apposition is more frequent and there is a higher degree of variation between -i and -is with the arguments of prepositions. It is possible that phonetic factors—in particular the instability of final -s and the shared vowel—have played a role here.2⁴ These results lead one to think that the overextension of the -e ending in -astem nouns and of -i in -o- stem nouns does not depend on phonetic processes but is the consequence of analogical mechanisms which restructure and at the same time reduce the number of cells in the paradigm, changes which operate at a level of autonomous morphology. Although the data come from written documents, they may support the hypothesis that in the spoken registers that lie behind them, 2⁴ The possible influence of the phonetic factor is also found with singular specifiers. It is particularly typical of Alhoini, a notary whose command of language is marked by uncertainties and hypercorrections.

324

rosanna sornicola

-e was in the process of generalizing to all the plural cells of -a- stem nouns. It is less clear whether a similar conclusion is justified for the -i ending of -o- stem nouns. In both instances it is possible that the reduction in the number of cells compared to the paradigms of the classical language was a factor which favoured the extension of the forms in -e and -i via an intermediate stage in which nouns in general only had two or at most three cases (the oscillation between two and three cases varies from notary to notary).

11.10 Preservation of morphological properties of class III nouns In the plural consonant stem and -i- stem nouns tend to extend the inflexions -es, -e, -is, -i to different contexts (subject, object, object of preposition) while the form in -ibus is restricted to the prepositional object function. Given that in classical Latin consonant stem nouns had -ēs in both the nominative and the accusative, while -i- stems had -ēs in the nominative and -īs in the accusative, we can hypothesize that the latter pattern was disturbed in two different ways at different times: either the levelling of both -ēs endings as a result of the accusative being attracted to the nominative or levelling of -īs due to the nominative being attracted to the accusative (Leumann et al. 1972:I:437–440). A confusion of the inflexional properties of nouns of both classes could then have taken place in the lower sociolinguistic registers at a later date (Gaeng 1994). Such a development would explain the oscillations observable in our documents, and which cannot be explained, as they can be for documents from other regions, by orthographic oscillations between -e/-i and -o/-u, given that these are not frequent. Moreover, there are discernible differences between the lexical subtypes. Nouns in -i conform more strictly to the pattern with -is, -i (but the variant in -i is only really frequent in the type finis ‘boundary, limit’, of which there are a high number of occurrences of the various forms). The consonant stem nouns homo ‘man’ and heres ‘heir’ display a relatively conservative tripartite division with -es for subject, object, and object of preposition, -um for the specifier and -ibus with a preposition, presumably in virtue of the formulaic contexts in which they occur with the greatest frequency (cf. §11.7). On balance, then, and with some exceptions, one can recognize the relative permanence of the paradigms of classical Latin for nouns of the third class.2⁵ It is not without interest that, while the old forms of the genitive plural are very rare for all classes of noun, the forms of the dative are more resilient.

2⁵ It must be remembered that the analysis of the morphological coding of subjects and objects is constrained by the lack of examples of plurals for some lexemes and by the predominance of nonsubject contexts in the available texts.

polymorphism and polyfunctionalism in diachrony

325

11.11 Differentiation and hypodifferentiation of paradigm cells We have already seen that the nominal paradigms we can reconstruct vary according to noun class, and that there is a detectable external variable, especially for class I and II nouns, in the cultural level of the notaries. As we saw in §11.5, the difference across the two case cells is minimal in the singular of class I and II nouns with the morphology of the old accusative sporadically preserved in some object and prepositional contexts: cum ipso germanum meum ‘with himself.abl brother.acc my.acc (= with my brother himself)’ (868, Salerno, LI, 25, 10). Alternation between cells is more evident in class I nouns (-e/-as) and especially class II (-i/-os). For the latter it is even possible to reconstruct a triple case pattern -i/-os/-is, with the predominant -i forms still in competition with forms in -is which in places still survive. Generally speaking, then, we can say that nouns of the first two classes have underdifferentiated paradigms and a low level of polymorphism. Things are very different when it comes to both consonant and vowel stem class III nouns. As the number of cells increases so does the extent of biunique form–function relations. Set out below are the alternating forms characteristic of some groups of nouns. Singular A. Lexemes with more than three forms varying according to the cells of the paradigm. In the usage of most notaries these items tend to display a biunique correspondence between form and syntactic function: genitor ‘parent’: genitor (ienitor), genitori, genitorem, genitore iudex ‘judge’: iudex, iudices, iudici, iudicem, iudice uxor ‘wife’: uxor, uxoris, uxori, uxorem, uxore emptor ‘buyer’: emptores, emptoris, emptori, emptorem, emptore lex ‘law’: lex, legis, legem, lege B. Lexemes with three or fewer than three forms varying according to the cells of the paradigm: nepos ‘grandchild, nephew, niece’: nepotes, nepotem, nepote parens ‘relative’: parentes, parenti pars ‘part’: pars, partem, parte pater ‘father’: pater, patri portio ‘piece’: portionem, portione sors ‘share’: sortem, sorte sortio ‘inherited share’: sortionem, sortione

326

rosanna sornicola

C. Lexemes displaying no alternations: paries ‘wall’: pariete mater ‘mother’: matris Plural A. Lexemes with more than three forms varying according to the cells of the paradigm. These items display a more or less evident biunique polymorphism of syntactic function: emptor ‘buyer’: emptoris, emptori, emptores iudex ‘judge’: iudices, iudici, iudicibus paries ‘wall’: parietes, parieti, parietibus parens ‘relative’: parentes, parenti, parentibus pars ‘part’: partis, parti, partibus

The lexeme parens displays polymorphism as subject, in apposition, and as the object of a preposition. It is also characterized by polyfunctionalism. The paries type is also polymorphic when object of a preposition. B. Lexemes with two forms varying according to the cells of the paradigm: arbor ‘tree’: arbori, arboribus portio ‘piece’: portiones, portione sortio ‘inherited share’: sortiones, sortioni C. Lexemes displaying no alternations: lex ‘law’: legibus sors ‘share’: sorti termes ‘bough’: termiti

Some lexemes (frater ‘brother’, genetrix ‘mother, female parent’, portio ‘piece’, testis ‘witness’) display clear polymorphism with respect to syntactic function. In the singular, the forms genetrix/genetrices/genitrices are all associated with the subject function. The forms frater/frates/fratres found in the documents of some notaries are also always associated with subject. Conversely, portiones/portionem/portione are all used to mark the object, often with a partitive value. In the plural clear polymorphism is to be seen with the nouns pars, testis, iudex, nepos; in particular pars is highly polymorphic when it is the argument of a preposition with the forms partis/parti/partibus occurring freely.2⁶ The variants 2⁶ In this context there is perhaps also one occurrence of the atrophied form pars: de duobus pars ‘of two parts’ (859, Salerno, LI, 17, 5).

polymorphism and polyfunctionalism in diachrony

327

testes/testis and nepotis/nepotibus are also found in prepositional phrases. These facts suggest two hypotheses. First, the PP is a context which particularly favours polymorphism with this synchronic property in turn playing a role in diachrony in promoting paradigmatic reorganization. Second, there is a difference in status between the conservative forms in -ibus which reflect classical norms and the forms in -is, -es, -i, which are innovations at different chronological stages and which might have underlain further developments such as the Italo-Romance plurals in -i and -e for class III nouns (Sornicola 2016a). The lexemes finis, heres, and homo, which recur all the time in formulaic constructions, have a high incidence of diversified forms for the singular and plural with a pattern of variation according to functional context, which is of particular interest. We can reconstruct the following systems: Singular fine: the dominant alternant in lists finem (sporadically) or fines (only in one notary): alternants both in lists and PPs fini: minority alternant in lists and PPs Plural finis: alternant that occurs very occasionally as subject in some formulae finis, fini: prevalent alternant as object or in PPs fines, fine: minority alternant as object or in PPs and absolute locatives finibus: alternant only found in locative formulae

In sum, fine is polyfunctional with respect to number, while finis, fini, and fines are polymorphic variants as objects and in PPs. Singular (h)eredes: sporadic subject form erede: predominant object form Plural (h)eredes: sporadic subject form (h)eredes: predominant object form heredum: extremely sporadic form as specifier (h)eredibus: crystallized form in various formulae, sometimes preceded by ad ‘to(wards)’ Singular (h)omo: functions relative to the subject including one incidence of a ‘hanging nominative’ (h)omini: indirect object or specifier (h)ominem: argument of preposition (h)omine: argument of preposition

328

rosanna sornicola

Plural (h)omines: argument of preposition2⁷ (h)ominum: as a partitive2⁸ (h)ominibus: regular contexts following classical norms

These data suggest that, with occasional exceptions, some notaries were aware of the inflexional properties of some of the lexemes recurring in the legal formulae, not however in a productive way but as traditional patterns which had been learnt by heart. We can therefore postulate that for these lexemes there were ‘inert’ paradigms that existed side by side with the morphologically productive paradigms.

11.12 Interchangeability of forms and merging of paradigm cells The morphological cell of the classical Latin second declension genitive maintains its formal and functional properties only in the documents of some of the notaries as for example: uxori eiusdem Maioni germani mei ‘wife.dat same.gen Maionus.gen brother.gen my.gen (= to the wife of that same Maione, my brother)’ (858, Salerno, LI, 15, 23); fine terra Iohannemari p(res)b(ite)ri ‘edge land Iohannismarus.gen priest.gen (= at the edge of the land belonging to priest Iohannismarus)’ (882, Nocera, LII, 16, 11). Generally speaking, however, all the paradigms reconstructible on the basis of these documents display, both for class II and III nouns, forms of the genitive appearing in place of the dative and vice versa: tibi X presbiteri ‘you.dat X priest.gen (= to you priest X)’ (880, Nocera, LII, 9, 28); nongentos soli(dos) eidem mulieris […] conp(oneret) ‘900 soldi.acc same.dat woman.gen […] pay.pst.sbjv.3sg (= he should pay 900 soldi to the same woman)’ (894, [Salerno], LII, 29, 23). In sum, the forms of the dative and genitive singular are interchangeable; otherwise put, the functions of specifier and indirect object are not distinguished morphologically. For nouns of the third declension this phenomenon is particularly linked to the loss of the final -s, although in some instances this may simply be a question of orthography even if ultimately linked to sound change. This pattern is found in other corpora of Lombard documents (Petracco Sicardi 1978:132; Larson 2000:153; 2012:67) and in the Chronicon Salernitanum (Westerbergh 1956:126). Such fluctuations of form and function justify us in postulating the hypodifferentiation of paradigm cells which can be linked back to

2⁷ There is one instance of this form in a personal construction with licet ‘it is allowed’. The occurrence of homines with prepositions governing different cases is well attested in legal Latin (Sornicola 2013). 2⁸ Thus: neque a nullos quempias hominum ‘nor by any men whatsoever’ (860, Rota, LI, 21, 10).

polymorphism and polyfunctionalism in diachrony

329

ruptures in the relation between the genitive and the dative already attested in the third century (Herman 1998).2⁹

11.13 Syntactic factors For class I nouns, grammatical functions are determined on the basis of the syntactic configuration and even more on the semantic context rather than morphology and agreement. The prepositions de ‘of ’ and ad ‘to(wards)’ play a fundamental role in codifying respectively the specifier and the indirect object. They sit side by side with contexts in which the inflexional forms of the genitive and the dative survive, more frequently for class III than class I and II nouns and with the latter in particular being restricted to documents written by the notaries whose command of language was most sophisticated. Some special remarks are in order about the overextension of forms of the accusative in subject and related functions. For class I nouns, in rare instances the -as inflexion extends its boundaries to include equatives such as hoc sunt pezzas nobe ‘here are nine pieces’ (837, Salerno, L, 15, 3) and constructions with monovalent predicates such as passives and intransitives, the so-called extended accusative. In some studies this phenomenon has been linked to deeper changes in the syntactico-semantic structure and so-called unaccusativity. It has been argued that these developments became entangled with the changes in the structure of the nominal paradigm and thus prepared the way for the emergence of the Romance vernaculars. However, our studies suggest that the effect of these syntactic contexts is very limited (Sornicola 2015; 2016a).3⁰ As far as class III nouns are concerned, even in those contexts where the classical nominative is usually retained, there are occasional instances of the accusative or forms in -e in functions related to the subject such as NPs constructed with the be verb, intransitives, and equatives. It should however be noted that in such instances there may be independent local or more distant syntactic factors which might have triggered the presence of the accusative or an oblique case.31 The overextension of the accusative and obliques in general is frequent in partitive–appositive contexts which have been associated since early times with either the genitive or the accusative (Leumann et al. 1972:II:57f.). In these contexts we also frequently observe the free choice between forms in -em and -e:

2⁹ There is a considerable literature on this topic and a variety of interpretations. In addition to the works already cited, see Norberg (1943:43–45), Lo¨fstedt (1959:128, 133), and with an eye on the reconstruction of proto-Romance Dardel (1999) and Zamboni (1998; 2000). See too the balanced discussion in Stotz (1996-2004:4, 93, 267). 3⁰ Nouns ending in -is are also found in these contexts as a function of special local syntactic factors. 31 For example, the occurrence of portione in the following example might be justified by its role as subject of an infinitival clause: nec novis nec ad alios homine dicimus remanere portjone ‘neither to us nor to anyone else do we say that a part remains’ (823, Salerno, L, 9, 9).

330

rosanna sornicola

nec mihi [[…]] neque ad alio homine nihil reserbavi portjione ‘neither to myself nor to any other man have I reserved anything of the piece’ (819, Sarno, L, 6, 6, Leone) compared with nihil reserbabimus portjionem ‘we will reserve none of the piece’ (872, Salerno, LI, 34, 11). These oscillations of form are also found with the lexemes pars ‘part’, portio ‘piece’, sors ‘share’, sortio ‘inherited share’ when they occur as arguments of prepositions, a context in which agent (da ‘by’ + NP), specifier (de ‘from’ + NP), indirect object (a(d) ‘to(wards)’ + NP) and various adverbial roles converge, something which suggests that these contexts had a particularly important role in the breakdown of the classical Latin system of nominal inflexion.

11.14 Hypercharacterization of grammatical relations By hypercharacterization we mean the overlap of synthetic and analytic structures. An inflected form, which in itself already expresses the appropriate grammatical relation, is combined with a preposition, as for example ad heredibus ‘to heir.dat.pl’. This is redundant, since the dative heredibus already conveys the intended meaning and deviates from the classical norm, which requires accusative with ad. The fact that this phenomenon is attested most often in formulae and in class III plural forms in -ibus suggests that such forms were simply memorized and had lost their original syntactic and semantic value.

11.15 Polyfunctionalism of number and syntactic function In the documents we can observe some forms which are polyfunctional when it comes to their syntactic function and/or number. The same inflexion can be associated in the singular with the functions of specifier or object and in the plural with subject, object, object of a preposition, and in apposition. These patterns, just like polymorphism within the cells, attest to the instability of the paradigms in question. For class I nouns, polyfunctionalism of number is almost non-existent, but there is a certain degree of polyfunctionalism when it comes to syntax. By contrast, -ostem nouns show polyfunctionalism of both number and syntactic function, so that for example -os can mark subject and specifier in the singular and object and prepositional object in the plural. Some class III nouns also have forms which are used interchangeably in the singular and the plural, as for example iudices, iudici ‘judge(s)’. This is particularly the case for the noun emptor ‘buyer’ with the forms em(p)toris, emtores, and em(p)tori all serving both functions. This pattern may be due to the analogical reorganization of the nominative singular built on the oblique stem. For the lexemes

polymorphism and polyfunctionalism in diachrony

331

nepos ‘grandson, nephew’, pars ‘part’, sortio ‘inherited share’, testis ‘witness’ on the other hand number marking is relatively compartmentalized. That the same is true for finis, heres, and homo is not unconnected with the properties of these lexemes discussed in §11.11.

11.16 Towards a diachronic model of polymorphism and polyfunctionalism In different ways the instances of polymorphism and polyfunctionalism evidenced in the documents we have studied attest to the fragmentation and collapse of the inflexional paradigms of nouns in classical Latin, and thereby reveal the shift towards a series of micro-systems which are not yet identical to those that ultimately appear in Romance, although it is already possible to see some of the patterns that will be crystallized in the Italian vernaculars. Polymorphism and polyfunctionalism thus represent two different ways in which the traditional inflexions may achieve distributional freedom with respect to classical norms, and allow us to glimpse the diachronic dynamics at work. The emergent patterns are not yet regular systems of the kind we see in many of the highly standardized modern languages, but rather reveal to us a complex reservoir of forms and functions which do not form a unified system associated with a particular community. At the same time, the patterns do not simply reflect the different usages of individual notaries, but evidence changes within the system of nominal inflexion, some of which go back to periods before that of our documents but which were still active at that time. To take a single example, the absence of the final nasal in objects and arguments of some prepositions is a phenomenon famously present already in the inscriptions from Pompeii of the first century ce and earlier. In the written language of some of our lawyers, this nasal is intermittently present and absent, and testifies to a kind of development that we can call ‘cyclic’, emerging and re-emerging at various times over the centuries. The picture that emerges is one where there are several variants each having a certain degree of distributional freedom and not seen as threatening the clarity of communication of the individual scribes. The co-presence of different sets of rules and different degrees of competence on the part of those who were drafting the documents does not, however, mean that we cannot detect certain shared tendencies. These patterns of fragmentation and restructuring within different groups of speakers and within a circumscribed geographical area are consistent with the more general hypotheses that can be advanced concerning the different transformations of Latin paradigms which took place in different macro-regions of the Romance-speaking territory. That said, we are now in a position to sketch some general conclusions that emerge from the data we have been examining:

332

rosanna sornicola

(a) The inflexional micro-paradigms which we can reconstruct for our corpus are differentiated by lexical class, with the organization of cells being determined by syntactic function and number. The polymorphism of the cells can be seen as shining a light on the underlying changes, some of which had taken place at an earlier stage while others are ongoing at the time of our texts. (b) The neutralization of the case and/or number features of the forms in our documents when compared with those of classical Latin implies that the scribes, even when they were aware of the classical rules, did not always feel obliged to apply them and hence the forms in question were in effect in free variation. (c) When investigating the case/number neutralizations we need to bear in mind the syntagmatic refunctionalization these forms had undergone compared to the classical norms. On the basis of the extended syntactic contexts of these forms in synchrony we can derive some hypotheses concerning the corresponding diachronic trajectories. (d) The polymorphism attested in the documents can be represented as the copresence of forms in free variation within the individual paradigmatic cells. From the diachronic perspective it can be seen as an increase in the number of forms within a given cell counterbalanced by an overall reduction in the number of available cells. Synchronically, it is indicative of a ‘sense’ of the functional equivalence or identity of the forms in question notwithstanding the differences in usage among the scribes. (e) The polyfunctionalism of cells, particularly that which is associated with lexemes with restructured stems, is consistent with the refunctionalizations discernible across the full range of syntactic contexts. (f) The disappearance of the classical Latin forms can be modelled diachronically as a reduction in the number of cells in the paradigm. This in turn implies the need to rethink the definition of the inherently synchronic concept of cell in such a way as to make it more readily applicable in diachrony. (g) The polymorphism and polyfunctionalism we have considered here demonstrate the importance of a level of autonomous morphology both synchronically and diachronically speaking.

PART V

INFLEXION AND ITS INTERFACES

12 Thematic and lexico-aspectual constraints on V–S agreement Evidence from northern Italo-Romance Delia Bentley and Michela Cennamo

12.1 Introduction According to Moravcsik (1978a:365) ‘there is no language where, given a constituent class including both members that precede the verb and also members that follow the verb, the verb would agree with some or all of those members following it but with none of those preceding’. This observation finds support in verb–subject inversion constructions in many languages, where, under specific conditions (Corbett 2006:180–183), the agreement inflexion on the verb fails to spell out one or all of the relevant agreement features (Samek-Lodovici 2002). A condition on verb–subject agreement which has not received due attention (though see Parry 2000; Corbett 2006:186) is the thematic role of the controller argument, which is to be determined on the basis of the lexico-aspectual properties of the predicate (Van Valin and LaPolla 1997:113–129). The study of the dialects of Italy has played an important role in bringing to light this condition.1 In particular, in many of the dialects of the North, verb agreement with a postverbal argument in subject function varies in a way that is reminiscent of the unaccusative–unergative divide, but can only be properly captured considering the lexico-aspectual constraints on argument realization (Parry 2000; 2013; Bentley 2018).2 Relevant evidence from Milanese is provided in (1a–b). 1 We refer to the Romance dialects of Italy, which are languages in their own right, but have low sociopolitical status and limited degrees of standardization (see Maiden and Parry 1997:2f.). We dedicate this chapter to Martin Maiden, with immense gratitude and the highest respect for his lifelong commitment and invaluable contribution to the study and fostering of the Romance languages, and especially the lesser-known members of this language family. 2 For previous work highlighting the fact that agreement in VS inversion tends to be found with unergatives, and may be missing with unaccusatives, we refer to Saccon (1993); Parry (1997a:243; 2000); Savoia (1997:232); Tortora (1997; 2014); Manzini and Savoia (2005:III); Cennamo and Sorace (2007); Cardinaletti and Repetti (2010:131–133); Bentley, Ciconte, and Cruschina (2015:119–129); Romagnoli (2018). Delia Bentley and Michela Cennamo, Thematic and lexico-aspectual constraints on V–S agreement. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Delia Bentley and Michela Cennamo (2022). DOI: 10.1093/oso/9780198870807.003.0013

336

delia bentley and michela cennamo

(1) a. An ciamà i tò gent / tanti malà. (Mil.) have.3pl call.ptcp the.pl your people many patients ‘Your parents / Many patients have called.’ [Unergative intransitive] b. In / gh’ è rivà i tò suréi / di be.3pl –agr.cl= be.3sg arrive.ptcp the your sisters part pac. packets ‘Your sisters / some packets have arrived.’ (Bentley 2018:1262–1263) [Unaccusative intransitive] Disregarding the past participle, since the focus of this chapter will be on finite (i.e., person and number) agreement inflexion, we note that the perfect auxiliary patently agrees in number with the plural postverbal argument in the unergative constructions in (1a), whereas such agreement is optional in the unaccusative constructions in (1b). We shall call S the postverbal argument of the constructions under investigation, regardless of whether it controls agreement. Primary evidence from eight dialects of Piedmont, Lombardy, and Liguria reported in Bentley (2018) supports the observation that there are mismatches between verb–subject agreement and the better-known split intransitivity diagnostics. To begin with, verb–subject agreement is obligatory when the subject is a personal pronoun, regardless of verb class, with most dialects subsuming third-person pronouns among the obligatory controllers.3 (2) a. Po¨ si / **gh’ è rivà vui alter. (Mil.) then be.2pl –agr.cl= be.3sg arrive.ptcp you.pl other ‘Then you arrived.’ b. Po¨ in / **gh’ è rivà lur. (Mil.) then be.3pl –agr.cl= be.3sg arrive.ptcp they ‘Then they arrived.’ (Bentley 2018:1259) Although the examples in (2a–b) are classified as unaccusative, since the perfect auxiliary esse ‘be’ is selected in unaccusative domains in Milanese, verb–subject agreement is obligatory because the postverbal subject of these examples, unlike that of (1b), is a personal pronoun. As previously noted by Parry (2013), in some dialects the set of verbs which fail to exhibit verb–subject agreement is larger than the set of verbs associated with unaccusativity. In the Lombard dialect of Grosio, the third-person singular and plural of have is syncretic. Assuming, however, that finite agreement features can also be borne by the subject clitic accompanying the verb, we note that the alternation of the third-person plural subject clitic i with its third singular counterpart l in (3) indicates that number agreement is optional in VS order with the verb call, 3 Note that in VS constructions where S is a first- or second-person pronoun the verb is inflected for both number and person (cf. (2a–b)).

thematic & lexico-aspectual constraints

337

even though the relevant construction is unergative, as is indicated by the selection of the perfect auxiliary habere ‘have’. (3) I / l’ a ciama¨ al te pa e la toa scl.3pl= –agr.scl.3sg= have.3 call.ptcp the your dad and the your mama / tanč mala¨. (Grosio, Lmb.) mum many.pl patients ‘Your mum and dad / Many patients called.’ (Bentley 2018:1267–1268) In this chapter we report the findings of a survey of verb–subject agreement in thetic broad focus in Emilian-Romagnol, a group of northern Italo-Romance dialects (Hajek 1997). The survey was devised to test the claim that verb–subject agreement is sensitive to discourse and thematic properties of the controller, the latter type of property being determined by the position taken by the argument in the lexical–aspectual structure of its predicate (§12.2). Following Bentley and Cruschina (2018) and Bentley (2020a), we define thetic broad focus as an all-new construction in which a spatio-temporal or situational topic or argument satisfies an aboutness constraint on the predication (Bianchi 1993), while the thematic postverbal argument is part of the predication and may very well not exhibit any subject behaviour.⁴ We restrict the bulk of the analysis to number agreement on the finite form of the verb, i.e., the form which agrees with a preverbal subject in person and number. Although this form is a perfect auxiliary, in our data sample, following a well-established tradition (Rizzi 1986; Brandi and Cordin 1989), we consider the subject clitics occurring in northern Italian dialects to be extended exponents of finite agreement. Therefore, such clitics will also be within the scope of the investigation. In addition, in the qualitative analysis, we will mention agreement on the perfect participle, when the participial inflexion spells out this kind of agreement overtly. Quantitative analysis of our primary evidence suggests that the pronominal status and the thematic status of S, the latter being defined as the position of the argument in the predicate’s lexico-aspectual structure, are the most important predictors of V–S agreement. While qualitative analysis suggests that all definite Ss, rather than pronominal ones alone, may be more strongly associated with V–S agreement than indefinite ones, this hypothesis needs to be verified statistically against a larger data sample. Our findings also suggest that individual verb classes (for example, verbs of emission), determined not only in lexicoaspectual terms, but also in accordance with finer semantic properties, may stand out within their lexico-aspectual class, thus further supporting the assumption that lexico-semantic structure has a bearing on V–S agreement. Assuming that verb

⁴ In contrast with others (Sluckin, Cruschina, and Martin 2021), we assume that the postverbal argument of thetic broad focus can be a personal pronoun: although the discourse referent of this kind of pronoun is familiar to the interlocutors, or has been introduced in the previous text, the proposition in which the pronoun occurs can nonetheless constitute entirely new information.

338

delia bentley and michela cennamo

agreement is a subject diagnostic in the dialects under investigation, our results support the distinction between aboutness and thematic subjecthood, which has previously emerged from work of various theoretical persuasions (Bianchi 1993; Cardinaletti 2004; Rizzi 2005; Bentley 2018; 2020a; 2020b; Bentley and Cruschina 2018; Sluckin, Cruschina, and Martin 2021).

12.2 Theoretical preliminaries: lexico-aspectual structure and thematic roles Rather than relying on intuitive thematic role labels, we follow Van Valin and LaPolla (1997:113–129) in assuming that the thematic properties of an argument depend on the position that the argument takes in the lexico-semantic structure of its predicate, which is in turn largely determined by the lexico-aspectual properties of the predicate. We therefore begin this section with a brief discussion of lexicoaspectual structure, and then move on to the subtypes of the main lexico-aspectual classes which we factored into our investigation. Given that in all the VS constructions investigated the predicate was a verb, we shall discuss lexico-aspectual structure in terms of properties of verbs and verb classes. A number of standard tests are applied to determine the lexico-aspectual properties of a verb (Dowty 1979 and subsequent literature), and thus classify it as one of the four Vendlerian types: activity, state, achievement, and accomplishment (Vendler 1967).⁵ Both achievements and accomplishments have a result state in their lexico-aspectual structure, since they describe, respectively, punctual and non-punctual scalar change of state (Van Valin and LaPolla 1997:93; Beavers 2013). Therefore, a key distinction will be drawn between state-based and non-statebased lexico-aspectual classes, the former type comprising states, achievements, and accomplishments, the latter being the class of activities. This distinction is illustrated in Table 12.1 with the English translational equivalents of some of the verbs featuring in our questionnaire.⁶

⁵ Although individual verbs can fall into more than one Vendlerian class, depending on constructional input, an issue which is at the core of the projectionist-vs-constructionist debate (see Levin and Rappaport Hovav 2005 for an overview), such meaning alternations are not unconstrained. While this important issue goes beyond the scope of the present chapter, we should note that in our survey we classified the verbs within their construction, which ruled out most of the potential uncertainties. ⁶ Despite non-trivial cross-linguistic correspondences in the lexico-aspectual structure of verbs with comparable idiosyncratic meanings, such correspondences cannot be taken for granted. We constructed our questionnaire in Italian, providing a context for each entry, in order to facilitate a specific interpretation for each example. We asked our bilingual Italian/Emilian-Romagnol native speaker informants to translate the questionnaire entries into their dialect, which means that the informants themselves chose the most appropriate dialect verb for each given entry. We hope in this way to have maximally constrained any potential cross-dialectal mismatches in the lexico-aspectual structure of the relevant verbs.

thematic & lexico-aspectual constraints

339

Table 12.1 State-based vs non-state-based lexico-aspectual classes Non-state-based

State-based

Activities dance work phone

States remain be unwell suffice

Achievements arrive fall split

Accomplishments go (up, down…) become darker/black wither

As noted above, we take the thematic role of an argument to be diagnosed from its position in the lexico-semantic structure of its verb, which in turn largely depends on which Vendlerian class the verb belongs to. Following Van Valin and LaPolla (1997:127), there are five thematically relevant and grammatically salient positions, as shown in (4). (4) Arg of DO > 1st arg of do´ (x,…) > 1st arg of pred´ (x,y) > 2nd arg of pred´ (x,y) > arg of state pred´ (x). For our current purposes it is essential to note that the divide between activities, on the one hand, and state-based classes, on the other (see Table 12.1), corresponds to the threshold between the second leftmost position in (4), which identifies the argument of an activity, and the remaining three positions to its right, which are positions in the structure of states, including result states. There is a class of accomplishments which describes activities leading to a change of state or location (e.g., go, come (down, up …)). The higher argument of these accomplishments figures twice in the lexico-semantic structure of its verb, both as the 1st arg of do´ (x,…) and as the arg of state pred´ (x), and these verbs are included within the state-based classes, on a par with all accomplishments (Van Valin and LaPolla 1997:100–102; Van Valin 2018). Another syntactically relevant distinction, which concerns the verbs classified above as achievements and accomplishments, is the one between gradual completion verbs, or verbs of non-quantized scalar change (become darker/old(er looking), rot, decrease, etc.), and verbs of quantized scalar change, which are inherently telic (arrive, go/come up/down/in/out, die, be born, etc.) (Dowty 1979; Bertinetto and Squartini 1995; Hay, Kennedy and Levin 1999; Beavers 2008; 2011; 2013). According to Beavers (2013), only verbs of quantized scalar change lexicalize a specific final goal state, whereas the meaning of verbs of non-quantized scalar change only entails that a final state exists. We included both types in our survey to test whether a specific final goal state in the lexico-semantic structure of the verb is a predictor in V–S agreement. With respect to the result state lexicalized by the verb, another potentially relevant issue is whether this state is a goal location. This issue is known to be

340

delia bentley and michela cennamo

relevant to Sorace’s (2000; 2004) Auxiliary Selection Hierarchy (see also Cennamo 2001; 2008; 2010) and to patterns of V–S agreement in the Piedmontese dialect of Borgomanero (Tortora 1997; 2014). Therefore, we considered both verbs of inherently directed motion (Levin 1993:263f.) (arrive, go/come up/down/in/out) and other change of state verbs (die, be born, wither, grow, decrease, etc.). Lastly, among activities, we included verbs of emission, such as smell and drip, which have been shown to be of particular interest in perfect auxiliary selection (Cennamo and Sorace 2007).⁷ To summarize, thematic roles are defined as positions in the verb’s lexicosemantic structure, which reflects its lexico-aspectual structure (cf. (4)). In the following sections we report the results of our investigation of the sensitivity of V– S agreement to the status of the controller in discourse and thematic structure. We also explore any agreement mismatches in VS constructions with arguments occupying the same position in (4), since such mismatches can provide key insights into the role of the lexicon in grammar.

12.3 The case of Emilian-Romagnol 12.3.1 Our survey We conducted questionnaire-based interviews with 20 bilingual (Italian/EmilianRomagnol) native speakers of dialects of six provinces of Emilia Romagna (Bologna, Ferrara, Modena, Piacenza, Parma, Reggio Emilia).⁸ Our questionnaire consisted of 52 entries, which were syntactic minimal pairs with and without V–S agreement between the verb and its postverbal argument (see note 6). An appropriate context was given to convey an all-new interpretation of the questionnaire entries (see note 5). The informants were asked to select one of the given options, or both, depending on their acceptability judgement in the given context. If both options were deemed to be acceptable, the informants were asked to establish an order of preference between the two. The predictors—or independent variables—factored into the design of the questionnaire were the presence or lack of a state in lexico-aspectual structure ([±state], see Table 12.1) and the more fine-grained verb classifications discussed in §12.2, ⁷ The classification of verbs of emission as activities (Van Valin 2005) is controversial. Rappaport Hovav and Levin (2000) argue that these verbs are states, whereas Cennamo and Sorace (2007:82) view them as having an intermediate status between states and activities, with their variable behaviour reflecting also the degree of control of their subject. ⁸ These are traditionally classified as Emilian dialects on phonetic grounds, although Ferrarese and Eastern Bolognese have a number of Romagnol traits (Hajek 1997:273). We wish to thank Maria Cerullo, Lorenzo Filipponio, Nicola Grandi, and Cristina Guardiano for administering the questionnaire on our behalf, as well as our anonymous informants for their time and their invaluable support for our linguistic research. We also thank Giuliano Bernini, Nicola Grandi, Mair Parry, and Diego Pescarini for fruitful discussions of the data.

thematic & lexico-aspectual constraints

341

as well as the [±def(inite)] and [±pro(nominal)] status of S. Although the questionnaire was originally designed to elicit binary judgements (±AGR in VS order), we obtained a third effect, which we shall call +AGR_SV. This effect consisted in the informants translating the questionnaire entry as an SV construction, despite the suggested all-new interpretation of the example, which was expected to trigger VS order. In SV order, S–V agreement figured without exception. In §12.3.2 we describe the results of our investigation in qualitative terms, placing particular emphasis on the cross-dialectal variation attested. In §12.3.3 we report the results of quantitative analysis of a sample of the Emilian-Romagnol data.

12.3.2 Cross-dialectal variation In this section we investigate the agreement patterns of the verb and the related distribution of agreeing/non-agreeing subject clitics (on which, see Cennamo 1997:153–155; Parry 2000; 2013; Manzini and Savoia 2005:I, Chapter 2; Pescarini 2012; 2016; 2019; Bentley 2013; 2018; Poletto and Tortora 2016) in a number of Emilian-Romagnol dialects. In some varieties the non-agreeing pattern either prevails or is the only form available with a postverbal S. In such cases the finite verb lacks a subject clitic and occurs in the default third-person singular (detectable when no syncretism occurs between the third-person singular and plural forms of the verb).⁹ In turn, and in accordance with dialect variation, the verb is preceded by a non-agreeing clitic, an invariant vocalic particle a/e functioning as a thetic marker (preceding all other clitics), and/or by an expletive clitic l (identical with the third-person masculine singular clitic pronoun), which also occurs in existential and impersonal constructions (Benincà 1983; Bernini 2012; Bentley and Ciconte 2016; Pescarini 2016:745–747). The perfect participle in compound tenses occurs in the masculine singular (when gender and number differences are not overtly expressed; see §§12.3.2.1–12.3.2.2). In other varieties the agreeing and non-agreeing patterns alternate according to a clear opposition between activity (i.e., [–state]) verbs and change of state(-location) verbs, which include a state component in their semantics. This finding confirms the role played by lexico-aspectual and thematic parameters in this grammatical domain, which interact with pragmatic notions such as the (in)definiteness of S, e.g., whether this is pronominal and, when lexical, whether it is (in)definite (see Bentley 2018, and §12.3.3).

⁹ Among the dialects investigated, third-person syncretism of the finite verb, characteristic of Romagna and areas of Emilia (Hajek 1997:276; Loporcaro 2013:110), obtains only in Ferrara.

342

delia bentley and michela cennamo

12.3.2.1 Agreement variation with [–state] verbs In Noceto (Parma) non-agreement was found with [–state] verbs, irrespective of whether S was definite or not (5). (5) A ciamè i tó genitor / un sac ed pasjent. (Nct.) have.3sg call.ptcp the your parents a lot of patients ‘Your parents called/A lot of patients called.’ By contrast, in conjunction with a personal pronoun such as lor ‘they’, the highest ranking position on the Definiteness Hierarchy (Farkas 2002), only the agreeing form was accepted (6). (6) I an picè lor. (Nct.) +agr.scl.3pl= have.3pl knock.ptcp they ‘They knocked.’ With verbs of uncontrolled processes such as smell and substance emission, agreement variation reflects the more stative-like (e.g., spusar ‘smell’) or activitylike (e.g., gosar ‘drip’) nature of the verb. Thus, agreement with have was the pattern selected with both verb classes in (7)–(8), although be was also selected as an option with smell in (7), whereas the non-agreeing option was only found with be in conjunction with drip and an indefinite S in (8). (7) I an spusè/ en spusedi d’ fum +agr.scl.3pl= have.3pl smell.ptcp be.3pl smell.ptcp.fpl of smoke par di gioren chil giachi lé. (Nct.) for part days those jackets there ‘Those jackets smelled of smoke for some days.’ (8) I an /è +agr.scl.3pl= have.3pl be.3sg di gioren du part days two ‘Two taps dripped for some days.’

(**a) gosè par have.3sg drip.ptcp for rubinètt. (Nct.) taps

In the province of Bologna and surroundings, the agreeing and non-agreeing patterns with a postverbal non-pronominal S also appear to reflect the (in)definite nature of the latter to varying degrees in different dialects. In Lizzano in Belvedere, for example, both agreement options were found with a(n in)definite postverbal S (9a–b), though there was a preference for agreement among some informants. (9) a. I an ciamà +agr.scl.pl= have.3pl call.ptcp b. L’ a ciamà –agr.scl= have.3sg call.ptcp ‘Your parents/many patients called.’

i the i the

tóo / yours tóo / yours

dimóndi many dimóndi many

amalà. (Liz.) patients amalà. (Liz.) patients

thematic & lexico-aspectual constraints

343

With some verbs such as tosir ‘cough’ a postverbal (in)definite S, whether with or without agreement, was excluded, and only the SV order was accepted (10). (10) I mee fioo /Du amalà i an tosì. (Liz.) the my children two patients +agr.scl.3pl= have.3pl cough.ptcp ‘My children/two patients coughed.’ VS order is also impossible with emission verbs indicating smell (e.g., puzar ‘smell’) and substance (e.g., gociar ‘drip’) whenever the subject is definite, as in (11) for the verb drip. (11) El giacche i’ an puzà d’ fummo per the jackets +agr scl.pl= have.3pl smell.ptcp of smoke for trii dì. (Liz.) three days ‘The jackets smelled of smoke for three days.’ With other verbs, e.g., lavorar ‘work’, the agreeing option was the only form available, even with an indefinite subject. (12) I an lavorà in tanti /**l’ a +agr scl.3pl= have.3pl work.ptcp in many –agr.scl= have.3sg lavorà in tanti. (Liz.) work.ptcp in many ‘Many people worked.’ When the subject is the tonic pronoun lorre ‘they’ (13), a prototypical definite subject, only the agreeing option is available. (13) I an bussà lorre. (Liz.) +agr scl.pl= have.3pl knock.ptcp they ‘They knocked.’ Thus, in this variety both agreement possibilities occur in conjunction with a postverbal (in)definite S. The agreeing option is associated with definite Ss with some verbs, and is the only pattern available with pronominal subjects. SV order occurred with some verbs, whether the subject is definite or not, including verbs of emission with a definite subject (cf. also §§12.3.2.3–12.3.2.3). For some speakers of other varieties in the province of Bologna (e.g., Calderara di Reno), the non-agreeing option generally occurred with an indefinite S, sometimes alternating with the agreeing form (14a). By contrast, the agreeing option was most typically found in conjunction with preverbal (in)definite subjects (including personal pronouns) (14b), but only occurred with postverbal subjects if indefinite (14c).

344

delia bentley and michela cennamo

(14) a. L’ à viazè poca zaint/ I –agr.scl= have.3sg travel.ptcp few people +agr scl.3pl= 'äin viazè in puc. (Cld.) have.3pl travel.ptcp in few ‘Few people travelled.’ b. I tu gentor / lour / dimóndi malè the your parents they many patients 'äin ciamè / busè. (Cld.) i +agr.scl.3pl= have.3pl call.ptcp knock.ptcp ‘Your parents/many patients/they called/knocked.’ 'äin telefonè c. Incù i dimónd cliéint. (Cld.) today +agr scl.3pl= have.3pl phone.ptcp many ‘Many customers phoned today.’

customers

Verbs of emission such as smell and drip, on the other hand, did license agreement with a postverbal definite subject in conjunction with auxiliary have (15). 'äin (15) I puzè d fómm par di dé stî +agr scl.3pl= have.3pl smell.ptcp of smoke for part days these gia¨c. (Cld.) jackets ‘These jackets smelled of smoke for some days.’

To sum up, in this dialect non-agreement occurred with an indefinite postverbal S, while agreement with a definite subject was only accepted with verbs of emission, i.e., the subset of activity verbs ranking lowest in agentivity (cf. §§12.3.2.3–12.3.3). In the dialect of San Giovanni in Persiceto, also from the province of Bologna, the agreeing option was the only pattern found with postverbal pronominal subjects (16a). With a definite lexical postverbal S (16b), the agreement option was reported by our speaker in his forties as more typical, albeit perceived as less dialectal and employed for emphasis or contrast, but in any case alternated with the non-agreeing form (without a clitic) in (16c): (16) a. I an batò lour. (Prs.) +agr scl.3pl= have.3pl knock.ptcp they ‘They knocked.’ b. I an ciamè/ tusò i tu genitur. (Prs.) +agr scl.3pl= have.3pl call.ptcp cough.ptcp the your parents c. A ciamè /tusò i tu genitur. (Prs.) have.3sg call.ptcp cough.ptcp the your parents ‘Your parents called/coughed.’ Variation obtained, however, with an indefinite S, in that some verbs (e.g., ciamar ‘call’, tosir ‘cough’, viazer ‘travel’) only allowed the non-agreement form (17a), whereas others (e.g., balar ‘dance’) also permitted the agreement option (17b).

thematic & lexico-aspectual constraints

345

(17) a. A ciamè dimóndi clieint. (Prs.) have.3sg call.ptcp many customers ‘Many customers called.’ b. (A la festa) i an balé dimóndi at the party +agr.scl.3pl= have.3pl dance.ptcp many invidè. (Prs.) guests ‘Many guests danced at the party.’ With the stative-like verb of emission smell (18a) agreement was found (together with the invariable clitic a, which forms a composite form with the agreeing subject clitic gli; cf. Manzini and Savoia 2005:I:122), whereas with the activity-like verb drip only the non-agreeing option was found (18b). (18) a. A gli an fat pozza ed fomm a= +agr.scl.3pl= have.3pl make.ptcp smell.inf of smoke par di dé cal giac lè. (Prs.) for part days those jackets there ‘Those jackets smelled for some days.’ b. A guzlè par di dé du rubinett. (Prs.) have.3sg drip.ptcp for part days two taps ‘Two taps dripped for some days.’ Thus, in this variety agreement was the only option with pronominal subjects. With a(n in)definite lexical S either option was possible, with a preference for non-agreement in non-contrastive contexts. A different pattern of variation was found in the province of Piacenza (Ponte dell’Olio), where some speakers only accepted the non-agreeing form (without a non-agreeing clitic) in conjunction with a postverbal (in)definite S (19a), including with verbs of emission such as smell in (19b). The agreement option, by contrast, only occurred with SV order (20). ' i to (19) a. Alla festa a balä genitur / botta at.the party have.3sg dance.ptcp the your parents many ' . (Pnt.) invidä guests ‘Your parents/many guests danced at the party.’ ' par di dé b. A spusä chil giacchi lè. (Pnt.) have.3sg smell.ptcp for part days those jackets there ‘Those jackets smelled for some days.’ ' i (20) Alla festa i to genitur / botta invidä at.the party the your parents many guests +agr scl.3pl= ' . (Pnt.) han balä have.3pl dance.ptcp ‘Your parents/many guests danced at the party.’

346

delia bentley and michela cennamo

Agreement with a postverbal subject was only possible when pronominal (cf. §12.3.2.3), as in (21a), whereas for our informant in his forties agreement occurred with definite subjects (21b), although it was perceived as less authentic and influenced by Italian. (21) a. I an +agr.scl.3pl= have.3pl lur. (Pnt.) they ‘They have rung.’ b. I an +agr.scl.3pl= have.3pl ‘Your parents called.’

' ' lur /** A sunä sunä ring.ptcp they have.3sg ring.ptcp

' i to ciamä genitur. (Pnt.) call.ptcp the your parents

Thus, in this variety only the non-agreement pattern is accepted with VS order, with the exception again of pronominal subjects. In Modena speakers almost exclusively chose the agreeing pattern (22a), the only option with pronominal subjects. In this structure, at times alternating with the corresponding SV pattern when the subject is definite, some informants used the invariable clitic a, which forms a unit with the agreeing clitic (22b–c), in contrast to [+state] verbs where a functions as a thetic marker (cf. §12.3.2.2). ' tu (22) a. I an ciamä pèder e tu +agr.scl.3pl= have.3pl call.ptcp your father and your ' . (Mod.) mèdra / dimóndi malä mother several patients ‘Your parents/several patients called.’ b. A gli an tossì i me ragazol. (Mod.) a= +agr.scl.3pl= have.3pl cough.ptcp the my children ‘My children coughed.’ ' a gli c. I du malä an tossì tota the two patients a= +agr.scl.3pl= have.3pl cough.ptcp all la not. (Mod.) the night ‘The two patients coughed all night.’

In the rare examples of non-agreement (23a), the verb was preceded by the nonagreeing clitic l with some informants (23b). ' tu (23) a. A ciamä pèder have.3sg call.ptcp your father ‘Your father and mother called.’ ' b. L’ a ciamä –agr.scl= have.3sg call.ptcp ‘Your parents called.’

e tu mèdra. (Mod.) and your mother i to zenidor. (Mod.) the your parents

thematic & lexico-aspectual constraints

347

The verbs of emission puzzar ‘smell’ and gozzular ‘drip’ were only found with full agreement and variously alternating between the auxiliaries be/have, as shown in (24). (24) Soun puzzè ed fam par di giaurn quel giachi lè / be. 3pl smell.ptcp of smoke for part days those jackets there ' et fum I an spusä pur tri dè +agr.scl.pl= have.3pl smell.ptcp of smoke for three days quel giachi lè. (Mod.) those jackets there ‘Those jackets smelled for some/three days.’ In Ferrara, where agreement is signalled by an agreeing subject clitic on account of syncretism of the third-person singular/plural forms of the finite verb, both agreement options alternated with a definite S (25a), although one of our older informants (aged 70) preferred the non-agreeing variant in this case. However, the agreeing option was not allowed with some verbs (e.g., ciamar ‘call’) if S was indefinite (25b). Pronominal subjects (25c), by contrast, always licensed full agreement. (25) a. I a ciamà i to genitori / a +agr.scl.3pl= have.3 call.ptcp the your parents have.3 ciamà i to genitori. (Fer.) call.ptcp the your parents ‘Your parents/they have called,’ b. **I a ciamà purasà pazient. (Fer.) +agr.scl.3pl= have.3 call.ptcp several patients ‘Several patients have called.’ c. I a pigià lor. (Fer.) +agr.scl.3pl= have.3 knock.ptcp they ‘They have knocked.’ Our informant in her mid-fifties allowed the non-agreeing option with an indefinite S and tended to reserve agreement for definite subjects, which may occur in preverbal position without inversion. By contrast, two other informants in their mid-forties only allowed agreement with a postverbal (in)definite subject. With emission verbs only the agreement option obtained (26a–b): (26) a. I a sgullà par di gioran du +agr.scl.3pl= have.3 drip.ptcp for part days two rubinit. (Fer.) taps ‘Two taps dripped for some days.’

348

delia bentley and michela cennamo

b. I a puzà da fum par di gioràn +agr.scl.3pl= have.3 smell.ptcp of smoke for part days chill gabann. (Fer.) those jackets ‘Those jackets stank of smoke for some days.’ A less varied picture emerges in Reggio Emilia. Some informants (in their fifties) only used the non-agreeing form (27a), while others accepted either option with a(n in)definite postverbal S (27a–b), viewing the non-agreeing pattern as characteristic of older speakers. With pronominal subjects only the agreeing pattern was found (27c). (27) a. A balèe i to / na mocia d invidè. (Reg.) have.3sg dance.ptcp the yours a lot of guests ‘Your parents/many guests danced.’ an b. I ciamèe i too / tant pasieint. (Reg.) +agr.scl.3pl= have.3pl call.ptcp the yours many patients ‘Your parents/many patients called.’ c. I an ciamèe lor. (Reg.) +agr.scl.pl= have.3pl call.ptcp they ‘They called.’ Variation occurred with verbs of emission: the stative-like verb pussar ‘smell’ licensed the agreeing pattern with definite subjects (28a), whereas the activity-like verb sgussler ‘drip’ allowed either option with an indefinite S (28b). (28) a. I an pussèe ed fom per di gioren +agr.scl.3pl= have.3pl smell.ptcp of smoke for part days cal giachi lè. (Reg.) those jackets there ‘Those jackets stank of smoke for two days.’ b. I an sgusslèe/ A l’ a +agr.scl.3pl= have.3pl drip.ptcp a= –agr.scl=have.3sg sgusslèe per du dè du rubinet. (Reg.) drip.ptcp for two days two taps ‘Two taps dripped for two days.’ In the dialect of Correggio, on the other hand, the non-agreeing option was the only possibility with a(n in)definite S (29a), the exception being pronominal subjects which only licensed full agreement. Agreement was also the only option with preverbal (in)definite subjects (29b), including with verbs of emission. (29) a. A ciamè i to / dimóndi malé. (Crg.) have.3sg call.ptcp the yours many patients ‘Your parents/many patients called.’

thematic & lexico-aspectual constraints

349

b. I mé putein / du malé ij an the my children two patients +agr.scl3pl= have.3pl tusù tota not. (Crg.) cough.ptcp all night ‘My children/two patients coughed all night.’

12.3.2.2 Agreement variation with [+state] verbs Verbs with a state component in their lexico-aspectual structure (i.e., states, accomplishments, and achievements) show a lesser amount of variation with a postverbal S than activity verbs, and are sensitive to the (in)definiteness and pronominal/lexical distinctions in some varieties. In Noceto (Parma), for example, agreement obtained with pronominal subjects (30a), but not with a lexical (in)definite S (30b): (30) a. I en rivè lor. (Nct.) +agr.scl.3pl= be.3pl arrive.ptcp they ‘They arrived.’ b. È rivè il to soreli / socuant pac. (Nct.) be.3sg arrive.ptcp the your sisters many parcels ‘Your sisters/many parcels have arrived.’ In Lizzano in Belvedere, in the non-agreeing case we found either the thetic marker e (an invariable clitic; Manzini and Savoia 2005:I:71), which co-occurs with the etymologically locative clitic ghe (31a), or the expletive clitic l (31b). Alongside these patterns, we also found the possibility of agreement, which was in free variation with the non-agreeing form when S was indefinite (31a), but was preferred when the subject was definite (31b). With other verbs such as sparir ‘disappear’, the agreement/non-agreement options correlated with the definite/indefinite distinction (31c). (31) a. E gh’ è e –agr.cl= be.3sg I gh’ en +agr.scl.pl –agr.cl= be.3pl ‘Many people stayed.’ b. L’ è arvanzà –agr.scl= be.3sg remain.ptcp I en arvanzà +agr.scl.pl be.3pl remain.ptcp ‘Your students remained.’

arvanzà remain.ptcp arvanzà remain.ptcp i the i the

tòo your tòo your

tanta many in in

gente / people tanti. (Liz.) many

student / students. student. (Liz.) students

350

delia bentley and michela cennamo

c. I en sparì +agr.scl.pl= be.3pl disappear.ptcp L’ è sparì –agr.scl= be.3sg disappear.ptcp ‘Your sisters/some money disappeared.’

el the di part

too sorel / your sisters baiocchi. (Liz.) money

With pronominal subjects, by contrast, only the agreement option was found. In Calderara di Reno a slightly different picture emerges: (i) agreement was generally not found with definite subjects (32a), which only occurred preverbally (32b), but was found in conjunction with indefinite postverbal subjects (32c); (ii) there was a preference for non-agreement with an indefinite postverbal S, where the verb occurred either together with the thetic marker a and an etymologically locative dummy clitic j (Benincà 2007; Pescarini 2016:748), as in (32d, f), or with the expletive clitic l (32e); (iii) only rarely was non-agreement with a definite postverbal S found (32f): (32)

a. ** I ein stè +agr.scl.3pl= be.3pl be.ptcp tu nun. (Cld.) your grandparents b. I tu nun the your grandparents ein stè mèl. (Cld.) be.3pl be.ptcp ill ‘Your grandparents have been ill.’

mèl i unwell the

i +agr scl.3pl

c. I ein murt dimóndi inuzent. (Cld.) +agr.cl.3pl be.3pl die.ptcp many innocents ‘Many innocent people died.’ d. A j è arivè dimóndi lettar. (Cld.) a= –agr.cl= be.3sg arrive.ptcp many letters ‘Many letters arrived.’ e. L’ è bastè tarʃant gramm ed –agr.scl= be.3sg suffice.ptcp three.hundred grams of farina. (Cld.) flour ‘Three hundred grams of flour were enough.’ f. A j è gnó so in pajais i tu a= –agr.cl= be.3sg come.ptcp up in village the your nun. (Cld.) grandparents ‘Your grandparents went up to the village.’ By contrast, pronominal subjects always occurred in preverbal position (33).

thematic & lexico-aspectual constraints

351

(33) Lour i ein arrivè. (Cld.) they +agr.scl.3pl= be.3pl arrive.ptcp ‘They arrived.’ Summing up, in this variety with [+state] verbs definite lexical subjects tend to occur in preverbal position, the only position in which pronominal subjects can occur. In postverbal position, full agreement is excluded with a definite S whereas an indefinite S allows both options, though with a preference for non-agreement (cf. §12.3.3 for a quantitative analysis of the data). In the dialect of San Giovanni in Persiceto only the non-agreement option obtained with [+state] verbs. Generally, the agreement option was only possible with pronominal subjects (34a), although it was marginally accepted with definite lexical subjects in postverbal position in pragmatically marked contexts (emphasis or contrast), thereby ruling out a sentence-focus interpretation with this pattern. For example, in (34b) agreement was acceptable only if uttered in a situation in which the speaker underlines the fact that it was his or her grandparents who went to the village, and not somebody else’s. The non-agreeing variant in (34c), by contrast, yielded the broad focus reading. (34) a. I ein arrivè lour. (Prs.) +agr.scl.3pl= be.3pl arrive.ptcp they ‘They arrived.’ b. I ein andè so in paeis i tu non. (Prs.) +agr.scl.3pl= be.3pl go.ptcp up in village the your grandparents c. A j è andè so in paeis i tu non. (Prs.) a= –agr.cl= be.3sg go.ptcp up in village the your grandparents ‘Your grandparents have gone up to the village.’ In the non-agreeing pattern the finite verb may also be accompanied by the thetic marker a, the form most typically used (35). (35) A j è fiurè / a= –agr.cl= be.3sg blossom.ptcp piant. (Prs.) plants ‘Three plants blossomed.’

è fiurè trei be.3sg blossom.ptcp three

In Ponte dell’Olio (Piacenza) agreement with a(n in)definite S was not allowed with VS order (36a), where only the non-agreement option was possible (36b). Once again, the exception here was a pronominal S which may optionally license agreement (36c). With the verb arrive, in the non-agreeing pattern a locative adverb, e.g., chè ‘here’, là, ‘there’, must also occur with a pronominal S (36d). (36) a. Tant lettar many letters /**I enn +agr.scl.pl= be.3pl

i enn +agr.scl.3pl= be.3pl ' tant (a)rivä arrive.ptcp many

' (a)rivä arrive.ptcp lettar. (Pnt.) letters

352

delia bentley and michela cennamo

' tant lettar. (Pnt.) b. È (a)rivä be.3sg arrived many letters ‘Many letters arrived.’ ' c. I enn (a)rivä lur. (Pnt.) +agr.scl.3pl= be.3pl arrive.ptcp they ‘They arrived.’ ' ché / lé d. È (a)rivä lur. (Pnt.) be.3sg arrive.ptcp here there they ‘They arrived here/there.’

In Modena, on the other hand, agreement with a postverbal subject was well attested with [+state] verbs (37a), including without a subject clitic (37b). Our informants in their thirties overwhelmingly used full agreement, whereas our informants in their seventies alternated between both patterns, although full agreement was the only form accepted with some achievement and accomplishment verbs (an issue for future investigation). ' i to (37) a. I ein arrivä fradel. (Mod.) +agr.scl.3pl= be.3pl arrive.ptcp the your brothers ‘Your brothers arrived.’ b. Soun salì in paes i to non. (Mod.) be.3pl go.up.ptcp in village the your grandparents ‘Your grandparents went up to the village.’

In this dialect the thetic marker a and the expletive clitic l alternated in the nonagreeing construction, sometimes with one and the same verb, as in (38a–b). The invariable clitic a occurred in a wide range of structures (e.g., existentials, presentatives, meteorological contexts), and as a part of a composite (form of) subject clitic (38c). (38) a. A j è arrivé al tò sureli. (Mod.) a= –agr.cl= be.3sg arrive.ptcp the your sisters ‘Your sisters arrived.’ b. L’ è arrivé un quelc pac. (Mod.) –agr.cl= be.3sg arrive.ptcp a some parcel ‘Some parcels arrived. c. A gli ein turnedi al to sureli. (Mod.) a= +agr.scl.f.3pl= be.3pl return.ptcp.pl the your sisters ‘Your sisters came back.’ We found more variation in Ferrara. The non-agreeing pattern—variously instantiated by either the lack of a subject clitic or by the presence of the thetic marker a (with third-person distinctions on the finite verb neutralized)—was well attested

thematic & lexico-aspectual constraints

353

with our informants in their seventies with a(n in)definite S (39a), while the agreeing pattern was only marginally allowed in accordance with restrictions relating to the (in)definiteness of the postverbal S. In particular, agreement obtained mainly with definite subjects (39b), and was the only option with pronominal subjects (39c). (39) a. È rivà il to surèli / soquant pach. (Fer.) be.3 arrive.ptcp the your sisters many parcels ‘Your sisters/many parcels arrived.’ b. I è finidi il bust / **soquant +agr.scl.3pl= be.3 finish.ptcp.pl the envelopes many pach. (Fer.) parcels ‘We have run out of envelopes / of many parcels.’ c. I è rivà lor / **è rivà lor. (Fer.) +agr.scl.3pl= be.3 arrive.ptcp they be.3 arrive.ptcp they ‘They arrived.’ With an indefinite S, however, only non-agreement was generally accepted (40a–b). (40) a. È mort purasà be.3 die.ptcp many ‘Many innocents died.’ b. È arivà di be.3 arrive.ptcp part ‘Some parcels arrived.’

innuzent. (Fer.) innocents pac. (Fer.) parcels

Our informants in their mid-fifties alternated between agreement and nonagreement with a(n in)definite S, showing a slight preference for the latter. With vanzar ‘remain’, we found non-agreement (marked by the thetic marker a and the etymologically locative clitic gh) with indefinite Ss (41a). Definite subjects occurred in preverbal position with agreement (41b). (41) a. A gh è vanzà purasà a= –agr.cl= be.3 remain.ptcp many ‘Many people stayed.’ b. I tò student i è the your students +agr.scl.3pl= be.3 ‘Your students remained here.’

gent. (Fer.) people vanzà chi. (Fer.) remain.ptcp here

On the other hand, our informants in their mid-forties did not accept the nonagreeing form and otherwise preferred full agreement with (in)definite subjects. This included an agreeing subject clitic and optional agreement of the perfect participle with the postverbal subject (42).

354

delia bentley and michela cennamo

(42) Li è rivadi / arrivà i to sureli / +agr.scl.3pl= be.3 arrive.ptcp.pl arrive.ptcp.msg the your sisters soquant pac. (Fer.) many parcels ‘Your sisters/many parcels arrived.’ Thus, there appears to be a seemingly generational gradient (Nicola Grandi, p.c.), summarized in (43), moving from non-agreement to agreement, involving the presence of a subject clitic and past participle agreement with the postverbal subject, with intermediate patterns displaying an agreeing subject clitic and non-agreement of the perfect participle.1⁰ (43) è rivà > j è rivà > j è arivadi (Fer.) In Reggio Emilia, one of our two informants chose the non-agreeing form with a(n in)definite S, either in conjunction with the thetic marker a and the expletive clitic l (44a) or without a (44b). The other informant preferred agreement (44c). (44) a. A l’ è rivèe al a= –agr.scl= be.3sg arrive.ptcp the pac. (Reg.) parcels ‘Your sisters/some parcels have arrived.’ b. L’ è sparìi al –agr.scl= be.3sg disappear.ptcp the ‘Your sisters disappeared.’ c. I in rivèe al +agr.scl.3pl= be.3pl arrive.ptcp the pac. (Reg.) parcels ‘Your sisters/some parcels have arrived.’

to soreli / soquant your sisters some

to soreli. (Reg.) your sisters to soreli / soquant your sisters some

With a pronominal S speakers either alternated between agreeing and nonagreeing forms (45) or used only the former. (45) I in rivèe (lì) lor / A l’ è +agr.scl.3pl= be.3pl arrive.ptcp (there) they a= –agr.scl= be.3s rivèe (lì) lor. (Reg.) arrive.ptcp (there) they ‘They arrived.’ 1⁰ This issue needs to be investigated on a wider range of data in order to establish the existence of age-related variation and to detect whether the apparent generational change in progress towards the use of agreement with postverbal subjects (as in standard Italian), might in fact be a case of ‘agegrading’ (Sankoff and Blondeau 2007:562). We thank J. C. Smith for raising this important point. It must be stressed, however, that the influence of the standard language on Italian dialects and the existence of related age differences is undeniable also in this syntactic domain (Parry 2005 for Piedmontese and Ligurian; Cennamo and Sorace 2007 for Paduan).

thematic & lexico-aspectual constraints

355

In the dialect of Correggio, by contrast, non-agreement was the typical option, instantiated by the verb in the third-person singular (46a) and by the thetic marker a when the etymologically locative clitic ghe occurred (46b). (46) a. È sparì al to sureli/ un pò ed besi. (Crg.) be.3sg disappear.ptcp the your sisters a bit of money ‘Your sisters/A fair amount of money disappeared.’ b. A gh’ è armes i tò sculler. (Crg.) a= –agr.cl= be.3sg remain.ptcp the your students ‘Your students remained.’

12.3.2.3 Qualitative analysis: summary The high degree of variation displayed by the Emilian-Romagnol varieties investigated reveals a major division between [–state] and [+state] verbs in: (i) the distribution of agreement patterns; and (ii) the type of non-agreeing pronominal form(s) appearing in the non-agreeing pattern. More specifically, the following tendencies emerge: • [–state] verbs display a strong correlation between agreement and definite subjects, on the one hand, and non-agreement and an indefinite S on the other, while [+state] verbs are more strongly associated with non-agreement, regardless of the (in)definiteness of S. With pronominal subjects, however, agreement is the norm (cf. §12.3.3). • In some areas (Ponte Dell’Olio, Correggio), with both verb types only the non-agreement pattern is attested with a postverbal (lexical) S. The agreement pattern obtains, by contrast, with preverbal subjects (the +AGR_SV effect) and with pronominal postverbal subjects. • With [–state] verbs the non-agreeing pronominal form occurring in the nonagreeing pattern is instantiated by the expletive clitic l, whereas [+state] verbs combine with the invariable (vocalic) clitics a/e in thetic marker function and/or the expletive clitic in accordance with dialectal variation. The range of agreement possibilities analysed reflects the interplay of the lexicoaspectual characteristics of the relevant verb classes and the thematic properties of the postverbal S and its (in)definiteness. The latter parameter shapes the distribution of agreement, together with more fine-grained lexico-aspectual and/or thematic distinctions within the different verb classes, as with verbs of emission,11 which almost exclusively exhibit the agreement pattern, even in varieties where non-agreement is the main/only pattern with a postverbal S (cf. also §12.3.3). 11 A similar sensitivity of VS agreement to the lexico-aspectual properties of verbs and the thematic properties of S among [–state] verbs is discussed by Parry (2000; 2013:515) for some Piedmontese dialects in relation to the occurrence of the non-agreeing pattern with verbs denoting a lower degree of dynamicity and agentivity of S, e.g., sleep and work (= ‘be employed’).

356

delia bentley and michela cennamo

Pronominal Ss exhibit a higher incidence of agreement, whereas in the case of a lexical S agreement proves sensitive to (in)definiteness, as shown by the high degree of variability in the occurrence of the two patterns (cf. §12.3.3). The different behaviour of pronominal vs lexical Ss can be better understood if the observed synchronic variation is projected along the diachronic axis, and viewed as instantiating aspects of a change in progress leading to the gradual retrenchment of non-agreement to the advantage of the agreeing pattern, and seemingly affecting core definite Ss in its initial stages (e.g., Correggio, Ponte dell’Olio). This process appears to be more advanced in some varieties (e.g., Ferrara and Modena), and represents a topic which, however, needs much more thorough investigation on much larger bodies of historical data.

12.3.3 Quantitative analysis Excluding the questionnaires which were incomplete, the full Emilian-Romagnol data set was reduced to 881 tokens, produced by 15 speakers, and then to 809 tokens, restricting the sample to the preferred option for each questionnaire entry (although when no preference had been given we included both accepted alternatives as separate tokens in the latter data set). Using this reduced data set, we analysed the findings as follows. First, we calculated the overall incidence of the three effects, or dependent variables, which in percentage terms amounted to 51.6 for –AGR, 27.8 for +AGR, and 20.5 for +AGR_SV. An overall tendency towards the lack of V–S agreement emerges from this result. We then considered how each dependent variable correlated with each of the following predictors: [±pro] (whether S was a personal pronoun), [±def] (whether S was definite), [±state] (whether S was the argument of a verb with a stative component in its lexical meaning), and, finally, lexical aspect (whether S was the argument of an activity, a state, an achievement, or an accomplishment). The most significant finding was the one shown in Figure 12.1, which suggests that [±pro] is a predictor of V–S agreement in this group of dialects, although admittedly we only had a very small number of tokens with a pronominal S (this will be shown in Table 12.2 in due course).12 Importantly, 86.2% of the [+pro] tokens exhibited V–S agreement, while only 6.8% lacked such agreement, in stark contrast with the figures relating to the tokens with a non-pronominal S: only 23% of these exhibited V–S agreement, compared with 55% with no agreement. The percentages of AGR_SV were 6.8 with [+pro] and 21.5 with [–pro]. By comparison with [±pro], the [±def] feature of S turned out to be less significant, notwithstanding the fact that the [+pro] tokens were included within the [+def] class. The lesser significance of [±def] is suggested by the comparison of Figure 12.2 with Figure 12.1. 12 The reader should note that the significance of [±pro] was calculated without distinguishing between verb classes.

thematic & lexico-aspectual constraints 100 90 80 70 60 50 40 30 20 10 0

NoAGR

AGR NonPRO

357

AGR_SV PRO

Fig. 12.1 Incidence of three effects according to [±pro] predictor

60 50 40 30 20 10 0

NoAGR

AGR INDEF

AGR_SV DEF

Fig. 12.2 Incidence of three effects according to [±def] predictor

Whereas 55.3% of the [–def] tokens correlated with –AGR, 25.5% exhibited V–S agreement (+AGR) and 19% S–V agreement (+AGR_SV). As for the [+def] tokens, the percentages were 43.4 for –AGR, 33.8 for +AGR and, lastly, 22.6 for +AGR_SV. V–S agreement correlated more strongly with activities (non-state-based verbs: 35% for +AGR and 26.7% for +AGR_SV) than with states, achievements, and accomplishments (state-based verbs: 25.1% for +AGR and 18.2% for +AGR_SV), as can be seen in Figure 12.3. Conversely, –AGR correlates more strongly with [+state] verbs (56.5%) than with [–state] (38.2%). These findings suggest that the lexico-aspectual properties of the verb do have a bearing on V–S agreement, an issue to which we return below. In Figure 12.4, we illustrate the incidence of the three effects in the four subsets of the data representing the Vendlerian classes introduced in §12.2.

358

delia bentley and michela cennamo 60 50 40 30 20 10 0

NAGR

AGR STATE_B

AGRSV NonSTATE_B

Fig. 12.3 Incidence of three effects according to [±state] predictor 70 60 50 40 30 20 10 0

ST

ACC NoAGR

ACH AGR

ACT

AGR_SV

Fig. 12.4 Incidence of three effects according to lexical aspect

Although the incidence of –AGR exceeded that of each of the other effects with all lexico-aspectual classes, –AGR was more frequently found with states, achievements, and accomplishments than with activities, in accordance with the results illustrated in Figure 12.3.13 In addition, the combined incidence of +AGR and +AGR_SV only exceeded that of –AGR in the case of activities. Within the statebased classes, –AGR was a more prominent effect with states and achievements than with accomplishments, whereas +AGR_SV correlated more prominently 13 The relevant percentages are as follows: (i) (ii) (iii) (iv)

ACT(ivities): 38.2 for –AGR; 35 for +AGR; 26.7 for +AGR_SV. ST(ates): 58.4 for –AGR; 20.7 for +AGR; 20.7 for +AGR_SV. ACC(omplishments): 53.9 for –AGR; 28.2 for +AGR; 17.8 for +AGR_SV. ACH(ievements): 60.6 for –AGR; 21.8 for +AGR; 17.5 for +AGR_SV.

thematic & lexico-aspectual constraints

359

with states than with achievements and accomplishments. The score of +AGR_SV, which was even higher with activities than with states, may depend on the overall resistance of these two classes to verb–subject inversion (Bentley and Cruschina 2018; Bentley 2020a), and will not be discussed any further. Building on these initial findings, we ran a logistic regression analysis using Rbrul (Johnson, available at http://www.danielezrajohnson.com/Rbrul_manual. html). In order to code the dependent variable categorically as ±AGR, in this analysis we left out the complete sets of responses provided by five informants, which included the +AGR_SV effect. In addition, we restricted this statistical analysis to two sections of the questionnaire, the first balanced in terms of the number of entries per each Vendlerian class, and the second focusing on the entries with a pronominal S. A one-level analysis with the two predictors that had previously turned out to be the most significant, [±pro] and [±state], plus QUESTION as a random effect, yielded the results in Tables 12.2 and 12.3, respectively. The log odds and factor weights in Table 12.2, support the hypothesis that [±pro] is a predictor of V–S agreement.1⁴ Indeed, for this factor we obtained a significant p value (p=2.42e-06). Even though the findings in Table 12.3 are much more nuanced, crucially, we also obtained a significant p value for [±state] (p=0.0125).1⁵ Table 12.2 Significance of [±pro] in one-level logistic regression with [±pro] and [±state] Factor Log odds Tokens ± AGR Factor weight +pro 1.444 22 0.909 0.809 –pro –1.444 196 0.347 0.191

Table 12.3 Significance of [±state] in one-level logistic regression with [±pro] and [±state] Factor Log odds Tokens ± AGR Factor weight –state 0.384 77 0.532 0.595 +state –0.384 141 0.333 0.405

1⁴ Both log odds and factor weights measure the strength of the relationship between a factor (an independent variable) and the studied effect (the dependent variable). If the log odds figure is negative, there is a negative correlation between the independent and dependent variables; if it is above 0, the correlation is positive: the higher the value, the higher the correlation. As for factor weights, they measure the strength of the effect size within a 0 to 1.00 range. If the measure is 0 for log odds and 0.50 for factor weights, the correlation is neutral. 1⁵ Since the four variables are not independent of each other, with [–state] equating [activity], and [+pro] being included within [+def], we do not report the results of the logistic regression analyses in which we factored in all the variables. We simply note that a step-up analysis with the four variables [±pro], [±state], [±def], and verb class (activity, state, achievement, accomplishment), plus QUESTION as a random effect, revealed the best model to be with predictors [±pro] (p=1.48e-07) and [±state] (p=0.0125), where the p value for [±state] remains as reported above.

360

delia bentley and michela cennamo

We also note in passing that in the small sample considered in this analysis the independent variables provenance, age, and speaker turned out not to be significant. Although these initial findings should be tested against evidence from a larger data set, we take them to support the hypothesis that both [±pro] and [±state] are predictors of V–S agreement in Emilian-Romagnol. We should mention some additional results, which lend some support to the hypothesis that thematic structure, and ultimately verb semantics, play a role in V–S agreement. To begin with, we investigated accomplishments further, in order to test the relevance of a specific final goal state in the lexico-semantic structure of the verb (§12.2). We obtained higher scores for –AGR with verbs of quantized scalar change than with verbs of gradual completion (57.5% in the former class in contrast to 53.1% in the latter). By contrast, we did not detect a significant difference between verbs which lexicalize a locative goal and other verbs of scalar change (the percentages of –AGR were 55.6 with the former type of verb and 56.7 with the latter). Among the verbs which we considered to be activities, verbs of emission stood out as more closely associated with V–S agreement, their scores being 40% for +AGR and 37.5% for +AGR_SV, compared with 29.2% for +AGR and 26.3% for +AGR_SV with other activities.1⁶ Albeit subtle, these results would seem to suggest that the lexico-semantic properties of the verb play a role in V–S agreement. We comment on the broader implications of our results in the next section.

12.4 Theoretical considerations and conclusions In this section we briefly summarize the main findings of our research and reflect on their significance and theoretical repercussions. Despite a great deal of variation, and a general preference for lack of V–S agreement, two main predictors emerged from our quantitative analysis: [±pro] and [±state]. The significance of the former variable supports Lehmann’s (1982:240) claims that agreement grammaticalizes first in the syntactic domains in which an anaphoric relation with an antecedent is established and that ‘the signs of syntactic agreement, especially in their less grammaticalized forms, often signify the definiteness of the NP agreed with’. The diachronic prediction of this claim is that the VS constructions in which S is a personal pronoun will acquire V–S agreement before those in which it is not, since personal pronouns have an antecedent in text or discourse, constituting the core of a definiteness hierarchy.

1⁶ If compared with states (see note 7) verbs of emission stand out even more clearly, since with states we obtained 20.7% for +AGR and also for +AGR_SV (see Figure 12.4).

thematic & lexico-aspectual constraints

361

As for the latter independent variable, [±state], we offer the following comment on its significance with reference to the notion of subject. The subject of the clause grammaticalizes an aboutness relation and a thematic relation between the predicate and one of its arguments. The aboutness relation can be conceived of as a principle, a feature, or a criterion (Bianchi 1993; Cardinaletti 2004; Rizzi 2005), which, in SV order, is normally satisfied by the argument that ranks the highest on the independently established universal thematic scale in (4) (Van Valin and LaPolla 1997:175).1⁷ In thetic broad focus, the aboutness and thematic relations are decoupled: the aboutness relation is established between the whole predication and a spatio-temporal or situational topic or argument (Erteschik-Shir 1997; Pinto 2007; Klein 2008; Sheehan 2010; Parry 2013; Corr 2016; Bentley 2018; 2020a; 2020b; Bentley and Cruschina 2018). Depending on the analytic assumptions made, this can be conceived of as a silent argument in a subject position or as a topic directly retrieved from discourse (see also the syntactic account in Sluckin, Cruschina, and Martin (2021) for a distinction between a Subject of Predication and a preverbal topic). In the languages where finite agreement is fully grammaticalized (e.g., Italian), the agreement inflexion systematically bears the person and number features of the thematically prominent postverbal argument, thus spelling out thematic subjecthood. The significance of [±state] vis-à-vis V– S agreement in Emilian-Romagnol reveals that, in other languages, the thematic properties of the postverbal argument have a bearing on whether the construction does have a thematic subject, viz. a controller of V–S agreement. We have argued that more fine-grained properties of lexico-semantic structure may also play a relevant role. Finally, we note that the statistical significance of [±pro] suggests that the establishment of an anaphoric relation between the agreement inflexion and an antecedent outranks the thematic constraints on V–S agreement, and hence thematic subjecthood in Emilian-Romagnol, a result which fully corroborates Bentley’s (2018) findings on other northern Italo-Romance dialects. By way of conclusion to this section, we note that our findings require an account in which the lexicon is given recognition as an independent module of grammar, while at the same time, constructional properties like theticity are allowed to interact with general principles such as those operating in subject agreement.

1⁷ While the scale in (4) is claimed to be universal, the alignment of the privileged syntactic relation with the high end of the scale is only a characteristic of accusative alignment. In absolutive alignment, the privileged syntactic relation is associated with the low end of the same scale.

13 Conditioned epenthesis in Romance Mark Aronoff and Lori Repetti

13.1 Introduction Most modern discussion of epenthesis has concentrated on types that are phonetically or phonologically motivated. We show that the factors governing individual phenomena may go beyond phonetics and phonology to morphology and morphosyntax. We first discuss cases from Romance languages where a language has more than one epenthetic segment. Here, although the motivation for epenthesis may be phonological, the choice of one segment over another is determined by further factors: it can be morphologically conditioned or influenced by morphosyntax. Further up the scale, the insertion of a meaningless syllable—including stem extenders in Formentera Catalan, Italian , and Spanish antesuffixal interfixes—may have little if any phonological motivation, but is influenced by morphology. We conclude that the notion of epenthesis should be broadened from the purely phonological to include morphological and morphosyntactic conditions. Many of the phenomena that we discuss have been accounted for in terms of allomorphy, with allomorphs that are lexically listed. However, an analysis based on allomorphy is always descriptive and never explanatory. The purely allomorphic approach gives up on the possibility of finding more widely applicable constraints on the form and presence of the epenthetic segment, and hence misses a broad generalization: when more than one epenthetic segment is possible, the choice is predictable within specific morphological or morphosyntactic contexts.1 Epenthesis is ‘the interposition of a letter or syllable in the midst of a word’ (Smith 1656, the earliest citation in OED). The term and concept are ancient, dating to Classical Greek rhetoric. Although most modern treatments have concentrated on cases of epenthesis that are phonologically motivated, we show in this chapter

1 More than anyone else, Martin Maiden has revealed the power of the morphological explanation of complex phenomena throughout the Romance languages, most notably in his monumental volume, The Romance Verb (Maiden 2018a). We are delighted to honour him with this short contribution, whose focus is a morphological explanation of a phenomenon that has usually been viewed through a phonological lens.

Mark Aronoff and Lori Repetti, Conditioned epenthesis in Romance. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Mark Aronoff and Lori Repetti (2022). DOI: 10.1093/oso/9780198870807.003.0014

conditioned epenthesis in romance

363

that the factors governing it range from the purely articulatory to those that lie beyond phonology: morphology, morphosyntax, and as far as the lexicon. Here we exclude lexical conditions, for the sake of brevity. The insertion of an ‘unetymological’ segment is rampant throughout the world’s languages and goes under a wondrous plethora of ancient names, from epenthesis (AGk. ἐπένθεσις < ἐπί ‘in addition’ + ἐν ‘in’ + θέσις ‘placing’) to anaptyxis (also , , AGk. ανάπτυξις < ανα ‘up’ + πτύξις ‘folding’ ‘unfolding’, reserved for vowels) to svarabhakti (Sanskrit svara ‘vowel’ bhakti ‘separation’ ‘vowel separation’) and its consonantal counterpart vyanjanabhakti. In the phonetic tradition, there are less ancient terms for the phenomenon: transitional segment, especially transitional vowel, and most recently, two new terms, excrescent vowel (Levin 1987) and intrusion vowel (Hall 2006). In this contribution, we make the case for a unified approach to this cornucopia, while concentrating our attention on a limited set of data from Romance languages that lie on the morphological part of the spectrum. The conditions under which all these insertions occur run the gamut from purely phonetic to morphosyntactic. What unifies the phenomena is the fact that something is inserted between segments. What makes for the diversity are the sorts of conditions under which the insertion takes place. Our innovative contribution is the observation that both the type of condition and the nature of what is inserted are tied to what we used to call the linguistic level of the insertion.

13.1.1 Intrusive vowels One of the more remarkable types of insertion is found at the articulatory level and is attributed to the articulatory intrusion between consonants. Hence the term intrusive, as with the intrusive stops in English words like sense ([sεns] or [sεnts]) and temse ([tεms] or [tεmps]) ‘sieve’. Martin Maiden (1995:242) notes the same insertion of [t] in sequences of sonorant + [s] in various Romance varieties spoken in central and southern Italy, e.g., Umbrian [pέntso] < *pεnso ‘I think’, [fa´ltso] < *falsu ‘false’. The fact that these intrusions are specific to certain varieties shows that they are not automatic, but part of the linguistic system. Hall (2006) shows in some detail that intrusive vowels (as in the pronunciation of arm as [a´rәm] in some varieties of English) often do not rise to the status of segments: they do not count as syllabic nuclei in the phonological timing calculus. ‘Ordinary epenthetic vowels, however, are syllable nuclei’ (Hall 2006:388). Bertil Malmberg (1950) was the first to document intrusive vowels (in Argentinian Spanish), and they have since been observed in other varieties of Spanish and in French (e.g., Colantoni and Steele 2004). Miatto (2020) shows that certain word-final inserted vowels in Italian also qualify as intrusive in this sense, not having segmental status in the phonology. Hall emphasizes that the main difference between intrusive and canonical epenthetic vowels is that the latter count as full segments. While both types can

364

mark aronoff and lori repetti

have a single value throughout the language, so-called copy vowels, where the value of the vowel is taken from that of a nearby full vowel, may be more common in cases of intrusion. This is in line with the instability of excrescent vowels observed by Levin (1987). They are often optional or correlated with the speed of speech (less likely in faster speech). Researchers find that native speakers are also often unaware of the presence of intrusive vowels (Colantoni and Steele 2004).

13.1.2 Phonological epenthesis Phonological epenthesis, by contrast with intrusion, is the insertion of meaningless phonological material whose appearance is motivated by phonology (to repair an illegal structure), and whose quality is usually unmarked in the language (Kitto and de Lacy 1999). Consider the example in (1), where phonologically impermissible word-initial s-stop clusters are repaired by epenthesizing an initial vowel [e], which is unmarked in the language.2 (Throughout this chapter, the epenthetic segment is underlined.) (1)

Spanish: [eestóp]

As with intrusion, the quality of the epenthetic vowel may vary depending on a number of phonological and prosodic factors. Factors include its position, e.g., English loanwords in Bengali (Broselow 2015): school > [iiskul], glass > [geela´ʃ]; the surrounding phonetic environment, e.g., Afrikaans loanwords in Sotho (Rose ɷléke] ‘tin can’, /truwn/> [tɩɩróni] ‘throne’; and the and Demuth 2006):/blɪk/> [bɷ neighbouring vowels in ‘copy epenthesis’, e.g., English borrowings in American varieties of Italian (Repetti 2012): washtub [veʃʃeetúbbu], cocktail [kokkootέlla], popcorn [pappaakɔ´rno], among others. In all of these cases, the choice of the inserted segment is determined phonologically. In a number of cases, however, the quality of the inserted segment does not seem to be determined phonologically (Moradi 2017). Put another way, the presence of a vowel is motivated phonologically, but the exact quality of the epenthesized vowel is determined by other factors. For example, in the Romance variety of San Marino (Michelotti 2008), we find two possible epenthetic vowels to satisfy restrictions on word-final clusters: [ɪ]/[ɐ]. (2)

San Marino a. /ojm/ > [ójmɪɪ] ‘elm’ b. /dɔrm/ > [dɔ´:rmɐɐ] ‘sleep.prs.ind.3sg/pl’

2 Archangeli (1988) and others argue that /e/ is the maximally underspecified vowel in Spanish, and others show that /e/ is the most frequent phoneme in Spanish (Guirao and García Jurado 1990). Thanks to a reviewer for pointing this out.

conditioned epenthesis in romance

365

In San Marino, [ɪ] is the default epenthetic vowel (2a),3 but [ɐ] occurs in wordfinal position with certain verbs ending in unacceptable clusters (2b). The choice between [ɪ] and [ɐ] is made on the basis of morphosyntactic considerations: [ɪ] is the usual epenthetic vowel, except in certain morphosyntactic contexts when epenthetic [ɐ] in employed instead (see §13.2.2 for more details). We see from this example that the factors conditioning epenthesis are richer than previously noted and go beyond phonological factors to morphological and morphosyntactic factors. In what follows, we present a number of examples from Romance languages where the motivation for epenthesis is phonological, but the choice of segment is determined by these factors. This chapter is organized as follows. In §13.2 we introduce five case studies of morphologically conditioned epenthesis from Romance languages. We provide a review of previous accounts of the data in §13.3, including allomorphic accounts (§13.3.1) and epenthesis accounts (§13.3.2). We present our analysis in §13.3.3, and then discuss other types of insertion (§13.4). We summarize our proposal and conclude in §13.5.

13.2 Case studies in Romance languages In this section we present five case studies from Romance languages in which two phonologically distinct epenthetic segments may be used to satisfy phonological constraints. Crucially, the choice between the two segments is influenced by morphology or morphosyntax rather than phonology.

13.2.1 Brazilian Portuguese Brazilian Portuguese chooses between two epenthetic segments in resolving vowel hiatus: [j] is the default (3a), except when hiatus results from the addition of an affix, where [z] is used instead at the morphological juncture (3b) (Bachrach and Wagner 2007; Garcia 2017). [i] is the default epenthetic vowel in Brazilian Portuguese: it is used to break up impermissible morpheme-internal clusters (psicologia ‘psychology’ can be pronounced [piisikoloʒı´ɐ]), and some cases of consonant-final loanwords are adapted with insertion of final [i]: Eng. blog > [blɔ´gi] (Xavier 2013 reported in Artes 2016:114). The default epenthetic consonant [j] is the non-vocalic counterpart of [i]. At some level, we can think of them as the same element. (3)

Brazilian Portuguese a. Correa [koréjja] ‘(name)’ b. /sofa´ + iɲu/ > [sofazz´ĩɲu] ‘sofa (dim)’ /kafé + al/ > [kafezza´l] ‘coffee grove’⁴

3 The quality of the vowel is either [ɪ] or [i] which occurs in free variation (Michelotti 2008:332). 2⁵⁶ The /al/ suffixes following the name of a plant means the place where the plant is cultivated.

366

mark aronoff and lori repetti

Epenthetic [z] appears instead of the expected [j] in certain hiatus contexts between morphemes, e.g., BrPt./kafé + ĩɲu/ ‘coffee’ + dim > [kafezz´ĩɲu], **[kafe´ĩɲu] ‘little coffee (espresso)’, and only in hiatus contexts, e.g., BrPt. /zebr-a + iɲa/‘zebra’ + dim > [zebr´ĩɲa], [zebrazzĩɲa], **[zebrzz´ĩɲa] ‘little zebra’. Bachrach and Wagner (2007:7) show that the use of forms without epenthesis ([zebr´ĩɲa]) or with epenthesis ([zebrazz´ĩɲa]) is not only influenced by phonology, but also reflects the syntactic attachment site of the suffix. They provide details about the morphology, syntax, and semantics of the diminutive and other suffixes. Thinking more within the tradition of generative phonology, we may portray the difference between the members of pairs like [zebr´ĩɲa] and [zebrazz´ĩɲa] as differences in lexical levels (I and II in Lexical Phonology) or boundaries (+ and #) in SPE terms. Going further back, they are analogous to Sapir and Bloomfield’s two English -er suffixes, where one attaches as if to a word ([sı´ŋәɹ] vs [lɔ´ŋɡәɹ]) (Sapir [1925] 1949:43).⁵ The most important point for the moment is that the epenthetic consonant between two morphemes in hiatus, [z], is not the default purely phonologically conditioned segment in the same hiatus context ([j]). The addition of the morphological condition changes the environment. We will return to the Brazilian Portuguese case at the end of §13.3, where we discuss additional morphological complexities.

13.2.2 San Marino The Romance language spoken in the independent republic of San Marino in northern Italy chooses between two epenthetic vowels in word-final position, [ɪ] and [ɐ] (see note 3), as illustrated in (2). Both epenthetic vowels are used with unsyllabifiable consonant clusters word-initially that result from syncope of pretonic front vowels (Michelotti 2008:126): [i] is inserted within a consonant–sonorant– consonant cluster /krsu:/ > [kriisú:] ‘grown’; [ɐ] is used in initial position before a sonorant–consonant sequence /lgε:/ > [ɐɐlgέ:] ‘to tie’. Word-finally, [ɪ] is inserted with nouns (/lε:dr/ > [lέ:drɪɪ] ‘thief ’), adjectives (/sɐlba:dg/ > [sɐlba´:dgɪɪ] ‘wild’), and adverbs (/sεim ̯ pr/ > [sέim ̯ prɪɪ] ‘always’). The word-final situation with verbs is more complex and depends on the person, number, and conjugation class of the verb form. Verbs normally have no suffix marking the first-/second-person singular of the present indicative, regardless of their conjugation class (4).⁶ (4)

a. 1st conjugation class: [ba:l] ‘dance.prs.ind.1/2sg’ b. 4th conjugation class: [fnɪs] ‘finish.prs.ind.1/2sg’

⁵ Sapir attributes the analysis to a personal communication from Bloomfield. The distinction does not hold in many regional varieties. ⁶ Michelotti (2008:331) categorizes verbs into four conjugation classes. For the phenomena discussed here, the second, third, and fourth conjugation classes pattern the same way, and we will use data from the fourth conjugation class to illustrate the patterns of the three classes. In addition, we are not including subject clitic pronouns which are obligatory in some contexts.

conditioned epenthesis in romance

367

However, if the verb ends in an illegal cluster, epenthetic [ɪ] is inserted at the end of the verb (5). (5)

a. 1st conjugation class: [ú:rlɪɪ] ‘scream.prs.ind.1/2sg’ b. 4th conjugation class: [ı´:rvɪɪ] ‘open.prs.ind.1/2sg’

For third-person singular and plural present indicative verbs only, the first conjugation class differs from all others. The theme vowel [ɐ] marks the third-person singular and plural of first conjugation verbs, while the third-person singular and plural of the other conjugation class verbs have a different theme vowel and no suffix (6). (6)

a. 1st conjugation class:

[ba´:lɐ] ‘dance.prs.ind.3sg/pl’ [ú:rlɐ] ‘scream.prs.ind.3sg/pl’ b. 4th conjugation class: [fnɪs] ‘finish.prs.ind.3sg/pl’

However, if the verb ends in an illegal cluster, epenthetic [ɐ] rather than the expected [ɪ] is inserted at the end of the verb in the other conjugation classes, as in (7). (7)

4th conjugation class: [ı´:rvɐɐ] (**[ı´:rvɪɪ]) ‘open.prs.ind.3sg/pl’

A summary of the relevant verb forms is provided in Table 13.1. The ɐ-bearing cells are shaded. Table 13.1 San Marino verbs

1sg/2sg 3sg/3pl

1st conj. class

/bal/ [ba:l]

/url/ [ú:rlɪɪ]

4th conj. class 1st conj. class

/fnɪs/ [fnɪs] /bal + ɐ/ [bá:lɐ]

/irv/ [í:rvɪɪ]

4th conj. class

/fnɪs/ [fnɪs]

/irv/ [í:rvɐ]

/url + ɐ/ [ú:rlɐ]

We analyse the [ɪ] of both [ú:rlɪɪ] ‘scream.prs.ind.1/2sg’ (first conjugation class) and [ı´:rvɪɪ] ‘open.prs.ind.1/2sg’ (fourth conjugation class) as epenthetic. Furthermore, the [ɐ] of [ú:rlɐ] ‘scream.prs.ind.3sg/pl’ (first conjugation class) is an inflexional morpheme (cf. [ba´:lɐ] ‘dance.prs.ind.3sg/pl’), while the [ɐ] of [ı´:rvɐɐ] ‘open.prs.ind.3sg/pl’ (fourth conjugation class) is an epenthetic vowel. In San Marino, [ɪ] is the default epenthetic vowel, but [ɐ] is used to satisfy phonological constraints on word-final position in specific morphosyntactic contexts, namely with second-, third-, and fourth-conjugation third-person singular and plural verbs ending in an unacceptable cluster. The data in (6) clearly show that [ɐ] does not mark third-person singular and plural in general, but only for firstconjugation verbs. However, with non-first-conjugation verbs, if an epenthetic

368

mark aronoff and lori repetti

vowel is needed, [ɐ] is used. Its presence is phonologically motivated, but its quality is not. The choice between canonical [ɪ] and non-canonical [ɐ] is made on the basis of morphosyntactic considerations: [ɪ] is the usual epenthetic vowel, except with third-person singular and plural verbs, which instead employ epenthetic [ɐ] in final position. The next examples similarly illustrate cases in which the choice between two epenthetic vowels is made based on morphosyntactic considerations.

13.2.3 Paduan Paduan, a Romance variety spoken in the province of Padua in the Veneto region of northern Italy, also chooses between two epenthetic segments to satisfy syllable constraints: [e] is the default vowel used in most cases (8a–b), except at the right edge of a phrase consisting of a verb + pronoun, when [o] is used (8c–d) (Cardinaletti and Repetti 2007; 2008). The choice of epenthetic vowel ([e] or [o]) is determined by its position within a morphosyntactic context. (8) a. b. c. d.

Paduan /l maɲa/ > [eel ma´ɲa] ‘he eats’ /t maɲi/ > [tee ma´ɲi] ‘you eat’ /maɲa l/ > [ma´ɲe loo] ‘eats he’ > ‘does he eat?’ /maɲi t/ > [ma´ɲi too] ‘eat you’ > ‘do you eat?’

In Paduan, verb+clitic and clitic+verb structures form a Phonological Phrase (see Selkirk 1995 and Peperkamp 1997 for the prosodization of clitics) and are subject to certain constraints (Cardinaletti and Repetti 2008). In (8a–b) the default epenthetic vowel [e] is used to syllabify the proclitic. Its position before the /l/ in (8a) is determined by an alignment constraint that accounts for the ‘peripherality of epenthesis’ whereby a host and its clitic must be adjacent (Bonet and Lloret 2005). However, in (8b) its position after the /t/ is determined instead by (more general) constraints on coda consonants: /t/ is never an acceptable syllable coda in Paduan: **[eet ma´ɲi]. Crucially, epenthetic [e] is used in both contexts. In (8c–d), however, the special epenthetic vowel [o] is used at the end of a verb + enclitic pronoun unit.⁷ The default epenthetic vowel [e] is avoided in these contexts (**[ma´ɲe lee], **[ma´ɲi tee]), and a special morphosynctactically restricted epenthetic vowel [o] is used. Its presence satisfies phonological constraints, but its quality is not the usual one. Importantly, we are not claiming that the final [o] of [ma´ɲe loo] and [ma´ɲi too] is a morpheme; we are arguing that it is an epenthetic vowel selected for a special morphologically circumscribed position. As with the data from San Marino, a special epenthetic vowel is used in a restricted morphosyntactic position. ⁷ Paduan does not allow final /t/ or /l/, only final nasals (Zamboni 1981:34).

conditioned epenthesis in romance

369

13.2.4 Italian Italian uses two epenthetic segments to repair consonant cluster violations, and the choice between them is determined by morphosyntactic considerations: [i] is the default vowel used in most cases (9a), including with the masculine singular definite article (9b). However, at the right edge of the masculine singular definite article, epenthetic [o] is used instead of [i] (9c) (Cardinaletti and Repetti 2008; Repetti 2012; 2021). (9)

Italian a. spoken varieties: /psikɔ´logo/ > [piisikɔlogo] ‘psychologist’ historical change: alisna > [lésiina] ‘awl’ fixed spoken phrases: /per skritto/ > [per iiskrı´tto] ‘written’ (Di Vita 1931) American varieties of Italian: box [bókiisa] b. definite article: /l kane/ > [iil ka´ne] ‘the dog’ c. definite article: /l spεkkjo/ > [loo spέkkjo] ‘the mirror’

This analysis of the Italian masculine singular definite article rests on the assumption that the input form of the definite article is /l/, and the distribution of the three surface forms—[l], [il], [lo]—is predictable (for more details, see Repetti 2021). (10)

Italian masculine singular definite article a. [l] before a vowel: l’amico ‘the friend’ b. [il] before a single consonant (except those in (c)) and certain clusters (Cl, Cr, Cj, Cw): il bambino ‘the child’ c. [lo] before /t͡s/, /d͡z/, /ʃ/, /ɲ/, /λ/ and sC clusters: lo specchio ‘the mirror’

If we posit an input form /l/, we can easily account for the form in (10a) since no changes to the input form are necessary: the /l/ can be syllabified as the onset of the syllable: /l amiko/ > [la.mı´.ko] ‘the friend’. An epenthetic vowel is necessary in (10b) since an onset cluster cannot begin with /l/: /l bambino/ > [iil.bam.bı´.no], and the default epenthetic vowel is used. The position of the epenthetic vowel is accounted for by means of the alignment constraint discussed above, resulting in peripheral epenthesis. In (10c), an epenthetic vowel is needed, and it is placed after the /l/, in violation of the alignment constraint, because of more highly ranked constraints on syllable structure:⁸ /l spεkkjo / > [loos.pέk.kjo] ‘the mirror’. However, the default vowel [i] is not used. Instead, we find [o]. As in San Marino and Paduan, the motivation for vowel epenthesis is phonological, but the quality of the ⁸ The same analysis holds for the ‘inherently long’ consonants in (10c): /l t͡sμ io/> [looot. t͡sí.o] ‘the uncle’. See Repetti (2021) for details.

370

mark aronoff and lori repetti

epenthetic segment is not the usual one.⁹ Of course, the choice of the vowel [o] is no accident. The default masculine singular ending is this same vowel. The close parallel with San Marino is telling. In that language, the default third-person singular present verbal ending [ɐ] overrides the default epenthetic vowel [i] under morphologically circumscribed conditions. Here, the vowel of the default masculine singular nominal ending -o overrides the default epenthetic vowel [i] (under different morphologically circumscribed conditions). In both cases, a morphologically specific epenthetic vowel that is phonologically identical to a more general morpheme overrides a more general phonological default.

13.2.5 Catalan varieties Alguerès, the variety of Catalan spoken in Sardinia, uses two epenthetic vowels to satisfy syllable constraints: [i] is the default value used between words (11a), and [u] is used before the masculine plural suffix /s/ if an epenthetic vowel is needed (11b) (Loporcaro 1997c). Note that if an epenthetic vowel is not needed, the [u] does not appear before the plural marker: /mɔlt + s/ > [mɔlts] ‘dead.mpl’. (11)

Alguerès a. /amik meu/ lit. ‘friend my’ > [amı´k ii méu] ‘my friend’ b. /fresk + s/ ‘cool.m + pl’ > [fréskuus] ‘cool (mpl)’

Another variety of Catalan, Pallarès, spoken in Pallars in Catalonia, similarly chooses between two epenthetic vowels: [e] and [o] (Artes 2016). The former is the default vowel used to satisfy syllable constraints in most cases (12a). Its quality in central Catalan is [ә], but in Pallarès it is usually [e] (Artes 2016:148f.) (see Jiménez 2008 and Lloret and Jiménez 2008 for more on epenthetic vowel quality in Catalan varieties). The special epenthetic vowel [o] ([u] in central Catalan) is used before the masculine plural suffix /s/ (namely, with sibilant-final nouns and adjectives) (12b). As in Alguerès, Pallarès epenthetic [o] is not used with the masculine plural suffix if an epenthetic vowel is not needed: /gat + s/ > [gats] (**[gatoos]) ‘cats’. (12)

Pallarès Catalan a. [ee]Spielberg ‘Spielberg (film director)’ b. /gos + s/ > [gósoos] ‘dogs’

In the case studies above, we have seen that the choice between two epenthetic segments is made based on morphosyntactic considerations (morpheme boundaries §13.2.1, verb classes §13.2.2, phrases involving clitic pronouns §13.2.3) and ⁹ Note that there is a difference between the Italian (9)–(10) and Paduan (8) patterns. In Italian, the right edge of the definite article requires the selection of the special epenthetic vowel: /l/ > [loo] (10c), even though it is not at the right edge of the phrase, while in Paduan, the right edge of the subject pronoun only triggers the selection of the special epenthetic vowel when it is at the end of the phrase: /l/> [loo] (8c) and /t/> [too] (8d), but not phrase-internally: /l/> [eeel] (8a) and /t/> [teee] (8b).

conditioned epenthesis in romance

371

on specific morphemes (definite article §13.2.4, plural suffix §13.2.5). In these cases, the default epenthetic vowel is not used, but instead a special segment is selected. In the next section we discuss possible analyses of these phenomena.

13.3 Accounts of the phenomena Various analyses of the phenomena described above have been proposed in the literature. We consider two approaches: allomorphy (§13.3.1) and epenthesis (§13.3.2), and we conclude that the inserted segment is indeed epenthetic, and its quality can be morphologically conditioned (§13.3.3).1⁰

13.3.1 Allomorphic solutions Many of the phenomena outlined in §13.2 have been addressed in the literature as cases of allomorphy. For example, the San Marino first- and second-person singular verbs end either in Ø or/ ɪ/ (Table 13.1), and Michelotti (2008:332) refers to /ɪ/ as ‘a phonologically conditioned allomorph of the desinence occurring in complementary distribution with -Ø’. Similarly, third-person singular and plural verbs of the fourth conjugation class end either in Ø or /ɐ/, and Michelotti (2008:335) refers to /ɐ/ as ‘a complementary allomorph of the desinence -Ø’.11 The different Paduan proclitic vs enclitic forms (8) have also been attributed to allomorphy: some argue that there is a proclitic paradigm and an enclitic paradigm of pronouns (Munaro 1999) or an ‘interrogative conjugation’ (Zamboni 1974), while others claim the proclitic is a pronoun, and the enclitic is an inflexional suffix (Zamboni 1974; Benincà and Vanelli 1982; Benincà 1983; Poletto 2000). The Italian case (10) has also been described in terms of allomorphy. Many scholars have accounted for the distribution of the three forms of the masculine singular definite article by positing various listed forms: /l, il, lo/ (Dressler 1985; McCrary 2004; McCrary Kambourakis 2007; Garrapa 2009; 2011), /il, lo/ (Davis 1990; Marotta 1993; Morelli 1999; Kra¨mer 2009), /l, il/ (Vanelli 1992; Mascaró 1996; Tranel and Del Gobbo 2002). The appropriate allomorph is selected based on the phonological 1⁰ Many authors propose historical accounts of the synchronic facts, including Michelotti (2008) for San Marino (§13.2.2), Vanelli (1984; 1987) for Paduan (§13.2.3), Gro¨ber (1877), Ambrosini (1978), Vanelli (1992), Renzi (1993) for Italian (§13.2.4), Loporcaro (1997c) for Alguerès (§13.2.5), and Artes (2016) for Pallarès (§13.2.5). A discussion of these approaches is beyond the scope of this chapter. 11 Another possible account of the San Marino data is that this is an example of heteroclisis whereby the inflexional paradigm of a particular lexeme involves more than one inflexional class (Stump 2006). In the case of San Marino, this would mean that some second-, third-, and fourth-conjugation verbs adopt the inflexional suffix of first-conjugation verbs for the third-person singular and plural of the present indicative. However, this approach does not appear to apply because the unexpected final vowel only occurs in a particular phonological context, namely with an illicit final cluster. Thanks to a reviewer for pointing out this option to us.

372

mark aronoff and lori repetti

context. As for the Catalan data (§13.2.5), Bonet, Lloret, and Mascaró (2007) have proposed allomorphs of the masculine noun suffixes (‘gender allomorphy’). The listed and ranked lexical entries for masculine gender include /Ø > u/.12 The first allomorph is the one used, unless the plural suffix /s/ cannot be adjacent to the noun stem for phonological reasons, in which case the second is selected. Phonologically conditioned allomorphy is the most common account of the phenomena described in §13.2 and can be invoked to describe each case. However, this approach suffers from a serious problem: it fails to take into account the fact that the restricted ‘allomorph’ falls in with a larger morphological pattern in the language. The fact that [ɐ] is used in the restricted context in San Marino, [o] in Paduan and Italian, and [u]/[o] in Catalan is predictable, but simply listing all the allomorphs from which to choose results in an unnecessarily complex grammar. We expand on this in §13.3.2. We are not arguing that the listing of allomorphs does not ever exist. Allomorphs whose forms are unpredictable from other aspects of the language must clearly be listed in the lexicon, as with the English indefinite article a/an, definite article [ðә]/[ði], and suppletive forms of the verb be. But when the restricted ‘allomorph’ falls in with a larger pattern in the language, lexical listing prevents us from seeing that pattern. In general methodological terms, lexically listed allomorphy is a brute-force tool for describing the language. Listing allomorphs can never have any explanatory value. Positing a restricted allomorph should always be the solution of last resort, to be called upon only when all else fails.

13.3.2 Epenthesis The phenomena outlined in §13.2 have also been accounted for as instances of epenthesis. In the case of Brazilian Portuguese (§13.2.1), Bachrach and Wagner (2007:8), Garcia (2017:47), and others argue for a [z]-insertion process when the addition of an affix results in hiatus,13 but they do not address its quality. In other cases of hiatus resolution, [j] is used. The Catalan facts (§13.2.5) have also been accounted for as cases of epenthesis. Alguerès uses two epenthetic segments which Loporcaro (1997c) refers to as ‘iepenthesis’ vs ‘u-epenthesis’. The quality of epenthetic [i] is only recently stabilized from a more variable mid-central vowel quality reported less than a century ago (Loporcaro 1997c:215–217). On the other hand, he accounts for the quality of [u] as a case of ‘reanalysis leading to rule inversion’, i.e., historical [u]-deletion was blocked in contexts where deletion would have resulted in a disallowed cluster. 12 Bonet, Lloret, and Mascaró (2007) espouse a framework in which allomorphs are not simply listed, but ranked. The ranking does not follow simple elsewhere principles but must be specified. 13 Bachrach and Wagner (2007:7) further argue that the forms with or without the /z/ differ in the point of attachment.

conditioned epenthesis in romance

373

At that point it was reanalysed as [u]-insertion (Loporcaro 1997c:217). Mascaró (1985a; 1985b), Lloret and Viaplana (1992), Wheeler (2005), Bonet et al. (2007), Artes (2016), and others have also proposed that Catalan [u]/[o] is epenthetic but they argue that its quality is morphologically conditioned. In another approach, Zimmermann (2019) refers to Catalan [u] as a ‘ghost’ whose appearance is lexically determined. Various proposals regarding the role of morphology in determining the quality of this special epenthetic segment have been offered. Artes’s (2016) detailed proposal is that the [o] in (12b) is an inflexional morpheme whose floating features of the input are parsed only when phonotactic requirements make that necessary (p. 269). The Italian definite article [lo] (§13.2.4) has been analysed as the result of epenthesis by Vanelli (1992) and Tranel and Del Gobbo (2002) who conclude that the quality of the vowel is determined morphologically, and by Cardinaletti and Repetti (2007; 2008), Repetti (2012), and Repetti (2021) who come to the same conclusion for both the Paduan data (§13.2.3) and the Italian definite article (§13.2.4). The latter authors invoke the privileged position at the end of a word or phrase where inflexional information is located to account for the Paduan and Italian facts. They start with the premise that a vowel representing a morphologically marked category such as pl is avoided as an epenthetic vowel in final position. The vowels /e/ and /i/ mark plural number in Paduan and Italian nominals and so are avoided as epenthetic vowels in final position, where inflexional markers are usually found. For the Italian definite article, they posit the underlying form of the article as /l/, and the surface forms ([l], [iil], and [loo]) derive from /l/ by means of independently attested processes. The fact that [o] is used at the end of the article (rather than usual epenthetic vowel [i]) is justified on the grounds that [o] is the morphologically neutral final vowel in Italian nominals (Ferrari 2005), and non-inflexional [o] is found at the end of other morphemes in the nominal domain: indefinite and negative pronouns and adjectives, as well as adverbs (uno ‘one’, qualcuno ‘someone’, altro ‘other’, nessuno ‘no one’, ciascuno ‘each one’, tutto ‘all’, tanto ‘so much’, poco ‘little’, molto ‘much’) and the predicate clitic pronoun lo, as in Maria è simpatica, e anche Giovanna lo è ‘Maria is nice, and Giovanna is [lit. “is it”] too’ (Cardinaletti and Repetti 2007). Similarly, in Paduan a morphologically marked vowel in final position ([e] which marks feminine plural nouns) is avoided, and instead a different vowel is used: [o]. In the next section we build on these analyses.

13.3.3 Interim conclusion: morphologically conditioned epenthetic segment quality Our interim conclusion is that morphology and morphosyntax play a role in selecting the epenthetic segment used in a particular position; however, that segment is an epenthetic segment and not a morph.

374

mark aronoff and lori repetti

For San Marino, [ɪ] is the default epenthetic segment, but [ɐ] is used instead in a particular context: at the end of the third-person singular and plural verb because, we contend, [ɐ] is the most frequent vowel found in that context. Michelotti (2008:335) argues for San Marino that the [ɐ] in [i:rvɐɐ] ‘open.prs.ind.3sg/pl’ is an allomorph of Ø (as in [fnɪs] ‘finish.prs.ind.3sg/pl’) that is used after a final unsyllabifiable consonant cluster. He justifies the use of this particular segment by saying that it is used by analogy with the third-person singular/plural suffix /ɐ/ of the first conjugation class. However, this approach misses a more general point: the restricted ‘allomorph’—[ɐ]—happens to be the most frequent vowel used in final position with all third-person singular/plural verbs in all conjugation classes and tense/aspect/mood specifications: [ba´:lɐ] ‘dance.prs.ind.3sg/pl’ (first conjugation class), [durmı´:vɐ] ‘sleep.ipfv.ind.3sg/pl (all conjugation classes), [mɔ´:rɐ] ‘die.prs.sbjv.3sg/pl’ (all conjugation classes), [sɐrı´:ɐ] ‘be.cond.3sg/pl’ (all conjugation classes): [ɐ] is not the final segment in the third-person singular/plural imperfect subjunctive, perfect indicative, and future forms. With Paduan enclitics and the Italian definite article, [o] is used instead of the default epenthetic vowel ([e] and [i], respectively), because [o] is the morphologically neutral vowel in final position outside the verbal domain, as Ferrari (2005) illustrates for Italian, and Zamboni (1988) for northern Italian varieties. Zamboni (1988) reports a pattern of restitution of historically deleted word-final vowels in north-eastern Italian dialects: the ‘restored’ vowel is [o], the ‘morphological free and neutral vowel’ (Zamboni 1988:254), and it is found not only at the end of masculine nouns and adjectives, but also feminine nouns, third-person singular verbs, adverbs, and particles.1⁴ Two clarifications are in order. First, we repeat that we are not claiming that the final [o] of the Paduan enclitics and the Italian definite article is a morpheme; we are arguing that it is an epenthetic vowel selected for a special position. As with the data from San Marino, a special epenthetic vowel is reserved for phrase-final position, a morphosyntactically salient position. Furthermore, we are not claiming that all cases of final [o] are epenthetic: in most cases, the final [o] of a noun or adjective represents a morpheme and is not epenthetic. We analyse the Catalan facts in a similar way: the [o]/[u] found in some cases with the plural suffix is the most neutral vowel available for stem-final position in nouns, while the default epenthetic segment [e]/[ә] is a marked vowel in nominals. In Pallarès, [e] is the second most common vowel marking feminine nouns (after [a]) and is used before plural [s] with feminine nouns: cas[a]/cas[e]s ‘house/houses’ (Artes 2016:118). In other varieties of Catalan, [ә] marks feminine nominals. To sum up our proposal: if an epenthetic segment is needed for phonological reasons in a morphologically salient position, the usual epenthetic segment might not be used, and a special vowel can be employed instead. In San Marino, the special epenthetic vowel used in final position with third-person singular/plural verbs 1⁴ See Tekavčić (1977:459–467) for morphosyntactic conditioning of [o] restoration after apocope.

conditioned epenthesis in romance

375

is [ɐ] because that is the most frequent vowel in that position, and in Paduan, Italian, and Catalan, the special epenthetic vowel used in final position with nominals is [o]/[u] because [o]/[u] does not represent a morphologically marked category in final position in nominals. In each of these cases, we agree with Artes (2016:266f.) that ‘the use of regular epenthesis is discarded in word-final position (the location for inflection) to avoid mismatches between phonological and morphological structure’. We propose that the quality of the restricted epenthetic segment is the same as the morphologically neutral morph or the most frequent one in that particular context, but the inserted segment is not a morph. The one case that we have not discussed is Brazilian Portuguese /z/. Here we have no parallels to fall back on: /z/ does not seem to enjoy any special morphological status outside hiatus contexts (though see the discussion of Spanish antesuffixes in §13.4.3 for a possible etymology). According to Bachrach and Wagner (2007:8), /z/ is epenthesized ‘whenever adding an affix creates a hiatus’ and is thus morphologically conditioned. They note that an allomorphic analysis ‘would require postulating two allomorphs for all affixes’, showing that /z/ is indeed epenthetic. The complexity of the Brazilian Portuguese phenomenon lies elsewhere. First, if a noun lacks a theme vowel, the [z] is inserted even though it does not fall between vowels and is therefore not in a true phonological hiatus context: BrPt. flor ‘flower’, let’ florzinha let . Bachrach and Wagner posit an abstract theme vowel here, zinha ‘flowerlet whose sole justification is to save the hiatus analysis. Notably, though, only sibilant and rhotic consonants are allowed word-finally in the language. A brute force analysis would extend the ‘hiatus’ context to these consonants as well or might morphologize it: [z] is inserted between a noun stem and a Bloomfield–Sapirian word-level vowel-initial suffix. Evidence for this morphologized analysis is found in an intriguing observation that lies at the heart of Bachrach and Wagner’s article, revealed in pairs like BrPt. zebrinha and zebrazinha, both ‘little zebra’, though slightly different in connotation and distribution: the word zebra in Brazilian Portuguese can denote an unexpected outcome (most often of a football game). An informal survey of internet citations reveals that zebrinha is more likely to denote a small zebra and is often found in child-directed contexts, while zebrazinha is more likely to denote an unexpected outcome. As noted above, in cases like zebrazinha, in Bloomfield–Sapirian terms, the suffix appears to be attached to a whole word, as shown by the retention of the theme vowel -a, while in zebrinha the affix is attached ‘inside’ the word, to its root or stem (depending on one’s theoretical predilections), with concomitant loss or absence of the theme vowel. We also find two possible diminutive derivatives with athematic nouns like BrPt. flor ‘flower’: florzinha, where the attachment is to the whole word, and florinha, where it is attached ‘word-internally’.1⁵ The minimal 1⁵ Florzinha is by far the more frequent of the two. Florinha occurs as a proper name in Brazil, though it is rare.

376

mark aronoff and lori repetti

pair reveals that the epenthesis of /z/ does not take place at just any morpheme boundary, as Bachrach and Wagner originally observe, but only at the word level (however we encode that fact in our theory). Brazilian Portuguese /z/ epenthesis is thus shown to be morphologically complex in its conditioning, though in a different way from the other cases that we have discussed.

13.4 Other types of insertion The factors conditioning the insertion of semantically empty segments given above can be extended to other types of insertion, including Catalan stem extenders (§13.4.1), the Italian augment (§13.4.2), and Spanish antesuffixes (§13.4.3). In each of these cases, the presence of the inserted syllable satisfies a phonological constraint, but its distribution is regulated by the morphology. We do not account for the segmental make-up of the inserted syllables.

13.4.1 Catalan stem extenders In some dialects of Catalan, second-person singular imperatives of conjugation classes 2 and 3 have an ‘extension’ when enclitics are added. The ‘extension’ has no semantic role, but its presence is driven by a prosodic constraint: a moraic trochee is built at the right edge of the verb+enclitic unit (Bonet and Torres-Tamarit 2009). (13)

Formentera Catalan stem extenders ([ә] extension) a. [pέɾt] ‘lose.2sg’ [pәɾð-ә´ә´-lә] ‘lose=it.f!’ gә ´´-lә] ‘learn-it.f!’ ([gә] extension) b. [әpɾә´n] ‘learn.2sg’ [әpɾәŋ-gә c. [búλ] ‘boil.2sg!’ ([iɣә] extension) [buλ-iɣә´-lә] ‘boil=it.f!’

The form of the stem extender is taken from other forms of the imperative verb: it is the material that appears between the root and the person–number markers in the first- and second-plural imperatives. Compare the extensions in (13) with the first-person singular and plural forms in (14). (14)

Formentera Catalan first- and second-person plural imperatives a. [pέɾt] ‘lose.2sg!’ [pәɾð-ә´ә´-m] ‘lose.1pl!’ (with [ә]) [pәɾð-ә´ә´-w] ‘lose.2pl!’ gә ´´-m] ‘learn.1pl!’ (with [gә]) b. [әpɾә´n] ‘learn.2sg!’ [әpɾәŋ-gә ´´-w] ‘learn.2pl!’ [әpɾәŋ-gә gә ´´-m] ‘boil.1pl!’ c. [búλ] ‘boil.2sg!’ [buλ-iɣә (with [iɣә]) iɣә ´´-w] ‘boil.2pl!’ [buλ-iɣә iɣә

While there is no consensus on the morphological status of this material, it is clear that its presence is driven by phonology, but its form is determined by morphology.

conditioned epenthesis in romance

377

13.4.2 Italian Italian has a meaningless augment, , with some fourth-conjugation-class verbs,1⁶ in the present indicative, present subjunctive, and imperative.1⁷ This is illustrated in (15) with the verb finire ‘finish’. (The forms are given in the orthography of Italian, except for stress, which is indicated here, though it is not in Italian orthography. before a back vowel is pronounced [isk], and before a front vowel is pronounced [iʃʃ].) (15)

finire ‘to finish’ a. Present Indicative fin-´ı´ısc sc-o ‘1sg’ sc fin-´ı´ısc sc-i ‘2sg’ sc sc fin-´ı´ısc sc-e ‘3sg’ fin-´ı´ısc sc-ono ‘3pl’ sc b. Present Subjunctive fin-´ı´ısc sc-a ‘1sg’ sc sc fin-´ı´ısc sc-a ‘2sg’ fin-´ı´ısc sc-a ‘3sg’ sc fin-´ı´ısc sc-ano ‘3pl’ sc c. Imperative fin-´ı´ısc sc-i sc

fin-ia´mo (**fin-isc isc-ia isc ´mo) ‘1pl’ fin-ı´te (**fin-isc isc-ı ‘2pl’ isc ´te)

fin-ia´mo (**fin-isc isc-ia isc ´mo) ‘1pl’ isc ´te) fin-ia´te (**fin-isc isc-ia ‘2pl’

isc ´mo) ‘1pl’ fin-ia´mo (**fin-isc isc-ia ‘2sg’ fin-ı´te (**fin-isc isc-ı ‘2pl’ isc ´te)

Some argue that the presence of is phonologically motivated: it is only used with verb forms when it can be stressed, and is absent if it would be unstressed (Burzio and DiFabio 1994; Vogel 1994).1⁸ In other words, the augment regularizes the paradigm so that no verb forms are stressed on the stem.1⁹ In (16a) the augment is present between verb stem /fin/ and the first-person singular suffix /o/, and it is stressed; however, it is absent in (16b) because the stressed suffix /ı´/ would not allow stress to be realized on the augment. (16)

sc sc-o **fı´n-o ‘I finish’ a. fin-´ı´ısc isc isc b. **fin-isc-ı´ fin-ı´ ‘(s)he finished’

1⁶ Note that fourth-conjugation verbs with the augment are much more numerous than those without: 85% ~ 15% (Zamboni 1997:156). 1⁷ The etymology of the augment is to be found in Latin inchoative infix -sc- which was reanalysed in Italian and other Romance languages as a meaningless augment. 1⁸ Some claim that this fact about stress is a consequence of the presence of (Di Fabio 1990:4), and others, most notably Maiden (2004a:33) argue that ‘the augment is not sensitive to stress; rather, stress and the augment are independently sensitive to the same, abstract, paradigmatic distribution’. 1⁹ In some Romance varieties the phonological conditioning is gone, and all forms have the augment: Nemi (Lazio) feniscémo ‘finish.1pl’, feniscéte ‘finish.2pl’ (Rohlfs 1968:243f.).

378

mark aronoff and lori repetti

Hoberman and Aronoff (2003:74) show that the same augment found in Italian verbs is borrowed into Maltese ( in Maltese). It is found with the same verbs; however, the infix is distributed in different cells in the paradigms of the two languages. In Maltese, as in Italian, the augment is found ‘in just those cases where the stem would otherwise be stressed’. The data in (17) show the distribution of the augment in the verb suggest in the present and perfect indicative in Italian and Maltese. (17)

Italian Maltese sc (**suggér-o) ni-ssuġġer-íxx íxx-i sugger-´ı´ısc sc-o íxx (**ni-suggér-i) ‘I suggest’ i-ssuġġer-íí (**i-sugger-ixx-í) ‘to suggest’ sugger-ı´ (**sugger-isc-ı´) i-ssuġġer-íet (**i-ssuġġer-ixx-íet) ‘she suggested’ íxx-a i-ssuġġer-íxx íxx (**i-ssuġġér-a) ‘he suggested’

Hoberman and Aronoff (2003:74) point out that ‘this is a case of borrowing a phonological condition on a morphological rule’.

13.4.3 Spanish antesuffixes Spanish antesuffixes or interfixes are inserted elements (, , etc.) whose distribution is morphologically conditioned (i.e., after certain stems and before certain derivational suffixes). These antesuffixes probably have the same etymology as the Brazilian Portuguese /z/. Latin had an -ulus/-culus alternation for diminutives (the latter formed from nouns ending in -cus + -lus diminutive). These elements have no meaning or connotative value (although in Italian they may; Prati 1942), but their distribution is phonologically influenced: short bases favour insertion of the antesuffix, as do bases ending in a sonorant consonant (except /l/) (Dressler and Merlini Barbaresi 1994). (18)

Spanish papel ‘paper’ papel-ito ‘(dim)’ ec-ita ‘(dim)’ madr-e ‘mother’ madr-ec ec comadr-e ‘godmother’ comadr-ita ‘(dim)’

The antesuffixes are productive, although unpredictable, so that the same base and suffix allow for various forms (Aguero-Bautista 1998), and the interfixed/noninterfixed form can have a special lexical meaning, or the two can be synonymous. Related antesuffixal interfixes can be found in other Romance languages as well: It. boccon-c-ino ‘little bite’ (Dressler and Merlini Barbaresi 1994; Napoli and Reynolds 1995). These are yet other examples of inserted segments that are sensitive to phonological structure, but their form and distribution are not the purview of phonology.

conditioned epenthesis in romance

379

13.5 Findings The factors conditioning the insertion of semantically empty material are given in Table 13.2. We see that in canonical epenthesis, the motivation for insertion is phonological, and morphology plays no role in the distribution and quality of the segment. In the cases we examined, phonology similarly is the motivation for insertion, but morphology plays a role in its quality and distribution.

Table 13.2 Types of epenthesis Inserted Elements

intrusive purely phonovowel logical epenthesis

counts as segment(s) presence is phonologically motivated distribution is influenced by morphology (or morphosyntax) quality is influenced by morphology

− + depends + on one’s definition − −

−

−

morphologically conditioned epenthesis special syllable insertion epenthetic segment + + + + +

+

+

±

We have shown that insertion of semantically vacuous material (segment or syllable) lies along a cline from phonetic to phonological to morphological conditioning, as diagrammed in Figure 13.1.

PHONETICS

PHONOLOGY

MORPHOLOGY

MORPHOSYNTAX

Fig. 13.1 Factors conditioning the insertion of semantically vacuous material

The traditional view of epenthesis, going back to the first uses of the term, leads us to see it as a quintessentially phonological phenomenon. Research over the last several decades has uncovered its phonetic and even physiological roots in the classic notion of ease of articulation. In this contribution, we have used data from a number of Romance languages to show how epenthesis has found its way into what more traditionally might be called the grammar of these languages: individual

380

mark aronoff and lori repetti

morphemes, more abstract purely morphological structures, and morphosyntax. What epenthesis retains throughout is its dependence on pure form. Even in cases where entire strings of segments are epenthesized, they have no meaning. If what characterizes human language above all else is its double articulation (Martinet 1949), then epenthesis provides striking demonstration of the persistence, power, and beauty of this property.

14 Koinéization and language contact The social causes of morphological change in and with Portuguese Tom Finbow and Paul O’Neill

14.1 Introduction Linguists have long been aware of contact as a major factor in language change, whether this contact is between speakers of different languages or between speakers of what are perceived as varieties of the same language. In this chapter, we focus on the diachronic morphological implications of two fundamentally contact-related processes: the reorganization of pronominal and verbal paradigms during the koinéization of standard Spanish and Portuguese in Iberia and of Portuguese varieties in Brazil; and linguistic hybridization through Brazilian Portuguese morphosyntactic influence on the grammar of the Tupi–Guarani language Yẽgatú (also known as Lı´ngua geral amazônica ‘Amazonian lingua franca’). These cases exemplify the central role played by contact in driving language evolution on all levels, from the minute idiolectal accommodations and microvariation in interactions between individuals (Aboh 2015; D’Alessandro 2021) which power the selection of competing features (Mufwene 1994; 2007) and project identities (Eckert 2012), up to the emergent macro-level phenomena with which linguists have more commonly been engaged. We emphasize the commonalities in two cases of koinéization and hybridization, despite their appearing to occur in widely divergent ecologies.

14.2 Koinéization in Romance When speakers of mutually intelligible dialects come into intense and extended contact, linguistic accommodation between speakers may typically become routine and, if it is prolonged, koinéization can occur (Siegel 1985; Trudgill 1986; 2001; 2010a, 2010b; Kerswill 2000; 2013). This process often leads to quite rapid, and

Tom Finbow and Paul O’Neill, Koinéization and language contact. In: Periphrasis and Inflexion in Diachrony. Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent, Oxford University Press. © Tom Finbow and Paul O’Neill (2022). DOI: 10.1093/oso/9780198870807.003.0015

382

tom finbow and paul o’neill

occasionally dramatic, change, usually characterized by reduced morphological complexity, trends towards morphological transparency and predictability, levelling of allomorphy, and even the loss of certain morphological distinctions. The idea is that when speakers of different dialects interact, they are confronted with different ways to say the same thing. The result is that either the least marked or most transparent variant wins out, or different variants are assigned different sociolinguistic, stylistic, or even grammatical functions. We shall refer to the latter process as reallocation (Trudgill 1986; Britain and Trudgill 2005), although others have called it refunctionalization (Smith 2011a), exaptation (Lass 1990; 1997; Vincent 1995; Norde 2002), and even regrammaticalization (Greenberg 1991); see Traugott (2004) for an overview. Lodge (2010) has hypothesized that koinés (new dialects born from koinéization) underlie standard French, and Brown (2020) makes a similar claim for standard Italian. Spanish, however, is the Romance language to which the theory of koinéization has been most widely applied (see, for instance, Penny 2000; Tuten 2003), due to the sociopolitical history of the Romance traditionally referred to as ‘Castilian’ that was spoken in Old Castile (centre–north of the Iberian Peninsula). The Christian reconquest of the Muslim-controlled territories further south brought Castilians into contact with substantial numbers of speakers of other mutually intelligible northern Romance varieties (e.g., Asturian, Leonese, and Navarrese), who had also migrated south into New Castile and Andalusia. This population movement resulted in a new (koiné) variety, phonologically and morphologically different from the contributing varieties, albeit named after just one of them (Castilian). This dialect mixing occurred several times, through migrations to and rapid growth of certain cities (e.g., Burgos, Toledo, and Seville), which resulted in different linguistic changes (see Tuten 2003 for an overview). For example, in the Toledo phase, speakers from diverse central regions of the Peninsula migrated to that city, bringing arrays of possessive pronouns. Table 14.1 presents the hypothesized possessive pronouns present at that time for a secondperson singular non-deferential addressee. As the modern Spanish system developed, most variants were levelled; however, 3, 4, and 7 were reallocated. Monosyllabic tu became proclitic to nouns, whereas the stressed disyllabic tuyo/tuya ‘your (msg)/(fsg)’ occurred elsewhere, e.g., as a postposed adjective (el amigo tuyo ‘def.art.msg friend.msg 2sg.poss.msg’), as a pronoun (los tuy-o-s ‘def.mpl 2sg.poss-m-pl’), and in a predicative function with ser ‘be’ (¿es tuy-o? be.prs.ind.3sg 2sg.poss-msg ‘Is it/he yours?’). Invariable, proclitic tu ‘your’ emerges due to its frequency, markedness, and simplicity: the most frequent phonological sequence in all forms was /tu-/, since in rapid speech before a vowel-initial noun, synæresis made several forms readily perceptible as tu (to amigo [tu̯a.'mi.ɣ̞o] ‘your.msg friend’, tua amiga [tu̯ami.ɣ̞a] ‘your.fsg friend’, tue amiga [tu̯a.mi.ɣ̞a] ‘your.fsg friend’, tuo hombre [tu̯om.bɾe] ‘your.msg man’). Accordingly, /tu-/ appeared morphologically simpler, since in

koinéization and language contact

383

Table 14.1 Different second-person singular possessive pronouns in the Toledo phase and a historical explanation of their forms (Tuten 2003:204–213)

1 2 3 4 5 6 7

variant morphological forms tua (f) tue (f) tu (f) tuya (f) to (m) tuo (m) tuyo (m)

source etymological < tua(m) sound change (weakening: a > e) sound change (apocope: tue > tu) analogy with tuyo (see 7 below) etymological < tuu(m) analogy with tua (see 1 above) sound change (yod epenthesis: cf. creo > creio ‘believe.prs.ind.1sg’, oe > oye ‘s/he listen.prs.ind.3sg’)

such contexts it seemed uninflected for gender and therefore had a greater inflexional range than the other forms. The resulting situation is shown in Table 14.2, where modern Spanish stands out from its neighbours, Portuguese, Galician, and Catalan in having two sets of possessive pronouns with different syntactic, morphological, and phonological properties: the enclitic stressed set is inflected for gender and number, the proclitic unstressed set is invariable for gender.1 Table 14.2 Comparison of Catalan, Portuguese, Galician, and Spanish possessive pronouns Catalan Portuguese Galician Spanish

‘your friend.msg’ el teu amic o teu amigo o teu amigo tu amigo

‘your friend.fsg’ la teva amiga a tua amiga a tua amiga tu amiga

‘it’s yours.msg’ és teu é teu é teu es tuyo

‘it’s yours.fsg’ és teva é tua é túa es tuya

The development of the Spanish pronominal system illustrates how koinéization does not necessarily result in a morphologically simpler system but can increase complexity through the reallocation of variants to different (socio)linguistically determined domains. Two different types of reallocation are identified in the literature: socio-stylistic reallocation and structural reallocation (Britain and Trudgill 2005:187). In the former, the original variants become indicative of some stylistic or social feature in the new dialect—compare allophonic reflexes of Latin /f-/ (Ø and [h]), e.g., fumu(m) > ['u.mo] versus ['hu.mo] ‘smoke’, correlating with education levels across Latin America (Penny 2000:54f.). We now illustrate how 1 Note that diachronically Catalan did have proclitic forms and these are still preserved in some varieties. However, the standard language of Catalonia only preserves these proclitic forms (mon ‘my’, ton ‘your’, son ‘his/her/their’) with kinship terms, e.g., mon pare ‘my.msg father’, and a small number of other nouns, e.g., ta vida ‘your.fsg life’.

384

tom finbow and paul o’neill

the same dialect-mixing processes help explain changes in the pronominal and verbal systems of European Portuguese and certain varieties of Brazilian Portuguese, and how the concept of stylistic reallocation is crucial to explaining why the morphology of planned and unplanned discourse (Keenan 1977:6) can be so different.2

14.3 Koinéization in the history of Portuguese While koinéization has frequently been invoked in Spanish historical linguistics, historical studies of Portuguese rarely refer to this notion; notable exceptions are Cardeira and Fernandes (2008) and Cardeira (2009) for European Portuguese, and Pessoa (2003) for Brazilian Portuguese, but none of these focuses particularly on morphology. The main reasons for this are the lack of historical morphological grammars for Portuguese and the emphasis within Brazilian Portuguese on the debate around whether Brazilian Portuguese emerged via some complex process of creolization due to the high number of African slaves and indigenous peoples (Guy 1981; 1989; Lucchesi 1999; 2001; 2003; 2017) or not (Naro and Scherre 2003; 2007). Space restrictions do not permit a thorough analysis of the arguments but, leaving terminological distinctions aside, the effects of contact-induced change can certainly be more intense the more varieties differ from each other, when the participants are adult learners, and when there is only partial exposure to a particular linguistic input (Trudgill 2003:437f.; 2010b; 2011a; 2011b).

14.3.1 Development of second-person plural pronominal and verbal forms in European Portuguese The sociopolitical histories of Peninsular Portuguese and Spanish differ because the Spanish standard became based on the central–northern varieties of Toledo and Madrid, not the southern dialect of Seville, while the Portuguese standard was based on the Lisbon variety from halfway up the western seaboard, south of Toledo and Madrid but north of Seville. The capture of Lisbon (1147) and Seville (1248) and their occupation by large numbers of more northerly Romance speakers from the respective conquering kingdoms brought about dialect mixing and acquisition of Romance by non-native speakers: first, Arabic-speaking Muslims and Jews who remained in their ancestral

2 Planned discourse is produced with sufficient time for reflection and organizational preparation, whereas unplanned discourse lacks such forethought.

koinéization and language contact

385

homes,3 and later African slaves, who were already numerous in both cities around 1492.⁴ The verbal paradigm of these cities’ speech was morphologically simpler than that of more northerly varieties, as the deferential/non-deferential address distinction in the plural had been lost. Reflexes of Latin nominative uos (2pl) ‘you’ were no longer used for second-person plural non-deferential address, being replaced by reflexes of Pt. vossa mercê/Cst. vuestra merced ‘your.2pl Grace’ (> Pt. você. Cst. usted) (see Table 14.3). Table 14.3 Comparison of second-person plural address (‘you are’) in standard Spanish, Galician, and Portuguese, and in the speech typical of Seville non-deferential

standard Galician vos sodes

standard Spanish vosotros sois

deferential

vostedes son

ustedes son

standard EuPt

Seville

vocês são

ustedes son

We maintain that these differences result directly from koinéization processes in Seville and Lisbon. Before analysing the evolution of the European Portuguese forms, however, we should note that there is a marked distinction between what appears in prescriptive grammars and actual usage. Table 14.4 displays the European Portuguese pronouns and verb agreements. The forms in italics represent the commonest variants in the speech of Lisbon and in the majority of European Portuguese speakers’ unplanned discourse. Focusing exclusively on the forms in italics and comparing them with the corresponding standard Galician forms in Table 14.5, which historically underlie the European Portuguese forms, it is clear that European Portuguese developed in line with known koinéization processes. The morphological system has simplified, merging the two paradigms for second-person plural address and removing the distinction between deferential and non-deferential address.⁵ Moreover, within the reorganization of the paradigm generally the less morphologically complex forms have won out. This is most apparent with vós, which not only has a greater inflexional range (Croft 1990) than the direct object pronoun lhe, but also lacks the allomorphy of the object pronouns which etymologically correspond to vocês (-os/-as, -nos/-nas, -los/-las). 3 Hebrew was not a vernacular language in Jewish communities in medieval times. It was a liturgical language learned to fluency mainly through formal instruction. The Jewish communities would have spoken local Romance and Andalusian Arabic varieties as their vernaculars. ⁴ During the period around Columbus’s 1492 voyage it has been estimated that up to 150,000 subSaharan African slaves entered Portugal. In Seville, approximately 300 arrived in 1495–1496 (see Sweet 1997:163f. and fn. 72). ⁵ Note that this distinction was then recovered by using the term of address os senhores, lit. ‘the gentlemen’.

386

tom finbow and paul o’neill

Table 14.4 Second-person plural forms of address for European Portuguese with respective verbal agreement and corresponding pronouns Standard European Portuguese second-person plural address Subject with + direct object indirect object pronoun vocês convosco os, nos, los lhes vós con vocês as, nas, las vos vos Verbal agreement

Imperative

3pl

cantem ~ cantai comam ~ comei durmam ~ dormi digam ~ dizei vão ~ ide

Present indicative cantam ~ cantais comem ~ comeis dormem ~ dormis dizem ~ dizeis vão ~ ides

Imperfect indicative cantavam ~ cantáveis comiam ~ comíeis dormiam ~ dormíeis diziam ~ dizíeis iam ~ íeis

possessive seu(s) sua(s) vosso(s) vossa(s) Preterite cantaram ~ cantastes comeram ~ comestes dormiram~ dormistes disseram ~ dissestes foram ~ fostes

While frequency and greater inflexional range may have favoured seu(s)/sua(s) over vosso(s)/vossa(s) for ‘your.m(pl)/f(pl)’, the triumph of the latter clearly stems from the formal relationship vocês—vos—vosso. Likewise, although desire for morphological transparency could have militated against convosco, paradigmatic support for this synthetic form from comigo, contigo, consigo, conosco surely contributed to its survival in unplanned discourse alongside periphrastic com vocês. The other second-person plural variants have not been levelled but have undergone stylistic and geographic reallocation due to their preponderance in older written texts, especially translations of the Bible, and in less koinéized European Portuguese varieties. Thus, in northern Portugal the forms not in italics can appear in unplanned discourse alongside the standard variants, as in (1). Observe also in (2) the formulaic questions asked of the bride and groom by a priest at a wedding, which show how the etymological vós paradigm is still present throughout Portugal, albeit strongly associated with liturgical contexts. Regarding stylistic reallocation in the mixed standard system, prescriptive grammarians have noticed the diachronic attestation of the morphological variants corresponding etymologically to both vós and vocês and have incorporated them into language manuals. Their subsequent legitimization means that native speakers implement these pronouns in planned discourse. Witness the message in (3), which appears (noted in 2019) on passenger screens of the Portuguese-owned airline TAP. Here vos is used for the indirect object, with -los as the direct object.

Table 14.5 Second-person plural forms of address for Galician with the respective verbal agreement and corresponding pronouns Subject

3pl Verbal agreement

canten coman durman digan vaian

cantan comen dormen din van

cantaban comían dormían dicían ían

possessive seu(s) súa(s) Preterite

cantaron comeron dormiron dixeron foron

Galician second-person non-deferential address Subject with + direct object indirect object pronoun vós convosco vos vos 2pl Verbal agr

Imperative Present indicative

Imperfect indicative

cantade comede durmide dicide ide

cantabades comiades durmiades diciades iades

cantades comedes durmides dicides ides

possessive voso(s) vosa(s) Preterite

cantastes comestes durmistes dixestes fostes

koinéization and language contact

vostedes

Galician second-person deferential address with direct object indirect +pronoun object os, nos, los lles con vostedes as, nas, las Imperative Present Imperfect indicative ind

387

388

tom finbow and paul o’neill

(1)

Vocês[you-3pl] andais[2pl] a gastar dinheiro; vocês[you-3pl] tenham[3pl] cuidado, sois[2pl] educadas para isso. ‘You’ve been out spending money; be careful, you’re too well brought up for that’. (Aguiar and Conceição de Paiva 2017:147)

(2)

- Viestes[2pl] aqui para celebrar o vosso[2pl] Matrimónio. É de vossa[2pl] livre vontade e de todo o coração que pretendeis[2pl] fazê-lo? - Vós[you-2pl] que seguis[2pl] o caminho do Matrimónio, estais[2pl] decididos a amar-vos[2pl] e a respeitar-vos[2pl], ao longo de toda a vossa[2pl] vida? ‘- You have come here to celebrate your Marriage. Is it of your own free will and with all your heart that you wish to undertake this?’ ‘- You who pursue the path of Matrimony, are you resolved to love and respect each other as long as you both shall live?’

(3)

senhores passageiros desejamos que a viagem vos[2pl] tenha agradado e esperamos vê-los[3pl] de novo, em breve. Obrigado por voarem[3pl] na TAP. ‘Ladies and gentlemen, we hope that you have had a pleasant journey and hope to see you again soon. Thank you for flying TAP.’

Such discrepancies between common spontaneous variants and those prescribed in didactic grammars can cause much uncertainty for foreign students and native speakers of Portuguese. The Portuguese website ciberdúvidas da lı´ngua portuguesa,⁶ when trying to resolve a student’s doubts about alternative pronominal forms, posted the response in (4) regarding possessive vosso/seu. (4)

embora muitos gramáticos não aceitem vosso como possessivo de vocês, o certo é que este uso está de tal modo generalizado, que parecerá afectado dizer «deixem aqui as suas malas» em vez de «deixem aqui as vossas malas». ‘although many grammarians do not accept vosso as the possessive form of vocês, it is true that this usage has become so widespread that it probably sounds affected to say deixem aqui as suas malas (“leave your bags here”) instead of deixem aqui as vossas malas (“leave your bags here”).’

The site makes the valid point that the variation in verbal agreement (cant-am sing-prs.ind.3pl vs cant-ais sing-prs.ind.2pl) is geographical and notes that all pronominal allomorphs (those in Table 14.4) are accepted as normatively correct. However, the choice of one form or another and their perceived correctness can even cause problems for native speakers. In one post, a native European Portuguese ⁶ https://ciberduvidas.iscte-iul.pt/consultorio/perguntas/a-forma-de-tratamento-voces-opronome-atono-vos-e-o-possessivo-vosso-ii/34,887

koinéization and language contact

389

speaker observes how she translated English nice to meet you (with second-person plural interpretation) as prazer em conhecer-vos[2pl.obj] and was questioned by the (foreign) commissioner of the translation who doubted the level of formality, assuming -vos was non-deferential and the deferential address would be prazer em conhecê-los/las[3pl.obj.m/f]. The native speaker turned to the website for advice on which form was correct. Copied in (5) is the response. (5)

Em Portugal há muita hesitação quanto a avaliar o grau de formalidade do pronome vós, pelo que muitas vezes surge como forma de pronome complemento correspondente quer a vocês (informal: «vou chamar-vos») quer a os senhores/as senhoras («vou chamar-vos», em alternativa a «vou chamálos»). É uma área em que há uma grande instabilidade e acerca da qual as gramáticas de referência não são claras. ‘In Portugal there is considerable uncertainty regarding the level of formality of the pronoun vós, such that it often appears as the form of the object pronoun corresponding either to vocês “you[3pl]” (informal: vou chamar-vos “I shall call you [2pl] [informal]”) or to os senhores/as senhoras “the [=you] gentlemen/the [=you] ladies]” (vou chamar-vos “I shall call you [2pl] [informal]” instead of vou chamá-los “I shall call you [3pl] [formal]”). This is an area of great instability and with respect to which the reference grammars are unclear.’

Had these forms not been preserved in writing and seized upon by grammarians, they would in all probability have been levelled in southern European Portuguese. However, their preservation has meant their reallocation as quasi-stylistic formality markers in planned discourse. The website interprets the overabundance of forms as the standard language needing time to settle. Observe the comments in (6) about the variation of pronouns and verbal agreement for second-person plural address. (6)

o que acabo de descrever é uma situação de hesitação e instabilidade. É, pois, possível que dentro de algumas dezenas de anos estes usos se definam mais claramente e sejam aceites pela norma europeia.⁷ ‘what I have just described is a situation of hesitance and instability. It is therefore possible that within a few decades these usages will be defined more clearly and will be accepted by the European standard.’

The situation regarding the pronominal system and corresponding verb forms for second-person singular non-deferential address in Brazilian Portuguese is similar. However, mixed systems and hesitations are often perceived by Brazilians and

2⁷⁸ https://ciberduvidas.iscte-iul.pt/consultorio/perguntas/as-formas-de-tratamento-vos-e-vocese-os-possessivos/17448

390

tom finbow and paul o’neill

other lusophones as the result of Brazilians not being able to speak Portuguese properly, rather than the language not having settled. Note that generally in unplanned speech, Brazilians alternate the unstressed clitic pronouns with the stressed invariable subject forms in postverbal position or following a prepositional complement, e.g., ‘I love you’: te=am-o (2sg.obj=loveprs.ind.1sg) vs am-o você (love-prs.ind-1sg you [3sg]); ‘I spoke to you’: te=fal-ei (2sg.obj=speak-prt/prf.1sg) vs fal-ei para você (speak-prt/prf.1sg to you [3sg]). Given space limitations, only the unstressed clitics will be considered here.

14.3.2 Second-person singular non-deferential address in unplanned discourse of two Brazilian Portuguese varieties We shall focus on two geographically distinct but fairly well-established systems of second-person singular address in two Brazilian Portuguese varieties: the speech of the city of São Paulo (Table 14.6) and the southern city of Porto Alegre (Table 14.7). Both cities experienced dramatic population growth at different times in their history, during which speakers of different varieties of Portuguese formed a koiné. One must not forget that both cities were also home to numerous speakers of other languages who acquired Portuguese as adult L2 speakers. Contact in both cities caused the etymological paradigms for tu and você, both intact in European Portuguese and given in Tables 14.8 and 14.9, to merge and reorganize, eliminating the morphological distinction between deferential and non-deferential address.⁸ In São Paulo, você triumphed; in Porto Alegre, tu won out. These different outcomes undoubtedly arose from the frequency of the subject pronouns used by the Portuguese-speaking communities in the respective geographic areas. Table 14.6 New você paradigm for second-person singular non-deferential address in the spoken Brazilian Portuguese of the city of São Paulo (forms in italics correspond etymologically with the subject pronoun você) Subject você 3sg verbal agreement

with + pronoun com você Imperative canta come dorme diz vai

direct object

indirect object

possessive

te Present indicative canta come dorme diz vai

te Imperfect indicative cantava comia dormia dizia ia

seu(s) sua(s) Preterite cantou comeu dormiu disse foi

⁸ As with European Portuguese, this distinction was reintroduced as o senhor/a senhora ‘sir’/‘madam’ and their corresponding morphological forms.

koinéization and language contact

391

Table 14.7 New tu-paradigm for second-person singular non-deferential address in spoken Brazilian Portuguese of the city of Porto Alegre (forms in italics correspond etymologically with the subject pronoun você) Subject tu verbal agreement

with + pronoun contigo Imperative canta come dorme diz vai

direct object

indirect object

possessive

te Present indicative canta come dorme diz vai

te Imperfect indicative cantava comia dormia dizia ia

teu(s) tua(s) Preterite cantou comeu dormiu disse foi

Table 14.8 Standard European Portuguese second-person singular non-deferential address (no forms correspond to você) Subject tu 2sg verbal agreement

with + pronoun contigo Imperative canta come dorme diz vai

direct object te Present indicative cantas comes dormes dizes vais

indirect object te Imperfect indicative cantavas comias dormias dizias ias

possessive teu(s) tua(s) Preterite cantaste comeste dormiste dixeste foste

Table 14.9 Standard European Portuguese second-person singular deferential address (all forms correspond to você) Subject você 3sg verbal agreement

with + pronoun com você

direct object

indirect object

possessive

lhe

seu(s) sua(s)

Imperative

o, no, lo a, na, la Present indicative

Preterite

cante coma durma diga vá

canta come dorme diz vai

Imperfect indicative cantava comia dormia dizia ia

cantou comeu dormiu disse foi

It is striking how the same koinéization principles operate in the development of both of these new paradigms and are similar to those which determined the novel mixed paradigm for second-person plural address in European Portuguese. Thus, in unplanned discourse in São Paulo, the subject pronoun você alternates with the

392

tom finbow and paul o’neill

object pronoun te and not the pronouns to which it etymologically corresponds (direct object o(s), a(s), -no(s), -na(s), -lo(s), -la(s); indirect object lhe(s)), since, like vos in European Portuguese, te is allomorph-free and thus displays a greater inflexional range. Likewise, the etymological imperative which corresponds to você and is syncretic with the third-person singular present subjunctive forms (cante ‘sing!’, coma ‘eat!’, durma ‘sleep!’) is disfavoured over the imperative which corresponds to the pronoun tu (canta ‘sing!’, come ‘eat!’, dorme sleep!’),⁹ since the form associated with tu is the most autonomous/most basic verb form (Bybee and Brewer 1980; Hock 1986:214–237; Schilling-Estes and Wolfram 1994:276). Similarly, in the speech of Porto Alegre, the subject pronoun tu alternates with its etymological object pronoun (te) and etymological forms of the imperative (canta ‘sing!’, come ‘eat!’, dorme ‘sleep!’) as opposed to the inflexions corresponding to você. However, for the remaining verbal agreements the etymological tuforms (cant-as ‘sing-prs.ind.2sg’, cant-av-as ‘sing-pst.ipfv-2sg’, cant-aste ‘singprt/prf.2sg’) lose out to those corresponding to você (cant-a ‘sing-prs.ind.3sg’, cant-av-a ‘sing-pst.ipfv-3sg’, cant-ou ‘sing-prt/prf.3sg’), since the latter set is more frequent (it is found with all 3sg subjects, as well as você), and is one phoneme shorter. Thus, in Porto Alegre, one hears onde tu nasc-eu? (where you.sg be.born-prt/prf.3sg) ‘where were you born?’, onde tu foi ontem? (where you-sg go.prt/prf.3sg) ‘where did you go yesterday?’. The formal relationship of possessive teu (2sg.poss.m) with the subject pronoun tu (2sg) and the clitic object pronoun te (2sg) was probably significant in its trumping more frequent seu (3sg.poss.m) in Porto Alegre. Analytical, morphologically transparent com tu ‘with you’ never prospered, because syncretic contigo was fortified by its paradigmatic relation to comigo ‘with me’—conosco ‘with us’ and its formal similarity to subject tu (2sg)—clitic object te (2sg)—tonic object ti (2sg.obl). As with the remnants of the European Portuguese second-person plural address system, the morphological variants in Brazilian Portuguese systems escaped levelling, since their prevalence historically in writing and in European Portuguese, the basis for the Brazilian republican standard (Pagotto 1998; Faraco 2001; 2008; Faraco and Zilles 2017), made grammarians incorporate these ‘former’ paradigms into several prescriptive grammars. Thus, the etymological paradigms were reallocated to planned discourse, whereas the new paradigms prospered in everyday speech. In southern Brazilian speech, where the tu-paradigm in Table 14.7 predominates, a number of studies show that teu-possessives correlate strongly with

⁹ While this is the general tendency, it must be noted that there is substantial variation and this can at times be correlated with a particular lexeme, e.g., the imperative form veja, etymologically a você imperative, for ver ‘see’.

koinéization and language contact

393

concepts of solidarity and social proximity and also asymmetrical power relations of superior to inferior, such that seu/teu contrast as deferential/non-deferential markers (Menon 1995:102; Loregian-Penkal 2004; Arduin 2005), distinct from the more common informal você (3sg) ‘you’—seu (3sg.poss.msg) ‘your’ and formal o senhor (3sg) ‘you/sir’—seu (3sg.poss.msg) ‘your’ alternations. Morphological variant reduction and reallocation also take place in other verbal persons due to scope for confusion between subject pronouns and their corresponding verbal forms. For example, in first-person plural forms, nós ‘we’ (< Lat. nos) and a gente ‘we’ (lit. ‘the people’, see also §14.4.1.1) compete. The latter etymologically requires third-person singular agreement, stimulating innovations such as nós cant-ou ‘1pl sing-prt/prf.3sg’, and, less frequently, a gente cant-amos ‘the people/we sing-prt/prf.1pl’ (Scherre et al. 2018), alongside etymological nós cant-amos ‘1pl sing-prt/prf.1pl’ and a gente cant-ou ‘the people/we singprt/prf-3sg’. Likewise, in vocative contexts, non-deferential vocês alternates with the nouns gente ‘people’ and pessoal ‘people (lit. personnel)’, which usually display singular agreement. Therefore, in positive imperative contexts there is much variation between the etymological vocês imperative (façam isso! ‘do(3pl) that’), the etymological tu imperative (faz(2pl) isso! ‘do(2pl) that!’), and an innovation based on the third-person plural present indicative (fazem isso! ‘do(3pl) that’).1⁰ Such variation is present elsewhere in Brazil in second-person singular address, as detailed dialectological studies such as Scherre (2012) and references therein record. Observe the percentages in Table 14.10 for the use of você and tu and the corresponding verbal forms for a number of cities in the northern region of Brazil (data taken from da Costa 2013:68). Table 14.10 Percentages of the use of subject pronouns and verbal agreement in a number of cities in Brazil City

Você

Tu

Belém Boa Vista Macapá Manaus Porto Velho Rio Branco

29.2 51.7 52.5 31.5 62.9 34.5

70.8 48.3 47.5 68.5 37.1 65.5

verbal agreement of tu with etymological forms 57.7 75.8 70.8 80.5 54.5 44.9

In Salvador, Brazil’s first capital, the second-person singular subject pronoun is always você, as in the speech of São Paulo, but, in contrast to the paulistano variety, the etymological você paradigm is more frequent, e.g., cant-e ‘sing-prs.sbjv.3sg (= 1⁰ This form is most probably due to the verb ir ‘go’, whose etymological vocês imperative vão is syncretic with the 3pl.prs.ind.

394

tom finbow and paul o’neill

sing!, (that) he/she/it may sing!)’, com-a ‘eat-prs.sbjv.3sg (= eat!, (that) he/she/it may eat)’, durm-a ‘sleep.prs.sbjv.3sg (= sleep!, (that) he/she/it may sleep)’, dig-a ‘say-prs.sbjv.3sg (= say!, (that) he/she/it may say)’. The tu-imperative, i.e. cant-a ‘sing-imp.2sg (sing!)’, com-e ‘eat-imp.2sg (= eat!)’, dorm-e ‘sleep-imp.2sg (sleep!)’, diz ‘say.imp.2sg (= say!)’, that predominates in São Paulo is only used 25% of the time (Sampaio, 2001; Scherre 2012). Often koinéization is conceived as a bottom-up type of change which emerges naturally from linguistic interaction and, after temporary instability, ends up focused and stable (Trudgill 1986; Trudgill et al. 2000). Such natural, contactbased change is often contrasted with the more prescriptive and superficial top-down changes involved in standardization. The Portuguese and Brazilian data raise questions about these assumptions and show how bottom-up and topdown processes interact with each other at different periods (see also Brown 2020) to produce a standard language rich in morphological variation, regulated not only by grammatical constraints but also by imprecise, ill-defined stylistic constraints.

14.4 Brazilian Portuguese as morphosyntactic model for indigenous languages In the second part of this chapter, we turn from the intralinguistic phenomenon of koinéization, as exemplified by changes in the paradigms and agreement patterns associated with the pronoun system of pluricentric Portuguese, to interlinguistic contact—in particular, the morphosyntactic effects of contact between languages which are typologically quite distinct. Our focus here is the influence of Brazilian Portuguese on Nheengatu/Yẽgatu, a Tupi–Guarani language, historically the most widespread group of indigenous languages spoken in Brazil. In the first part of this chapter, we described population movements that brought diverse Ibero-Romance linguistic variants into competition and discussed the top-down and bottom-up pressures that favoured or inhibited their expansion. Similar developments accompanied Portugal’s exploitation of its South American colonies, during which hundreds of thousands of indigenous people were forcibly uprooted and thrown into profoundly alien linguistic and sociocultural ecologies (Monteiro 2001; Monteiro 2012; Langfur 2014; Roller 2014). Crucially, for most of Brazil’s history, Portuguese was not the first language of the majority of the population. Multilingualism was the norm, profoundly influencing the formation of Brazilian Portuguese (see for instance Mufwene 2004; 2008; Negrão and Viotti 2008; Lucchesi, Baxter, and Ribeiro 2009; Negrão and Viotti 2010; 2011a; 2011b; 2012; 2014a; 2014b; Lucchesi 2015). Alongside autochthonous languages and those of African slaves and freedmen and their descendants, African and Tupi–Guarani-based lingua francas arose and

koinéization and language contact

395

were very widely spoken (Rodrigues 1996; Bonvini and Petter 1998; Freire and Rosa 2003; Rodrigues 2003; Noll and Dietrich 2010; Nobre 2013; 2016). Today, however, Brazilian Portuguese has been the hegemonic lingua franca for over a century and has profoundly influenced the indigenous language that has coexisted longest with it: Nheengatu or Yẽgatú (/iẽ̯ .ɡaˈtu/[ȷ˜ẽ.ʔẽ.ɡaˈtu] ~ [ɲẽ.ʔẽ.ŋɡ aˈtu]) ‘good speech’ (< old Tupi/jẽ'ʔẽ(ŋ)/‘speak’ +/ka'tu/‘good’) or Pt. lı´ngua geral (‘lingua franca’, lit. ‘general language’).11 There is no doubt that Yẽgatú can be called a contact language, given the conditions of its emergence in multilingual, multi-ethnic missionary villages and colonial settlements and its use as a second language alongside many other indigenous languages over a vast region. Historical references suggest a certain degree of nineteenth-century dialectalization (Hartt 1872; Magalhães 1876; Stradelli 1929), especially a marked Arawak phonological substrate in the Rio Negro (Taylor 1985; Moore, Fagundes, and Pires 1994; Cruz 2011; Moore 2014). Since 2002, Yẽgatú, Tukano (East Tukanoan), Baniwa (Northern Arawak), and Yanomani (isolate) are co-official languages of the municipality of São Gabriel da Cachoeira in Brazil’s Amazonas state. Despite the ever-increasing shift to Portuguese, Yẽgatú is still spoken by a multiethnic, multilingual indigenous population of perhaps some 7–10,000 people on the upper Rio Negro and its main tributaries in north-west Brazil, Colombia, and Venezuela. A small community of traditional speakers exists on the Middle Amazon among the Sateré-Mawé and Maragua´ peoples (Moore, Fagundes, and Pires 1994; Schwade 2014). Several Amazonian communities whose ancestral languages died out undocumented through language shift to lı´ngua geral, are revitalizing their indigenous identity through learning Yẽgatú, which often ceased to be spoken within living memory, replaced by Brazilian Portuguese.

14.4.1 Old Tupi and Portuguese Mutually intelligible varieties of old Tupi (Tupi–Guarani branch, Tupian family) were spoken along most of Brazil’s Atlantic coast in 1500. The early Portuguese colonists’ descendants spoke their indigenous mothers’ old Tupi natively. Missionaries adopted this lı´ngua brası´lica ‘Brazilian language’ as the medium for 11 Old Tupi examples are in a broad phonemic transcription. Morpheme-final consonants elided in compounding appear between parentheses. Where relevant, phonotactic anaptyxis of vowels and glides is shown in square brackets. Yẽgatú examples use the phonemic FOIRN/UFAM orthography adopted in the Rio Negro in 2016 (the Tapajós and Baixo Amazonas communities use other orthographies but the FOIRN/UFAM norm is the official norm of the largest community of native speakers), to which we have added an acute accent to mark word stress, following Casasnova’s unofficial orthography (2006), to assist readers unfamiliar with Yẽgatu, as stress is not marked in the FOIRN/UFAM norm. Where tonic vowels are nasal, a tilde marks both functions. Nineteenth-century examples retain the original spellings between angled brackets.

396

tom finbow and paul o’neill

evangelization, writing grammars (Anchieta 1595; Figueira 1621), catechisms (Anchieta 1730, Araújo 1618; 1686[2]; Bettendorff 1687 [1800]), and religious poetry and theatre (Anchieta 1583; 1583–1594). Crown recognition of the de facto norm came in late 1689. The indigenous demographic collapse on the coast from c.1550 (Hemming 1987:139–160) led to the introduction of African slaves (see, for instance, Boxer 1973:236–254), profoundly altering the demography and thus the linguistic composition of the southern colony, outside the captaincy of São Paulo.12 Yẽgatú evolved from Tupinamba´ old Tupi spoken around Belém, Portugal’s first Amazonian foothold (1616). From the 1650s, missionary villages were founded on the lower Amazon, at first peopled mainly by Tupi–Guarani speakers. Horrific indigenous mortality drove the Portuguese ever further into the interior in search of new peoples to subdue and bring downriver to work the land and collect forest products for export. Hundreds of thousands of ethnically and linguistically diverse people were forcibly integrated into a colonial system that spoke an old Tupi-based lingua franca (Baena 1831:247; Raiol 1900:132; Freire 1983:50, 68; Barros 2003:89). A century later, the Jesuit João Daniel (1727–1776) observed that a ‘corrupt’ lı´ngua geral spoken by all had substituted the ‘ancient, true lı´ngua geral of the Tupinamba´’, preserved only among their remnants and the missionaries (Daniel 1976:I,269; II,226; Barros 2003:86–87). Cabral and Rodrigues (2011), Monserrat (2003), and Barros (2003) observe little substantive change between sixteenth and seventeenth-century old Tupi and eighteenth-century written language. Despite official prohibition in 1722, use of lı´ngua geral remained near-universal in Brazilian Amazonia, as many nineteenth-century travellers observed (see Hemming 1995:465–491). However, in urban areas on the lower Amazon, especially after 1850, shift to Portuguese was increasingly noticeable (Freire 2003), stimulating a folkloric interest that produced notable works of documentation, e.g., Faria (1858), Hartt (1872), Magalhães (1876), Sympson (1877), Rodrigues ([1890] 2017), Tastevin (1923), Amorim (1926), Stradelli (1929). In Amazonas state, Portuguese finally overtook Yẽgatú around 1870, following major demographic change during the rubber boom (Freire 2003), that saw a huge influx of monolingual Portuguese-speakers, mainly from north-eastern Brazil. Portuguese influence is very apparent in nineteenth- and twentieth-century Yẽgatú when compared to the pre-1800 language. Modern descriptions (e.g., Taylor 1985; Moore, Fagundes, and Pires 1994; Grenand and Ferreira 1989; Cruz

12 In the south, the língua geral paulista (Rodrigues 1996) arose independently among the mestizos of São Paulo and their Tupiniquim, old Tupi- and Guarani-speaking slaves. Sparsely documented, it was spoken by small populations of paulistas scattered over a vast area of central south-western and south-eastern Brazil in the second half of the eighteenth century. Already moribund in the city of São Paulo at the beginning of the nineteenth century, it probably became extinct around mid-century in the interior of São Paulo province.

koinéization and language contact

397

2011; Moore 2014) also reveal marked changes in some features since the nineteenth century. Considerable social change in the upper Rio Negro since the 1960s has drawn the region closer to mainstream Brazilian society—for instance, exposure to radio and television media, greater access to primary and secondary education, different commercial relations, large numbers of monolingual Brazilian Portuguese-speaking incomers, and faster, more affordable river and air links to Manaus and beyond.

14.4.1.1 Demonstratives and pronouns Tupi–Guarani demonstratives typically exhibit tripartite distance/SAP/person distinctions, specific/generic reference and visible/invisible, as illustrated in Table 14.11. Table 14.11 Old Tupi demonstratives (Rodrigues 2010:27, adapted) proximal speaker 1st person

visible

(i-)'ko (i-)'ke (i-)'nã(n) kɨ'βõ (i-)'ʔã(ŋ)

non-proximal proximal addressee 2nd person eβo-'kʷe(j) emo-'nã(n) eβo-'wĩ(ŋ) eβ(o)-a'nõj

a-'kʷe(j) a-'mõ

specific referent a-j-'po [heard]

invisible (i-)ã(ŋ)

generic referent

distal 3rd person 'kʷe(j) 'wĩ(j)(ŋ) 'mõ e'rik

eβ(o)-a-'po

a-'ʔe a-'wã a-'po a-'nõj

m-a-'ʔe ‘something’ m-a-'mõ ‘somewhere yonder (vis.)’ m-a-'nõj ‘from somewhere over there’ u'm-ã(n) ‘somewhere’ m-i-'rã(m) ‘at some future time’

These forms underlie numerous old Tupi locative, temporal, and contextual expressions (Rodrigues 2010). Nominalizing suffixes -βa'ʔe and -a derived pronominal forms, e.g., i-ko-βa'ʔe ‘this visible one’, a-'kʷej-a ‘yonder invisible one’. Yẽgatú iké ‘here’ is perhaps the only unchanged survival. Old Tupi a'ʔe became Yẽgatú third-person singular personal pronoun aé. Yẽgatú exhibits only the noun phrase proclitic kua´ ‘this’ (< ?i-ko-a ‘this visible one’) (9) and yaã ‘that’ (8), mirroring popular BrPt. esse ‘this’ and aquele ‘that’, respectively. Two independent demonstrative pronouns aitekua´ and aiteyaã grammaticalized from aé ‘3sg’+ te ‘focus’ + kua´ ‘this’/yaã ‘that’ (7).

398

tom finbow and paul o’neill

(7)

se=r-amu˜ya mayé=taa´ ya-serúka aitekua´? 1sg.poss=r2∖grandfather be.like=q 1pl.act.sbj-call dem.prox.foc ‘Grandfather, what do we call this one?’ (Cruz 2011:148, adapted)

(8)

a-piripãna yaã úka. 1sg.act.sbj-buy dem.dist house ‘I bought that house.’

(9)

(Cruz 2011:149, ex. 198, adapted)

ya-pudéri ya-pisı´ka kua´ yẽga Português. 1pl.act.sbj-can 1pl.act.sbj-get dem.prox language Portuguese ‘We can get this language, Portuguese.’ (Cruz 2011:158, ex. 116c, adapted)

Radical paradigmatic reduction is typical of intense contact situations and some scholars, such as Schmidt-Riese (2003), Lee (2005), and Nobre (2016), claim Yẽgatú is a creole language. However, all nineteenth-century descriptions mention the same forms, and parallels with Brazilian Portuguese demonstratives that occur far from historically Yẽgatú-speaking areas suggest Portuguese pressure from an early date. Tupi–Guarani languages typically distinguish exclusive and inclusive firstperson plural (o're versus ja'ne). Yẽgatú has generalized first-person plural inclusive yãdé, ya- (Cabral and Oliveira 2013), a loss certainly connected to the absence of the contrast in Portuguese (Moore, Fagundes, and Pires 1994; Moore 2014), as the inclusive/exclusive distinction in first-person plural is common in Amazonian languages. Losing o're caused restructuring of first-person plural yãdé into a periphrastic inclusive (payẽ) yãdé (payẽ) ‘we/us-all’. Some suggest (e.g., Schmidt-Riese 1998:326 n.60), that the Tupi–Guarani distinction between an inclusive and an exclusive first-person plural potentially stimulated the emergence of Brazilian Portuguese a gente, initially as an inclusive first-person plural, in contrast to ancestral Romance nós, that expressed exclusive first-person plural (see §14.3.2). Conclusive evidence regarding this hypothesis is lacking, although, impressionistically, a gente and nós do seem to have a different distribution in Rio Negro Portuguese from that found in major south-eastern varieties, e.g., Rio de Janeiro and São Paulo. Plural is generally unexpressed on Tupi–Guarani nouns. However, eta´ ‘be.numerous’ grammaticalized into a postnominal plural suffix, -ita´ (Cruz 2015) and Yẽgatú developed a third-person plural inflexion ta(u)- from fusing the reduced variant third-person plural pronoun ta´ (< aĩta´ < a'ʔe-e'ta dem.dist.ivsblbe.numerous) with u- (3.sbj.a) (Cruz 2015).

14.4.1.2 Portuguese influence on Yẽgatú verb typology Old Tupi was an active–stative or split ergative language with person hierarchy (Seki 1990; 2000; Reich 2003; Mithun 2008a; 2008b; Birchall 2014). Tables 14.12– 14.18 compare old Tupi and Yẽgatú verb paradigms. Thus, in old Tupi, intransitive predicates are divided into two classes, one of which (active intransitive) contains

Table 14.12 Old Tupi sentence with active verbs, first-, second-, or third-person agent and third-person patient or reciprocal or reflexive complements

3 3

koinéization and language contact

3 3

399

400

tom finbow and paul o’neill

Table 14.13 Old Tupi stative verb

3 3

3

Table 14.14 Old Tupi transitive verb with third-person Agent and first- or second-person Patient Verb object index se= 1sg.stat= ja'ne= 1pl.incl.stat= o're= 1pl.excl.stat= ne= 2sg.stat= pe'je= 2pl.stat=

Subject regular root pluriform root pɨ'sɨk ‘catch’ r-e'pʲak ‘r2∖see’

ja'wara ‘jaguar’ pi'ra ‘fish’ a'ʔe dem.dist.ivsbl

forms in which the subject is closer to a prototypical agent (i.e., it exhibits characteristics such as high animacy, volition, or topicality), for instance, so ‘go’, jãn ‘run’, ʔɨ'taβ ‘swim’, jẽ'ʔẽŋ ‘speak’; these take the same personal affixes as transitive predicates (Table 14.12). In the other class of intransitive predicates (stative or inactive), the subject is prototypically less agentive (i.e., an experiencer, or patient-like), and is marked with a different set of personal indices (Table 13). Active–stative distinctions are lexicalized and are not always obvious, e.g., -kaj ‘burn.intr’, is active, but -pɨ'tu ‘breathe.intr’, is stative. Verbs agree in number and person with any subject whenever the object is third person, reflexive, or reciprocal which also is indexed on the verb (Table 14.12).

koinéization and language contact

401

Table 14.15 Old Tupi transitive verb with first-person Agent and second-person Patient subject+object index index oro1.act.sbj >2sg.objopo1.act.sbj >2pl.obj-

Verb root (regular, pluriform) pɨ'sɨk catch e'pʲak see

Table 14.16 Old Tupi transitive verb with second-person Agent and first-person Patient Verb object index se= 1sg.stat= ja'ne= 1pl.incl.stat= o're= 1pl.excl.stat=

Subject root (regular, pluriform) pɨ'sɨk catch r-e'pʲak r2∖see

je'pe 2sg.ag; sbj peje'pe 2pl.ag; sbj

The phenomenon of person hierarchy refers to a relative ranking of discourse participants which determines which will be indexed. Thus, first-person forms rank higher than second-person forms, and both discourse participants outrank any third-person forms. So, in Table 14.14, the verb indexes the higher-ranking direct object with the third-person subject unindexed. In Table 14.15, the firstperson subject is historically the main index, although the second-person object is also referenced, as the indices have fused into a portmanteau morpheme. In Table 14.16, the higher-ranking first-person direct object is again indexed, and an independent pronoun expresses the lower-ranking second-person subject. Tables 14.17–14.18 show the Yẽgatú transitive active and intransitive stative paradigms. As in old Tupi, Yẽgatú active intransitives, e.g., su ‘go’, yãna ‘run’, wita´ ‘swim’, yeẽ ‘speak’, take the same subject indices as transitive verbs (Table 14.17). Old Tupi nouns, verbs, and postpositions could be ‘pluriform’, i.e., exhibit allomorphy in the form of grammaticalized root-initial mutations, depending on syntactic status. One form (r1, with /t-/, /s-/, or Ø-) occurs when a root is the argument of some predicate, as in (10). (10)

a. 't-uβ-a pi'ra-Ø o-Ø-'ʔu. r1∖father-nlzr fish-nlzr 3.act.sbj-3.obj-eat ‘His/her/its/someone’s father eats a/the fish.’

402

tom finbow and paul o’neill Table 14.17 Yẽgatú transitive verb with first-, second-, or third-person Agent and Patient (Subject) (ixé 1sg) (yãdé 1pl) ('ĩdé 2sg) (peyẽ 2pl) (aé 3sg) ([aĩ]tá 3pl) kuyã ‘woman’ (n)

Verb subject index a- 1sg.actya- 1pl.actre- 2sg.actpe- 2pl.actu- 3sg.actta(u)3pl.actu- 3sg.act-

Object root

(-yu-) (rflx/rcip)

pisíka ‘catch’ xipiá ‘look at’

ixé 1sg yãdé 1pl 'ĩdé 2sg peyẽ 2pl aé 3sg [aĩ]tá 3pl pirá ‘fish’ (n)

b. pi'ra-Ø 't-uβ-a o-Ø-'ʔu. fish-nlzr r1∖father-nlzr 3.act.sbj-3.obj-eat ‘A/the fish eats his/her/its/someone’s father.’ c. Ø-'pe-Ø i=po'rãŋ. r1∖path-nlzr 3.stat.sbj=be.beautiful. ‘A/the path is beautiful.’ d. ku'jã-mu'ku Ø-u'ru-Ø o[w]-er(o)-'ur. woman-long r1∖vessel-nlzr 3.act.sbj-caus.com-come ‘A/the young woman brought a/the vessel.’ Another form (r2, with a prothetic /r-/, /rV-/, or /rVC-/) appears under three conditions: (i) when a pluriform root is immediately preceded by a first- or second-person stative/object pronoun, (11a); (ii) when a first- or second-person possessive pronoun or a noun possessor immediately precedes the pluriform root (11b, d); or (iii) with a first- or second-person pronoun or noun complement of a pluriform postposition (11c). (11)

a. se=/ja'ne=/o're=/ne=/pe= r-epʲak pi'ra-Ø 1sg=/1pl.incl=/1pl.excl=/2sg=/2pl.stat= r2∖see fish-nlzr ‘A/the fish sees me/us/you.’ b. se=/ja'ne=/o're=/ne=/pe=/ku'jã-Ø 1sg.poss=/1pl.incl.poss=/1pl.excl.poss=/2sg.poss=/2pl.poss=/ woman-nlzr 'r-uβ-a pi'ra-Ø o-Ø-'ʔu r2∖father-nlzr fish-nlzr 3-p-eat ‘My/our/your/the woman’s father eats a/the fish.’ c. se=/ja'ne=/o're=/ne=/pe=/ku'jã-Ø 1sg.stat=/1pl.incl.stat=/1pl.excl.stat=/2sg.stat=/2pl.stat=/ woman-nlzr r-e'se r2∖about ‘About me/us/you/a/the woman.’

koinéization and language contact

403

d. ɨ'[ɣ]ar-a ra-'pe-Ø. canoe-nlzr[poss] r2∖path-nlzr ‘path of the canoe(s).’ e. ku'jã-mu'ku ja'ne=rep-u'ru-Ø woman-long 1pl.incl-poss=r2∖vessel-nlzr o[w]-er(o)-ur. 3.act.sbj-caus.com-come ‘A/the young woman brought our vessel.’ A third form (r3, with /s(V)(C)-/ or /t(V)(C)-/) appears in the absence of any immediate antecedent, i.e., when the referent is a third-person pronoun or a noun elsewhere in the discourse (12a–e): (12)

a. a'βa s-orɨβ. person r3∖be.happy ‘A/the person is happy.’ b. s-orɨβ. r3∖be.happy ‘He/she/it/someone is happy.’ (NB a'βa r-orɨβ-a (person r2∖be.happy-nlzr) ‘a/the person’s happiness’.) c. t-uβ. r3∖have.a.father ‘He/she/it/someone has a father.’ d. s-e'se. r3∖about ‘About him/her/it/someone.’ o-s-apek. e. kunumĩ s-o'ʔo-Ø boy r3∖flesh-nlzr 3.act.sbj-r3.obj-toast ‘A/the boy toasts the/some meat’ (lit. ‘something’s flesh’, as flesh is inalienably possessed). (NB pluriform transitive verbs take /-s-/ (r3.obj) as their object index, rather than regular /-i-, -jo-/.)

Table 14.17 shows Yẽgatú lacks person hierarchy. Although Yẽgatú still uses special forms for the subjects of stative predicates, for the objects of postpositions, and for possession, it no longer uses these stative/possessive forms for direct object marking, as old Tupi did (see Tables 14.12, 14.13, and 14.15). Instead, like vernacular Brazilian Portuguese, Yẽgatú has one set of personal pronouns to express nominative and accusative case, distinguished largely by position, i.e., preverbal = subject, postverbal = object. Old Tupi transitive predicates exhibited agent and patient indices (Table 14.12). Yẽgatú transitive and active intransitive verbs have a single personal prefix that agrees with the subject in number and person, like Brazilian Portuguese.

404

tom finbow and paul o’neill Table 14.18 Yẽgatú stative verbs

The Tupi–Guarani active–stative distinction survives (Table 14.18), although considerably reduced. Yẽgatú’s inherited inflected stative predicates form a closed class under pressure from the active intransitives and a class of uninflected stative verbs, absent in old Tupi, that grows through borrowing Portuguese adjectives (Cruz 2011:193).

14.4.1.3 Grammaticalization of rãm A much-commented feature of Tupi–Guarani languages is ‘nominal tense’ (see for instance Guasch 1956; Gregores and Sua´rez 1967; Bertinetto 2020), in which two originally stative roots are compounded onto nominal phrases: 'pʷer ‘former’, ‘ex’, ‘denatured’ (cf. Pt. caduco ‘decrepid, ancient’; Taylor 1985; Cruz 2011:179) expresses ‘past’ (13) and 'rãm ‘prospective’ expresses ‘future’ (14)–(15) and also deontic modality. (13)

A-roɨ̯'rõ se=r-eko-'pʷer-a 1sg.act.sbj-hate 1sg.poss=r2∖custom-former-nmlz ‘I detest my former customs.’ (Anchieta, Poemas, 104; Navarro 2013:408)

(14)

A-'ju=ne i'se pe=r-emi-ʔu-'rãm-a 1sg.act.sbj-come=pot;irr 1sg 2pl.poss=r2∖res.p.nmlz-eat-fut-nmlr ‘I shall come, your [future] food!’ (Staden, Duas Viagens, 87, cit. Navarro 2008:138, §183)

koinéization and language contact

405

Old Tupi 'pʷer > Yẽgatu =kuéra, had similar functions (15) or was fossilized (16). (15)

kuyã ména Ø kuéra woman[poss] husband cop denatured[pst] < ku'jã- Ø mẽn-a woman.nlzr[poss] husband-nlzr i='pʷer 3sg.sbj.stat=denatured[pst] ‘The woman’s husband is no more.’

(16)

a. kãwéra ‘bone’


third person hierarchy to textbook nominative–accusative. Likewise, the two languages’

412

tom finbow and paul o’neill

symmetrical personal pronoun paradigms, with argument roles distinguished primarily by position relative to the verb and agreement only with subjects, is clearly a case of Brazilian Portuguese, itself the result of intense intralinguistic and interlinguistic contact, impinging on the Tupi–Guarani distinctions. Similar pressures are evident in the rapid growth of the go-future and the emergence of loaned pre-main verb modal auxiliaries (§14.4.1.4). Nevertheless, the process is not straightforwardly linear but can often exhibit circularities—for instance, the loss of o're seems to have stimulated the creation of periphrastic yãdé payẽ ‘we-all’. Converting 'rãm ‘nominal future’ into a final subordinator and a benefactive/dative postposition may have eased the emergence of future su from the go-periphrases extant in the Rio Negro’s Arawak substrate and Ibero-Romance superstrate. The latter has itself become increasingly adstratic to Yẽgatú as the predominant means of daily communication for many Yẽgatú speakers outside the home and thus provides the main source of linguistic innovation. At all times, the constant interplay of pragmatic concerns drives individuals’ adaptive responses to their own communicative needs, as they negotiate their way around an intensely multilingual environment. In the case of Yẽgatú, the result is a hybrid language with a predominantly Tupi– Guarani lexicon that retains many key features of Tupi–Guarani morphosyntax, e.g., active and stative verb classes; pluriform root allomorphy; inalienable possession marking of body parts, family members and some objects; a postverbal relative complementizer; postpositions; incorporation of main verbs into enclitic modal and aspectual expressions. Nevertheless, as we have seen, Yẽgatú parallels vernacular Brazilian Portuguese closely in many other grammatical features, and the ongoing trends in bilingualism and language shift favour increased structural closeness. Increasingly fine-grained investigations will doubtless uncover many more subtle associations, such as those revealed in the koinéization of the far better-known Portuguese varieties. However, the underlying processes are likely to be broadly similar, as, in a statistical analysis of more than 2,000 languages, Lupyan and Dale (2010) found that morphological complexity was strongly correlated with sociohistorical and demographic factors relating to the number of speakers, geographical extension of the language and the degree of language contact (see also Nettle 2012). To a very great extent, therefore, one might argue that, in practice, since all languages are collections of idiolects brimming with variation, all languages are emergent phenomena, and as such are morphosyntactically hybrid and koinéized to some degree, and that it is precisely speakers’ on-the-hoof selections from among the different structures available to express themselves that both provide the grist for the mill of language change and power its wheels. Thus, contact lies at the very heart of language and is central to linguistic investigation.

References and bibliographical abbreviations Aboh, Enoch O. (2015). The Emergence of Hybrid Grammars: Language Contact and Change. Cambridge: Cambridge University Press. Ackerman, Farrell and Gregory Stump (2004). ‘Paradigms and periphrastic expression: a study in realization-based lexicalism’. In Louisa Sadler and Andrew Spencer (eds), Projecting Morphology. Stanford, CA: CSLI Publications, pp. 111–157. Acquaviva, Paolo (2002). ‘Il plurale in -a come derivazione lessicale’. Lingue e linguaggio 2: 295–326. Acquaviva, Paolo (2005). ‘The morphosemantics of transnumeral nouns’. In Geert Booij, Emiliano Guevara, Angela Ralli, Salvatore Sgroi, and Sergio Scalise (eds), Morphology and Linguistic Typology: On-line Proceedings of the Fourth Mediterranean Morphology Meeting (MMM4), Catania, 21–23 September 2003, pp. 251–265. Available at: http:// www.lilec.it/mmm/wp/wp-content/uploads/2012/09/251-265-Acquaviva-MMM4.pdf. Acquaviva, Paolo (2008). Lexical Plurals: A Morphosemantic Approach. Oxford: Oxford University Press. Adams, J. N. (1991). ‘Some neglected evidence for Latin habeo with infinitive: the order of the constituents’. Transactions of the Philological Society 89: 131–196. Adams, J. N. (2013). Social Variation and the Latin Language. Cambridge: Cambridge University Press. Adams, James and Wolfgang de Melo (2016). ‘Ad versus the dative: from early to late Latin’. In J. N. Adams and Nigel Vincent (eds), Early and Late Latin: Continuity or Change? Cambridge: Cambridge University Press, pp. 87–131. Aguero-Bautista, Calixto (1998). ‘Cyclic and identity effects in Spanish diminutives and augmentatives’. MS, Massachusetts Institute of Technology. Aguiar, Joana and Maria da Conceição de Paiva (2017). “‘Vocês tenham cuidado, sois educadas para isso”: second person pronouns in Braga speech’. In Pilar Barbosa, Maria da Conceição de Paiva, and Celeste Rodrigues (eds), Studies on Variation in Portuguese. Amsterdam: Benjamins, pp. 136–152. Aguiar, José Lourenço Costa (1898). Doutrina christã destinada aos naturaes do amazonas em nhihingatu (com traducção portugueza em face). Petrópolis, RJ: Pacheco, Silva & Cº. Aikhenvald, Alexandra Y. (1998). ‘Warekena’. In Desmond C. Derbyshire and Geoffrey K. Pullum (eds), Handbook of Amazonian Languages. Vol. 4. Berlin: Mouton de Gruyter, pp. 225–440. Aikhenvald, Alexandra Y. (2008). The Manambu Language of East Sepik, Papua New Guinea. Oxford: Oxford University Press. AIS = Karl Jaberg and Jakob Jud (1928–1940). Sprach- und Sachatlas Italiens und der Su¨dschweiz. Zo¨fingen: Ringier. Albano Leoni, Federico, Rosanna Sornicola, Eleonora Stenta Krosbakken, and Carolina Stromboli (eds) (2001). Dati empirici e teorie linguistiche, Atti del XXXIII Congresso della Società di Linguistica Italiana, Napoli, 28–30 ottobre 1999. Rome: Bulzoni. Alboiu, Gabriela and Virginia Motapanyane (2000). ‘The generative approach to Romanian grammar: an overview’. In Virginia Motapanyane (ed.), Comparative Studies in Romanian Syntax. Amsterdam: Elsevier, pp. 1–48.

414

references and bibliographical abbreviations

Aleza Izquierda, Milagros (1987). ser con participio de perfecto en construcciones activas no oblicuas (español medieval). València: Universitat de València. Alibèrt, Loı´s (1935). Gramatica occitana segón los parlars lengadocians. Toulouse: Societat d’estudis occitans. Alinei, Mario (2009). ‘Sull’etimologia di andare (e affini)’. Bollettino dell’Atlante lessicale degli antichi volgari italiani 2: 9–15. Allières, Jacques (1954). ‘Un exemple de polymorphisme phonétique: le polymorphisme de l’–s implosif en gascon garonnais’. Via Domitia 1: 69–103. Ambrosini, Riccardo (1978). ‘Il e lo nell’italiano, soprattutto antico’. Linguistica e letteratura 3: 9–33. Amenta, Luisa (2006). ‘La perifrasi aviri a/da + infinito nel siciliano contemporaneo. Analisi di un campione di dati dell’ALS’. Rivista italiana di dialettologia 30: 59–74. Amenta, Luisa (2010). ‘Perifrasi verbali in siciliano’. In Jacopo Garzonio (ed.), Studi sui dialetti della Sicilia (Quaderni di lavoro ASIt 11). Padua: Unipress, pp. 1–18. Amenta, Luisa and Nicolò Paesano (2010). ‘Strutture analitiche e sintetiche: modalità e temporalità nel siciliano contemporaneo’. Cuadernos de filologı´a italiana 17: 11–27. Amenta, Luisa and Erling Strudsholm (2002). ‘«Andare a + infinito» in italiano: parametri di variazione sincronici e diacronici’. Cuadernos de filologı´a italiana 9: 11–29. Amorim, Antônio Brandão de (1926). ‘Lendas em nheengatú e em português’. Revista do Instituto Histórico e Geogra´fico Brasileiro 100(154): 9–478. Anchieta, José de (1595). Arte de grammatica da lı´ngoa mais usada na costa do Brasil. Coimbra: Antonio de Mariz [5th edition, facsimile, Salvador, BA: EDUFBA, 2014]. Anchieta, José de (1730). Dia´logo da fé (ed. Armando Cardoso). São Paulo, SP: Loyola [2nd edition, 1988]. Anderson, Gregory D. S. (2006). Auxiliary Verb Constructions. Oxford: Oxford University Press. Anderson, John M. (2011). The Substance of Language. Vol. II: Morphology, Paradigms, and Periphrasis. Oxford: Oxford University Press. Anderson, Stephen R. (1982). ‘Where’s morphology?’. Linguistic Inquiry 13: 571–612. Anderson, Stephen R. (1992). A-morphous Morphology. Cambridge: Cambridge University Press. Anderson, Stephen R. (2008). ‘Phonologically conditioned allomorphy in the morphology of Surmiran (Rumantsch)’. Word Structure 1: 109–134. Anderson, Stephen R. (2011). ‘Stress-conditioned allomorphy in Surmiran (Rumantsch)’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 13–35. Anderson, Stephen R. (2013). ‘Stem alternations in Swiss Romance’. In Cruschina, Maiden, and Smith (eds), pp. 8–23. Anderson, Stephen R. (2018). ‘A short history of morphological theory’. In Audring and Masini (eds), pp. 19–33. Andrews, Avery D. and Christopher D. Manning (1999). Complex Predicates and Information Spreading in LFG. Stanford, CA: CSLI Publications. Andriani, Luigi (2017). The Syntax of the Dialect of Bari. Unpublished Ph.D. thesis, University of Cambridge. Available at: https://www.repository.cam.ac.uk/handle/ 1810/269405. Andriani, Luigi (2018). ‘Instability and change: a parametric approach to Barese auxiliary selection’. In Gabriela Pană Dindelegan, Adina Dragomirescu, Irina Nicula, and Alexandru Nicolae (eds), Comparative and Diachronic Perspectives on Romance Syntax. Newcastle: Cambridge Scholars, pp. 361–397.

references and bibliographical abbreviations

415

Andriani, Luigi, Kim Groothuis, and Giuseppina Silvestri (2020). ‘Pathways of grammaticalization in Italo-Romance’. In Adam Ledgeway and Ian Roberts (eds), Diachronic Romance Linguistics, special issue of Probus 32: 327–366. Anthonissen, Lynn, Astrid De Wit, and Tanja Mortelmans (2016). ‘Aspect meets modality: a semantic analysis of the German am-progressive’. Journal of Germanic Linguistics 28: 1–30. Apothéloz, Denis (2009). ‘La quasi-synonymie du passé composé et du passé surcomposé dit “régional”’. Pratiques 141/142: 98–120. Apothéloz, Denis (2010). ‘Le passé surcomposé et la valeur de parfait existentiel’. Journal of French Language Studies 20: 105–126. Araújo, Antônio de (1686). Catecismo brası´lico da doutrina cristaã (2nd edition, corrected by Bartolomeu de Leão). Lisbon: Miguel Deslandes [facsimile, ed. Júlio Platzmann. Leipzig: Teubner, 1898]. Archangeli, Diana (1988). ‘Aspects of underspecification theory’. Phonology 5: 183–207. Arduin, Joana (2005). A variação dos pronomes possessivos de segunda pessoa do singular teu/seu na região sul do Brasil. Unpublished dissertação de mestrado, Universidade Federal de Santa Catarina. Arnauld, Antoine and Claude Lancelot (1660). Grammaire générale et raisonnée, etc. Paris: Le Petit. Aronoff, Mark (1992). ‘Stems in Latin verbal morphology’. In Mark Aronoff (ed.) Morphology Now. Albany, NY: SUNY Press, pp. 5–32. Aronoff, Mark (1994). Morphology by Itself: Stems and Inflectional Classes. Cambridge, MA: MIT Press. Arroyo Vega, Paloma (2001). La dia´tesis verbal en el castellano del siglo XV. València: Universitat de València. Arroyo Vega, Paloma (2003). ‘La dia´tesis en castellano del siglo XV de la Corona de Aragón’. In Francisco Sanchez Miret (ed.), Actas del XXIII Congreso internacional de lingu¨´ıstica y filologı´a roma´nica, Salamanca, 24–30 de septiembre de 2001. Vol. 2, tomo 1. Berlin: de Gruyter, pp. 69–75. Artés, Eduard (2016). The Influence of Phonology on Inflection: the interplay between syllabification and lexical insertion in Pallarese Catalan. Unpublished doctoral thesis, Universitat Autònoma de Barcelona. Arvinte, Vasile (1993). ‘Construcții perifrastice în Biblia de la București (1688), în ms. 45 și ms. 4389’. Studii și cercetări lingvistice 44: 177–181. Ascoli, Graziadio Isaia (1896). ‘Un problema di sintassi comparata dialettale’. Archivio glottologico italiano 14: 453–468. Ascoli, Graziadio Isaia (1901). ‘Appendice all’articolo “Un problema di sintassi comparata dialettale”’. Archivio glottologico italiano 15: 221–225. Ashdowne, Richard (2007). Names and Addresses: aspects of address in Latin and Romance. Unpublished D.Phil. thesis, University of Oxford. Audring, Jenny and Francesca Masini (eds) (2018). The Oxford Handbook of Morphological Theory. Oxford: Oxford University Press. Avram, Larisa (1999). Auxiliaries and the Structure of Language. Bucharest: Editura Universității din București. Avram, Larisa and Virginia Hill (2007). ‘An irrealis be auxiliary in Romanian’. In Raúl Aranovich (ed.), Split Auxiliary Systems: A Cross-linguistic Perspective. Amsterdam: Benjamins, pp. 47–64.

416

references and bibliographical abbreviations

Ayres-Bennett, Wendy and Janice Carruthers (1992). “‘Une regrettable et fort disgracieuse faute de français”? The description and analysis of the French surcomposés from 1530 to the present day’. Transactions of the Philological Society 90: 219–257. Bach, Xavier (2019). The Origins of Inflectional Classes. Unpublished D.Phil. thesis, University of Oxford. Bachrach, Asaf and Michael Wagner (2007). ‘Syntactically driven cyclicity vs output– output correspondence: the case of adjunction in diminutive morphology’. University of Pennsylvania Working Papers in Linguistics 10.1. Badia i Margarit, Antoni M. (1994). Gramàtica Històrica Catalana. Valencia: Tres i Quatre. Baena, Antônio Luis M. ([1831] 1901). Representação ao Conselho Geral da provı´ncia do Para´ sobre a especial necessidade de um novo regulamento promotor da civilização dos ´ındios da mesma provı´ncia. Belém, PA: ABAPP 2, pp. 241–292. Baerman, Matthew (2009). ‘Case syncretism’. In Andrew Spencer and Andrej Malchukov (eds), The Oxford Handbook of Case. Oxford: Oxford University Press, pp. 219–230. Baerman, Matthew (2015a). ‘Paradigmatic deviations’. In Baerman (ed.), pp. 141–159. Baerman, Matthew (ed.) (2015b). The Oxford Handbook of Inflection. Oxford: Oxford University Press. Baerman, Matthew, Dunstan Brown, and Greville G. Corbett (2017). Morphological Complexity. Cambridge: Cambridge University Press. Baker, Mark (1985). ‘The Mirror Principle and Morphosyntactic Explanation’. Linguistic Inquiry 16: 373–415. Baker, Mark C. (2010). The Syntax of Agreement and Concord. Cambridge: Cambridge University Press. Baños Baños, José Miguel (2000). ‘Vulgarismos sinta´cticos en Plauto II: quae ad patrem uis nuntiari (Cap. 360)’. In Benjamı´n Garcı´a-Herna´ndez (ed.), Latı´n vulgar y tardı´o: homenaje a Veikko Va¨a¨na¨nen (1905–1997). Madrid: Ediciones Cla´sicas, pp. 1–15. Barcelo, Gérard Jean (2003). ‘L’occitan contemporain et l’expression verbo-temporelle du futur: la proximité de la langue d’oc avec le français, son opposition avec le catalan’. In Hervé Lieutard and Marie-Jeanne Verny (eds), Nouvelle recherche en domaine occitan. Montpellier: Presses universitaires de la Méditerranée, pp. 167–184. Barros, Maria Cândida Drummond M. de (2003). ‘Notas sobre a polı´tica jesuı´tica da lı´ngua geral na Amazônia (séculos XVII–XVIII)’. In José Ribamar Bessa Freire and Maria Carlota Rosa (eds), Lı´nguas gerais: polı´tica linguı´stica e catequese na América do Sul no perı´odo colonial. Rio de Janeiro, RJ: EDUERJ, pp. 85–112. Barros, Maria Cândida Drummond M. de, Luis Carlos Borges, and Ma´rcio Meira (1996). ‘Lı´ngua geral como uma identidade construı´da’. Revista de antropologia 39: 191–219. Battaglia, Salvatore and Vincenzo Pernicone (1965). La grammatica italiana. Turin: Loescher. Baudouin de Courtenay, Jan (1972). A Baudouin de Courtenay Anthology: The Beginnings of Structural Linguistics, ed. and trans. Edward Stankiewicz. Bloomington, IN: Indiana University Press. Bauer, Brigitte (1995). The Emergence and Development of SVO Patterning in Latin and French. Oxford: Oxford University Press. Bauer, Brigitte (2006). “‘Synthetic” vs “Analytic” in Romance: the importance of varieties’. In Randall Gess and Deborah Arteaga (eds), Historical Romance Linguistics: Retrospective and Perspectives. Amsterdam: Benjamins, pp. 287–304. Beard, Robert (1995). Lexeme–Morpheme Base Morphology: A General Theory of Inflection and Word Formation. Albany, NY: SUNY Press.

references and bibliographical abbreviations

417

Beavers, John (2008). ‘Scalar complexity and the structure of events’. In Johannes Do¨lling and Tatjana Heyde-Zybatow (eds), Event Structures in Linguistic Form and Interpretation. Berlin: Mouton de Gruyter, pp. 245–265. Beavers, John (2011). ‘On affectedness’. Natural Language & Linguistic Theory 29: 335–370. Beavers, John (2013). ‘Aspectual classes and scales of change’. Linguistics 51: 681–706. Benincà, Paola (1983). ‘Il clitico “a” nel dialetto padovano’. In Paola Benincà, Manlio Cortelazzo, Aldo Prosdocimi, Laura Vanelli, and Alberto Zamboni (eds), Scritti Linguistici in Onore di Giovan Battista Pellegrini. Pisa: Pacini, pp. 25–35. Benincà, Paola (1983–1984). ‘Un’ipotesi sulla sintassi delle lingue romanze medievali’. Quaderni patavini di linguistica 4: 3–19 [reprinted in Paola Benincà, La variazione sintattica. Bologna: il Mulino, 1994, pp. 177–194]. Benincà, Paola (1993). ‘Sintassi’. In Alberto Sobrero (ed.), Introduzione all’italiano contemporaneo. Le strutture. Bari: Laterza, pp. 256–263. Benincà, Paola (1997). ‘Conjunctions’. In Maiden and Parry (eds), pp. 131–136. Benincà, Paola (2007) ‘Clitici e ausiliari: gh ò, z è’. In Delia Bentley and Adam Ledgeway (eds), Sui dialetti italoromanzi. Saggi in onore di Nigel B. Vincent (special supplement number 1 to The Italianist 27). King’s Lynn: Biddles, pp. 27–47. Benincà, Paola (2014). ‘Friulian linguistics’. In Rosa Mucignat (ed.), The Friulian Language: Identity, Migration, Culture. Newcastle: Cambridge Scholars, pp. 30–52. Benincà, Paola and Guglielmo Cinque (1992). ‘Sur l’ambiguïté structurale des verbes météorologiques en italien’. In Liliane Tasmowski-De Ryck and Anne Zribi-Hertz (eds), De la Musique à la linguistique. Ghent: Communication and Cognition, pp. 154–162. Benincà, Paola, Adam Ledgeway, and Nigel Vincent (eds) (2014). Diachrony and Dialects: Grammatical Change in the Dialects of Italy. Oxford: Oxford University Press. Benincà, Paola, Mair Parry, and Diego Pescarini (2016). ‘The dialects of northern Italy’. In Ledgeway and Maiden (eds), pp. 185–205. Benincà, Paola and Laura Vanelli (1982). ‘Appunti di sintassi veneta’. In Manlio Cortelazzo (ed.), Guida ai dialetti veneti IV. Padua: CLEUP, pp. 7–38. Bentein, Klaas (2011). ‘Towards the identification of verbal periphrasis in ancient Greek: a prototype analysis’. Acta Classica 54: 1–25. Bentley, Delia (1997). ‘Modalità e futuro nel siciliano antico e moderno’. In Mari D’Agostino (ed.), Aspetti della variabilità. Palermo: Centro di studi filologici e linguistici siciliani, pp. 49–66. Bentley, Delia (2013). ‘Subject canonicality and definiteness effects in Romance there sentences’. Language 89: 675–712. Bentley, Delia (2018). ‘Grammaticalization of subject agreement on evidence from ItaloRomance’. Linguistics 56: 1245–1301. Bentley, Delia (2020a). ‘On thetic broad focus’. Studii și cercetări lingvistice 71: 5–23. Bentley, Delia (2020b). ‘Breaking down the subject function: VS constructions in ItaloRomance’. Paper presented at the Romance Linguistics Seminar, Research Centre for Romance Linguistics, University of Oxford, 23 January 2020. Bentley, Delia and Francesco Maria Ciconte (2016). ‘Copular and existential constructions’. In Ledgeway and Maiden (eds), pp. 847-859. Bentley, Delia, Francesco Maria Ciconte, and Silvio Cruschina (2015). Existentials and Locatives in Romance Dialects of Italy. Oxford: Oxford University Press. Bentley, Delia and Silvio Cruschina (2018). ‘The silent argument of broad focus: typology and predictions’. Glossa: a journal of general linguistics 3(1): p. 118. doi: https://doi.org/ 10.5334/gjgl.677.

418

references and bibliographical abbreviations

Bentley, Delia and Thórhallur Eythórsson (1999). “‘Have” is not BE: counterarguments to the decompositional account’. MS, University of Manchester. Bentley, Delia and Thórhallur Eythórsson (2001). ‘Alternation according to person in ItaloRomance’. In Laurel J. Brinton (ed.), Historical Linguistics 1999. Selected Papers from the 14th International Conference on Historical Linguistics, Vancouver, 9–13 August 1999. Amsterdam: Benjamins, pp. 63–74. Benucci, Franco (2001). ‘Aspect prefixes in verbal periphrases in Italian and other Romance languages’. In Cinque and Salvi (eds), pp. 67–96. Berchem, Theodor (1968). ‘Considérations sur le parfait périphrastique VADO + infinitif en catalan et gallo-roman’. In Antonio Quilis, Ramón Blanco Carril, and Margarita Cantarero (eds), Actas del XI Congreso Internacional de Lingu¨´ıstica y Filologı´a Roma´nicas. Vol. 3. Madrid: Revista de filologı´a española, anejo 8, pp. 1159–1170. Berea-Găgeanu, Elena (1972). ‘Observații privind structura și evoluția viitorului în limba română’. Limba română 21: 503–515. Berizzi, Mariachiara (2013). ‘The preposition ma in the central Italian dialects’. In Diego Pescarini (ed.), Studi sui dialetti delle Marche (Quaderni di lavoro ASIt 15). Padua: Unipress, pp. 1–21. Bernini, Giuliano (2012). ‘Il clitico a nell’italo-romanzo settentrionale: osservazioni metodologiche’. In Vincenzo Orioles (ed.), Per Roberto Gusmani. 2: Linguistica storica e teorica. Studi in ricordo. Udine: Forum, pp. 269–282. Bertinetto, Pier Marco (1987). ‘Structure and origin of the “narrative” imperfect’. In Anna Giacalone Ramat, Onofrio Carruba, and Giuliano Bernini (eds), Papers from the 7th International Conference on Historical Linguistics. Amsterdam: Benjamins, pp. 71–86. Bertinetto, Pier Marco (1990). ‘Perifrasi verbali italiane, criteri di identificazione e gerarchie di perifrasticità’. In Giuliano Bernini and Anna Giacalone Ramat (eds), La temporalità nell’acquisizione di lingue seconde. Milano: Franco Angeli, pp. 331–350. Bertinetto, Pier Marco (2000). ‘The progressive in Romance, as compared with English’. In Östen Dahl (ed.), Eurotyp. Vol. 6: Tense and Aspect in the Languages of Europe. Berlin: de Gruyter Mouton, pp. 559–604. Bertinetto, Pier Marco (2020). ‘On nominal tense’. Linguistic Typology 24: 311–352. Bertinetto, Pier Marco and Mario Squartini (1995). ‘An attempt at defining the class of “gradual completion” verbs’. In Pier Marco Bertinetto, Valentina Bianchi, James Higginbotham, and Mario Squartini (eds), Temporal Reference, Aspect and Actionality, 1: Semantic and Syntactic Perspectives. Turin: Rosenberg & Sellier, pp. 11–26. Bertinetto, Pier Marco and Mario Squartini (2016). ‘Tense and aspect’. In Ledgeway and Maiden (eds), pp. 937–953. Bertocci, Davide and Silvia Rossi (2011). ‘Perifrasi aspettuali ed azionali in alcune varietà piemontesi’. In Jacopo Garzonio (ed.), Studi sui dialetti del Piemonte (Quaderni di lavoro ASIt 13). Padua: Unipress, pp. 49–70. Bettendorff, Joam Phelippe (João Felipe/Johann Philipp) (1687 [sic]). Compêndio da doutrina cristã na lı´ngua portuguesa e brası´lica, etc. Lisboa: Offic. de Miguel Deslandes. Reimpresso por Frei José Mariano da Conceição Vellozo, Lisboa: Offic. de Simão Thaddeu Ferreira, 1800. Bianchi, Valentina (1993). ‘Subject positions and e-positions’. Quaderni del laboratorio di linguistica 7. Pisa: Scuola Normale Superiore, pp. 51–69. Biondelli, Bernardino (1853). Saggio sui dialetti galloitalici. Milan: Bernardoni [Anastatic reprint, Bologna: Forni, 1970].

references and bibliographical abbreviations

419

Birchall, Joshua (2014). ‘Verbal argument marking patterns in South American languages’. In Loretta O’Connor and Pieter Muysken (eds), The Native Languages of South America. Cambridge: Cambridge University Press, pp. 223–249. Blake, Barry J. (2001). Case (2nd edition). Cambridge: Cambridge University Press. Blanche-Benveniste, Claire (1990). Le Français parlé: études grammaticales. Paris: CNRS. Blasco Ferrer, Eduardo (1984). Grammatica storica del catalano e dei suoi dialetti con speciale riguardo all’Algherese.Tu¨bingen: Narr. Blasco Ferrer, Eduardo (2005). ‘Medieval Occitan va(i) + infinitive reconsidered’. In Dominique Billy and Ann Buckley (eds), Études de langue et de littérature médiévales offertes à Peter T. Ricketts à l’occasion de son 70ème anniversaire. Turnhout: Brepols, pp. 595–602. Blevins, James (2015). ‘Inflectional paradigms’. In Baerman (ed.), pp. 87–111 Blevins, James P. (2016). Word and Paradigm Morphology. Oxford: Oxford University Press. Blevins, James, Farrell Ackerman, and Rob Malouf (2018). ‘Word and paradigm morphology’. In Audring and Masini (eds), pp. 265–284. Bobaljik, Jonathan (2017). ‘Distributed Morphology’. Oxford Research Encyclopedia of Linguistics. doi: 10.1093/acrefore/9780199384655.013.131. Bollée, Annegret and Philippe Maurer (2016). ‘Creoles’. In Ledgeway and Maiden (eds), pp. 447–466. Bonami, Olivier (2015). ‘Periphrasis as collocation’. Morphology 25: 63–110. Bonet, Eulàlia and Francesc Torres-Tamarit (2009). ‘Allomorphy in pre-clitic imperatives in Formenteran Catalan: an output-based analysis’. In Sonia Colina, Antxon Olarrea, and Ana Maria Carvalho (eds), Romance Linguistics 2009: Selected Papers from the 39th Linguistic Symposium on Romance Languages (LSRL), Tucson, Arizona, March 2009. Amsterdam: Benjamins, pp. 337–352. Bonet, Eulàlia and Maria-Rosa Lloret (2005). ‘More on alignment as an alternative to domains: the syllabification of Catalan clitics’. Probus 17: 37–78. Bonet, Eulàlia, Maria-Rosa Lloret, and Joan Mascaró (2007). ‘Allomorph selection and lexical preferences: two case studies’. Lingua 117: 903–927. Bonvini, Emilio and Maria Margarida Tadoni Petter (1998). ‘Portuguais du Brésil et langues africaines’. Langages 130: 68–83. Booij, Geert (1994). ‘Against split morphology’. In Geert Booij and Jaap van Marle (eds), Yearbook of Morphology 1993. Dordrecht: Kluwer, pp. 27–49. Booij, Geert (1996). ‘Inherent versus contextual inflection and the split morphology hypothesis’. In Geert Booij and Jaap van Marle (eds), Yearbook of Morphology 1995. Dordrecht: Kluwer, pp. 1–16. Bo¨rjars, Kersti, Nigel Vincent, and Carol Chapman (1997). ‘Paradigms, periphrases and pronominal inflection: a feature-based account’. In Geert Booij and Jaap van Marle (eds), Yearbook of Morphology 1996. Dordrecht: Springer, pp. 155–180. Bossong, Georg (2016). ‘Classifications’. In Ledgeway and Maiden (eds), pp. 63–72. Bottasso, Enzo (ed.) (1953). ‘Comedia de l’homo’. In L’opera piacevole di Gian Giorgio Alione. Bologna: Palmaverde, pp. 1–35. Bourciez, Edouard (1956). Éléments de linguistique romane (4th edition). Paris: Klincksieck. Boxer, Charles R. (1973). Salvador de Sa´ e a luta pelo Brasil e Angola, 1602–1686, trans. Oliveiro de Oliveiro Pinto. São Paulo, SP: Editora Nacional/EDUSP Brandi, Luciana and Patrizia Cordin (1989). ‘Two Italian dialects and the null subject parameter’. In Jaeggli and Safir (eds), pp. 111–142. Bres, Jacques and Emmanuelle Labeau (2013a). ‘The narrative construction va + infinitive in Contemporary French’. Diachronica 30: 295–322.

420

references and bibliographical abbreviations

Bres, Jacques and Emmanuelle Labeau (2013b). ‘Aller et venir: des verbes de déplacement aux auxiliaires aspectuels-temporels-modaux’. Langue française 179: 13–28. Bres, Jacques and Emmanuelle Labeau (2015). ‘Venir de (+ infinitive): an immediate anteriority marker in French’. Diachronica 32: 530–570. Britain, David and Peter Trudgill (2005). ‘New dialect formation and contact-induced reallocation: three case studies from the English Fens’. International Journal of English Studies 5: 183-209. Broselow, Ellen (2015). ‘The typology of position–quality interactions in loanword vowel insertion’. In Yuchau E. Hsiao and Lian-Hee Wee (eds), Capturing Phonological Shades within and across Languages. Newcastle: Cambridge Scholars, pp. 292–319. Brown, Dunstan (2007). ‘Peripheral functions and overdifferentiation: the Russian second locative’. Russian Linguistics 31: 61–76. Brown, Dunstan, Marina Chumakina, and Greville Corbett, G. (eds) (2013). Canonical Morphology and Syntax. Oxford: Oxford University Press Brown, Dunstan, Marina Chumakina, Greville Corbett, Gergana Popova, and Andrew Spencer (2012). ‘Defining “periphrasis”: key notions’. Morphology 22: 233–275. Brown, Dunstan and Andrew Hippisley (2012). Network Morphology: A Defaults-based Theory of Word Structure. Cambridge: Cambridge University Press. Brown, Josh (2020). ‘Language history from below: standardization and koinéization in Renaissance Italy’. Journal of Historical Sociolinguistics 6: 1–27. Brucale, Luisa and Egle Mocciaro (2019). ‘Possession and volition in the development of modal meanings: a case-study from Sicilian’. Journal of Contemporary Philology 2: 27–47. Buridant, Claude (2000). Grammaire nouvelle de l’ancien français. Paris: SEDES. Burzio, Luigi (1986). Italian Syntax. Dordrecht: Reidel. Burzio, Luigi and Elvira DiFabio (1994). ‘Accentual stability’. In Mazzola (ed.), pp. 19–34. Butt, Miriam (1997). ‘Complex predicates in Urdu’. In Alex Alsina, Joan Bresnan, and Peter Sells (eds), Complex Predicates. Stanford, CA: CSLI Publications, pp. 107–149. Bybee, Joan L. and Mary A. Brewer (1980). ‘Explanation in morphophonemics: changes in Provençal and Spanish preterite forms’. Lingua 52: 201–242. Bybee, Joan and William Pagliuca (1985). ‘Cross-linguistic comparison and the development of grammatical meaning’. In Jacek Fisiak (ed.), Historical Semantics and Historical Word-Formation. Berlin: Mouton de Gruyter, pp. 59–83. Bybee, Joan L., William Pagliuca, and Revere Perkins (1991). Back to the future. In E. Traugott and B. Heine (eds) Approaches to Grammaticalization, Vol. II. Amsterdam: John Benjamins. 17-58. Bybee, Joan L., Revere Perkins, and William Pagliuca (1994). The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago: University of Chicago Press. Cabral, Ana Suelly Arruda Câmara, and Gabriel Barros Viana de Oliveira (2013). ‘Sobre possı´veis influências bilateriais entre a lı´ngua geral amazônica falada no Baixo Amazonas e as lı´nguas tupı´-guaranı´ do subramo VIII, com foco especial na lı´ngua zo’é’. Anais do SILEL—Simposio Internacional de Letras e Linguı´stica 3: 1–13. Cabral, Ana Suelly Arruda Câmara, Aryon Dall’Igna Rodrigues, and Fa´bio Bonfim Duarte (eds) (2010). Lı´nguas e culturas tupi. Vol. 2. Campinas, SP: Biblioteca Curt Nimuendajú/Brası´lia, DF: Laboratório de Lı´nguas Indı´genas, Universidade de Brası´lia. Cabré, Teresa and Maria del Mar Vanrell (2011). ‘Accent i entonació en els vocatius de l’alguerès’. In La toponomàstica de les illes del Mediterrani occdental (XXXV Coŀloqui de la Societat d’Onomàstica, 10–11 maig de 2008, en L’Alguer). Valencia: Denes, pp. 331–341.

references and bibliographical abbreviations

421

Caha, Pavel (2018). ‘Notes on insertion in Distributed Morphology and Nanosyntax’. In Lena Baunaz, Karen De Clercq, Liliane Haegeman, and Eric Lander (eds), Exploring Nanosyntax. Oxford: Oxford University Press, pp. 57–87. Calabrese, Andrea (2008). ‘On absolute and contextual syncretism: remarks on the structure of paradigms and on how to derive it’. In Asaf Bachrach and Andrew Nevins (eds), Inflectional Identity. Oxford: Oxford University Press, pp. 156–205. Calabrese, Andrea (2012). ‘Allomorphy in the Italian passato remoto: a Distributed Morphology analysis’. Language and Information Society 18: 1–75. Calabrese, Andrea (2013). ‘The irregular forms of the Italian “Passato Remoto”: a synchronic and diachronic analysis’. In Sergio Bauuw, Frank Drijkoningen, Luisa Meroni, and Manuela Pinto (eds), Romance Languages and Linguistic Theory 2011. Selected Papers from ‘Going Romance’ Utrecht 2011. Amsterdam: Benjamins, pp. 17–58. Calabrese, Andrea (2015). ‘Irregular morphology and athematic verbs in Italo-Romance’. Isogloss 2: 61–93. Calabrese, Andrea (2016). ‘On the morphophonology of metaphonic alternations in Altamurano’. In Francesc Torres-Tamarit, Kathrin Linke, and Marc Van Oostendorp (eds), Approaches to Metaphony in the Languages of Italy. Berlin: de Gruyter Mouton, pp. 89–126. Calabrese, Andrea (2019). ‘Morphophonological investigations: a theory of PF. From syntax to phonology in Italian and Sanskrit verb forms’. MS, University of Connecticut. Calabrese, Andrea (2020a). ‘Remarks on the role of the perfect participle in Italian morphology and on its history’. Probus 32: 209–269. Calabrese, Andrea (2020b). ‘The morphosyntax of andative forms in the Campiota vernacular. The synthetic behavior of restructuring roots’. MS, University of Connecticut. Campbell, Lyle (2001). ‘What’s wrong with grammaticalization?’. Language Sciences 23: 113–161. Camproux, Charles (1958). Étude syntaxique des parlers gévaudanais. Paris: Presses universitaires de France. Canale, Michael, Raymond Mougeon, and Monique Bélanger (1978). ‘Analogical leveling of the auxiliary être in Ontarian French’. In Margarita Suñer (ed.), Contemporary Studies in Romance Linguistics. Washington, DC: Georgetown University Press, pp. 41–61. Cappellaro, Chiara (2018). ‘Genesis and diachronic persistence of overabundance: data from Romance languages’. In Antje Dammel, Matthias Eitelmann, and Mirjam Schmuck (eds), Reorganising Grammatical Variation: Diachronic Studies in the Retention, Redistribution and Refunctionalisation of Linguistic Variants. Amsterdam: Benjamins, pp. 119–148. Caragiu-Marioțeanu, Matilda (1969). ‘Verbul’. In Ion Coteanu (ed.), Istoria limbii române. Vol. 2. Bucharest: Editura Academiei, pp. 254–278. Cardeira, Esperança A. (2009). ‘Revisitando a periodização do português: o português médio’. Revista eletrônica de linguı´stica 3: 103–120 Cardeira, Esperança A. and Maria Alice Fernandes (2008). ‘O português medieval: koinização e elaboração’. Veredas 9: 155–176. Carden, Guy and David Pesetsky. (1977). ‘Double-verb constructions, markedness, and a fake co-ordination’. In Woodford A. Beach, Samuel E. Fox, and Shulamith Philosoph (eds), Papers from the Thirteenth Regional Meeting of the Chicago Linguistic Society, pp. 82–92. Chicago: Chicago Linguistic Society. Cardinaletti, Anna (1997). ‘Subjects and clause structure’. In Lilianne Haegeman (ed.), The New Comparative Syntax. London: Longman, pp. 33–63.

422

references and bibliographical abbreviations

Cardinaletti, Anna (2004). ‘Toward a cartography of subject positions’. In Luigi Rizzi (ed.), The Structure of CP and IP (The Cartography of Syntactic Structures. Vol. 2). Oxford: Oxford University Press, pp. 115–165. Cardinaletti, Anna and Verner Egerland (2010). ‘I pronomi personali e riflessivi’. In Salvi and Renzi (eds), pp. 414–450. Cardinaletti, Anna and Giuliana Giusti (2001). ‘Semi-lexical motion verbs in Romance and Germanic’. In Norbert Corver and Henk Van Riemsdijk (eds), Semi-Lexical Categories: On the Function of Content Words and the Content of Function Words. Berlin: Mouton de Gruyter, pp. 371–414. Cardinaletti, Anna and Giuliana Giusti (2003). ‘Motion verbs as functional heads’. In Christina Tortora (ed.), The Syntax of Italian Dialects. Oxford: Oxford University Press, pp. 31–49. Cardinaletti, Anna and Giuliana Giusti (2019). ‘Multiple agreement in southern Italian dialects’. In Ludovico Franco and Paolo Lorusso (eds), Linguistic Variations: Structure and Interpretation. Studies in Honor of M. Rita Manzini. Berlin: de Gruyter, pp. 125–147. Cardinaletti, Anna and Lori Repetti (2007). ‘Vocali epentetiche nella morfologia dell’italiano e dei dialetti italiani’. In Roberta Maschi, Nicoletta Penello, and Piera Rizzolatti (eds), Miscellanea di studi linguistici offerti a Laura Vanelli da amici e allievi padovani. Udine: Forum, pp. 116–126. Cardinaletti, Anna and Lori Repetti (2008). ‘The phonology and syntax of preverbal and postverbal subject clitics in northern Italian dialects’. Linguistic Inquiry 39: 523–563. Cardinaletti, Anna and Lori Repetti (2010). ‘Proclitic vs enclitic pronouns in northern Italian dialects and the Null-Subject Parameter’. In D’Alessandro, Ledgeway, and Roberts (eds), pp. 119–134. Carratala´, Ernesto (1980). Morfosintaxis del castellano actual. Barcelona: Labor. Carruthers, Janice (1994). ‘The passé surcomposé régional: towards a definition of its function in contemporary spoken French’. Journal of French Language Studies 4: 171–190. Carruthers, Janice (1996). ‘The passé surcomposé général: on the relationship between a rare tense and discourse organization’. Romance Philology 50: 183–200. Carstairs, Andrew (1987). Allomorphy in Inflection. London: Croom Helm. Carstairs-McCarthy, Andrew (1994). ‘Inflection classes, gender, and the principle of contrast’. Language 70: 737–788. Casasnovas, Afonso (2006). Noções de lı´ngua geral ou nheengatú: grama´tica, lendas e vocabula´rio (2nd edition). Manaus, AM: UFAM/Faculdade Salesiana Dom Bosco. Castellani, Arrigo (1976). I più antichi testi italiani. Bologna: Pàtron. Cennamo, Michela (1997). ‘Passive and impersonal constructions’. In Maiden and Parry (eds), pp. 145–161. Cennamo, Michela (2001). ‘L’inaccusatività in alcune varietà campane: teorie e dati a confronto’. In Albano Leoni, Sornicola, Krosbakken, and Stromboli (eds), pp. 427–453. Cennamo, Michela (2002). ‘La selezione degli ausiliari perfettivi in napoletano antico: fenomeno sintattico o sintattico-semantico?’. Archivio glottologico italiano 87: 175–222. Cennamo, Michela (2008). ‘The rise and development of analytic perfects in Italo-Romance’. In Eythórsson (ed.), pp. 115–142. Cennamo, Michela (2010). ‘Perfective auxiliaries in the pluperfect in some southern Italian dialects’. In D’Alessandro, Ledgeway, and Roberts (eds), pp. 210–224.

references and bibliographical abbreviations

423

Cennamo, Michela and Antonella Sorace (2007). ‘Auxiliary selection and split intransitivity in Paduan. Variation and lexical-aspectual constraints’. In Raúl Aranovich (ed.), Split Auxiliary Systems: A Cross-linguistic Perspective. Amsterdam: Benjamins, pp. 65–99. Champion, James Joseph (1978). The Periphrastic Futures Formed by the Romance Reflex of Vado (ad) Plus Infinitive. Chapel Hill, NC: Atkins Library. Chumakina, Marina and Greville G. Corbett (eds) (2012). Periphrasis: The Role of Syntax and Morphology in Paradigms. Oxford: Oxford University Press/British Academy. CICA = Corpus Informatitzat del Català Antic. Available at: http://www.cica.cat/ index.php. Cinque, Guglielmo (1999). Adverbs and Functional Heads: A Cross-linguistic Perspective. Oxford: Oxford University Press. Cinque, Guglielmo (2004). ‘Restructuring and functional structure’. In Adriana Belletti (ed.), Structures and Beyond: The Cartography of Syntactic Structures, Volume 3. Oxford: Oxford University Press, pp. 132–191. Cinque, Guglielmo (2006). Restructuring and Functional Heads: The Cartography of Syntactic Structures, Volume 4. Oxford: Oxford University Press. Cinque, Guglielmo and Giampaolo Salvi (eds) (2001). Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi. Amsterdam: Elsevier. Ciorănescu, Alexandru (2002). Dicționar etimologic al limbii române. Bucharest: Saeculum. Clackson, James (2016). ‘Latin as a source for the Romance languages’. In Ledgeway and Maiden (eds), pp. 3–13. Cocchi, Gloria (1994). ‘Il problema della selezione dell’ausiliare nei tempi composti’. In Gianluigi Borgato (ed.), Teoria del linguaggio e analisi linguistica. XX Incontro di Grammatica Generativa. Padua: Unipress, pp. 179–194. Cocchi, Gloria (1995). La selezione dell’ausiliare. Padua: Unipress. Colantoni, Laura and Jeffrey Steele (2004). ‘Length cues to phonemic voicing in Romance stop-liquid clusters’. Paper presented at the Annual Conference of the Canadian Linguistic Association, Winnipeg, May 2004. Colella, Gianluca (2020). ‘Le perifrasi verbali’. In Dardano, Maurizio (ed.), Sintassi dell’italiano antico II. La prosa del Duecento e del Trecento. La frase semplice. Rome: Carocci, pp. 234–258. CoLFIS = Bertinetto, Pier Marco, Cristina Burani, Alessandro Laudanna, Lucia Marconi, Daniela Ratti, Claudia Rolando, and Anna Maria Thornton (2005). Corpus e Lessico di Frequenza dell’Italiano Scritto (CoLFIS). Available at: http://linguistica.sns.it/CoLFIS/ Home.htm [consulted via EsploraCoLFIS: Bambini, Valentina and Marco Trevisan. 2012. EsploraCoLFIS: Un’interfaccia Web per ricerche sul Corpus e Lessico di Frequenza dell’Italiano Scritto Quaderni del Laboratorio di Linguistica della Scuola Normale Superiore 11: 1–16. http://linguistica.sns.it/esploracolfis/home.htm]. Collins, Chris and Richard Kayne (2020). ‘Towards a theory of morphology as syntax’. MS, New York University. Colón, Germà (1978a). ‘El perfet perifràstic català “va + infinitiu”’. In Germà Colon (ed.), La llengua catalana en els seus textos, Vol. II. Barcelona: Curial, pp. 119–130. Colón, Germà (1978b). ‘Sobre el perfet perifràstic ‘vado + infinitiu’, en català, en provençal i en francès’. In Germà Colón (ed.), La llengua catalana en els seus textos. Vol. II. Barcelona: Curial, pp. 131–174. Company Company, Concepción (1983). ‘Sintaxis y valores de los tiempos compuestos en el español medieval’. Nueva revista de filologı´a hispa´nica 32: 235–257. Comrie, Bernard (1976). Aspect. Cambridge: Cambridge University Press.

424

references and bibliographical abbreviations

Comrie, Bernard (1981). ‘On Reichenbach’s approach to tense’. In Roberta A. Hendrick, Carrie S. Masek, and Mary Frances Miller (eds), Papers from the Seventeenth Regional Meeting of the Chicago Linguistic Society, April 30–May 1, 1981. Chicago: University of Chicago Press, pp. 24–30. Comrie, Bernard (1985). Tense. Cambridge: Cambridge University Press. Confais, Jean-Paul (1995). Temps, mode, aspect: les approches des morphèmes verbaux et leurs problèmes à l’exemple du français et de l’allemand. Toulouse: Presses universitaires du Mirail. Corbett, Greville (2005). ‘The canonical approach in typology’. In Zygmunt Frajzyngier, Adam Hodges, and David S. Rood (eds), Linguistic Diversity and Language Theories. Amsterdam: Benjamins, pp. 25–49. Corbett, Greville (2006). Agreement. Cambridge: Cambridge University Press. Corbett, Greville (2007a). ‘Canonical typology, suppletion and possible words’. Language 83: 8–42. Corbett, Greville (2007b). ‘Deponency, syncretism, and what lies between’. In Matthew Baerman, Greville Corbett, Dunstan Brown, and Andrew Hippisley (eds), Deponency and Morphological Mismatches. Oxford: Oxford University Press, pp. 21–43. Corbett, Greville G. (2009). ‘Canonical inflectional classes’. In Fabio Montermini, Gilles Boyé, and Jesse Tseng (eds), Selected Proceedings of the 6th Décembrettes. Sommerville, MA: Cascadilla Proceedings Project, pp. 1–11. www.lingref.com,document#2231. Corbett, Greville (2012). ‘Periphrasis and possible lexemes’. In Marina Chumakina and Greville Corbett (eds), Periphrasis. The Role of Syntax and Morphology in Paradigms. Oxford: British Academy / Oxford University Press, 169-190. Corbett, Greville G. (2015). ‘Morphosyntactic complexity: a typology of lexical splits’. Language 91: 145–193. Corbett, Greville G. (2019). ‘Pluralia tantum nouns and the theory of features: a typology of nouns with non-canonical number properties’. Morphology 29: 51–108. CORDE = Corpus Diacrónico del Español. Available at: http://corpus.rae.es/cordenet.html. Cordin, Patrizia (1988). ‘I pronomi personali: caratteristiche generali’. In Lorenzo Renzi, Giampaolo Salvi, and Anna Cardinaletti (eds), Grande grammatica italiana di consultazione. Vol. I: La frase. I sintagmi nominale e preposizionale. Bologna: il Mulino, pp. 535–549. Cornilescu, Alexandra (1995). ‘Rumanian genitive constructions’. In Guglielmo Cinque and Giuliana Giusti (eds), Advances in Rumanian Linguistics. Amsterdam: Benjamins, pp. 1– 54. Cornilescu, Alexandra (2000). ‘The double subject construction in Romanian’. In Virginia Motapanyane (ed.), Comparative Studies in Romanian Syntax. Amsterdam: Elsevier, pp. 83–133. Cornu, Maurice (1953). Les Formes surcomposées en français. Bern: Francke. Coromines, Joan (1988). Diccionari etimologic i complementari de la llengua catalana. Barcelona: Curial. Corr, Alice (2016). ‘Wide-focus subject–verb inversion in Ibero-Romance: a locative account’. Glossa: a journal of general linguistics 1(1), p. 11. doi: https://doi.org/ 10.5334/gjgl.85. Coseriu, Eugenio (1971). Essai d’une nouvelle typologie des langues romanes. Sinaia: Universitatea din București. Coseriu, Eugenio (1987). ‘Le latin vulgaire et le type linguistique roman (à propos de la thèse de Humboldt: “Es sanken Formen, nicht aber die Form”)’. In József Herman (ed.),

references and bibliographical abbreviations

425

Latin vulgaire—latin tardif. Actes du Ier colloque international sur le latin vulgaire et tardif (Pécs, 2–5 septembre 1985). Tu¨bingen: Niemeyer, pp. 53–64. Coseriu, Eugenio (1988). ‘Der romanische Sprachtypus: Versuch einer neuen Typologisierung der romanischen Sprachen’. In Jo¨rn Albrecht (ed.), Energeia und Ergon: Sprachliche Variation, Sprachgeschichte, Sprachtypologie. Band I, Schriften von Eugenio Coseriu (1965–1987). Tu¨bingen: Narr, pp. 207–224. Costa, Frederico Aguiar (1909). ‘Elementos necessa´rios para aprender o yẽgatú’. In Frederico Aguiar Costa, Carta pastoral de D. Frederico Costa, Bispo do Amazonas a seus amados diocesanos. [Fortaleza,] CE: Minerva, pp. 157–248. Costa, Lairson Barbosa da (2013). Variação dos pronomes ‘tu’/‘você’ nas capitais do norte. Unpublished dissertação de mestrado, Universidade Federal do Para´. Craig, Colette (1991). ‘Ways to go in Rama: a case study of polygrammaticalization’. In Elizabeth Closs Traugott and Bernd Heine (eds), Approaches to Grammaticalization. Vol. II. Amsterdam: Benjamins, pp. 455–492. Croft, William (1990). Typology and Universals. Cambridge: Cambridge University Press. Cruschina, Silvio (2013). ‘Beyond the stem and inflectional morphology: an irregular pattern at the level of periphrasis’. In Cruschina, Maiden, and Smith (eds), pp. 262–283. Cruschina, Silvio (in press). ‘Gone unexpectedly: pseudo-coordination and the expression of surprise in Sicilian’. In Vincenzo Nicolò Di Caro, Giuliana Giusti, and Daniel Ross (eds), Pseudo-Coordination and Multiple Agreement Constructions. Amsterdam: Benjamins. Cruschina, Silvio and Valentina Bianchi (2021). ‘Mirative implicatures at the syntax– semantics interface: a surprising association and an unexpected move’. In Andreas Trotzke and Xavier Villalba (eds), Expressive Meaning across Linguistic Levels and Frameworks. Oxford: Oxford: University Press, pp. 86–107. Cruschina, Silvio and Andrea Calabrese (2021). ‘Fifty shades of morphosyntactic microvariation: motion verb constructions in southern Italian dialects’. In Marc-Olivier Hinzelin, Natascha Pomino, and Eva-Maria Remberger (eds), Formal Approaches to Romance Morphosyntax: Linking Variation to Theory. Berlin: de Gruyter, pp. 145–198. Cruschina, Silvio, Adam Ledgeway, and Eva-Maria Remberger (eds) (2019). Italian Dialectology at the Interfaces. Amsterdam: Benjamins. Cruschina, Silvio, Martin Maiden, and John Charles Smith (eds) (2013). The Boundaries of Pure Morphology: Diachronic and Synchronic Perspectives. Oxford: Oxford University Press. Cruschina, Silvio and Melanie Rinollo (2013). ‘Between subject enclitic and agreement marker: the second person endings in Sicilian’. Transactions of the Philological Society 111: 259–273. Cruz, Aline (2011). Fonologia e grama´tica do yẽgatú: a lı´ngua geral falada pelos povos baré, warekene e baniwa. Utrecht: LOT. Cruz, Aline (2015). ‘The rise of number agreement in Yẽgatú’. Boletim do Museu Paraense Emı´lio Goeldi. Ciências Humanas 10: 419–439. CTILC = Corpus Textual Informatitzat de la Llengua Catalana. Available at: https:// ctilc.iec.cat/scripts/. Culbertson, Jennifer (2010). ‘Convergent evidence for categorial change in French: from subject clitic to agreement marker’. Language 86: 85–132. Cusic, David D. (1981). Verbal Plurality and Aspect. Unpublished Ph.D. dissertation, Stanford University.

426

references and bibliographical abbreviations

Cuzzolin, Pierluigi (2010). ‘A proposito di trent’anni di “Latino classico”’. Aion: Annali del Dipartimento di Studi del Mondo Classico e del Mediterraneo Antico. Sezione Linguistica 30: 299–341. Cuzzolin, Pierluigi (2013). ‘Ma esistono lingue “classiche”?’ In Nicola Grandi (ed.), Nuovi dialoghi sulle lingue e sul linguaggio. Bologna: Pàtron, pp. 99–111. D’Alessandro, Roberta (2021). ‘Syntactic change in contact: Romance’. Annual Review of Linguistics 7: 309–328. D’Alessandro, Roberta and Adam Ledgeway (2010). ‘The Abruzzese T-v system: feature spreading and the double auxiliary construction’. In D’Alessandro, Ledgeway, and Roberts (eds), pp. 201–209. D’Alessandro, Roberta, Adam Ledgeway, and Ian Roberts (eds) (2010). Syntactic Variation: The Dialects of Italy. Cambridge: Cambridge University Press. D’Alessandro, Roberta, and Ian Roberts (2010). ‘Past participle agreement in Abruzzese: split auxiliary selection and the null-subject parameter’. Natural Language & Linguistic Theory 28: 41–72. D’Alessandro, Roberta and Marc Van Oostendorp (2016). ‘Prosody, phi-features and deixis in southern Italian: what vocatives can tell us on the architecture of language’. In Ernestina Carrilho, Alexandra Fiéis, Maria Lobo, and Sandra Pereira (eds), Romance Languages and Linguistic Theory 10. Selected Papers from ‘Going Romance’ 28, Lisbon. Amsterdam: Benjamins, pp. 61–82. Dahl, Östen (1985). Tense and Aspect Systems. Oxford: Blackwell. Daniel, João (1976). Tesouro descoberto no ma´ximo Rio Amazonas. Rio de Janeiro, RJ: Biblioteca nacional [MS.,1757–1776, Biblioteca nacional, Rio de Janeiro, RJ]. Dardel, Robert de (1999). ‘L ’origine du génitif-datif ’. Vox romanica 58: 26–56. Darrigrand, Robert (1974). Initiation au gascon. Orthez: Per Noste. Davies, William and Carol Rosen (1988). ‘Unions as multi-predicate clauses’. Language 64: 52–88. Davis, Stuart (1990). ‘Italian onset structure and the distribution of il and lo’. Linguistics 28: 43–55 DeGraff, Michel (2008). ‘Morphology and word order in “creolization” and beyond’. In Guglielmo Cinque and Richard S. Kayne (eds), The Oxford Handbook of Comparative Syntax. Oxford: Oxford University Press, pp. 293–372. Del Prete, Fabio and Giuseppina Todaro (2020). ‘Building complex events: the case of Sicilian Doubly Inflected Construction’. Natural Language & Linguistic Theory 38: 1–41. De Mauro, Tullio, Federico Mancini, Massimo Vedovelli, and Miriam Voghera (1993). Lessico di frequenza dell’italiano parlato. Milan: Etas Libri. Denis, Derek and Sali A. Tagliamonte (2018). ‘The changing future: competition, specialization and reorganization in the contemporary English future temporal reference system’. English Language and Linguistics 22: 403–430. Densusianu, Ovid (1938). Histoire de la langue roumaine. Vol. 2: Le Seizième Siècle. Paris: Ernest Leroux. Densusianu, Ovid (1961). Istoria limbii române. Vol. 2: Secolul al XVI-lea. Bucharest: Editura Științifică. Deshaies, Denise and Éve Laforge (1981). ‘Le futur simple et le futur proche dans le français parlé dans la ville de Québec’. Langues et linguistiques 7: 23–37. Detges, Ulrich (2004). ‘How cognitive is grammaticalization? The history of the Catalan perfet perifràstic’. In Olga Fischer, Muriel Norde, and Harry Peridon (eds), Up and Down the Cline: The Nature of Grammaticalization. Amsterdam: Benjamins, pp. 211–227.

references and bibliographical abbreviations

427

Detges, Ulrich (2020). ‘Future markers in Western Romance: cyclic change, synchronic variation and diachronic competition’. Journal of Historical Pragmatics 21: 289–314. Devos, Maud and Jenneke van der Wal (eds) (2014). COME and GO off the Beaten Grammaticalization Path. Berlin: de Gruyter. D’Hulst, Yves, Martine Coene, and Larisa Avram (2004). ‘Syncretic and analytic tenses in Romanian: the Balkan setting of Romance’. In Olga Mišeska Tomic´ (ed.), Balkan Syntax and Semantics. Amsterdam: Benjamins, pp. 355–374. Di Caro, Vincenzo Nicolò (2019a). Multiple Agreement Constructions in Southern ItaloRomance: the syntax of Sicilian pseudo-coordination. Unpublished doctoral thesis, Ca’ Foscari University of Venice. Di Caro, Vincenzo Nicolò (2019b). ‘The inflected construction in the dialects of Sicily: parameters of microvariation’. In Cruschina, Ledgeway, and Remberger (eds), pp. 63–78. Di Caro, Vincenzo Nicolò (2019c). ‘Perifrasi verbali deontiche e paradigmi difettivi nel dialetto di Delia’. Bollettino del Centro di studi filologici e linguistici siciliani 30: 217–244. Diémoz, Federica and Andres Kristol (2014). ‘L’Atlas linguistique audiovisuel du francoprovençal valaisan et les défis du polymorphisme’. In Fabio Tosques (ed.), 20 Jahre digitale Sprachgeographie. 20 Ans de géolinguistique numérique. 20 anni di geolinguistica digitale. 20 años de geografı´a lingu¨´ıstica digital. Berlin: Humboldt-Universita¨t zu Berlin, Institut fu¨r Romanistik, 163–183. Diez, Friedrich (1843). Grammatik der romanischen Sprachen. Dritter Theil. Bonn: Weber [5th edition, Bonn: Weber, 1882]. Di Fabio, Elvira G. (1990). The Morphology of the Verbal Infix /isk/ in Italian and in Romance. Unpublished Ph.D. dissertation, Harvard University. Dimitrescu, Florica (ed.) (1978). Istoria limbii române. Bucharest: Editura Didactică și Pedagogică. Di Vita, Rosario (1931). ‘Corruzione di vocaboli inglesi sicilianizzati’. Rivista italiana di letteratura dialettale 3: 266-273. Dobrovie-Sorin, Carmen (1994). The Syntax of Romanian. Berlin: Mouton de Gruyter. Dowty, David R. (1979). Word Meaning and Montague Grammar. Dordrecht: Reidel. Dragomirescu, Adina (2012). ‘Verbul’. In Gheorghe Chivu, Gabriela Pană Dindelegan, Adina Dragomirescu, Isabela Nedelcu, and Irina Nicula (eds), Studii de istorie a limbii române: morfosintaxa limbii literare în secolele al XIX-lea și al XX-lea. Bucharest: Editura Academiei Române, pp. 185–222. Dragomirescu, Adina (2014). ‘L’accord du participe passé en ancien roumain’. Linguistica Atlantica 33: 38–51. Dragomirescu, Adina (2016). ‘The past participle and the participial clause’. In Pană Dindelegan (ed.), pp. 259–270. Dragomirescu, Adina and Alexandru Nicolae (2013). ‘Urme ale selecției auxiliarului de perfect compus în română’. In Oana Balaș and Coman Lupu (eds), Hommages offerts à Florica Dimitrescu et Alexandru Niculescu. Bucharest: Editura Universității din București, pp. 340–355. Dragorimescu, Adina and Alexandru Nicolae (2014). ‘The multiple grammaticalization of Romanian veni “come”: focusing on the passive construction’. In Devos and van der Wal (eds), pp. 69–100. Dressler, Wolfgang (1985). ‘On the definite Austrian and Italian articles’. In Edmund Gussman (ed.), Phono-Morphology: Studies in the Interaction of Phonology and Morphology. Lublin: Catholic University, pp. 36-47.

428

references and bibliographical abbreviations

Dressler, Wolfgang and Lavinia Merlini Barbaresi (1994). Morphopragmatics: Diminutives and Intensifiers in Italian, German, and Other Languages. Berlin: de Gruyter. Drude, Sebastian (2008). ‘Tense, aspect and mood in Aweti verb paradigms: analytic and synthetic forms’. In K. David Harrison, David S. Rood, and Arienne Dwyer (eds), Lessons from Documented Endangered Languages. Amsterdam: Benjamins, pp. 67–110. Duarte, Inés and Gabriela Matos (2000). ‘Romance clitics and the Minimalist Program’. In João Costa (ed.), Portuguese Syntax. Oxford: Oxford University Press, pp. 116–142. Du Cange, C. Du Fresne (1733). Glossarium mediae et infimae latinitatis. Niort: Favre. Eckardt, Regine (2006). Meaning Change in Grammaticalization: An Enquiry into Semantic Reanalysis. Oxford: Oxford University Press. Eckert, Penelope (2012). ‘Three waves of variation study: the emergence of meaning in the study of variation’. Annual Review of Anthropology 41: 87–100. Egerland, Verner and Michela Cennamo (2010). ‘Frasi subordinate all’infinito’. In Salvi and Renzi (eds), pp. 815–859. Einhorn, E. (1974). Old French: A Concise Handbook. Cambridge: Cambridge University Press. Elcock, W. D. (1960). The Romance Languages. London: Faber & Faber. Elson, Mark J. (1992). ‘A Latin source for the conditional auxiliary in Romanian’. Zeitschrift fu¨r romanische Philologie 108: 360–375. Embick, David (2000). ‘Features, syntax, and categories in the Latin perfect’. Linguistic Inquiry 31: 185–230. Embick, David (2007). ‘Blocking effects and analytic/synthetic alternations’. Natural Language & Linguistic Theory 25: 1–37. Embick, David and Maurice Halle (2005). ‘On the status of stems in morphological theory’. In Twan Geerts, Ivo van Ginneken, and Haike Jacobs (eds), Romance Languages and Linguistic Theory 2003. Amsterdam: Benjamins, pp. 37–62. Embick, David and Rolf Noyer (2001). ‘Movement operations after syntax’. Linguistic Inquiry 32: 555–595. Engel, Dulcie M. (1985). ‘The survival of the French passé simple: a reply to van Vliet’. Word 36: 77–81. Engel, Dulcie M. (1996). ‘L’expression du temps et la variation linguistique’. Revue romane 31: 215–233. Ernout, Alfred (1945). Morphologie historique du latin. Paris: Klincksieck. Erteschik-Shir, Nomi (1997). The Dynamics of Focus Structure. Cambridge: Cambridge University Press. Esher, Louise (2014). ‘Autonomous morphology and extramorphological coherence’. Morphology 24: 325–350. Fabricius-Hansen, Cathrine (1980). ‘Lexikalische Dekomposition, Bedeutungspostulate und wieder’. In Dieter Kastovsky (ed.), Perspektiven der lexikalischen Semantik. Bonn: Bouvier, pp. 26–40. Fabricius-Hansen, Cathrine (2001). “‘Wi(e)der” and “Again(st)”’. In Caroline Féry and Wolfgang Sternefeld (eds), Audiatur Vox Sapientiae: A Festschrift for Arnim von Stechow. Berlin: Akademie Verlag, pp. 101–130. Faggin, Giorgio (1985). Vocabolario della lingua friulana. Udine: Del Bianco. Faggin, Giorgio (1997). Grammatica friulana. Udine (Campoformido): Ribis. Faraco, Carlos Alberto (2001). ‘A questão da lı´ngua: revisitando Alencar, Machado de Assis e Cercanias’. Lı´nguas e instrumentos linguı´sticos 7: 33–51. Faraco, Carlos Alberto (2008). Norma culta brasileira: desatando alguns nós. Rio de Janeiro, RJ: Civilização Brasileira.

references and bibliographical abbreviations

429

Faraco, Carlos Alberto (2017). História sociopolı´tica da lı´ngua portuguesa. São Paulo, SP: Para´bola. Faraco, Carlos Alberto and Ana Maria S. Zilles (2017). Para conhecer norma linguı´stica. São Paulo, SP: Contexto. Faria, Francisco R. Corrêa de (1858). Compêndio da lı´ngua indı´gena brası´lica para uso dos que a elle se quizerem dedicar. Belém do Para´, PA: Santos Filho [2nd edition, Anais da Biblioteca e Arquivo Público do Para´, 1903, pp. 297–337]. Farkas, Donka F. (2002). ‘Varieties of indefinites’. Proceedings of SALT 12: 59–83. Ferrari, Franca (2005). A Syntactic Analysis of the Nominal Systems of Italian and Luganda: how nouns can be formed in the syntax. Unpublished Ph.D. dissertation, New York University. Ferreiro, Manuel (1999). Grama´tica histórica galega. I: Fonética e morfosintaxe. Santiago de Compostela: Laiovento. Ferrero, Carlo B. (1976). La bassa Russia. Turin: Viglongo. Fink, Robert O. (1972). ‘Person in nouns: Is the vocative a case?’. American Journal of Philology 93: 61–68. von Fintel, Kai and Sabine Iatridou (2008). ‘How to say ought in foreign: the composition of weak necessity modals’. In Jacqueline Guéron and Jacqueline Lecarme (eds), Time and Modality. Dordrecht: Springer, pp. 115–141. Fiorentino, Fernando (1998). Vernacolo salentino. Galatina: Congedo Editore. Fischer, Olga (2007). Morphosyntactic Change: Functional and Formal Perspectives. Oxford: Oxford University Press. Fleischman, Suzanne (1982). The Future in Thought and Language: Diachronic Evidence from Romance. Cambridge: Cambridge University Press. Fleischman, Suzanne (1990). Tense and Narrativity: From Medieval Performance to Modern Fiction. Austin, TX: University of Texas Press. Fleischman, Suzanne and Linda R. Waugh (eds) (1991). Discourse-Pragmatics and the Verb: Evidence from Romance. London: Routledge. Floricic, Franck (2011). ‘Le vocatif et la périphérie du système des cas: entre archaïsmes et innovations’. Mémoires de la Société de Linguistique de Paris, nouvelle série 19: 103–134. Fogarasi, Miklós (1969). Grammatica italiana del novecento: sistemazione descrittiva. Budapest: Tanko¨nyvkiadó. Formentin, Vittorio (2001). ‘L’ausiliazione perfettiva in antico napoletano’. Archivio glottologico italiano 86: 79–117. Formentin, Vittorio (2002). ‘Tra storia della lingua e filologia: note sulla sintassi della “Cronica” d’Anonimo romano’. Lingua e Stile 37: 203–250. Formentin, Vittorio (2004). ‘Quattro note sintattiche dal “Tristano Veneto”’. In Maurizio Dardano and Gianluca Frenguelli (eds), SintAnt: la sintassi dell’italiano antico. Atti del Convegno internazionale di studi (Università di Roma Tre, 18–21 settembre 2002). Rome: Aracne, pp. 175–196. Formentin, Vittorio (2020). ‘Forme verbali doppie negli antichi volgari italiani. Frammenti di una Stellungsregel italoromanza’. Lingua e stile 45: 183–228. Foulet, Lucien (1919). Petite Syntaxe de l’ancien français. Paris: Champion. Fradin, Bernard (2003). Nouvelles approches en morphologie. Paris: Presses universitaires de France. Fraenkel, Eduard (1932). ‘Kolon und Satz: Beobachtungen zur Gliederung des antiken Satzes I’. Nachrichten der Go¨ttinger Gesellschaft der Wissenschaften, Philosophischhistorische Klasse, 197–213.

430

references and bibliographical abbreviations

Fraenkel, Eduard (1933). ‘Kolon und Satz: Beobachtungen zur Gliederung des antiken Satzes II’. Nachrichten der Go¨ttinger Gesellschaft der Wissenschaften, Philosophischhistorische Klasse, 319–354. Fraenkel, Eduard (1965). ‘Noch einmal Kolon und Satz’. Bayerische Akademie der Wissenschaften, Philosophisch-historische Klasse, Sitzungsberichte 65(2): l–73. Frâncu, Constantin (1970). ‘Conjunctivul perfect românesc, privire diacronică’. Studii și cercetări lingvistice 21: 205–224. Frâncu, Constantin (1983–1984). ‘Geneza și evoluția timpurilor verbale supracompuse în limba română’. Anuar de lingvistică și istorie literară 29: 23–62. Frâncu, Constantin (2009). Gramatica limbii române vechi (1521–1780). Iași: Demiurg. Frâncu, Constantin (2010). Conjunctivul românesc și raporturile lui cu alte moduri. Iași: Demiurg Plus. Frâncu, Constantin (2012). ‘Despre formele verbale supracompuse de tipul am fost cântat, am fost cântând’. Analele Universității ‘Alexandru Ioan Cuza’ din Iași, Secțiunea III, Lingvistică 58: 115–125. Frâncu, Constantin (2014). ‘Despre unele forme supracompuse’. In Gheorghe Chivu and Oana Uță Bărbulescu (eds), Ion Coteanu, in memoriam. Bucharest: Editura Universității din București, pp. 145–156. Freire, José R. Bessa (2003). Rio Babel: a história das lı´nguas na Amazônia. Rio de Janeiro, RJ: EDUERJ. Freire, José R. Bessa and Maria Carlota Rosa (2003). Lı´nguas gerais: polı´tica linguı´stica e catequese na América do Sul no perı´odo colonial. Rio de Janeiro, RJ: EDUERJ. Friedman, Victor A. (1986). ‘Evidentiality in the Balkans: Bulgarian, Macedonian, and Albanian’. In Wallace Chafe and Johanna Nichols (eds), Evidentiality: The Linguistic Coding of Epistemology. Norwood, NJ: Ablex, pp. 168–187. Gaeng, Paul (1994). ‘À propos de la genèse des pluriels italiens et roumains du type cancîini: témoins d’une différenciation territoriale du latin de l’Empire’. In Gerold Hilty (ed.), Actes du XXe Congrès International de Linguistique et Philologie Romanes. Tome II, Section III: La Fragmentation linguistique de la Romania. Tu¨bingen: Francke, pp. 367–374. Galante, Maria (1997). Chartae Latinae Antiquiores. 2nd series. Ninth Century, 50: Italy 22. Dietikon-Zu¨rich: Urs Graf. Galante, Maria (1998). Chartae Latinae Antiquiores. 2nd series. Ninth Century, 52: Italy 24. Dietikon-Zu¨rich: Urs Graf. Garcia, Guilherme (2017). ‘Weight gradience and stress in Portuguese’. Phonology 34: 41– 79. Gardani, Francesco, Michele Loporcaro, and Alberto Giudici (eds) (in press). Romance Languages and the Others: In and around the Balkans. Special issue of Journal of Language Contact 14. Gardani, Francesco, Michele Loporcaro, and Alberto Giudici (eds) (in press). ‘In and around the Balkans: Romance languages and the making of multi-layered languages’. In Gardani, Loporcaro, and Giudici (eds). Garrapa, Luigia (2009). Vowel Elision in Florentine Italian. Unpublished doctoral dissertation, Universita¨t Konstanz. Garrapa, Luigia (2011). Vowel Elision in Florentine Italian. Bern: Lang. Gasca Queirazza, Giuliano (ed.) (1965). Documenti di antico volgare in Piemonte. I: Le ‘Recomendaciones’ del Laudario di Saluzzo. Turin: Bottega d’Erasmo. Gasca Queirazza, Giuliano (ed.) (1966). Documenti di antico volgare in Piemonte. II: Gli ordinamenti dei Disciplinati e dei Raccomandati di Dronero. Turin: Bottega d’Erasmo.

references and bibliographical abbreviations

431

Gatti, Tiziana (1990). Confronto tra fenomeni sintattici nell’italiano e nel dialetto trentino: participio passato, accordo e ausiliari. Unpublished tesi di laurea. University of Trento. Gentile, Salvatore (ed.) (1988). Giovanni Brancati, Vita e favole di Esopo. Volgarizzamento del secolo XV (Glossario di Rosa Franzese). Naples: Liguori. Gheorghe, Mihaela (2013). ‘The structure of complex clauses: subordination, argument clauses’. In Pană Dindelegan (ed.), pp. 466–473. Gheție, Ion (ed.) (1997). Istoria limbii române literare. Epoca veche (1532–1780). Bucharest: Editura Academiei Române. Giacalone Ramat, Anna (1995). ‘Sulla grammaticalizzazione di verbi di movimento: andare e venire + gerundio’. Archivio glottologico italiano 80: 168–203. Giacalone Ramat, Anna (2000). ‘On some grammaticalization patterns for auxiliaries’. In John Charles Smith and Delia Bentley (eds), Historical Linguistics 1995: Selected Papers from the 12th International Conference on Historical Linguistics, Manchester, August 1995. Vol. 1: General Issues and Non-Germanic Languages. Amsterdam: Benjamins, pp. 125– 154. Giacalone Ramat, Anna (2001). ‘Emergent auxiliaries and the theory of grammaticalization’. In Schaner-Wolles, Rennison, and Neubarth (eds), pp. 121–131. Giacalone Ramat, Anna and Andrea Sansò (2014). ‘Venire (“come”) as a passive auxiliary in Italian’. In Devos and van der Wal (eds), pp. 21–44. Gildea, Spike (1998). On Reconstructing Grammar: Comparative Cariban Morphosyntax. Oxford: Oxford University Press. Gildea, Spike (ed.) (2000). Reconstructing Grammar: Comparative Linguistics and Grammaticalization. Amsterdam: Benjamins. Gili Y Gaya, Samuel (1964). Curso superior de sintaxis española. Barcelona: Bibliograf. Giorgi, Alessandra (2010). About the Speaker: Towards a Syntax of Indexicality. Oxford: Oxford University Press. Giorgi, Alessandra and Fabio Pianesi (1997). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Giorgi, Alessandra and Fabio Pianesi (2004). ‘Complementizer deletion in Italian’. In Luigi Rizzi (ed.), The Structure of CP and IP. Oxford: Oxford University Press, pp. 190–210. Gisborne, Nikolas and Andrew Hippisley (2017). ‘Defaults in linguistics’. In Nikolas Gisborne and Andrew Hippisley (eds), Defaults in Morphological Theory. Oxford: Oxford University Press, pp. 1–17. Giusti, Giuliana and Rossella Iovino (2016). ‘Latin as a split-DP language’. Studia Linguistica 70: 221–249. Givón, Thomas J. [Talmy] (1971). ‘Historical syntax and synchronic morphology: an archeologist’s field trip’. In Papers from the Seventh Regional Meeting of the Chicago Linguistic Society, April 16–18, 1971. Chicago: University of Chicago Press, pp. 394–415. Goossens, Louis (1995). ‘Metaphtonymy: the interaction of metaphor and metonymy in figurative expressions for linguistic action’. In Louis Goossens, Paul Pauwels, Brygida Rudzka-Ostyn, Anne-Marie Simon-Vandenbergen, and Johan Vanparys (eds), By Word of Mouth: Metaphor, Metonymy and Linguistic Action in a Cognitive Perspective. Amsterdam: Benjamins, pp. 159–174. Gougenheim, Georges (1929). Étude sur les périphrases verbales de la langue française. Paris: Les Belles Lettres. Grandgent, C. H. (1905). An Outline of the Phonology and Morphology of Old Provençal. Boston: Heath.

432

references and bibliographical abbreviations

Grandi, Nicola (2015). ‘The place of evaluation within morphology’. In Nicola Grandi and Livia Ko¨rtvélyessy (eds), The Edinburgh Handbook of Evaluative Morphology. Edinburgh: Edinburgh University Press, pp. 74–90. Grandi, Nicola and Livia Ko¨rtvélyessy (2015). ‘Introduction: why evaluative morphology?’. In Nicola Grandi and Livia Ko¨rtvélyessy (eds), The Edinburgh Handbook of Evaluative Morphology. Edinburgh: Edinburgh University Press, pp. 3–20. Greenberg, Joseph (1991). ‘The last stages of grammatical elements; contractive and expansive desemanticization’. In Elizabeth Closs Traugott and Bernd Heine (eds), Approaches to Grammaticalization. Volume I. Amsterdam: Benjamins, pp. 301-315. Gregores, Emma, and Jorge Sua´rez. 1967. A Description of Colloquial Guaranı´. The Hague: Mouton. Grenand, Françoise and Epimonides H. Ferreira (1989). Pequeno diciona´rio da lı´ngua geral. Manaus: SEDUC. Grevisse, Maurice and André Goosse (2016). Le Bon Usage (16th edition). Brussels: De Boeck and Duculot. Griffiths, Bruce and Dafydd Glyn Jones (1995). Geiriadur yr Academi/The Welsh Academy Dictionary. Cardiff: University of Wales Press. Gro¨ber, Gustav (1877). ‘Grammatisches II: lo, li-il, i, im Altitalienischen’. Zeitschrift fu¨r romanische Philologie 1: 108–110. Guasch, Antonio (1956). El idioma guaranı´: grama´tica y antologı´a de prosa y verso. Asunción: Casa Américas. Guirao, Miguelina and Marı´a A. Garcı´a Jurado (1990). ‘Frequency of occurrence of phonemes in American Spanish’. Revue québécoise de linguistique 19: 135–149. Guy, Gregory R. (1981). Linguistic Variation in Brazilian Portuguese: aspects of phonology, syntax and language history. Unpublished Ph.D. dissertation, University of Pennsylvania. Guy, Gregory R. (1989). ‘On the nature ond origins of popular Brazilian Portuguese’. In Estudios sobre el español de América y linguı´stica afroamericana. Bogota´: Instituto Caro y Cuervo, 227-245. Haase, Martin (1995). ‘Tense, aspect and mood in Romanian’. In Rolf Thieroff (ed.), Tense Systems in European Languages II. Tu¨bingen: Niemeyer, pp. 135–165. Hacken, Pius ten (2014). ‘Delineating derivation and inflection’. In Rochelle Lieber and Pavol Štekauer (eds), The Oxford Handbook of Derivational Morphology. Oxford: Oxford University Press, pp. 10–25. Hajek, John (1997). ‘Emilia-Romagna’. In Maiden and Parry (eds), pp. 271–278. Hall, Nancy (2006). ‘Cross-linguistic patterns of vowel intrusion’. Phonology 23: 387–429. Hall, Robert A. Jr (1976). Proto-Romance Phonology. New York: Elsevier. Halle, Morris and Alec Marantz (1993). ‘Distributed Morphology and the pieces of inflection’. In Kenneth Hale and Samuel Jay Keyser (eds), The View from Building 20. Cambridge, MA: MIT Press, pp. 111–176. Halle, Morris and Alex Marantz (1994). ‘Some key features of Distributed Morphology’. MIT Working Papers in Linguistcs 21: 275–288. Harley, Heidi and Rolf Noyer (2003). ‘Distributed Morphology’. In Lisa Cheng and Rint Sybesma (eds), The Second Glot International State-of-the-Article Book: The Latest in Linguistics. Berlin: Mouton de Gruyter, pp. 462–496. Harris, Alice C. and Lyle Campbell (1995). Historical Syntax in Cross-linguistic Perspective. Cambridge: Cambridge University Press. Harris, Martin (1978). The Evolution of French Syntax: A Comparative Approach. London: Longman.

references and bibliographical abbreviations

433

Harris, Martin (1982). ‘The “past simple” and the “present perfect” in Romance’. In Nigel Vincent and Martin Harris (eds), Studies in the Romance Verb: Essays Offered to Joe Cremona on the Occasion of His 60th birthday. London: Croom Helm, pp. 42–70. Hartt, Charles Frederick (1872). ‘Notes on the Lingoa Geral or Modern Tupi of the Amazonas’. Transactions of the American Philological Association 3: 58–76. Haspelmath, Martin (2000). ‘Periphrasis’. In Geert E. Booij, Christian Lehmann, and Joachim Mugdan (eds) (in collaboration with Wolfgang Kesselheim and Stavros Skopeteas), Morphologie: ein internationales Handbuch zur Flexion und Wortbildung/ Morphology: An International Handbook on Inflection and Word-Formation. Berlin: Walter de Gruyter, pp. 654–664. Haspelmath, Martin (2002). Understanding Morphology. London: Hodder Arnold. Hay, Jennifer, Christopher Kennedy, and Beth Levin (1999). ‘Scalar structure underlies telicity in “Degree Achievements”’. In Tanya Matthews and Devon Strolovitch (eds), Proceedings of Semantic and Linguistic Theory IX, Ithaca, NY: CLC Publications, Cornell University, pp. 127–144. Heidinger, Steffen (2015). ‘The auxiliary selection of French monter “move upward” from the 16th to the 20th century’. In Rolf Kailuweit and Malte Rosemeyer (eds), Auxiliary Selection Revisited: Gradience and Gradualness. Berlin: de Gruyter, pp. 277–300. Heine, Bernd (1993). Auxiliaries: Cognitive Forces and Grammaticalization. Oxford: Oxford University Press. Heine, Bernd (1995). ‘Agent-oriented vs epistemic modality: some observations on German modals’. In Joan Bybee and Suzanne Fleischman (eds), Modality in Grammar and Discourse. Amsterdam: Benjamins, pp. 17–53. Heine, Bernd (1997). Possession: Cognitive Sources, Forces and Grammaticalization. Cambridge: Cambridge University Press. Heine, Bernd (2003). ‘Grammaticalization’. In Joseph and Janda (eds), pp. 575–601. Heine, Bernd, Ulrike Claudi and Friederike Hu¨nnemeyer (1991). ‘From cognition to grammar: evidence from African languages’. In Elizabeth Closs Traugott and Bernd Heine (eds), Approaches to Grammaticalization. Vol. 1. Amsterdam: Benjamins, pp. 149–187. Heine, Bernd and Tania Kuteva (2005). Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Hemming, John (1987). Red Gold: The Conquest of the Brazilian Indians. London: Papermac [1st edition, London: Macmillan, 1978]. Hemming, John (1995). Amazon Frontier: The Defeat of the Brazilian Indians. London: Papermac [1st edition, London: Macmillan, 1987]. Henrichsen, Arne-Johan (1966). ‘La périphrase anar + infinitif en ancien occitan’. In Iorgu Iordan (ed.), Omagiu lui Alexandru Rosetti la 70 de ani. Bucharest: Editura Academiei Republicii Socialiste România, pp. 357–363. Herman, József (1998). ‘La chronologie de la transition: un essai’. In József Herman and Luca Mondin (eds), La transizione dal latino alle lingue romanze. Tu¨bingen: Niemeyer, pp. 5–26. Hjelmslev, Louis (1928). Principes de grammaire générale. Copenhagen: Kgl. Danske Videnskabernes Selskab. Hoberman, Robert and Mark Aronoff (2003). ‘The verbal morphology of Maltese’. Language Acquisition and Language Disorders 28: 61–78. Hock, Hans Henrich (1986). Principles of Historical Linguistics. Berlin: Mouton de Gruyter. Hock, Hans Henrich and Brian D. Joseph (1996). Language History, Language Change, and Language Relationship: An Introduction to Historical and Comparative Linguistics. Berlin: Mouton de Gruyter.

434

references and bibliographical abbreviations

Hockett, Charles F. (1958). A Course in Modern Linguistics. New York: Macmillan. Hopper, Paul (1991). ‘On some principles of grammaticization’. In Elizabeth Closs Traugott and Bernd Heine (eds), Approaches to Grammaticalization. Vol. 1. Amsterdam: Benjamins, pp. 17–35. Hopper, Paul and Sandra Thompson (1980). ‘Transitivity in grammar and discourse’. Language 56: 251-299. Hopper, Paul J. and Elizabeth Closs Traugott (1993). Grammaticalization. Cambridge: Cambridge University Press. Hopper, Paul J. and Elizabeth Closs Traugott (2003). Grammaticalization (2nd edition). Cambridge: Cambridge University Press. Hornstein, Norbert (1990). As Time Goes By: Tense and Universal Grammar. Cambridge, MA: MIT Press. Humboldt, Wilhelm von (1836). Über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluß auf die geistige Entwickelung des Menschengeschlechts. Berlin: Druckerei der Ko¨niglichen Akademie der Wissenschaften. Hummel, Martin (2013). ‘Sincronı´a y diacronı´a de los llamados adjetivos adverbializados y de los adverbios en -mente’. Anuario de Letras 1: 215–281. Hummel, Martin (2014). ‘The adjective–adverb interface in Romance and English’. In Petra Sleeman, Freek Van De Velde, and Harry Perridon (eds), Adjectives in Germanic and Romance. Amsterdam: Benjamins, pp. 35–72. IEC = Institut d’estudis catalans (2016). Gramàtica de la llengua catalana. Barcelona: Institut d’estudis catalans. Iliescu, Maria (1972). Le Frioulan (à partir des dialectes parlés en Roumanie). The Hague: Mouton. Iliescu, Maria (2008). Miscellanea Romanica (1956–2007). Cluj-Napoca: Clusium/Scriptor. Iorga Mihal, Ana-Maria (2013). ‘The indirect object’. In Pană Dindelegan (ed.), pp. 143– 150. Iovino, Rossella and Silvia Rossi (2014), ‘Alcune riflessioni sul “vocativo inverso” tra pragmatica e sintassi’. In Gianna Marcato (ed.), Le mille vite del dialetto. Padua: CLEUP, pp. 219–226. Ivănescu, Gheorghe (1980/2000). Istoria limbii române. Iași: Junimea. Jacobs, Bart (2011). ‘Present and historical perspectives on the Catalan go-past’. Zeitschrift fu¨r Katalanistik 24: 227–255. Jacobs, Bart and Hans Peter Kunert (2014). ‘Whatever happened to the Occitan go-past? Insights from the dialects of Gascony and Guardia Piemontese’. Revue romane 49: 177– 203. Jaeggli, Osvaldo and Kenneth J. Safir (eds) (1989). The Null Subject Parameter. Dordrecht: Kluwer. Janda, Richard D., and David Kathman (1992). ‘Shielding morphology from exploded INFL’. In Jeanette Marshall Denton, Grace P. Chan, and Costas P. Canakis (eds), Papers from the 28th Regional Meeting of the Chicago Linguistic Society 1992. Vol. 2: The Parasession: The Cycle in Linguistic Theory. Chicago: Chicago Linguistic Society, pp. 141–157. Jeanroy, Alfred (1941). ‘Le roman de Jaufré’. Annales du Midi 53: 212, 363–390. Jendraschek, Gerd (2014). ‘Future tense, prospective aspect, and irrealis mood as part of the situation perspective: insights from Basque, Turkish, and Papuan’. In Philippe De Brabantier, Mikhail Kissine, and Saghie Sharifzadeh (eds), Future Times, Future Tenses. Oxford: Oxford University Press, pp. 138–164. Jensen, Frede (1986). The Syntax of Medieval Occitan. Tu¨bingen: Niemeyer.

references and bibliographical abbreviations

435

Jensen, Frede (1990). Old French and Comparative Gallo-Romance Syntax. (Beihefte zur Zeitschrift fu¨r romanische Philologie 232). Tubingen: Narr. Jiménez, Jesús (2008). ‘Altres fenòmens vocàlics en el mot’. In Joan Solà, Maria Rosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds), Gramàtica del català contemporani (edició definitiva). Vol. 1. Barcelona: Empúries, pp. 169–194. Johnson, Daniel E. Rbrul Manual. Avaiable at: http://www.danielezrajohnson.com/ Rbrul_manual.html. Jolivet, Rémi (1984). ‘L ’acceptabilité des formes verbales surcomposées’. Le Français moderne 52: 159–176. Jolivet, Rémi (1986). ‘Le passé surcomposé: emploi “général” et emploi “régional”: examen des insertions dans le syntagme verbal surcomposé’. In Mélanges d’onomastique, linguistique et philologie offerts à M. Raymond Sindou. Tome II: Linguistique et philologie. Millau: Presses de la Société des imprimeurs Maury, pp. 109–116. Jones, Michael Allan (1993). Sardinian Syntax. London: Routledge. Joseph, Brian D. (2008). ‘Morphologization from syntax’. In Joseph and Janda (eds), pp. 472–492. Joseph, Brian D. (2011). ‘Grammaticalization: a general critique’. In Narrog and Heine (eds), pp. 193–205. Joseph, Brian D. (2020). ‘What grammaticalization is and what it is not’. Lecture delivered on 30/5/2020 as part of the Abralin Ao Vivo series. Available at: https://www.youtube. com/watch?v=1OOfcVMDVMU. Joseph, Brian D. and Richard D. Janda (1988). ‘The how and why of diachronic morphologization and demorphologization’. In Michael Hammond and Michael Noonan (eds), Theoretical Morphology: Approaches in Modern Linguistics. San Diego, CA: Academic Press, pp. 193–213. Joseph, Brian D. and Richard D. Janda (eds) (2008). The Handbook of Historical Linguistics. Oxford: Blackwell. Juge, Matthew L. (2006). ‘Morphological factors in the grammaticalization of the Catalan “go” past’. Diachronica 23: 313–339. Juge, Matthew L. (2008). ‘Narrative and the Catalan go-past’. Folia Linguistica Historica 29: 27–56. Kamp, Hans and Antje Rossdeutscher (1994). ‘DRS-construction and lexically-driven interference’. Theoretical Linguistics 20: 165–235. Karlsson, Keith E. (1981). Syntax and Affixation: The Evolution of mente in Latin and Romance. (Beihefte zur Zeitschrift fu¨r romanische Philologie 182). Tu¨bingen: Niemeyer. Kaye, Steven J. (2015). Conjugation Class from Latin to Romance: heteroclisis in diachrony and synchrony. Unpublished D.Phil. thesis, University of Oxford. Kayne, Richard S. (1989). ‘Null subjects and clitic climbing’. In Jaeggli and Safir (eds), pp. 239–261. Kayne, Richard S. (1993). ‘Toward a modular theory of auxiliary selection’. Studia Linguistica 47: 3–31. Keenan, Elinor (1977). ‘Why look at planned and unplanned discourse?’. Southern California Occasional Papers in Linguistics 5: 1–13 Keniston, Hayward (1937). The Syntax of Castilian Prose: The Sixteenth Century. Chicago: University of Chicago Press. Kerswill, Paul (2013). ‘Koinéization’. In J. K. Chambers and Natalie Schilling (eds), The Handbook of Language Variation and Change (2nd edition). Chichester: Wiley– Blackwell, pp. 519–536.

436

references and bibliographical abbreviations

Kerswill, Paul and Ann N. N. Williams, (2000). ‘Creating a New Town koiné: children and language change in Milton Keynes’. Language in Society 29: 65–115. King, Ruth and Terry Nadasdi (2003). ‘Back to the future in Acadian French’. Journal of French Language Studies 13: 323–337. Kiparsky, Paul (2005). ‘Blocking and periphrasis in inflectional paradigms’. In Geert. Booij and Jaap Van Marle (eds), Yearbook of Morphology 2004. Dordrecht: Springer, 113–135. Kiparsky, Paul (2015). ‘New perspectives in historical linguistics’. In Claire Bowern and Bethwyn Evans (eds), The Routledge Handbook of Historical Linguistics. London and New York: Routledge, pp. 64–102. Kitto, Catherine and Paul de Lacy (1999). ‘Correspondence and epenthetic quality’. In Catherine Kitto (ed.), Proceedings of AFLA VI, the Sixth Meeting of the Austronesian Formal Linguistics Association, held at the University of Toronto, April 16–18, 1999. Toronto: University of Toronto Department of Linguistics/The Hague: Holland Academic Graphics, pp. 181–200. Klein, Wolfgang (1994). Time in Language. London: Routledge. Klein, Wolfgang (2008). ‘The topic situation’. In Bernt Ahrenholz, Ursula Bredel, Wolfgang Klein, Martina Rost-Roth, and Romuald Skiba (eds), Empirische Forschung und Theoriebildung: Beitra¨ge aus Soziolinguistik, Gesprochene-Sprache- und Zweitspracherwerbsforschung: Festschrift fu¨r Norbert Dittmar zum 65 Geburtstag. Frankfurt am Main: Lang, pp. 287–305. Klein-Andreu, Flora (1991). ‘Losing ground: a discourse-pragmatic solution to the history of -ra in Spanish’. In Fleischman and Waugh (eds), pp. 165–178. Komatsu, Eisuke and George Wolf (eds) (1997). F. de Saussure Deuxième cours de linguistique generale (1908–1909), d’après les cahiers d’Albert Riedlinger et Charles Patois. Oxford: Pergamon Press. Korotkova, Natalia and Yuri Lander (2010). ‘Deriving affix ordering in polysynthesis: evidence from Adyghe’. Morphology 20: 299–319. Kra¨mer, Martin (2009). The Phonology of Italian. Oxford: Oxford University Press. Kristol, Andres (2018). ‘Variation diachronique et variation infra-dialectale: éclairages mutuels. Vers une grammaire du polymorphisme’. In Paolo Greco, Cesarina Vecchia, and Rosanna Sornicola (eds), Strutture e dinamismo della variazione e del cambiamento. Atti del Convegno DIA III, Università di Napoli Federico II, 24–27 novembre 2014. Naples: Giannini, pp. 331–350. Kroch, Anthony (1989). ‘Reflexes of grammar in patterns of language change’. Language Variation and Change 1: 199–244. Kroch, Anthony (2001). ‘Syntactic change’. In Mark Baltin and Chris Collins (eds), The Handbook of Contemporary Syntactic Theory. Oxford: Blackwell, pp. 699–729. Kruszewski, Mikołaj (1995). Writings in General Linguistics: On Sound Alternation (1881) and Outline of Linguistic Science (1883), ed. E. F. K. Koerner. Amsterdam: Benjamins. Kunert, Hans Peter (1994). ‘L’occitan en Calàbria’. Estudis occitans 16: 3–14. Kuryłowicz, Jerzy (1964). The Inflectional Categories of Indo-European. Heidelberg: Carl Winter. Kuryłowicz, Jerzy (1965). ‘The evolution of grammatical categories’. Diogenes 13: 55–71. Kuteva, Tania (2001). Auxiliation: An Enquiry into the Nature of Grammaticalization. Oxford: Oxford University Press. Kuteva, Tania A. (1995). ‘The auxiliation constraint and reference’. In Richard A. Geiger (ed.), Reference in Multidisciplinary Perspective: Philosophical Object, Cognitive Subject, Intersubjective Process. Hildesheim: Olms, pp. 374–386.

references and bibliographical abbreviations

437

Lacross, Elisabeth (2018). Variation in Future Temporal Reference in Southern France. Unpublished Ph.D. dissertation, University of Georgia. La Fauci, Nunzio (1988). Oggetti e soggetti nella formazione della morfosintassi romanza. Pisa: Giardini. La Fauci, Nunzio (1989). ‘Accordo del participio passato e ausiliari perfettivi in italiano ed in francese’. In Franco Foresti, Elena Rizzi, and Paola Benedini (eds), L’italiano tra le lingue romanze. Atti del XX Congresso internazionale di studi della SLI (SLI 27), Bologna, 25–27 settembre 1986. Rome: Bulzoni, pp. 213–242. 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: CSFLS, pp. 185–220 [also in Nunzio La Fauci Forme romanze della funzione predicativa. Pisa: Edizioni ETS, 2000, pp. 41–73]. La Fauci, Nunzio (2000). ‘Aggettivi e participi italiani (e romanzi) nella teoria della gemmazione predicativa’. In Nunzio La Fauci, Forme romanze della funzione predicativa. Pisa: Edizioni ETS, pp. 75–90. La Fauci, Nunzio (2004). ‘Armonia differenziale dell’ausiliazione perfettiva nel volgare di Dante’. In Maurizio Dardano and Gianluca Frenguelli (eds), SintAnt. La sintassi dell’italiano antico. Atti del Convegno internazionale di studi (Università ‘Roma Tre’ 18–21 settembre 2002). Rome: Aracne, pp. 237–252. La Fauci, Nunzio and Michele Loporcaro (1989). ‘Passifs, avancements de l’objet indirect et formes verbales périphrastiques dans le dialecte d’Altamura (Pouilles)’. Rivista di Linguistica 1: 161–196. La Fauci, Nunzio and Michele Loporcaro (1993). ‘Grammatical relations and syntactic levels in Bonorvese morphosyntax’. In Adriana Belletti (ed.), Syntactic Theory and the Dialects of Italy. Turin: Rosenberg & Sellier, pp. 155–203. Langacker, Ronald W. (1977). ‘Syntactic reanalysis’. In Charles N. Li (ed.), Mechanisms of Syntactic Change. Austin: University of Texas Press, pp. 57–139. Langacker, Ronald W. (1987). The Foundations of Cognitive Grammar. Vol. I. Theoretical Prerequisites. Stanford, CA: Stanford University Press. Langacker, Ronald W. (1990). ‘Subjectification’. Cognitive Linguistics, 1: 5–38. Langfur, Hal (ed.) (2014). Native Brazil: Beyond the Convert and the Cannibal, 1500–1900. Albuquerque, NM: University of New Mexico Press. Larson, Pa¨r (2000). ‘Tra linguistica e fonti diplomatiche: quello che le carte dicono e non dicono’. In József Herman, Anna Marinetti and Luca Mondin (eds), La preistoria dell’italiano. Tu¨bingen: Niemeyer, pp. 151–166. Larson, Pa¨r (2012). ‘Le carte alto-medievali come fonte di lingua: qualche esperienza personale’. In Rosanna Sornicola and Paolo Greco (eds), La lingua dei documenti notarili alto-medievali dell’Italia meridionale: bilancio degli studi e prospettive di ricerca. Naples: Tavolario, pp. 63–74. Lass, Roger (1990). ‘How to do things with junk: exaptation in language evolution’. Journal of Linguistics 26: 79–102. Lass, Roger (1997). Historical Linguistics and Language Change. Cambridge: Cambridge University Press. Lăzărescu, Paul (1984). ‘Subdialectul moldovean’. In Rusu (ed.), pp. 208–240. Ledgeway, Adam (1997). ‘Asyndetic complementation in Neapolitan dialect’. The Italianist 17: 231–273. Ledgeway, Adam (1998). ‘Avé(re) and esse(re) alternation in Neapolitan’. In Olga Fullana and Francesc Roca (eds), Studies on the Syntax of Central Romance Languages. Universitat de Girona, 123–147.

438

references and bibliographical abbreviations

Ledgeway, Adam (1999). ‘I tempi sovraccomposti nel napoletano antico’. L’Italia dialettale 60: 105–124. Ledgeway, Adam (2000). A Comparative Syntax of the Dialects of Southern Italy: A Minimalist Approach (Publications of the Philological Society 33). Oxford: Blackwell. Ledgeway, Adam (2009). Grammatica diacronica del napoletano. Tu¨bingen: Niemeyer. Ledgeway, Adam (2011). ‘Syntactic and morphosyntactic typology and change’. In Maiden, Smith, and Ledgeway (eds), pp. 382–471. Ledgeway, Adam (2012), From Latin to Romance: Morphosyntactic Typology and Change. Oxford: Oxford University Press. Ledgeway, Adam (2014). ‘Romance auxiliary selection in light of Romanian evidence’. In Gabriela Pană Dindelegan, Rodica Zafiu, Adina Dragomirescu, Irina Nicula, Alexandru Nicolae, and Louise Esher (eds), Diachronic Variation in Romanian. Newcastle: Cambridge Scholars, pp. 3–35. Ledgeway, Adam (2016a). ‘From coordination to subordination: the grammaticalization of progressive and andative aspect in the dialects of Salento’. In Pratas, Pinto, and Pereira (eds), pp. 157–184. Ledgeway, Adam (2016b). ‘Italian, Tuscan, and Corsican’. In Ledgeway and Maiden (eds), pp. 206–227. Ledgeway, Adam (2016c). ‘Functional categories’. In Ledgeway and Maiden (eds), pp. 761– 771. Ledgeway, Adam (2017a). ‘Syntheticity and analyticity’. In Andreas Dufter and Elisabeth Stark (eds), Manual of Romance Morphosyntax and Syntax (Manual of Romance Linguistics). Berlin: de Gruyter, pp. 837–884. Ledgeway, Adam (2017b). ‘The Romanian definite article in a comparative Romance perspective’. In Adina Dragomirescu, Alexandru Nicolae, Camelia Stan, and Rodica Zafiu (eds), Sintaxa ca mod de a fi. Omagiu doamnei profesoare Gabriela Pană Dindelegan, la aniversare. Bucharest: Editura Universități din Bucarești, pp. 231–247. Ledgeway, Adam (2019). ‘Parameters in the development of Romance perfective auxiliary selection’. In Michela Cennamo and Claudia Fabrizio (eds), Historical Linguistics 2015. Selected Papers from the 22nd International Conference on Historical Linguistics, Naples, 27–31 July 2015. Amsterdam: Benjamins, pp. 343–384. Ledgeway, Adam (2021). ‘Passive periphrases in the Romance languages’. In Michele Loporcaro and Francesco Gardani (eds), The Oxford Encyclopedia of Romance Linguistics. Oxford: Oxford University Press. Advance online publication, doi: https://doi.org/10. 1093/acrefore/9780199384655.013.639. Ledgeway, Adam (in press a). ‘Coming and going: Calabrian pseudo-coordination’. Revue roumaine de linguistique. Ledgeway, Adam (in press b). ‘Parametric variation in differential object marking in the dialects of Italy’. In Monica Irimia and Alexandru Mardale (eds), Differential Object Marking in Romance. Amsterdam: Benjamins. Ledgeway, Adam and Martin Maiden (eds) (2016). The Oxford Guide to the Romance Languages. Oxford: Oxford University Press. Lee, M. Kittiya (2005). Conversing in Colony: the Brası´lica and the Vulgar in Portuguese America 1500–1759. Unpublished Ph.D. dissertation, Johns Hopkins University. Lefebvre, Claire (2006). Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Legendre, Géraldine (2010). ‘A formal typology of person-based auxiliary selection in ItaloRomance’. In D’Alessandro, Ledgeway, and Roberts (eds), pp. 186–200.

references and bibliographical abbreviations

439

Lehmann, Christian (1982). ‘Universal and typological aspects of agreement’. In HansJakob Seiler and Franz Josef Stachowiak (eds), Apprehension. Das sprachliche Erfassen von Gegensta¨nden. Teil II: Die Techniken und ihr Zusammenhang in der Einzelsprachen. Tu¨bingen: Narr, pp. 201–267. Lehmann, Christian (1985). ‘Grammaticalization: synchronic variation and grammatical change’. Lingua e stile 20: 303–318. Lehmann, Christian (2015). Thoughts on Grammaticalization (3rd edition). Berlin: Language Science Press. Lenci, Alessandro and Pier Marco Bertinetto (2000). ‘Aspect, adverbs and events: habituality and perfectivity’. In James Higginbotham, Fabio Pianesi, and Achille C. Varzi (eds), Speaking of Events. Oxford: Oxford University Press. pp. 245–287. Leone, Alfonso (1973). ‘Vattel’a pesca, vieni a piglialo’. Lingua nostra 34: 11–13. Leone, Alfonso (1995). Profilo di sintassi siciliana. Palermo: CSFLS. Lepschy, Giulio. (1965). ‘Giusto Lipsio e il volgare nel VI secolo. “Torna, torna, frater” e l’“Instrumentum plenarie securitatis”’. In Italia medioevale e umanistica 8: 295–308 [reprinted in Giulio Lepschy, Saggi di linguistica italiana, Bologna: il Mulino, 1989, pp. 231–244]. Leumann, Manu, Johann Hofmann, and Anton Szantyr (1972). Lateinische Grammatik (2 vols). Munich: Beck. Levin, Beth (1993). English Verb Classes and Alternations: A Preliminary Investigation. Chicago: University of Chicago Press. Levin, Beth and Malka Rappaport Hovav (2005). Argument Realization. Cambridge: Cambridge University Press. Levin, Juliette (1987). ‘Between epenthetic and excrescent vowels’. Proceedings of the West Coast Conference on Formal Linguistics 6: 187–201. Lichtenberk, Frantisek (1991). ‘Semantic change and heterosemy in grammaticalization’. Language 67: 475–509. Lieber, Rochelle (1992). Deconstructing Morphology: Word Formation in Syntactic Theory. Chicago: University of Chicago Press. Lizio-Bruno, Letterio (1871). Canti popolari delle Isole Eolie. Messina: Tipi Ignazio d’Amico e Figli. Lloret, Maria-Rosa and Jesús Jiménez (2008). ‘Marcatge posicional i prominència en el vocalisme àton’. Caplletra: Revista internacional de filologia 45: 55–91. Lloret, Maria-Rosa and Joaquim Viaplana (1992). ‘Les vocals finals dels elements nominals en català’. In Antoni Ferrando and Albert G. Hauf (eds), Misceŀlània Joan Fuster. Vol. 5. Barcelona: Abadia de Montserrat, pp. 419–445. Lloyd, Paul M. (1993). Del latı´n al español. I. Fonologı´a y morfologı´a históricas de la lengua española. Madrid: Gredos. Lodge, Anthony (2010). ‘Standardisation, koinéisation et l’historiographie du français’. Revue de linguistique romane 74: 5–26. Lo¨fstedt, Bengt (1961). Studien u¨ber die Sprache der langobardischen Gesetze. Stockholm: Almqvist & Wiksell. Lo¨fstedt, Einar (1959). Late Latin. Oslo: Aschehoug. Lois, Ximena (1990), ‘Auxiliary selection and past participle agreement in Romance’. Probus 2: 233–255. Lombard, Alf (1974). La Langue roumaine: une présentation. Paris: Klincksieck. Lombardi, Alessandra (2007). ‘Posizione dei clitici e ordine dei costituenti nella lingua sarda medievale’. In Delia Bentley and Adam Ledgeway (eds), Sui dialetti italoromanzi: saggi in

440

references and bibliographical abbreviations

onore di Nigel B. Vincent (The Italianist 27, special supplement 1). Norfolk: Biddles, pp. 133–147. López Sanz, Rafael (1972). El baré: estudio lingu¨´ıstico. Caracas: Universidad Central de Venezuela. Loporcaro, Michele (1988). Grammatica storica del dialetto di Altamura. Pisa: Giardini. Loporcaro, Michele (1997a). ‘Puglia and Salento’. In Maiden and Parry (eds), pp. 338–348. Loporcaro, Michele (1997b). L’origine del raddoppiamento fonosintattico: saggio di fonologia diacronica romanza. Basel and Tu¨bingen: Francke. Loporcaro, Michele (1997c). ‘On vowel epenthesis in Alguer Catalan’. In Pier Marco Bertinetto, Livio Gaeta, Georgi Jetchev, and David Michaels (eds), Certamen Phonologicum III: Papers from the Third Cortona Phonology Meeting. Turin: Rosenberg & Sellier, pp. 205–227. Loporcaro, Michele (1998). Sintassi comparata dell’accordo participiale romanzo. Turin: Rosenberg & Sellier. Loporcaro, Michele (1999a). ‘Il futuro cantare-habeo nell’Italia meridionale’. Archivio glottologico italiano 84: 67–114. Loporcaro, Michele (1999b). ‘L’ausiliazione perfettiva nelle parlate di Zagarolo e di Colonna e lo studio della sintassi dei dialetti mediani’. Contributi di filologia dell’Italia Mediana 13: 203–226. Loporcaro, Michele (2001). ‘La selezione dell’ausiliare nei dialetti italiani: dati e teorie’. In Albano Leoni, Sornicola, Krosbakken, and Stromboli (eds), pp. 455–476. Loporcaro, Michele (2007). ‘On triple auxiliation in Romance’. Linguistics 45: 173–222. Loporcaro, Michele (2008). ‘L’allineamento attivo-inattivo e il rapporto fra lessico e morfosintassi’. In Emanuela Cresti (ed.), Prospettive nello studio del lessico italiano. Atti del IX Congresso SILFI (Firenze, 14–17 June 2006). Vol. 1. Florence: Firenze University Press, pp. 311–320. Loporcaro, Michele (2011). ‘A euroversal in a global perspective: auxiliation and alignment’. In Peter Siemund (ed.), Linguistic Universals and Language Variation. Berlin: Mouton de Gruyter, pp. 55–91. Loporcaro, Michele (2012). ‘Per lo studio della morfosintassi dei dialetti lucani: acquisizioni recenti e nuove prospettive’. In Patrizia Del Puente (ed.), Atti del II Convegno internazionale di dialettologia—Progetto A.L.Ba. Rionero in Vulture: Calice, pp. 176–198. Loporcaro, Michele (2013). Profilo linguistico dei dialetti italiani. Bari: Laterza. Loporcaro, Michele (2014). ‘Perfective auxiliation in Italo-Romance: the complementarity of historical and modern cross-dialectal evidence’. In Benincà, Ledgeway, and Vincent (eds), pp. 48–70. Loporcaro, Michele (2015). ‘Perfective auxiliation with reflexives in Medieval Romance: syntactic vs semantic gradients’. In Rolf Kailuweit and Malte Rosemeyer (eds), Auxiliary Selection Revisited: Gradience and Gradualness. Berlin: de Gruyter, pp. 43–77. Loporcaro, Michele (2016). ‘Auxiliary selection and participial agreement’. In Ledgeway and Maiden (eds), pp. 802–818. Loporcaro, Michele (2018). Gender from Latin to Romance: History, Geography, Typology. Oxford: Oxford University Press. Loporcaro, Michele and Ignazio Putzu (2013). ‘Variation in auxiliary selection, syntactic change, and the internal classification of Campidanese Sardinian’. In Giulio Paulis, Immacolata Pinto, and Ignazio Putzu (eds), Repertorio plurilingue e variazione linguistica a Cagliari. Milan: FrancoAngeli, pp. 200–244.

references and bibliographical abbreviations

441

Loregian-Penkal, Loremi (2004). (Re)ana´lise da referência de segunda pessoa na fala da Região Sul. Unpublished tese de doutorado, Universidade Federal do Parana´. Lorusso, Paolo (2019). ‘A person split analysis of the progressive forms in some Southern Italian varieties’. In Cruschina, Ledgeway, and Remberger (eds), pp. 203–236. Lucchesi, Dante (1999). ‘A questão da formação do português popular do Brasil: notı´cia de um estudo de caso’. A Cor da Letras 3: 73-100. Lucchesi, Dante (2001). ‘As duas grandes vertentes da história sociolingu¨´ıstica do Brasil (1500- 2000)’, D.E.L.T.A. Documentação e Estudos em Linguı´stica Teórica e Aplicada 17: 97-130. Lucchesi, Dante (2003). ‘O conceito de transmissão lingu¨´ıstica irregular e o processo de formação do português do Brasil’, in Cla´udia Roncarati and Jussara Abraçado (eds), Português brasileiro: contato lingu¨´ıstico, heterogeneidade e história. Rio de Janeiro, RJ: 7Letras, pp. 202–258. Lucchesi, Dante (2015). Lı´ngua e sociedade partidas: a polarização sociolinguı´stica do Brasil. São Paulo, SP: Contexto. Lucchesi, Dante (2017). ‘A periodização da história sociolinguı´stica do Brasil’. DELTA 33: 347–382. Lucchesi, Dante, Alan Baxter, and Ilsa Ribeiro (2009). O português afro-brasileiro. Salvador, BA: EDUFBA. Luı´s, Ana and Andrew Spencer (2004). ‘A paradigm function account of “mesoclisis” in European Portuguese’. Yearbook of Morphology 2004, pp. 177–228. Lunn, Patricia and Thomas D. Cravens (1991). ‘A contextual reconsideration of the Spanish -ra indicative’. In Fleischman and Waugh (eds), pp. 147–164. Lupyan, Gary and Rick Dale (2010). ‘Language structure is partly determined by social structure’. PLoS One 5: e8559. Lyons, John (1977). Semantics 1. Cambridge: Cambridge University Press. Magalhães, José Vieira Couto de (1876). O selvagem. I: Curso de lı´ngua geral segundo Ollendorf—textos de lendas indı´genas. Rio de Janeiro, RJ: Reforma. 1–281. Magistrale, Francesco (1998). Chartae Latinae Antiquiores. 2nd series. Ninth Century, 51: Italy 23. Dietikon-Zu¨rich: Urs Graf. Maiden, Martin (1991). Interactive Morphonology: Metaphony in Italy. London: Routledge. Maiden, Martin (1992). ‘Irregularity as a determinant of morphological change’. Journal of Linguistics 28: 285–312. Maiden, Martin (1995). A Linguistic History of Italian. London: Longman. Maiden, Martin (1998). Storia linguistica dell’italiano. Bologna: il Mulino. Maiden, Martin (2000). ‘Di un cambiamento intramorfologico: origini del tipo dissi, dicesti ecc., nell’italoromanzo’. Archivio glottologico italiano 85: 137–171. Maiden, Martin (2001a). ‘Di nuovo sulle alternanze velari nel verbo italiano e spagnolo’. Cuadernos de filologı´a italiana 8: 39–61. Maiden, Martin (2001b). ‘A strange affinity: “perfecto y tiempos afines”’. Bulletin of Hispanic Studies 78: 441–464. Maiden, Martin (2004a). ‘Verb augments and meaninglessness in early Romance morphology’. Studi di grammatica italiana 22: 1–61. Maiden, Martin (2004b). ‘When lexemes become allomorphs: on the genesis of suppletion’. Folia Linguistica 38: 227–256. Maiden, Martin (2005). ‘Morphological autonomy and diachrony’. In Geert Booij and Jaap van Marle (eds), Yearbook of Morphology 2004. Dordrecht: Springer, pp. 137–175.

442

references and bibliographical abbreviations

Maiden, Martin (2008). ‘Lexical nonsense and morphological sense: on the real importance of “folk etymology” and related phenomena for historical linguists’. In Eythórsson (ed.), pp. 307–328. Maiden, Martin (2009). ‘Where does heteroclisis come from? Evidence from Romanian dialects’. Morphology 19: 59–86. Maiden, Martin (2011a). ‘Morphophonological innovation’. In Maiden, Smith, and Ledgeway (eds), pp. 216–267. Maiden, Martin (2011b). ‘Morphomes and “stress-conditioned allomorphy” in Romansh’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 36–50. Maiden, Martin (2016a). ‘Romanian, Istro-Romanian, Megleno-Romanian, and Aromanian’. In Ledgeway and Maiden (eds), pp. 91–125. Maiden, Martin (2016b). ‘Morphomes’. In Ledgeway and Maiden (eds), pp. 708–721. Maiden, Martin (2017). ‘Romansh allomorphy (again!)’. In Claire Bowern, Laurence Horn, and Raffaella Zanuttini (eds), On Looking into Words (and Beyond): Structures, Relations, Analyses. Berlin: Language Science Press, pp. 189–211. Maiden, Martin (2018a). The Romance Verb: Morphomic Structure and Diachrony. Oxford: Oxford University Press. Maiden, Martin (2018b). ‘New thoughts on an old puzzle: the Italian alternation type dissi, dicesti; feci facesti’. Revue romane 53: 217–260. Maiden, Martin (2020a). ‘Folk etymology and contamination in the Romance languages’. Oxford Research Encyclopedia of Linguistcs. Retrieved 28 Feb. 2021, from https://oxfordre.com/linguistics/view/10.1093/acrefore/9780199384655.001.0001/ acrefore-9780199384655-e-457. Maiden, Martin (2020b). ‘Dalmatian (Vegliote)’. Oxford Research Encyclopedia of Linguistcs. Retrieved 28 Feb. 2021, from https://oxfordre.com/linguistics/view/ 10.1093/acrefore/9780199384655.001.0001/ acrefore-9780199384655-e-726. Maiden, Martin (in press). ‘Establishing contact: Slavonic influence on Romanian morphology’. In Gardani, Loporcaro, and Giudici (eds). Maiden, Martin, Adina Dragomirescu, Gabriela Pană Dindelegan, Oana Uță Bărbulescu, and Rodica Zafiu (2021). The Oxford History of Romanian Morphology. Oxford: Oxford University Press. Maiden, Martin and Paul O’Neill (2010). ‘On morphomic defectiveness: evidence from the Romance languages of the Iberian Peninsula’. In Matthew Baerman, Greville G. Corbett, and Dunstan Brown (eds), Defective Paradigms: Missing Forms and What They Tell Us. Oxford: Oxford University Press/British Academy, pp. 103–124. Maiden, Martin and Mair Parry (eds) (1997). The Dialects of Italy. London: Longman. Maiden, Martin, John Charles Smith, Maria Goldbach, and Marc-Olivier Hinzelin (eds) (2011). Morphological Autonomy: Perspectives from Romance Inflectional Morphology. Oxford: Oxford University Press. Maiden, Martin, John Charles Smith, and Adam Ledgeway (eds) (2011). The Cambridge History of the Romance Languages. Vol. 1: Structures. Cambridge: Cambridge University Press. Maiden, Martin, John Charles Smith, and Adam Ledgeway (eds) (2013). The Cambridge History of the Romance Languages. Vol. 2: Contexts. Cambridge: Cambridge University Press. Maiden, Martin, Andrew Swearingen, and Paul O’Neill (2009). ‘Imperative morphology in diachrony: evidence from the Romance languages’. In Monique Dufresne, Fernande Dupuis, and Etleva Vocaj (eds), Historical Linguistics 2007. Selected papers from the

references and bibliographical abbreviations

443

18th International Conference on Historical Linguistics, Montreal, 6–11 August 2007. Amsterdam: Benjamins, pp. 99–108. Malmberg, Bertil (1950). Études sur la phonétique de l’espagnol parlé en Argentine. Lund: Gleerup. Manasantvongs, Peter (2005). Explaining the Null Subject in the Romance Languages: finding a suitable framework. Unpublished Ph.D. thesis, University of Cambridge. Manea, Dana, Gabriela Pană Dindelegan, and Rodica Zafiu (2008). ‘Verbul’. In Valeria Guțu Romalo (ed.), Gramatica limbii române. I: Cuvântul. Bucharest: Editura Academiei Române, pp. 323–584. Manoliu, Maria (1959). ‘Une déviation du système de conjugaison romane: temps composés avec a fi “être” à la diathèse active en roumain’. In Recueil d’études romanes publiées à l’occasion du IXe Congrès International de Linguistique Romane à Lisbonne du 31 mars au 3 avril 1959. Bucharest: Editura Academiei, pp. 135–141. Manzari, Giovanni (2019). Microdiatopia periurbana: variazione fonetica e altri aspetti strutturali di sei dialetti dell’entroterra barese. Alessandria: Edizioni dell’Orso. Manzini, M. Rita, Paolo Lorusso, and Leonardo M. Savoia (2017). ‘A/bare finite complements in southern Italian varieties: monoclausal or bi-clausal syntax?’. Quaderni di Linguistica e Studi Orientali (QULSO) 3:11–59. Florence: Firenze University Press. Manzini, M. Rita and Leonardo M. Savoia (1998). ‘Clitics and auxiliary choice in Italian dialects: their relevance for the person ergativity split’. Recherches linguistiques de Vincennes 27: 115–138. Manzini, M. Rita and Leonardo M. Savoia (2005). I dialetti italiani e romanci: morfosintassi generativa (3 vols). Alessandria: Edizioni dell’Orso. Manzini, M. Rita and Leonardo M. Savoia (2007). A Unification of Morphology and Syntax: Investigations into Romance and Albanian Dialects. London: Routledge. Manzini, M. Rita and Leonardo M. Savoia (2011). Grammatical Categories: Variation in Romance Languages. Cambridge: Cambridge University Press. Marantz, Alec (1997). ‘No escape from syntax: don’t try morphological analysis in the privacy of your own lexicon’. In Alexis Dimitriadis, Laura Siegel, Clarissa Surek-Clark, and Alexander Williams (eds), University of Pennsylvania Working Papers in Linguistics 4(2): 201–225. Marcato, Gianna and Flavia Ursini (1998). Dialetti italiani: grammatica e storia. Padua: Unipress. Marchello-Nizia, Christiane (1980). La Langue française aux XIVe et XVe siècles. Paris: Nathan. Mardale, Alexandru (2009). Les Prépositions fonctionnelles du roumain: études comparatives sur le marquage casuel. Paris: L’Harmattan. Mardale, Alexandru (2015). ‘Differential object marking in the first original Romanian texts’. In Virginia Hill (ed.), Formal Approaches to DPs in Old Romanian. Leiden: Brill, pp. 200–245. Mărgărit, Iulia and Victorela Neagoe (1997). ‘Cercetări asupra graiurilor românești din nord-estul Bulgariei (regiunea Loveci)’. Fonetică și dialectologie 16: 75–96. Marin Maria (1985). ‘Formes verbales périphrastiques de l’indicatif dans les parlers dacoroumains’. Revue roumaine de linguistique 30: 459–468. Marin, Maria (2000). ‘Elemente arhaice sau învechite în morfosintaxa graiurilor românești din Ungaria’. Fonetică și dialectologie 19: 197–210. Marin, Maria (2005–2007). ‘Timpurile indicativului în graiurile dacoromâne actuale (II)’. Fonetică și dialectologie 24–26: 89–164.

444

references and bibliographical abbreviations

Marin, Maria (2012). ‘Din nou despre unele forme verbale perifrastice’. Fonetică și dialectologie 31: 20–24. Marin, Maria (2014). ‘Trăsături dialectale comune graiurilor muntenești și celor crișene’. Fonetică și dialectologie 33: 87–98. Marin, Maria, Mihai Conțiu, Bogdan Marinescu, Carmen-Ioana Radu, and Marilena Tiugan (2017). Graiurile din Sălaj: studiu lingvistic, texte dialectale, glosar. Bucharest: Editura Academiei Române. Marin, Maria, Iulia Mărgărit, and Victorela Neagoe (1998). ‘Graiuri românești din Ucraina și Republica Moldova’. Fonetică și dialectologie 17: 69–156. Markopoulos, Theodore (2009). The Future in Greek: From Ancient to Medieval. Oxford: Oxford University Press. Marotta, Giovanna (1993). ‘Selezione dell’articolo e struttura sillabica in italiano: un’interazione totale?’. Studi di grammatica italiana 15: 255–293. Martinet, André (1949). ‘La double articulation linguistique’. Travaux du Cercle Linguistique de Copenhague 5: 30–37. Martı´nez Cortés, Juan Pablo, Santiago J. Paricio Martı´n, and Francho Rodés Orquı´n (eds) (2021). Gramatica basica de l’aragonés. Zaragoza: Prensas de la Universidad de Zaragoza. Mascaró, Joan (1985a). Morfologia. Barcelona: Enciclopèdia Catalana. Mascaró, Joan (1985b). ‘Catalan nominal markers and vowel insertion’. Cornell Working Papers in Linguistics 7: 162–175. Mascaró, Joan (1996). ‘External allomorphy as emergence of the unmarked’. In Jacques Durand and Bernard Laks (eds), Current Trends in Phonology: Models and Methods. Salford: European Studies Research Institute, University of Salford, pp. 513–522. Matalon, Zuan Nazzi (1977). Marilenghe: gramatiche furlane. Udine: Institût di Studis Furlans. Matras, Yaron (2007). ‘The borrowability of structural categories’. In Yaron Matras and Jeanette Sakel (eds), Grammatical Borrowing in Cross-linguistic Perspective. Berlin: Mouton de Gruyter, pp. 31–73. Matthews, P. H. (1972). Inflectional Morphology. Cambridge: Cambridge University Press. Matthews, P. H. (2007). The Concise Oxford Dictionary of Linguistics (2nd edition). Oxford: Oxford University Press. Matthews, P. H. (2014). The Concise Oxford Dictionary of Linguistics (3rd edition). Oxford: Oxford University Press. Matushansky, Ora, Alec P. Marantz, Isabel Oltra-Massuet, Rolf Noyer, and Martha McGinnis (eds) (2013). Distributed Morphology Today: Morphemes for Morris Halle. Cambridge, MA: MIT Press. Maupas, Charles (1607). Grammaire françoise. Blois: n.p. Mayerthaler, Willi (1981). Morphologische Natu¨rlichkeit (Linguistische Forschungen 28). Wiesbaden: Athenaion. Mazzola, Michael L. (ed.) (1994). Issues and Theory in Romance Linguistics. Selected Papers from the Linguistic Symposium on Romance Languages XXIII, April 1–4 1993. Washington, DC: Georgetown University Press. Mazzoleni, Marco (1995). ‘Il vocativo’. In Lorenzo Renzi, Giampaolo Salvi, and Anna Cardinaletti (eds), Grande grammatica italiana di consultazione. Volume III. Bologna: il Mulino, pp. 377–402, 559–566. Mazzoleni, Marco (1997). ‘Vocativo diretto e inverso: i termini di parentela fra enunciato ed enunciazione’. In Franca Bizzoni and Mariapia Lamberti (eds), Palabras, poeta e ima´genes de Italia. Mexico, DF: Facultdad de filosofı´a y letras, Universidad Nacional Autónoma de México, pp. 115–129.

references and bibliographical abbreviations

445

McCrary, Kristie (2004). Reassessing the Role of the Syllable in Italian Phonology: an experimental study of consonant cluster syllabification, definite article allomorphy and segment duration. Unpublished PhD dissertation, UCLA. McCrary Kambourakis, Kristie (2007). Reassessing the Role of the Syllable in Italian Phonology: An Experimental Study of Consonant Cluster Syllabification, Definite Article Allomorphy and Segment Duration. London: Routledge. Meillet, Antoine (1936). Linguistique historique et linguistique générale. Vol. 2. Paris: Klincksieck. Melchior, Luca (2012). ‘Tra esperienzialità e iteratività: il “passé surcomposé à valeur spéciale” in francese (e in altri idiomi romanzi)’. Revue de linguistique romane 76: 65–98. Melo, Juliana Gonçalves (2009). Identidades fluidas: ser e perceber-se como Baré (Arawak) na Manaus contemporânea. Unpublished tese de doutorado, Universidade de Brası´lia. Menéndez Pidal, R. (1962). Manual de grama´tica histórica española (11th edition). Madrid: Espasa-Calpe. Menon, Odete Pereira da Silva (1995). ‘O sistema pronominal do Português do Brasil’. Revista Letras, 44: 91–106. Menoni, Viviana (1982). ‘Formazione e storia del gerundio composto nell’italiano antico’. Studi di grammatica italiana 11: 5–88. Mensching, Guido and Eva-Maria Remberger (2016). ‘Sardinian’. In Ledgeway and Maiden (eds), pp. 270–291. Merlini Barbaresi, Lavinia (2004). ‘Alterazione’. In Maria Grossmann and Franz Rainer (eds), La formazione delle parole in italiano. Tu¨bingen: Niemeyer, pp. 264–292. Meyer-Lu¨bke, Wilhelm (1899). Grammatik der romanischen Sprachen. Band III. Leipzig Reisland. Meyer-Lu¨bke, Wilhelm (1925). Das Katalanische: Seine Stellung zum Spanischen und Provenzalischen sprachwissenschaftlich und historisch dargestellt. Heidelberg: Winter. Miatto, Veronica (2020). ‘The syllabicity of paragogic vowels in Italian’. MS, Stony Brook University. Michalias, Régis (1906). Éléments abrégés de grammaire auvergnate: dialecte des environs d’Ambert (Puy-de-Dôme). Ambert: Migeon. Michelotti, Alexander (2008). The Position of the Sammarinese Dialects in the Romagnol Linguistic Group. Unpublished Ph.D. dissertation, University of Toronto. Miremont, Pierre (1976). La Syntaxe occitane du Périgord. Toulouse: Pierre Miremont. Mithun, Marianne (2008a). ‘Borrowed rhetorical constructions as starting points for grammaticalization’. In Alexander Bergs and Gabriele Diewald (eds), Constructions and Language Change. Berlin: Mouton de Gruyter, pp. 196–228. Mithun, Marianne (2008b). ‘The emergence of agentive systems’. In Mark Donohue and Søren Wichmann (eds), The Typology of Semantic Alignment Systems. Oxford: Oxford University Press, pp. 297–333. Mithun, Marianne (2012). ‘Morphologies in contact: form, meaning, and use in the grammar of reference’. In Thomas Stolz, Martine Vanhove, Hitomi Otsuka, and Anna Urdzu (eds), Morphologies in Contact. Berlin: Akademie Verlag. pp. 15–36. Mocciaro, Egle (2014). ‘Passive in motion: the early Italian auxiliary andare (“to go”)’. In Devos and van der Wal (eds), pp. 45–68. Mok, Quirinus I. M. (1991). ‘Concurrence de tornar + infinitif et re- en occitan’. In Dieter Kremer (ed.), Actes du XVIIIe Congrès International de Linguistique et de Philologie romanes, Université de Trèves (Trier) 1986, III. Tu¨bingen: Niemeyer, pp. 104–111. Moll, Francesc de B. (1969). Diccionari català—valencià—balear. Barcelona: Moll.

446

references and bibliographical abbreviations

Moll, Francesc de B. (1975). Gramàtica catalana referida especialment a les Illes Balears. Palma de Mallorca: Moll. Monachesi, Paola (1999). A Lexical Approach to Italian Cliticization. Stanford, CA: CSLI Publications. Monaci, Ernesto (1955). Crestomazia italiana dei primi secoli (nuova edizione). Rome: Società Editirice Dante Alighieri. Monserrat, Ruth (2003). ‘O tupi do século XVIII (tupi médio)’. In José R. Bessa Freire and Maria Carlota Rosa (eds), Lı´nguas gerais: polı´tica linguı´stica e catequese na América do Sul no perı´odo colonial. Rio de Janeiro, RJ: EDUERJ, pp. 185–194. Monteiro, John M. (2001). Tupis, tapuias e historiadores: estudos de história indı´gena e do indigenismo. Unpublished tese de livre docência, Universidade Estadual de Campinas. Monteiro, John M. (2012). ‘Rethinking Amerindian resistance and persistence in colonial Portuguese South America’. In John Gledhill and Patience A. Schell (eds), New Approaches to Resistance in Brazil and Mexico. Durham, NC: Duke University Press, pp. 25–43. Montes Giraldo, José Joaquı´n (1976). ‘Un arcaı´smo gramatical en Colombia: la construcción del pretérito compuesto de subjuntivo con ser’. Thesaurus (Boletı´n del Instituto Caro y Cuervo) 31: 51–52. Monville-Burston, Monique and Waugh, Linda R. (1985). ‘Le passé simple dans le discours journalistique’. Lingua 67: 121–170. Mooney, Damien (2020). ‘Future temporal reference in French and Gascon: aller/anar + infinitive periphrasis and structural transfer in the bilingual grammar’. In Wolfe and Maiden (eds), pp. 258–278. Moore, Denny (2014). ‘Historical development of Yẽgatú’. In Salikoko S. Mufwene (ed.), Iberian Imperialism and Language Evolution in Latin America. Chicago: University of Chicago Press, pp. 108–142. Moore, Denny, Sidney Facundes, and Na´dia Pires (1994). ‘Yẽgatú (lı´ngua geral amazônica), its history and the effects of language contact’. In Margaret Langdon (ed.), Survey of California and Other Indian Languages. Report 8: Proceedings of the Meeting of the Society for Indigenous Languages of the Americas, July 2–4, 1993, and the Hotian-Penutian Workshop, July 3, 1993, Ohio State University, Columbus, Ohio. Berkeley, CA: Survey of California and Other Indian Languages, pp. 93–118. Moradi, Sedigheh (2017). ‘Non-canonical epenthesis: epenthetic quality and the role of morphonology’. MS, Stony Brook University. Morais Silva, António de (1958). Grande diciona´rio da lı´ngua portuguesa. Lisbon: Confluência. Moravcsik, Edith A. (1978a). ‘Agreement’. In Joseph H. Greenberg (ed.), Universals of Human Language. Vol. 4. Stanford, CA: Stanford University Press, pp. 331–374. Moravcsik, Edith A. (1978b). ‘Language contact’. In Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik (eds), Universals of Human Language. Vol. 1: Method & Theory. Stanford, CA: Stanford University Press, pp. 93–122. Morelli, Frida (1999). The Phonotactics and Phonology of Obstruent Clusters in Optimality Theory. Unpublished Ph.D. dissertation, University of Maryland, College Park. Mufwene, Salikoko (1994). ‘New Englishes and criteria for naming them’. World Englishes 13: 21-31. Mufwene, Salikoko S. (1995). ‘Language evolution: the population genetics way’. In Guenther Hauska (ed.), Gene, Sprachen, und ihre Evolution. Regensburg: Universita¨tsverlag Regensburg, pp. 30–52.

references and bibliographical abbreviations

447

Mufwene, Salikoko (2004). ‘Language Birth and Death’. Annual Review of Anthropology 33: 201-222. Mufwene, Salikoko (2007). ‘Population movements and contacts in language evolution’. Journal of Language Contact 1: 63-92. Mufwene, Salikoko S. (2008). Language Evolution: Contact, Competition and Change. London: Continuum. Mufwene, Salikoko S. (2009). ‘Restructuring, hybridization and complexity in language evolution’. In Enoch O. Aboh and Norval Smith (eds), Complex Processes in New Languages. Amsterdam: Benjamins, pp. 367–400. Munaro, Nicola (1999). Sintagmi interrogativi nei dialetti italiani settentrionali. Padua: Unipress. Nagy C., Katalin (2010). ‘The pragmatics of grammaticalisation: the role of implicatures in semantic change’. Journal of Historical Pragmatics 11: 67–95. Napoli, Donna Jo and Bill Reynolds (1995). ‘Evaluative affixes in Italian’. In Geert Booij and Jaap van Marle (eds), Yearbook of Morphology 1995. Dordrecht: Kluwer, pp. 151–178. Naro, Anthony Julius and Maria Marta Pereira Scherre (2003). ‘O conceito de transmissão lingu¨´ıstica irregular e as origens estruturais do português brasileiro: um tema em debate’. In Cla´udia Roncarati and Jussara Abraçado (eds), Português brasileiro: contato lingu¨´ıstico, heterogeneidade e história. Rio de Janeiro, RJ: 7Letras, pp. 285–302. Naro, Anthony Julius and Maria Marta Pereira Scherre (2007). Origens do português brasileiro. São Paulo, SP: Para´bola. Narrog, Heiko and Bernd Heine (eds) (2011). The Oxford Handbook of Grammaticalization. Oxford: Oxford University Press. Narrog, Heiko and Bernd Heine (eds) (2018). Grammaticalization from a Typological Perspective. Oxford: Oxford University Press. Navarro, Eduardo de Almeida (2008). Método novo de tupi antigo (3rd edition). São Paulo, SP: Global. Navarro, Eduardo de Almeida (2011). Curso de lı´ngua geral (yẽgatú ou tupi moderno): a lı´ngua das origens da civilização amazônica. São Paulo, SP: Paym. Navarro, Eduardo de Almeida (2013). Diciona´rio de Tupi antigo: a lı´ngua cla´ssica do Brasil. São Paulo, SP: Global. Neagoe, Victorela (1984). ‘Subdialectul bănățean’. In Rusu (ed.), pp. 240–284. Neagoe, Victorela (1992a). ‘Considerații cu privire la poziția graiului din Ținutul Pădurenilor (Hunedoara) în ansamblul dacoromânei’. Fonetică și dialectologie 11: 77–86. Neagoe, Victorela (1992b). ‘Aspecte ale morfosintaxei graiului din Țara Moților’. Fonetică și dialectologie 11: 161–167. Negrão, Esmeralda Vailatti and Evani de Carvalho Viotti (2008). ‘Estratégias de impessoalização em português brasileiro’. In José Luis Fiorin and Margarida Maria Tadoni Petter (eds), África no Brasil: a formação da lı´ngua portuguesa. São Paulo, SP: Contexto, pp. 179–203. Negrão, Esmeralda Vailatti and Evani de Carvalho Viotti (2010). ‘A estrutura sinta´tica das sentenças absolutas no português brasileiro’. Revista linguı´stica da ALFAL 23: 37–58. Negrão, Esmeralda Vailatti and Evani de Carvalho Viotti (2011a). ‘A ergativização do português brasileiro: una conversa continuada com Carlos Franchi’. In Dermeval Da Hora and Esmeralda Vailatti Negrão (eds), Estudos da linguagem: casamento entre termos e perspectivas. João Pessoa, PB: Idéia Editora Universita´ria, pp. 37–61. Negrão, Esmeralda Vailatti and Evani De Carvalho Viotti (2011b). ‘Epistemological aspects of the study of the participation of African languages in Brazilian Portuguese’. In

448

references and bibliographical abbreviations

Margarida Maria Tadoni Petter and Marc Vanhove (eds), Études afro-brésiliennes. Paris: Karthala, pp. 13–44. Negrão, Esmeralda Vailatti and Evani de Carvalho Viotti (2012). ‘Em busca de uma história linguı´stica’. Revista estudos linguı´sticos, Belo Horizonte, 20: 309–342. Negrão, Esmeralda Vailatti and Evani de Carvalho Viotti (2014a). ‘Contato entre quimbundo e português classico: impactos na grama´tica de impessoalização do português brasileiro e angolano’. Linguı´stica 30: 289–330. Negrão, Esmeralda Vailatti and Evani de Carvalho Viotti (2014b). ‘Brazilian Portuguese as a transatlantic language: agents of linguistic contact’. InterDISCIPLINARY: Journal of Portuguese Diaspora Studies 3: 135–154. Nettle, Daniel (2012). ‘Social scale and structural complexity in human languages’. Philosophical Transactions of the Royal Society B 367: 1829–1836. Nevaci, Manuela and Aida Todi (2009). ‘The grammaticalization of perfect auxiliaries in Romanian: historical and dialectal aspects’. Revue roumaine de linguistique 54: 137–150. Newmeyer, Frederick J. (2001). ‘Deconstructing grammaticalization’. Language Sciences 23: 187–229. Nicolae, Alexandru (2015). Ordinea constituenților în limba română: o perspectivă diacronică. Bucharest: Editura Universității din București. Nicolae, Alexandru (2019). Word Order and Parameter Change in Romanian. Oxford: Oxford University Press. Nicolae, Alexandru (2020). ‘Grammaticalization as pattern formation: Romanian auxiliaries from a diachronic Romance perspective’. Revue roumaine de linguistique 45: 394-422. Nicula, Irina (2013). ‘The gerund (present participle)’. In Pană Dindelegan (ed.), pp. 245– 253. Niculescu, Dana (2008). Mijloace lingvistice de exprimare a posesiei în limba română. Bucharest: Editura Universității din Bucureti. Niculescu, Dana (2013). Particularități sintactice ale limbii române din perspectivă tipologică. Gerunziul. Bucharest: Editura Muzeului Național al Literaturii Române. Niculescu, Dana (2016). ‘The gerund and the gerundial construction’. In Pană Dindelegan (ed.), pp. 271–287. Nobre, Wagner Argolo (2013). ‘Colonização e lı´ngua geral: o caso do Sul da Bahia’. Papia: Revista brasileira de estudos crioulos e similares 23: 75–96. Nobre, Wagner Argolo (2016). ‘As lı´nguas gerais na história social-linguı´stica do Brasil’. Papia: Revista brasileira de estudos crioulos e similares 26: 7–52. Noll, Volker and Wolf Dietrich (2010). O português e o tupi no Brasil. São Paulo, SP: Contexto. Noonan, Michael (1985). ‘Complementation’. In Timothy Shopen (ed.), Language Typology and Syntactic Description. II: Complex Constructions. Cambridge: Cambridge University Press, pp. 42–140. Norberg, Dag (1943). Syntaktische Forschungen auf dem Gebiete des Spa¨tlateins und des fru¨hen Mittellateins (Uppsala Universitets Arsskrift 1943: 9). Uppsala: Lundequistska Bokhandeln. Norde, Muriel (2002) ‘The final stages of grammaticalization: affixhood and beyond’. In Ilse Wischer and Gabriel Diewald (eds), New Reflections on Grammaticalization. Amsterdam: Benjamins, pp. 45–81. Nordlinger, Rachel and Louisa Sadler (2004). ‘Nominal tense: a cross-linguistic perspective’. Language 80: 776–806.

references and bibliographical abbreviations

449

Nyrop, K. (1935). Grammaire historique de la langue française. Tome 1. (4th edition). Copenhagen: Gyldendalske Boghandel, Nordisk Forlag. Ojeda, Almerindo (1995). ‘The semantics of the Italian double plural’. Journal of Semantics 12: 213–257 Oliveira, Fa´tima (2013). ‘Tempo verbal’. In Eduardo Buzaglo Paiva Raposo et al. (eds), Grama´tica do português. Lisbon: Fundação Calouste Gulbenkian, pp. 509–553. O’Neill, Paul (forthcoming). ‘Morphologization and the boundary between morphology and phonology’. Oxford Research Encyclopedia of Romance Linguistics. OVI = The Opera del Vocabolario Italiano textual database. Available at: http://www. vocabolario.org or http://gattoweb.ovi.cnr.it/(S(se3fsd1kvoeagufqio3tu4uv))/Cat Form01.aspx or http://www.ovi.cnr.it/Interroga-il-Corpus.html. Paciaroni, Tania (2009). ‘Sull’ausiliazione perfettiva in maceratese’. In Bruno Moretti, Elena Maria Pandolfi, and Matteo Casoni (eds), Linguisti in contatto. Ricerche di linguistica italiana in Svizzera. Atti del Convegno dell’Osservatorio linguistico della Svizzera italiana (OLSI), Bellinzona, 16–17 novembre 2007. Bellinzona: OLSI, pp. 37–58. Postprint available at: https://www.zora.uzh.ch/. Pagotto, Emilio G. (1998). ‘Norma e condescendência: ciência e pureza’. Lı´nguas e instrumentos linguı´sticos 2: 49–68. Palay, Simin (1961). Dictionnaire du béarnais et du gascon modernes. Paris: CNRS. Palmer, Frank R. (2001). Mood and Modality (2nd edition). Cambridge: Cambridge University Press. Pamfil, Viorica (1973). ‘Formele de trecut indicativ în limba română din secolul al XVI-lea și frecvența lor’. Cercetări de lingvistică 18: 205–216. Pană Dindelegan, Gabriela (2012). ‘Verbul’. In Gheorghe Chivu, Gabriela Pană Dindelegan, Adina Dragomirescu, Isabela Nedelcu, and Irina Nicula (eds), Studii de istorie a limbii române. Morfosintaxa limbii literare în secolele al XIX-lea și al XX-lea. Bucharest: Editura Academiei Române, pp. 535–596. Pană Dindelegan, Gabriela (2013a). ‘Flexiunea cazuală: între analitic și sintetic. Cardinale și cuantificatori non-numerici în româna veche’. Limba română 62: 159–173. Pană Dindelegan, Gabriela (ed.) (2013b). The Grammar of Romanian. Oxford: Oxford University Press. Pană Dindelegan, Gabriela (ed.) (2016). The Syntax of Old Romanian. Oxford: Oxford University Press. Paoli, Sandra (2020). ‘Avér o gavér? Questo è il dilemma! Microvariazione negli esiti del latino habere nel nord d’Italia’. Revue romane 55: 283–310. Parodi, Ernesto G. (1898). ‘Studj liguri’. Archivio glottologico italiano 15: 1–110. Parry, Mair (1995). ‘Some observations on the syntax of clitic pronouns in Piedmontese’. In Smith and Maiden (eds), pp. 133–160. Parry, Mair (1997a). ‘Piedmont’. In Maiden and Parry (eds), pp. 237–244. Parry, Mair (1997b). ‘The reinterpretation of the reflexive in Piedmontese: “impersonal” SE constructions’. Transactions of the Philological Society 96: 63–116. Parry, Mair (2000). ‘Accordo e soggetti postverbali in piemontese’. In Actes du XXIIe Congrès International de Lingu¨istique e Philologie Romanes, Bruxelles 1998, VI: De la Grammaire des formes à la grammaire du sens. Tu¨bingen: Niemeyer, pp. 391–402. Parry, Mair (2005). Parluma ‘d Coiri. Sociolinguistica e grammatica del dialetto di Cairo Montenotte. Savona: Editrice Liguria.

450

references and bibliographical abbreviations

Parry, Mair (2013). ‘Variation and change in the presentational constructions of northwestern Italo-Romance Varieties’. In Elly Van Gelderen, Michela Cennamo, and Jóhanna Barðdal (eds), Argument Structure in Flux. Amsterdam: Benjamins, pp. 511–548. Partee, Barbara H. (2000). ‘A history of the UMass Linguistics Department to 1999’. In Paul de Lacy and Anita Nowak (eds), UMOP 24: Papers from the 25th Reunion. Amherst, MA: GLSA, pp. xiii–xli. Pellegrini, Giovan Battista (1977). Carta dei dialetti italiani. Pisa: Pacini. Penny, Ralph J. (2000). Variation and Change in Spanish. Cambridge: Cambridge University Press. Peperkamp, Sharon (1997). Prosodic Words. The Hague: Holland Academic Graphics. Pérez Saldanya, Manuel (2002). ‘Les relacions temporals i aspectuals’. In Joan Solà, MariaRosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds), Gramàtica del català contemporani. Vol. 3. Barcelona: Empúries, pp. 2567–2662. Perlmutter, David M. (1978). ‘Impersonal Passives and the Unaccusative Hypothesis’. Proceedings of the Fourth Annual Meeting of the Berkeley Linguistics Society (BLS 4): 157–189. Perlmutter, David M. (1989). ‘Multiattachment and the Unaccusative Hypothesis: the perfect auxiliary in Italian’. Probus 1: 63–119. Pescarini, Diego (2012). ‘Clitici espletivi nei dialetti settentrionali: i dati dell’Atlante Sintattico d’Italia (ASIt)’. Atti del Sodalizio Glottologico Milanese (nuova serie) 7: 42–56. Pescarini, Diego (2016). ‘Clitic pronominal systems: morphophonology’ in Ledgeway & Maiden (eds), pp. 742-757. Pescarini, Diego (2019). ‘Subject clitics and agreement: a reassessment’. MS, CNRS, Université Côte d’Azur, Nice. Pescarini. Diego and Michele Loporcaro (forthcoming). ‘Variation in Romance’. In Adam Ledgeway and Martin Maiden (eds), The Cambridge Handbook of Romance Linguistics. Cambridge: Cambridge University Press. Pessoa, Marlos de Barros (2003). Formação de uma variedade urbana e semi-oralidade: o caso do Recife, Brasil. Tu¨bingen: Niemeyer. Petracco Sicardi, Giulia (1978). ‘La lingua e le formule delle carte piacentine alto-medievali’. In Paola Galetti and Giulia Petracco Sicardi (eds), Le carte private della Cattedrale di Piacenza. Parma: Deputazione di storia patria per le province parmensi, pp. 107–187. Philippide, Alexandru (1927). Originea românilor, II. Iași: Viața Românească. Pinkster, Harm (1985). ‘The development of future tense auxiliaries in Latin’. Glotta 63: 186–208. Pinkster, Harm (1990). ‘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, pp. 195–209. Pinto, Manuela (1997). Licensing and Interpretation of Inverted Subjects in Italian. Utrecht: UiL OTS Dissertation series. Pisano, Simone (2010.) ‘L’utilizzo dei tempi sovraccomposti in alcune varietà sarde moderne’. Lingua e stile 45: 125–133. Plank, Frans (1991). ‘Of abundance and scantiness in inflection: a typological prelude’. In Frans Plank (ed.) Paradigms: The Economy of Inflection. Berlin: Mouton de Gruyter, pp. 1–39. Plank, Frans and Nigel Vincent (2019). ‘Suppletion: questions for history and theory’. Transactions of the Philological Society 117: 319–337. Poggi, Giuseppe (1992). Ginn-a de Sampedaenn-a (ed. Fiorenzo Toso). Genoa: Microart.

references and bibliographical abbreviations

451

Poletto, Cecilia (1992). ‘The analysis of the passé surcomposé: a hypothesis on sentence structure formation’. University of Venice Working Papers in Linguistics 1.7: 1–24. Poletto, Cecilia (2000). The Higher Functional Field: Evidence from Northern Italian Dialects. Oxford: Oxford University Press. Poletto, Cecilia (2001). ‘Complementizer deletion and verb movement’. In Cinque and Salvi (eds), pp. 265–286. Poletto, Cecilia (2008). ‘Syntactic encoding of aspect in some northern Italian dialects’. In Ángeles Carrasco Gutiérrez (ed.), Tiempos compuestos y formas verbales complejas. Madrid: Vervuet, pp. 499–517. Poletto, Cecilia (2009). ‘Double auxiliaries, anteriority and terminativity’. Journal of Comparative Germanic Linguistics 12: 31–48 Poletto, Cecilia and Christina Tortora (2016). ‘Subject clitics: syntax’. In Ledgeway and Maiden (eds), pp. 772–785. Pomino, Natascha and Eva-Maria Remberger (2019). ‘Verbal suppletion in Romance synchrony and diachrony: the perspective of Distributed Morphology’. In Nigel Vincent and Frans Plank (eds), The Diachrony of Suppletion. Special issue of Transactions of the Philological Society 117: 471–497. Popescu-Marin, Magdalena (2008). ‘Adjectivul’. In Valeria Guțu-Romalo (ed.), Gramatica limbii române. I: Cuvântul. Bucharest: Editura Academiei Române, pp. 141–179. Poplack, Shana and Danielle Turpin (1999). ‘Does the FUTUR have a future in (Canadian) French?’. Probus 11: 134–164. Posner, Rebecca (1996). The Romance Languages. Cambridge: Cambridge University Press. Posner, Rebecca (1997). Linguistic Change in French. Oxford: Clarendon Press. Pountain, Christopher (2011). ‘Latin and the structure of written Romance’. In Maiden, Smith, and Ledgeway (eds), pp. 606–659. Pratas, Fernanda, Clara Pinto, and Sandra Pereira (eds) (2016). Coordination and Subordination: Form and Meaning. Selected Papers from CSI Lisbon 2014. Newcastle: Cambridge Scholars. Prati, Angelico (1942). ‘Antisuffissi’. L’Italia dialettale 18: 75–166. Price, Glanville (1971). The French Language: Present and Past. London: Edward Arnold. Proietti, Domenico (2019). «Kelle terre». Storia, lingua e toponomastica nei giudicati campani del X secolo. Rome: Aracne. Pullum, Geoffrey K. and Arnold Zwicky (1988). ‘The syntax–phonology interface’. In Frederick J. Newmeyer (ed.), Linguistics: The Cambridge Survey. Vol. I: Linguistic Theory: Foundations. Cambridge: Cambridge University Press, pp. 255–280. Quer, Josep (2016). ‘Mood’. In Ledgeway and Maiden (eds), pp. 954–966. Raiol, Domingos Antônio (1900). Catechese dos ´ındios do Para´. Vol. 2. Belém, PA: ABAPP. Ramat, Paolo (2001). ‘Degrammaticalization or transcategorization?’. In Schaner-Wolles, Rennison, and Neubarth (eds), pp. 393–401. Raoult, Didier et al. (2006). ‘Evidence for louse-transmitted diseases in soldiers of Napoleon’s Grand Army in Vilnius’. Journal of Infectious Diseases 193: 112–120. Rappaport Hovav, Malka and Beth Levin (2000). ‘Classifying single argument verbs’. In Peter Coopmans, Martin B. H. Everaert, and Jane Grimshaw (eds) Lexical Specification and Insertion. Amsterdam: Benjamins, pp. 269–304. Rea, Béatrice (2020). ‘A comparative analysis of French auxiliation, with new evidence from Montréal’. In Wolfe and Maiden (eds), pp. 213–240.

452

references and bibliographical abbreviations

Rebotier, Aude (2015). ‘Le futur périphrastique français avec aller : un renvoi spécifique à l’avenir ou un temps en voie de grammaticalisation? Une approche contrastive’. Revue de sémantique et pragmatique 38: 11–34. Recasens i Vives, Daniel (2017). Fonètica històrica del català. Barcelona: Institut d’Estudis Catalans. Régnier-Desmarais, François-Séraphin (1705). Traité de la grammaire françoise. Paris: JeanBaptiste Coignard. Reich, Uli (2003). ‘Mudança sinta´tica e pragma´tica na lı´ngua geral amazônica (LGA): marcação de caso e sistema pronominal’. In José R. Bessa Freire and Maria Carlota Rosa (eds), Lı´nguas gerais: polı´tica linguı´stica e catequese na América do Sul no perı´odo colonial. Rio de Janeiro, RJ: EDUERJ, pp. 167–184. Reichenbach, Hans (1947). Elements of Symbolic Logic. New York: Macmillan & Co. Remacle, Louis (1956). Syntaxe du parler wallon de La Gleize. Tome 2: Verbes, adverbes, prépositions. Paris: Les Belles Lettres. Renzi, Lorenzo (1993). ‘Da dove viene l’articolo il?’. In Johannes Kramer and Guntram A. Plangg (eds), Verbum Romanicum: Festschrift fu¨r Maria Iliescu. Hamburg: Buske, pp. 215–230. Repetti, Lori (2012). ‘Consonant-final loanwords and epenthetic vowels in Italian’. Catalan Journal of Linguistics 11: 167–188. Repetti, Lori (2021). ‘The masculine singular definite article in Italian: the role of the syllable’. To appear in Italian Journal of Linguistics. Ricketts, Peter T. (2005). Concordance de l’occitan médiéval. Turnhout: Brepols. Rizzi, Luigi (1982). Issues in Italian Syntax. Dordrecht: Foris. Rizzi, Luigi (1986). ‘On the status of subject clitics in Romance’. In Osvaldo Jaeggli and Carmen Silva-Corvala´n (eds), Studies in Romance Linguistics. Dordrecht: Foris, pp. 391– 419. Rizzi, Luigi (2005). ‘On some properties of subjects and topics’. In Laura Brugè, Giuliana Giusti, Nicola Munaro, Walter Schweikert, and Giuseppina Turano (eds), Contributions to the XXX Incontro di Grammatica Generativa. Venice: Cafoscarina, pp. 203–224. Roberts, Ian (2009). ‘A formal account of grammaticalization in the history of Romance futures’. Folia Linguistica 26: 219–258. Roberts, Ian (2019). Parameter Hierarchies and Universal Grammar. Oxford: Oxford University Press. Roberts, Ian and Anna Roussou (2002). ‘The history of the future’. In David Lightfoot (ed.), Syntactic Effects of Morphological Change. Oxford: Oxford University Press, pp. 23–56. Roberts, Ian and Anna Roussou (2003). Syntactic Change: A Minimalist Approach to Grammaticalizaton. Cambridge: Cambridge University Press. Rodrigues, Aryon Dall’Igna (1996). ‘As lı´nguas gerais sul-americanas’. Papia: Revista brasileira de estudos crioulos e similares 4: 6–18. Rodrigues, Aryon Dall’Igna (2003). ‘Obra nova da lı´ngua geral de mina: a lı´ngua ewe nas Minas Gerais’. Papia: Revista brasileira de estudos crioulos e similares 13: 92–96. Rodrigues, Aryon Dall’Igna (2010). ‘Estrutura do Tupinamba´’. In Cabral, Rodrigues, and Duarte (eds), pp. 11–42. Rodrigues, Aryon Dall’Igna and Ana Suelly Arruda Câmara Cabral (2011). ‘A contribution to the linguistic history of the lı´ngua geral amazônica’. Alfa 55: 613–639. Rodrigues, João Barbosa (2017). Poranduba amazonense, ou kochiyma-uara porandub (ed. Tenório Telles). Manaus, AM: Valer [1st edition, Rio de Janeiro, RJ: Leuzinger, 1890]. Rohlfs, Gerhard (1966). Grammatica storica della lingua italiana e dei suoi dialetti. I: Fonetica. Turin: Einaudi.

references and bibliographical abbreviations

453

Rohlfs, Gerhard (1968). Grammatica storica della lingua italiana e dei suoi dialetti. II: Morfologia. Turin: Einaudi. Rohlfs, Gerhard (1969). Grammatica storica della lingua italiana e dei suoi dialetti. III: Sintassi e formazione delle parole. Turin: Einaudi. Rohlfs, Gerhard (1970). Le Gascon: études de philologie pyrénéenne. Tu¨bingen: Niemeyer. Rohlfs, Gerhard (1971). ‘Autour de l’accusatif prépositionnel dans les langues romanes: concordances et discordances’. Revue de linguistique romane 35: 312–334. Roller, Heather F. (2014). Amazonian Routes: Indigenous Mobility and Colonial Communities in Northern Brazil. Stanford, CA: Stanford University Press. Romagnoli, Serena (2018). Urbinate. Available at: https://www.dai.uzh.ch/DAI_Urbinate %20Overview.pdf [last accessed 8 November 2020]. Ronjat, Jules (1937). Grammaire istorique [sic] des parlers provencaux modernes, III. Montpellier: Société des Langues Romanes. Rose, Yvan and Katherine Demuth (2006). ‘Vowel epenthesis in loanword adaptation: representational and phonetic considerations’. Lingua 116: 1112–1139. Rosemeyer, Malte (2016). ‘The development of iterative verbal periphrases in Romance’. Linguistics 54: 235–272. Rosen, Carol (1982). ‘The unaccusative hypothesis and the “inherent clitic” phenomenon in Italian’. Proceedings of the Eigteenth Annual Meeting of the Chicago Linguistic Society (CLS 18): 530–541. Rosen, Carol (1984). ‘The interface between semantic roles and initial grammatical relations’. In David M. Perlmutter and Carol Rosen (eds), Studies in Relational Grammar 2. Chicago: Chicago University Press, pp. 38–77. Rosen, Carol (1997). ‘Auxiliation and serialization: on discerning the difference’. In Alex Alsina, Joan Bresnan, and Peter Sells (eds), Complex Predicates. Stanford, CA: CSLI Publications, pp. 175–202. Rosén, Hannah (1999). Latine loqui. Trends and Directions in the Crystallization of Classical Latin. Munich: Fink. Rosetti, Alexandru (1932). Limba română în secolul al XVI-lea. Bucharest: Cartea Românească. Rosetti, Alexandru (1946). Istoria limbii române, VI. Bucharest: Fundația Regală pentru Literatură și Artă. Rosetti, Alexandru (1986). Istoria limbii române. I. De la origini până la începutul secolului al XVII-lea. Bucharest: Editura Științifică și Enciclopedică. Rusu, Valeriu (ed.) (1984). Tratat de dialectologie românească. Craiova: Scrisul Românesc. Sabatini, Francesco (1965). ‘Esigenze di realismo e dislocazione morfologica in testi preromanzi’. Rivista di cultura classica e medievale 7: 972–998 Sabatini, Francesco (1968). ‘Dalla “scripta latina rustica” alle “scriptae” romanze’. Studi medievali 3: 320–358. Saccon, Graziella (1993). Post-verbal Subjects: a study based on Italian and its dialects. Unpublished Ph.D. dissertation, Harvard University. Salvi, Giampaolo (2011) ‘Morphosyntactic persistence’. In Martin Maiden, John Charles Smith, and Adam Ledgeway (eds), The Cambridge History of the Romance Languages. Cambridge: Cambridge University Press, 318–381. Salvi, Giampaolo and Lorenzo Renzi (eds) (2010). Grammatica dell’italiano antico (2 vols). Bologna: il Mulino. Samek–Lodovici, Vieri (2002). ‘Agreement impoverishment under subject inversion: a crosslinguistic analysis’. In Gisbert Fanselow and Caroline Féry (eds), Resolving Conflicts in Grammar (Linguistiche Berichte, Sonderheft 11), pp. 49–82.

454

references and bibliographical abbreviations

Sampaio, Dilcélia Almeida (2001). Modo imperativo: sua manifestação/expressão no português contemporâneo. Unpublished dissertação de mestrado, Universidade Federal da Bahia (UFBA). Sa´nchez Arrobo, Marı´a Elena (2010). ‘Auxiliares “ser” y “haber” en los tiempos compuestos del español antiguo’. In Memorias del VI Foro de estudios en lenguas internacional. Chetumal: Universidad de Quintana Roo, pp. 611–622. Sandfeld, Kristian (1930). Linguistique balkanique: problèmes et résultats. Paris: Champion. Sankoff, Gillian and Hélène Blondeau (2007). ‘Language change across the lifespan: /r/ in Montreal French’. Language 83: 560–588. Sankoff, Gillian and Pierrette Thibault (1977). ‘L’alternance entre les auxiliaires avoir et être en français parlé à Montréal’. Langue française 34: 81–198. Sankoff, Gillian and Pierrette Thibault (1980). ‘The alternation between the auxiliaries avoir and être in Montreal French’. In Gillian Sankoff (ed.), The Social Life of Language. Philadelphia: University of Pennsylvania Press, pp. 311–345. Sankoff, Gillian and Suzanne Evans Wagner (2006). ‘Age-grading in retrograde movement: the inflected future in Montréal French’. In Papers from NWAV 34. University of Pennsylvania Working Papers in Linguistics 12(2): 203–216. Sapir, Edward ([1925] 1949). ‘Sound patterns in language’. Language 1: 37–51 [reprinted in David B. Mandelbaum (ed.), Selected Writings of Edward Sapir in Language, Culture and Personality. Berkeley: University of California Press [1949], pp. 33–45]. Saussure, Louis de and Bertrand Sthioul (2012). ‘The surcomposé past tense’. In Robert I. Binnick (ed.), The Oxford Handbook of Tense and Aspect. Oxford: Oxford University Press, pp. 586–610. Savoia, Leonardo (1997). ‘The geographical distribution of the dialects’. In Maiden and Parry, pp. 225–234. Scaglione, Massimo (ed.) (1982). Teatro in piemontese: antologia di testi con note storiche e di regia. Turin: Piazza. Scalise, Sergio (1984). Generative Morphology. Dordrecht: Foris. Schaden, Gerhard (2012). ‘Modelling the “aoristic drift of the present perfect” as inflation: an essay in historical pragmatics’. International Review of Pragmatics 4: 261–292. Schaner-Wolles, Chris, John Rennison, and Friedrich Neubarth (eds) (2001). Naturally! Linguistic Studies in Honour of Wolfgang Ulrich Dressler Presented on the Occasion of His 60th Birthday. Turin: Rosenberg & Sellier. Scherre, Maria Marta Pereira (2012). ‘Padrões sociolinguı´sticos do português brasileiro: a importância da pesquisa variacionista’. Revista do Programa de Pós-Graduação em Estudo de Linguagens, Universidade do Estado da Bahia 4: 1–32. Scherre, Maria Marta Pereira, Lilian Coutinho Yacovenco, and Anthony Julius Naro (2018). ‘Nós e a gente no português brasileiro: concordâncias e discordâncias’. Estudios de lingu¨´ıstica galega (volume especial) 1: 13–27. Schifano, Norma (2018). Verb Movement in Romance: A Comparative Study. Oxford: Oxford University Press. Schilling-Estes, Natalie and Walt Wolfram (1994). ‘Convergent explanation and alternative regularization patterns: were/weren’t leveling in a vernacular English variety’. Language Variation and Change 6: 273–302. Schirru, Giancarlo (2018). ‘Avita fatta: non-etymological forms of auxiliary habere in southern Italian dialects’. In Roberta D’Alessandro and Diego Pescarini (eds), Advances in Italian Dialectology: Sketches of Italo-Romance Grammars. Leiden: Brill, pp. 188–213.

references and bibliographical abbreviations

455

Schlegel, August Wilhelm von (1818). Observations sur la langue et la littérature provençales. Paris: Librairie grecque-latine-allemande. Schmidt-Riese, Roland (1998). ‘Perspectivas diacrônicas brasileiras: o rastro das lı´nguas gerais’. Romanistisches Jahrbuch 40: 307–335. Schmidt-Riese, Roland (2003). ‘Condições da mudança em Yẽgatú: pragma´tica e contatos linguı´sticos’. In José R. Bessa Freire and Maria Carlota Rosa (eds), Lı´nguas gerais: polı´tica linguı´stica e catequese na América do Sul no perı´odo colonial. Rio de Janeiro, RJ: EDUERJ, pp. 147–168. Schogt, H. G. (1964). ‘L’aspect verbal en français et l’élimination du passé simple’. Word 20: 1–17. Schwade, Michéli Carolı´ni de Deus Lima (2014). Descrição fonético-fonológico do Nheengatu falado no Médio Rio Amazonas. Unpublished dissertação de mestrado, Universidade Federal do Amazonas (UFAM). Schwegler, Armin (1990). Analyticity and Syntheticity: A Diachronic Perspective with Special Reference to Romance Languages. Berlin: Mouton de Gruyter. Scorretti, Mauro (1994). Complementizer Deletion. Unpublished doctoral thesis, University of Amsterdam. Seifert, Eva (1930a). “‘Haber” y “tener” como expresiones de posesión en español’. Revista de filologı´a española 17: 233–276, 345–389. Seifert, Eva (1930b). ‘Tenere “haben” im Romanischen’. Zeitschrift fu¨r romanische Philologie 50: 1–28. Seifert, Eva (1935). tenere ‘haben’ im Romanischen. Florence: Olschki. Seifert, Eva (1957). ‘Die Verben habere und tenere im Katalanischen’. Estudis romànics 6: 1–74. Seki, Lucy (1990). ‘Kamaiura (Tupi-Guarani) as an active-stative language’. In Doris Payne (ed.), Amazonian Linguistics: Studies in Lowland South American Languages. Austin, TX: University of Texas Press, pp. 367–392. Seki, Lucy (2000). Grama´tica do kamaiura´: lı´ngua tupi-guarani do Alto Xingu. Campinas, SP: Editora da UNICAMP & São Paulo, SP: Imprensa Oficial. Selkirk, Elisabeth (1995). ‘The prosodic structure of function words’. In Jill N. Beckman, Laura Walsh Dickey, and Suzanne Urbanczyk (eds), Papers in Optimality Theory. University of Massachusetts Occasional Papers 18. Amherst: University of Massachusetts, Graduate Linguistic Student Association, pp. 439–469. Serianni, Luca (1988). Grammatica italiana (in collaboration with Alberto Castelvecchi). Turin: UTET. Setterberg-Jørgensen, Birgit (1950). Andare, venire et tornare. Verbes copules et auxiliaires dans la langue italienne. Aarhus: Universitetsforlaget. Sgroi, S. C. (2012). ‘Le vocatif et l’“allocution inverse” en italien’. Rivista italiana di linguistica e dialettologia 14: 65–88. Sheehan, Michelle (2010). “‘Free” inversion in Romance and the Null Subject Parameter’. In Theresa Biberauer, Anders Holmberg, Ian Roberts, and Michelle Sheehan (eds), Parametric Variation: Null Subjects in Minimalist Theory. Cambridge: Cambridge University Press, pp. 231–262. Sibille, Jean (2007). ‘Le prétérit périphrastique’. In Marcellin Richard (ed.), La Passion de saint André suivie d’une étude linguistique comparée par Jean Sibille. Paris: Champion, pp. 253–269. Siddiqi, Daniel (2018). ‘Distributed Morphology’. In Audring and Masini (eds), pp. 143– 165. Siegel, Jeff (1985). ‘Koinés and koinéization’. Language in Society 14: 357–378.

456

references and bibliographical abbreviations

Sluckin, Benjamin L., Silvio Cruschina, and Fabienne Martin (2021). ‘Locative inversion in Germanic and Romance: a conspiracy theory’. In Wolfe and Meklenborg (eds), pp. 165–194. Smith, John (1656). The Mysterie of Rhetorique Unvail’d. London: George Eversden. Smith, John Charles (1989). ‘Actualization reanalyzed: evidence from the Romance compound past tenses’. In Thomas J. Walsh (ed.), Synchronic and Diachronic Approaches to Linguistic Variation and Change (Georgetown University Round Table on Languages and Linguistics, 1988). Washington, DC: Georgetown University Press, pp. 310–325. Smith, John Charles (1991). ‘Problemi dell’accordo del participio passato coll’oggetto diretto nei tempi composti coniugati con avere in italiano, con speciale riferimento ai dialetti’. In Luciano Giannelli, Nicoletta Maraschio, Teresa Poggi Salani, and Massimo Vedovelli (eds), Tra Rinascimento e strutture attuali: saggi di linguistica italiana (Atti del Primo Convegno della Società Internazionale di Linguistica e Filologia Italiana, Siena, 28–31 marzo 1989. Vol. I). Turin: Rosenberg & Sellier, pp. 365–371. Smith, John Charles (1995). ‘Perceptual factors and the disappearance of agreement between past participle and direct object in Romance’. In Smith and Maiden (eds), pp. 161–180. Smith, John Charles (1996). ‘Surfonctionnalité et hyperanalyse: l’accord du participe passé dans les langues romanes à la lumière de deux théories récentes’. Faits de langues 8: 113– 120. Smith, John Charles (1997). ‘Types and tokens in language change: some evidence from Romance’. In Raymond Hickey and Stanisław Puppel (eds), Language History and Linguistic Modelling: A Festschrift for Jacek Fisiak on His 60th Birthday. Berlin: Mouton de Gruyter, pp. 1099–1111. Smith, John Charles (1999). ‘Markedness and morphosyntactic change revisited: the case of Romance past participle agreement’. In Sheila Embleton, John E. Joseph, and HansJosef Niederehe (eds), The Emergence of the Modern Language Sciences: Studies on the Transition from Historical–Comparative to Structural Linguistics in Honour of E. F. K. Koerner. Vol. 2: Methodological Perspectives and Applications. Amsterdam: Benjamins, pp. 203–215. Smith, John Charles (2000). ‘Reanalysis and extension: notes and queries’. In Irena Kovačič, Milena Milojevic´-Sheppard, Silvana Orel-Kos, and Janez Orezˇnik (eds), Linguistics and Language Studies: Exploring Language from Different Perspectives. Ljubljana: Filozofska Fakulteta Univerze v Ljubljani, pp. 45–60. Smith, John Charles (2001). ‘Markedness, functionality, and perseveration in the actualization of a morphosyntactic change’. In Henning Andersen (ed.), Actualization: Linguistic Change in Progress. Amsterdam: Benjamins, pp. 203–223. Smith, John Charles (2011a). ‘Change and continuity in form-function relationships’. In Maiden, Smith, and Ledgeway (eds), 268–317, 713–723. Smith, John Charles (2011b). ‘Variable analyses of a verbal inflection in (mainly) Canadian French’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 311–326 Smith, John Charles (2016). ‘French and northern Gallo-Romance’. In Ledgeway and Maiden (eds), pp. 292–318. Smith, John Charles (2020). ‘Old, Middle, and Modern: temporality and typology’. In Keith Allan (ed.), Dynamic Language Changes: Looking in and across Languages. Singapore: Springer Nature, pp. 183–200. Smith, John Charles (2021). ‘Have in English and Romance’. In Wolfe and Meklenborg (eds), pp. 495–531.

references and bibliographical abbreviations

457

Smith, John Charles and Martin Maiden (eds) (1995). Linguistic Theory and the Romance Languages. Amsterdam: Benjamins. Solà i Pujols, Jaume (2002). ‘Modificadors temporals i aspectuals’. In Joan Solà, MariaRosa Lloret, Joan Mascaró, and Manuel Pérez Saldanya (eds), Gramàtica del català contemporani, vol. 3. Barcelona: Empúries, pp. 2867–2936. So¨ll, Ludwig (1983). ‘De la concurrence du futur simple et du futur proche en français moderne’. In Franz-Josef Hausman (ed.), Études de grammaire française descriptive. Heidelberg: Groos, pp. 292–318. Sorace, Antonella (2000). ‘Gradients in auxiliary selection with intransitive verbs’. Language 76: 859–890. Sorace, Antonella (2004). ‘Gradience at the lexicon–syntax interface: evidence from auxiliary selection and implications for unaccusativity’. In Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert (eds), The Unaccusativity Puzzle: Explorations of the Syntax–Lexicon Interface. Oxford: Oxford University Press, pp. 243–268. Sornicola, Rosanna (1976). ‘Vado a dire, vaiu a ddicu: problema sintattico o problema semantico?’. Lingua nostra 37: 65–74. Sornicola, Rosanna (1997). ‘L’oggetto preposizionale in siciliano antico e in napoletano antico: considerazioni su un problema di tipologia diacronica’. Italienische Studien 18: 66–80. Sornicola, Rosanna (2003). ‘Polimorfismo e instabilità strutturale: un esame della dittongazione spontanea dell’area flegrea in una prospettiva romanza’. In Fernando Sa´nchez Miret (ed.), Actas del XXIII Congreso Internacional de Lingu¨´ıstica y Filologı´a Roma´nica (Salamanca, 24–30 settembre 2001). Tu¨bingen: Niemeyer, pp. 301–313. Sornicola, Rosanna (2006). ‘Expletives and dummies’. In Keith Brown (ed.), Encyclopedia of Language and Linguistics. Vol. 4 (2nd edition). Oxford: Elsevier, 399–410. Sornicola, Rosanna (2007). ‘Nominal inflection and grammatical relations in tenth-century legal documents from the South of Italy (Codex Diplomaticus Amalfitanus)’. In Roger Wright (ed.), Latin Vulgaire—Latin Tardif VIII. Actes du VIIIe Colloque international sur le latin vulgaire et tardif, Oxford, 6–9 Septembre 2006. Hildesheim: Olms, pp. 510–520. Sornicola, Rosanna (2012). ‘Potenzialità e problemi dell’analisi linguistica dei documenti notarili alto-medievali dei domini bizantini e longobardi’. In Rosanna Sornicola and Paolo Greco (eds), La lingua dei documenti notarili alto-medievali dell’Italia meridionale. Bilancio di studi e prospettive di ricerca. Naples: Tavolario, pp. 9–62. Sornicola, Rosanna (2013). ‘ “Vulgo dicitur”: vulgarisms in legal Latin’. Journal of Latin Linguistics 12: 269–299. Sornicola, Rosanna (2014). ‘On the history of European functionalism’. La Linguistique 2(50): 7–62, Sornicola, Rosanna (2015). ‘Curiales, notarii, presbyteri nella Campania alto-medievale. Alcuni problemi di sociolinguistica storica, con particolare riguardo alla morfosintassi’. In Carlo Consani (ed.), Contatto interlinguistico tra presente e passato. Milan: LED Edizioni Universitarie, pp. 237–282. Sornicola, Rosanna (2016a). ‘Una prospettiva sul cambiamento dei paradigmi morfologici: lo studio della formazione del plurale italo-romanzo in base all’analisi del polimorfismo’. In Patrizia Cordin and Alessandro Parenti (eds), Problemi e prospettive della linguistica storica. Atti del XL Convegno della Società Italiana di Glottologia (Trento, 22–24 ottobre 2015). Roma: Il Calamo, pp. 111–130. Sornicola, Rosanna (2016b). ‘La rappresentazione del polimorfismo tra teoria e storia nella linguistica romanza’. Paper presented to ‘Dia IV’ Conference (Zurich, 12–14 September 2016).

458

references and bibliographical abbreviations

Sornicola, Rosanna (2017a). “‘Transizione” e “transizioni” dal latino al romanzo: il progetto di analisi linguistica dei documenti cavensi del IX secolo’. In Sornicola, D’Argenio, and Greco (eds), pp. 13–25. Sornicola, Rosanna (2017b). ‘La morfologia nominale: polimorfismo e polifunzionalismo nei sistemi di flessione’. In Sornicola, D’Argenio, and Greco (eds), 85–134. Sornicola, Rosanna (2018). ‘The study of polymorphism and its relevance to historical linguistics’, invited lecture to Wiener Sprachgesellschaft (Österreichischer Akademie der Wissenschaften), 22 May 2018. Sornicola, Rosanna (2020). ‘Processi di pluralizzazione in siciliano: classi flessive, stampi prosodici e rappresentazioni morfologiche’. Bollettino del Centro di studi linguistici e filologici siciliani 31: 133–200. Sornicola, Rosanna, Elisa D’Argenio, and Paolo Greco (eds) (2017). Sistemi, norme, scritture. La lingua delle più antiche carte cavensi. Naples: Giannini. Spencer, Andrew (2001). ‘The paradigm-based model of morphosyntax’. Transactions of the Philological Society 99: 279–314. Spencer, Andrew (2003). ‘Periphrastic paradigms in Bulgarian’. In Uwe Junghanns and Luka Szucsich (eds), Syntactic Structures and Morphological Information. Oxford: Elsevier, pp. 249–293. Spencer, Andrew (2006). ‘Periphrasis’. In Keith Brown (ed.), The Encyclopedia of Language and Linguistics. Oxford: Elsevier, pp. 287–293. Spencer, Andrew (2013). Lexical Relatedness: A Paradigm-based Model. Oxford: Oxford University Press. Spencer, Andrew and Arnold M. Zwicky (1998). ‘Introduction’. In Andrew Spencer and Arnold M. Zwicky (eds), The Handbook of Morphology. Oxford: Blackwell, pp. 1–10. Spencer, Andrew and Gergana Popova (2015). ‘Periphrasis and inflection’. In Baerman (ed.), pp. 197–232. Squartini, Mario (1998). Verbal Periphrases in Romance: Aspect, Actionality, and Grammaticalization. Berlin: de Gruyter. Squartini, Mario (1999). ‘On the semantics of the pluperfect: evidence from Germanic and Romance’. Linguistic Typology 3: 51–90. Squartini, Mario (2005). ‘L ’evidenzialità in rumeno e nelle altre lingue romanze’. Zeitschrift fu¨r romanische Philologie 121: 246–268. Staden, Hans (1974). Duas viagens ao Brasil. Belo Horizonte, MG: Itatiaia and São Paulo, SP: EDUSP. Stan, Camelia (2013a), ‘Genitive and dative case-marking’. In Pană Dindelegan (ed.), pp. 262–272. Stan, Camelia (2013b). O sintaxă diacronică a limbii române vechi. Bucharest: Editura Universități din București. Stan, Camelia (2016). ‘Analytic marking: prepositions a, de, la’. In Pană Dindelegan (ed.), pp. 317–321. Stati, Sorin (1965). ‘Verbul’. In Alexandru Graur (ed.), Istoria limbii române. Vol. 1. Bucharest: Editura Academiei, pp. 166–197. Steele, Susan (1995). ‘Towards a theory of morphological information’. Language 71: 260– 309. Stéfanini, Jean (1954). ‘La tradition grammaticale française et les temps surcomposés’. Annales de la Faculté des Lettres d’Aix 28: 67–108. Stewart, Thomas and Gregory Stump (2007). ‘Paradigm function morphology and the morphology–syntax interface’. In Gillian Ramchand and Charles Reiss (eds), The Oxford Handbook of Linguistic Interfaces. Oxford: Oxford University Press, pp. 383–422.

references and bibliographical abbreviations

459

Štichauer, Pavel (2018). ‘Lexical splits within periphrasis: mixed perfective auxiliation systems in Italo-Romance’. Morphology 28: 1–23. Štichauer, Pavel (2019). ‘Mixed paradigms in Italo-Romance: a case of morphologization of auxiliary selection?’. In Cruschina, Ledgeway, and Remberger (eds), pp. 79–100. Stotz, Peter (1996–2004). Handbuch zur lateinischen Sprache des Mittelalters (4 vols). Munich: Beck. Stradelli, Ernano (1929). ‘Vocabula´rio da lı´ngua geral português-yẽgatú e yẽgatú-português, precedidos de um esboço de gramma´tica nheenga-umbuê-sa´ua’. Revista do Instituto Histórico e Geogra´fico Brasileiro 158 (104): 9–768 [2nd edition, Cotia, SP: Ateliê, 2014]. Stump, Gregory T. (1991). ‘A paradigm-based theory of morphosemantic mismatches’. Language 67: 675–725. Stump, Gregory (1993). ‘How peculiar is evaluative morphology?’. Journal of Linguistics 29: 1–36. Stump, Gregory T. (2001). Inflectional Morphology: A Theory of Paradigm Structure. Cambridge: Cambridge University Press. Stump, Gregory T. (2006). ‘Heteroclisis and paradigm linkage’. Language 82: 279–322. Stump, Gregory (2015). ‘Inflection classes’. In Baerman (ed.), pp. 113–139. Stump, Gregory (2016). Inflectional Paradigms: Content and Form at the Syntax– Morphology Interface. Cambridge: Cambridge University Press. Stump, Gregory (2018a). ‘Theoretical issues in inflection’. In Audring and Masini (eds), pp. 56–82. Stump, Gregory (2018b). ‘Paradigm function morphology’. In Audring and Masini (eds), pp. 285–304. Sturcken, H. T., Jr (1953). Studies in Thirteenth-Century Spanish Syntax. Unpublished Ph.D. dissertation, University of North Carolina at Chapel Hill. Svenonius, Peter (2012). ‘Spanning’. MS, University of Tromsø. Available at ling.auf.net/lingBuzz/001501. Svenonius, Peter (2016). ‘Spans and words’. In Heidi Harley and Daniel Siddiqi (eds), Morphological Metatheory. Amsterdam: Benjamins, pp. 201–222. Swearingen, Andrew (2011). ‘The Romance imperative: irregular morphology, syncretism, and the morphome’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 119–134. Sweet, James H. (1997). ‘The Iberian roots of American racist thought’. The William and Mary Quarterly 54: 143–166. Sympson, Pedro Luis (1877). Grammatica da lı´ngua brasileira (brası´lica, tupi ou nheéngatu). Manaos [Manaus], AM: Commercio do Amazonas [4th edition, Comissão Brasileira de Estudos Pa´trios, 1933]. Tagliamonte, Sali A., Mercedes Durham and Jennifer Smith (2014). ‘Grammaticalization at an early stage: future be going to in conservative British dialects’. English Language and Linguistics 18: 75–108. Tastevin, Constantin (1923). ‘Grammatica da lı´ngua tupy’. Revista do Muséu Paulista (Separata) 40: 535–763, 1277–1286 [1st edition, Vienna: Ko¨nigliche Akademie der Wissenschaften, 1910]. Taylor, Catherine (2011). ‘Periphrasis in Romance’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 401–416. Taylor, Gerald (1985). ‘Apontamentos sobre o nheengatu falado no rio Negro, Brasil’. Amérindia: revue d’ethnolinguistique amérindienne 10: 5–23. Tekavčic´, Pavao (1972). Grammatica storica dell’italiano. Vol. I: Fonematica. Bologna: il Mulino.

460

references and bibliographical abbreviations

Tekavčic´, Pavao (1977). ‘Interferenze linguistiche istroromanzo-venete: sulle vocali finali nell’istroromanzo’. In Alberto Varvaro (ed.), Atti del XIV Congresso Internazionale di Linguistica e Filologia Romanza, Napoli, 15–20 aprile 1974. Vol. 2. Amsterdam: Benjamins, pp. 447–468. Telmon, Tullio (1993). ‘Varietà regionali’. In Alberto A. Sobrero (ed.), Introduzione all’italiano contemporaneo. Vol. II: La variazione e gli usi. Bari: Laterza, pp. 93–149. Terracini, Benvenuto (1935–1936). ‘Di che cosa fanno la storia gli storici del linguaggio? Storia dei tipi Benio e Nerba nel latino volgare’. Archivio glottologico italiano 27: 133–152; 28: 1–31, 134–150 [reprinted in Terracini (1981), 175–231]. Terracini, Benvenuto (1937). ‘Minima. Saggio di ricostruzione di un focolare linguistico (Susa)’. Zeitschrift fu¨r romanische Philologie 57: 673–726 [reprinted in Terracini (1981), 265–323]. Terracini, Benvenuto (1981). Linguistica al bivio, ed. Gian Luigi Beccaria and Maria Luisa Porzio Gernia. Naples: Guida. Thomason, Sarah G. and Terrence Kaufman (1988). Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press Thornton, Anna (2016). ‘Un capitolo di storia della terminologia grammaticale italiana: il termine sovrabbondante’. In Francesco Dedè (ed.), Categorie grammaticali e classi di parole: Statuto e riflessi metalinguistici. Rome: Il Calamo, pp. 289–309. Thornton, Anna M. (2011). ‘Overabundance (multiple forms realizing the same cell): a noncanonical phenomenon in Italian verb morphology’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 358–381. Thornton, Anna M. (2012). ‘Reduction and maintenance of overabundance: a case study on Italian verb paradigms’. Word Structure 5: 183–207. Thornton, Anna M. (2018). ‘Troubles with flexemes’. In Olivier Bonami, Gilles Boyé, Georgette Dal, Hélène Giraudo, and Fiammetta Namer (eds), The Lexeme in Descriptive and Theoretical Morphology. Berlin: Language Science Press, pp. 303–321. Thornton, Anna M. (2019a). ‘Overabundance in morphology’. In Oxford Research Encyclopedia of Linguistics. Oxford University Press. doi: http://dx.doi.org/10.1093/acrefore/ 9780199384655.013.554. Thornton, Anna M. (2019b). ‘Overabundance: a canonical typology’. In Franz Rainer, Francesco Gardani, Wolfgang U. Dressler, and Hans Christian Luschu¨tzky (eds), Competition in Inflection and Word-Formation. Cham: Springer, pp. 223–258. Timberlake, Alan (1977). ‘Reanalysis and actualization in syntactic change’. In Charles N. Li (ed.), Mechanisms of Syntactic Change. Austin: University of Texas Press, pp. 141–177. Timotin, Emanuela (2016). ‘Presenting the corpus: typologizing, dating, and locating the texts’. In Pană Dindelegan (ed.), pp. 1–7. TLF = Trésor de la langue française: dictionnaire de la langue du XIXe et du XXe siècle,1789– 1960, publié sous la direction de Paul Imbs. 1971–1994. Paris: CNRS. Todaro, Giuseppina and Fabio Del Prete (2019). ‘The morphosyntax–semantics interface and the Sicilian Doubly Inflected Construction’. In Cruschina, Ledgeway, and Remberger (eds), pp. 131–154. Tortora, Christina (1997). The Syntax and Semantics of the Weak Locative. Unpublished Ph.D. dissertation, University of Delaware. Tortora, Christina (2014). A Comparative Grammar of Borgomanerese. Oxford: Oxford University Press. Toso, Fiorenzo (2004). ‘Il tabarchino: strutture, evoluzione storica, aspetti sociolinguistici’. In Carla Paciotto, Fiorenzo Toso, and Augusto Carli (eds), Il bilinguismo tra

references and bibliographical abbreviations

461

conservazione e minaccia: esempi e presupposti per interventi di politica linguistica e di educazione bilingue. Milan: FrancoAngeli, pp. 21–232. Tranel, Bernard and Francesca Del Gobbo (2002). ‘Local conjunction in Italian and French phonology’. In Caroline Wilshire and Joaquim Camps (eds), Romance Phonology and Variation. Amsterdam: Benjamins, pp. 191–218. Traugott, Elizabeth Closs (1995). ‘Subjectification in grammaticalisation’. In Dieter Stein and Susan Wright (eds), Subjectivity and Subjectivisation: Linguistic Perspectives. Cambridge: Cambridge University Press, pp. 31–54. Traugott, Elizabeth Closs (2004). ‘Exaptation and grammaticalization’. In Minoji Akimoto (ed.), Linguistic Studies Based on Corpora. Tokyo: Hituzi Syobo, pp. 133–156. Traugott, Elizabeth Closs and Richard B. Dasher (2001). Regularity in Semantic Change. Cambridge: Cambridge University Press. Traugott, Elizabeth Closs and Graeme Trousdale (2013). Constructionalization and Constructional Changes. Oxford: Oxford University Press. Trubetzkoy, Nikolaj (1969). Principles of Phonology (trans. Christiane Baltaxe). Berkeley: University of California Press. Trudgill, Peter (1986). Dialects in Contact. Oxford: Blackwell. Trudgill, Peter (2001). ‘Contact and simplification: historical baggage and directionality in linguistic change’. Linguistic Typology 5: 371–374. Trudgill, Peter (2003). ‘The impact of language contact and social structure on linguistic structure: focus on the dialects of Modern Greek’. In Bernd Kortmann (ed.), Dialectology Meets Typology: Dialect Grammar from a Cross-linguistic Perspective. Berlin: Mouton de Gruyter, pp. 435–452. Trudgill, Peter (2010a). Investigations in Sociohistorical Linguistics: Stories of Colonisation and Contact. Cambridge: Cambridge University Press. Trudgill, Peter (2010b). ‘Contact and sociolinguistic typology’. In Raymond Hickey (ed.), The Handbook of Language Contact. Chichester: Wiley–Blackwell, pp. 299–319. Trudgill, Peter (2011a). Sociolinguistic Typology: Social Determinants of Linguistic Complexity. Oxford: Oxford University Press. Trudgill, Peter (2011b). ‘Social structure, language contact and language change’. In Ruth Wodak, Barbara Johnstone, and Paul Kerswill (eds), The SAGE Handbook of Sociolinguistics. Los Angeles: SAGE, pp. 236–248. Trudgill, Peter, Elizabeth Gordon, Gillian Lewis, and Margaret A. Maclagan (2000). ‘Determinism in new-dialect formation and the genesis of New Zealand English’. Journal of Linguistics 32: 299–318. Tuten, Donald N. (2003). Koinéization in Medieval Spanish. Berlin: Mouton de Gruyter. Urițescu, Dorin (1984). ‘Subdialectul crișean’. In Rusu (ed.), pp. 284–320. Urițescu, Dorin (2007). ‘Dans la perspective de l’Atlas de Crișana (I). Le participe passé daco-roumain en -ă: mythe roumain ou innovation d’une langue romane?’. In Sanda Reinheimer Rîpeanu and Ioana Vintilă-Rădulescu (eds), Limba română, limbă romanică. Omagiu acad. Marius Sala la împlinirea a 75 de ani. Bucharest: Editura Academiei Române, pp. 555–566. Uță Bărbulescu, Oana (2018). ‘Scurte observații asupra evoluției formelor de genitiv-dativ în clasa demonstrativelor, indefinitelor și relativului’. In Gabriela Pană Dindelegan, Rodica Zafiu, and Isabela Nedelcu (eds), Studii lingvistice. Omagiu Valeriei Guțu Romalo. Bucharest: Editura Universității din București, pp. 341–346. Va¨a¨na¨nen, Veikko (1956). ‘La préposition latine de et le génitif: une mise au point’. Revue de linguistique romane 20: 1–20.

462

references and bibliographical abbreviations

Va¨a¨na¨nen, Veikko (1966). Le Latin vulgaire des inscriptions pompéiennes (3rd edition). Berlin: Deutsche Akademie der Wissenschaften. Vairel, Hélène (1981). ‘The position of the vocative in the Latin case system’. American Journal of Philology 102: 438–447. Van Gelderen, Elly (2011). The Linguistic Cycle: Language Change and the Language Faculty. Oxford: Oxford University Press. Van Valin, Robert D. Jr (2005). Exploring the Syntax–Semantics Interface. Cambridge: Cambridge University Press. Van Valin, Robert D. Jr (2018). ‘Some issues regarding (active) accomplishments’. In Rolf Kailuweit, Lisann Ku¨nkel, and Eva Staudinger (eds), Applying and Expanding Role and Reference Grammar. Freiburg im Breisgau: Freiburg Institute for Advanced Studies (FRIAS), Albert-Ludwigs-Universita¨t Freiburg, pp. 71–93. Van Valin, Robert D. Jr and Randy J. LaPolla (1997). Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Vanelli, Laura (1984). ‘Pronomi e fenomeni di prostesi vocalica nei dialetti italiani settentrionali’. Revue de linguistique romane 48: 281–295. Vanelli, Laura (1987). ‘I pronomi soggetto nei dialetti italiani settentrionali dal Medio Evo ad oggi’. Medioevo Romanzo 12: 173–211. Vanelli, Laura (1992). ‘Da “lo” a “il”: storia dell’articolo definito maschile singolare nell’italiano e nei dialetti settentrionali’. Rivista italiana di dialettologia 16: 29–66. Vanrell, Maria del Mar and Cabré, Teresa (2011). ‘Troncamento e intonazione dei vocativi in Italia centromeridionale’. In Barbara Gili Fivela, Antonio Stella, Luigina Garrapa, and Mirko Grimaldi (eds), Contesto comunicativo e variabilità nella produzione e percezione della lingua. Atti del 7 convegno AISV. Rome: Bulzoni, pp. 200–211. Varvaro, Alberto (2011). ‘Il DÉRom: un nuovo REW?’. Revue de linguistique romane 75: 297–304. Varvaro, Alberto (2013). ‘Latin and the making of the Romance languages’. In Maiden, Smith, and Ledgeway (eds), pp. 6–56. Vasiliu, Emanuel and Liliana Ionescu-Ruxăndoiu (1986), Limba română în secolele al XIIlea—al XV-lea. Bucharest: Universitatea din București, Facultatea de Filologie. Vecchio, Paola (2006). ‘L’ausiliazione perfettiva in napoletano. Studio di sintassi diacronica’. Revue de linguistique romane 70: 53–94. Vendler, Zeno (1967). ‘Verbs and times’. In Zeno Vendler, Linguistics in Philosophy. Ithaca, NY: Cornell University Press: pp. 143–160. Vennemann. Theo (1972). ‘Phonetic analogy and conceptual analogy’. In Theo Vennemann and Terence H. Wilbur (eds), Schuchardt, the Neogrammarians, and the Transformational Theory of Phonological Change: Four Essays by Hugo Schuchardt, Theo Vennemann, Terence H. Wilbur. Frankfurt am Main: Athena¨um, pp. 115–179. Vernet, Florian (2007). Mémento grammatical occitan: languedocien. Montpellier: Université Paul Valéry. Available at: http://www.univ-montp3.fr/occitan/images/ stories/memento_gramatical.pdf [accessed 13/09/2020]. Verratti, Vittore (1998). Fonologia e morfologia del volgare abruzzese. Con rimario-glossario. Lanciano: Itinerari. Vet, Co (2007). ‘The descriptive inadequacy of Reichenbach’s tense system: a new proposal’. In Louis De Saussure, Jacques Moeschler, and Genoveva Puskas (eds), Tense, Mood and Aspect: Theoretical and Descriptive Issues. Amsterdam: Rodopi, pp. 7–26. Vigna, Amedeo (1879). ‘Codice diplomatico delle colonie Tauro-Liguri durante la signoria dell’Ufficio di San Giorgio (1453–1475)’. Atti della Società Savonese di Storia Patria 7: 5–1003.

references and bibliographical abbreviations

463

Vincent, Nigel (1982). ‘The development of the auxiliaries habere and esse in Romance’. In Nigel Vincent and Martin Harris (eds), Studies in the Romance Verb. London: Croom Helm, pp. 71–96. Vincent, Nigel (1987). ‘The interaction of periphrasis and inflection: some Romance examples’. In Martin Harris and Paolo Ramat (eds), Historical Development of Auxiliaries. Berlin: Mouton, pp. 237–256. Vincent, Nigel (1995). ‘Exaptation and grammaticalization’. In Henning Andersen (ed.), Historical Linguistics 1993. Amsterdam: Benjamins, pp. 433–445. Vincent, Nigel (1997a). ‘Synthetic and analytic structures’. In Martin Maiden and Mair Parry (eds), The Dialects of Italy. London: Routledge, pp. 99–105. Vincent, Nigel (1997b). ‘The emergence of the D-system in Romance’. In Ans Van Kemenade and Nigel Vincent (eds), Parameters of Morphosyntactic Change. Cambridge: Cambridge University Press, pp. 149–169. Vincent, Nigel (1999). ‘The evolution of C-structure: prepositions and PPs from IndoEuropean to Romance’. Linguistics 37: 1111–1153. Vincent, Nigel (2011). ‘Non-finite forms, periphrases, and autonomous morphology in Latin and Romance’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 418–435. Vincent, Nigel (2014). ‘Similarity and diversity in the evolution of Italo-Romance morphosyntax’. In Benincà, Ledgeway, and Vincent (eds), pp. 1–21. Vincent, Nigel (2015). ‘Compositionality and change’. In Claire Bowern and Bethwyn Evans (eds), The Routledge Handbook of Historical Linguistics. London: Routledge, pp. 103–123. Vincent, Nigel (2016a). ‘A structural comparison of Latin and Romance’. In Ledgeway and Maiden (eds), pp. 37–49. Vincent, Nigel (2016b). ‘Continuity and change from Latin to Romance’. In J. N. Adams and Nigel Vincent (eds), Early and Late Latin: Continuity or Change? Cambridge: Cambridge University Press, pp. 1–13. Vincent, Nigel (in press). ‘Origins of Romance’. In Adam Ledgeway and Martin Maiden (eds), The Cambridge Handbook of Romance Linguistics. Cambridge: Cambridge University Press. Vincent, Nigel and Delia Bentley (2001).’ The demise of the Latin future periphrasis in -urus + esse’. In Claude Moussy (ed.), De lingua latina novae quaestiones. Actes du Xème Colloque international de linguistique latine Paris-Sèvres, 19–23 avril 1999. Leuven: Peeters, pp. 143–155. Vincent, Nigel and Kersti Bo¨rjars (2010). ‘Grammaticalization and models of language’. In Elizabeth Closs Traugott and Graeme Trousdale (eds), Gradience, Gradualness and Grammaticalization. Amsterdam: Benjamins, pp. 279–299. Vincent, Nigel and Frans Plank (eds) (2019). The Diachrony of Suppletion. Special issue of Transactions of the Philological Society 117 (3). Virdis, Maurizio (1987). ‘Appunti per una sintassi del sardo’. Biblioteca francescana sarda 1: 409–440. Vlad, Daciana (2012). ‘Le conditionnel en roumain: une forme modale’. Faits de langues 40: 61–68. Vliet, Edward R. van (1983). ‘The disappearance of the French passé simple: a morphological and sociolinguistic study’. Word 34: 85–115. Vogel, Irene (1994). ‘Phonological interfaces in Italian’. In Mazzola (ed.), pp. 109–125. Vulpe, Magdalena (1984). ‘Subdialectul maramureșean’. In Rusu (ed.), pp. 320–354. Wagner, Max Leopold (1997). La lingua sarda: storia, spirito e forma (new edition, ed. Giulio Paulis). Nuoro: Ilisso.

464

references and bibliographical abbreviations

Wagner, Suzanne E. and Gillian Sankoff (2011). ‘Age grading in the Montréal French inflected future’. Language Variation and Change 23: 275–313. Wakker, Gerry C. (2006). ‘Future auxiliaries or not?’. In Emilio Crespo, Jesús De La Villa, and Antonio R. Revuelta (eds), Word Classes and Related Topics in Ancient Greek. Louvain-la-Neuve: Peeters, pp. 237–255. Wa¨lchli, Bernhard (2006). ‘Typology of heavy and light “again”, or, the eternal return of the same’. Studies in Language 30: 69–113. Walkden, George (2014). Syntactic Reconstruction and Proto-Germanic. Oxford: Oxford University Press. Wanner, Dieter (1981). ‘Surface complementizer deletion: Italian che ~ Ø’. Journal of Italian Linguistics 6: 47–82. Waugh, Linda R. and Monville-Burston, Monique (1986). ‘Aspect and discourse function: the French simple past in journalistic usage’. Language 62: 846–877. Weigand, Gustav (1896). ‘Die Bildung des Imperfecti Futuri—Konditionalis, Optativi—im Ruma¨nischen’. Jahresbericht des Instituts fu¨r ruma¨nische Sprache zu Leipzig 3: 139–161. Weiss, Michael (2020). Outline of the Historical and Comparative Grammar of Latin (2nd edition). Ann Arbor: Beech Stave Press. Westerbergh, Ulla (1956). Chronicon Salernitanum. A Critical Edition with Studies on Literary and Historical Sources and on Language. Stockholm: Almqvist & Wiksell. Wheeler, Max W. (1988). ‘Occitan’. In Martin Harris and Nigel Vincent (eds), The Romance Languages. London: Routledge, pp. 246–278. Wheeler, Max W. (2005). The Phonology of Catalan. Oxford: Oxford University Press. Wheeler, Max W. (2011) ‘The evolution of a morphome in Catalan verb inflection’. In Maiden, Smith, Goldbach, and Hinzelin (eds), pp. 183–209. Wheeler, Max W. (2017). ‘Testimonis primitius i evolució gramatical del pretèrit perifràstic va- + infinitiu en occità i en català’. In Aitor Carrera and Isabel Grifoll (eds), Occitània en Catalonha: de tempses novèls, de novèlas perspectivas. Actes de l’XIen Congrès de l’Associacion Internacionala d’Estudis Occitans, Barcelona: Generalitat de Catalonha/Lleida: Institut d’Estudis Ilerdencs, pp. 279–295; English version ‘Stages in the grammaticalization of the periphrastic preterite va- + infinitive in Catalan and Occitan’, at: https://www.academia.edu/35650830/Stages_in_the_grammaticalization_ of_the_ periphrastic_preterite_va_infinitive_in_Catalan and https://www.researchgate. net/ publication/326654451_Stages_in_the_grammaticalization_of_the_periphrastic_ preterite_va-_infinitive_in_Catalan_and_Occitan Wheeler, Max W. (forthcoming). ‘La morfologia flexiva (II): la flexió verbal’. In Manuel Pérez-Saldanya and Josep Martines (eds), La gramàtica del català antic. València: Institut Interuniversitari de Filologia Valenciana. Wheeler, Max W., Alan Yates, and Nicolau Dols (1999). Catalan: A Comprehensive Grammar. London: Routledge. Willis, David (2007). ‘Syntactic lexicalization as a new type of degrammaticalization’. Linguistics 45: 271–310. Willis, David (2017). ‘Degrammaticalization’. In Adam Ledgeway and Ian Roberts (eds), The Cambridge Handbook of Historical Syntax. Cambridge: Cambridge University Press, pp. 28–48. Wilmet, Marc (1970). Le Système de l’indicatif en moyen français. Geneva: Droz. Wilmet, Marc (1976). Études de morpho-syntaxe verbale. Paris: Klincksieck. Wolfe, Sam (2018). Verb Second in Medieval Romance. Oxford: Oxford University Press. Wolfe, Sam (forthcoming). Syntactic Change in French. Oxford: Oxford University Press.

references and bibliographical abbreviations

465

Wolfe, Sam and Martin Maiden (eds) (2020). Variation and Change in Gallo-Romance Grammar. Oxford: Oxford University Press. Wolfe, Sam and Christine Meklenborg (eds) (2021). Continuity and Variation in Germanic and Romance. Oxford: Oxford University Press. Woods, Rebecca and Sam Wolfe (eds.) (2020). Rethinking Verb Second. Oxford: Oxford University Press. Wurzel, Wolfgang U. (1984). Flexionsmorphologie und Natu¨rlichkeit: ein Beitrag zur morphologischen Theoriebildung. Berlin: Akademie Verlag. Wurzel, Wolfgang Ulrich (1989). Inflectional Morphology and Naturalness. Dordrecht: Kluwer. Xavier, Carla (2013). ‘Is the production of epenthesis in loanwords influenced by English proficiency? A production study in Brazilian Portuguese’. Paper given at the Third International Conference on English Pronunciation: Issues and Practice (EPIP 3), Murcia, 8–10 May 2013. Yllera, Alicia (1980). Sintaxis histórica del verbo español: las perı´frasis medievales. Zaragoza: Departamento de Filologı´a. Yvon, Henri (1963). ‘Le passé simple est-il sorti d’usage?’, Le Français moderne 31: 161–176. Zafiu, Rodica (2013). ‘Mood, tense, and aspect’. In Pană Dindelegan (ed.), pp. 24–64. Zafiu, Rodica (2016). ‘The syntax of moods and tenses’. In Pană Dindelegan (ed.), pp. 14–52. Zafiu, Rodica (2017). ‘The auxiliary of the Romanian conditional: semantic and functional arguments concerning the reconstruction of a disputed grammaticalization process’. Diacronia 6: 1–20. Zamboni, Alberto (1974). Veneto. Pisa: Pacini. Zamboni, Alberto (1981). ‘Le caratteristiche essenziali dei dialetti veneti’. In Manlio Cortelazzo (ed.), Guida ai dialetti veneti. Padua: CLEUP, pp. 9–43. Zamboni, Alberto (1988). ‘Divergences and convergences among Neo-Latin systems in North-Eastern Italy’. Folia Linguistica Historica 8: 233–267. Zamboni, Alberto (1997). ‘Lessico(logia) e morfologia’. In Luisa Mucciante and Tullio Telmon (eds), Lessicologia e lessicografia. Atti del Convegno della SIG. Rome: Il Calamo, pp. 147–187. Zamboni, Alberto (1998). ‘Dal latino tardo al romanzo arcaico: aspetti diacronico-tipologici della flessione nominale’. In Paolo Ramat and Elisa Roma (eds), Sintassi storica. Atti del XXX Congresso della SLI (Pavia 26–28 settembre 1996). Rome: Bulzoni, pp. 127–146. Zamboni, Alberto (2000). Alle origini dell’italiano: dinamiche e tipologie della transizione dal latino. Rome: Carocci. Zamfir, Dana-Mihaela (2005). Morfologia verbului în dacoromâna veche (secolele al XVIlea-al XVII-lea). Vol. 1. Bucharest: Editura Academiei Române. Zamfir, Dana-Mihaela (2007). Morfologia verbului în dacoromâna veche (secolele al XVIlea-al XVII-lea). Vol. 2. Bucharest: Editura Academiei Române. Zimmermann, Eva (2019). ‘Gradient symbolic representations and the typology of ghost segments’. In Katherine Hout, Anna Mai, Adam McCollum, Sharon Rose, and Matthew Zaslansky (eds), Proceedings of Annual Meetings on Phonology 2018, LSA. https://doi.org/ 10.3765/amp. Zwicky, Arnold (1969). ‘Phonological constraints in syntactic descriptions’. Papers in Linguistics 1: 411–463. Zwicky, Arnold M. (1974). ‘Hey, Whatsyourname!’. In Michael W. La Galy, Robert A. Fox, and Anthony Bruck (eds), Papers from the 10th Regional Meeting of the Chicago Linguistic Society, April 19–21 1974. Chicago: Chicago Linguistic Society, pp. 787–801.

466

references and bibliographical abbreviations

Zwicky, Arnold (1986). ‘The general case: basic form versus default form’. In Vassiliki Nikiforidou, Mary VanClay, Mary Niepokuj, and Deborah Feder (eds), Proceedings of the Twelfth Annual Meeting of the Berkeley Linguistics Society: February 15–17, 1986 (BLS 12), pp. 305–314. Zwicky, Arnold (1996). ‘Syntax and phonology’. In Keith Brown and Jim Miller (eds), Concise Encyclopedia of Syntactic Theories. Oxford: Elsevier, pp. 300–305. Zwicky, Arnold M. and Geoffrey K. Pullum (1983). ‘Phonology in syntax: the Somali optional agreement rule’. Natural Language & Linguistic Theory 1: 385–402.

Index Aboh, Enoch 381 Abruzzese 148, 228 eastern 31, 78, 206 accusative (see also case: adverbal) 13, 15, 20, 55, 68, 306, 316, 319, 320, 323, 324, 325, 329, 330, 403 extended 329 Acireale (Sicily) 173, 174 Ackerman, Farrell 18, 21, 27, 28, 29, 31, 39, 74, 80, 84, 169 Acquaviva, Paolo 68, 307 active voice 2, 29–30, 41 Adams, J. N. 108, 123, 126, 277, 281 adstrate 409 adverbs 166–167, 226–227 in -ment(e) 56 position 86 affix 34, 38, 72, 94, 121, 146, 166, 173, 174, 176, 179, 184, 371 antesuffix 362, 375, 376, 378 African languages 166, 394 Afrikaans 364 age-grading 132 Aglianese (Piedmont) 148, 150 agreement (see also concord; participle: perfective) 62, 64–65, 313, 329 adjectival 66, 176, 319 gender 65, 67 lexical 67 markers 72 number 67–68, 336, 337 verbal 7, 31, 176, 319, 335, 385, 400, 412 Aguero-Bautista, Calixta 378 Aguiar, José 388 Aikhenvald, Alexandra 67, 409 Aktionsart (see also aspect) 65–66, 83–84, 111, 147, 153 Albanian 268 Alberobello (Apulia) 177–178, 179 Alboiu, Gabriela 269 Alguerès (Catalan, Sardinia) 70, 370–371, 372 Alibert, Lois 78 alignment (see also auxiliary: perfective: selection) 216 absolutive 361 active verb 399, 401, 403, 412

active–stative/–inactive 35, 149, 198–199, 216, 217, 218, 398, 400, 412 nominative–accusative 35, 361, 411 split ergative 398 stative verb 400, 401, 404, 412 Alinei, Mario 108 Allières, Jacques 308 allomorphy 362, 371–372, 382, 385, 388, 408 allophony 309, 312, 383 Altamurano (Apulia) 214, 215, 219, 226 Altomonte (Calabria) 202, 210 Amalfi 322 Amazon 395, 411 Amazonian languages 398 Ambertois, Ambert (Auvergne) 87 Ambrosini, Riccardo 371 Amenta, Luisa 183, 184 American structuralism 4 Amorim, Antônio 396, 410 analysability 51–52 analytic(ity), analysis 6, 11, 17, 37, 71, 255–258, 392, 411 expression 14 loss 262–268, 270 marking of obliques 272–302 vs periphrasis 12–21, 61 vs synthetic 77, 126, 241, 242, 302 anaptyxis 363, 395 Anchieta, José de 396 Andalusia 382 Anderson, Gregory 242, 266 Anderson, John 14, 18, 39 Anderson, Stephen R. 2, 4–5, 38, 62, 64, 71, 169 Andrews, Avery 164 Andriani, Luigi 16, 35, 38, 171, 172, 173, 174, 175, 177, 178, 179, 194, 200, 212, 222 animacy 56, 156, 182, 226, 287, 297, 298, 301, 400 Anthonissen, Lynn 257 antiperiphrasis 26 aoristic drift 28, 51, 76, 79, 83, 86 Apothéloz, Denis 78, 249 Aprosio, Sergio 146, 162 Apulia 35–36, 120, 173, 177, 180, 226, 230, 236, 237 central 214, 219–229

468

index

Aquilano, L’ Aquila (Abruzzo) 231 Arabic 384, 385 Aragonese 103, 117 Araújo 396 Arawak (South America) 395, 412 northern 409 Archangeli, Diana 364 Arduin, Joana 393 Ariellese, Arielli (eastern Abruzzo) 31 Arnauld, Antoine 130 Aromanian 275 Aronoff, Mark 2, 4, 7, 193, 319, 378 Artes, Eduard 365, 370, 371, 373, 374, 375 article 14 indefinite 214, 229, 297, 301, 372 definite 40, 42, 273, 274, 275, 294, 369–370, 371, 372, 373, 374 demonstrative article / cel 16 possessive article / al 299–230 Arvinte, Vasile 243 Ascoli, Graziadio 154, 172 Ashdowne, Richard 61, 69 aspect 22, 169–170 andative 83, 84, 130, 173, 175, 179, 181, 182 lexical see Aktionsart marker 36, 38, 51, 80, 166–167, 174, 175, 176, 184, 409 perfect 41, 83, 262 pre-hodiernal vs hodiernal contrast 28 prospective 139, 140, 144, 404, 409, 410 venitive 47, 83, 173, 175, 181 Asturian 382 attestation 54–56, 96, 122 augment 232, 372, 377–378 augmentative 63–64 auxiliary 36, 123, 408–409 come 47-48, 52, 83, 84, 145, 146, 171–177, 180 French venir de 48–49, 54, 79–80, 83, 94 conditional 26, 243, 266 want 244–245, 248, 255, 258–259, 266 continuative/progressive be 170 come 49–50, 146 go 45, 109–110 definition 216, 242 emergent 148 future 108 come 44, 108 go 26, 27, 42, 44, 53, 79–80, 83, 84, 89, 93, 117–118, 123, 124, 126, 126–132, 137, 170, 406, 407, 409–410, 411, 412 have 71, 73, 93, 96, 103–104, 126, 146, 183, 231, 244, 245, 246, 258, 259, 268

must 129, 130 want 44, 129, 130, 243–245, 246, 255, 258, 266, 268, 407 epistemic 256, 257, 261–261 go 44–45, 52, 53, 83, 84, 95, 108–118, 120–121, 123–144, 145, 146, 171–177, 178 have 95, 96–108, 217, 228 epistemic 183 deontic 98–99, 104, 107, 108, 183–187, 228, 231 vs lexical 119–120, 184 inchoative 126, 128, 129, 137–138, 139 iterative 42, 148, 157 pass 172, 180 passive 41, 164, 169–170, 329 be 38, 48, 170 come 38, 48–49, 58–59, 146 modal value 49 go 46–47, 84 modal value 46, 109 have 42 perfective 6, 26–27, 32, 34–35, 41, 76, 81–83, 342, 347 be 95, 151, 165, 231, 234, 246–251, 264, 336 uninflected form 250, 255, 264, 265, 267, 269 have 80, 87, 88, 95, 98, 104, 231 extension 151, 152, 164, 211 generalization 35, 197, 199, 201, 206, 246 selection 29–30, 34–35, 36, 41–42, 57, 151–153, 162, 193–212, 236, 264 active–stative / –inactive alignment 35, 149, 198–199, 216, 217 Auxiliary Selection Hierarchy 152, 340 free variation 206, 209, 215, 218, 221 mixed 194, 198, 199–204, 211–212, 218, 219, 230, 231, 233 reflexive predicates 201–204, 211, 217 (conditioned by) mood 42 person-driven (see also auxiliary: perfective: selection: mixed) 31, 42, 194–212, 213–236 phonology-driven 205, 208, 214, 219–229, 230, 236 (conditioned by) semantic/pragmatic variables 153 switch 148, 151, 162, 163–166, 167 (conditioned by) tense 42 triple systems 206, 212 presumptive 254, 255, 256, 257, 261, 267–268, 270 preterite

index go 124–125 Catalan 40, 42, 53, 84, 110–118, 125 Gascon 83, 118, 125 Occitan 59, 117–118, 123–125, 126, 130, 132–137 Guardiol 117–118, 125 progressive 32, 33, 35–36, 38, 39, 42, 45, 49, 84, 109, 120, 170, 171, 175, 178, 182, 188, 252, 254, 257, 270, prospective 139, 140, 144, 404, 409, 410 return 38, 50–51, 52, 84, 95, 145–167 semi-lexical 83, 172–173, 181 send 172, 180 stand 35–36, 38, 42, 120–121, 175, 178 auxiliation 87, 94, 123, 126, 146, 194, 198, 199, 201, 210 Avram, Larisa 244, 255, 256, 258, 262, 266, 268 Ayres-Bennett, Wendy 78, 242 Bach, Xavier 6, 195, 197, 211, 215, 216, 233 Bachrach, Asaf 365, 366, 372, 375, 376 back-formation 70 Badia i Margarit, Antoni 125 Baena, Antônio 396 Baerman, Matthew 38, 39, 195, 196, 198, 317 Baker, Mark 15, 176 Balkan 258, 268 languages 244 Sprachbund 245, 257 Baniwa (North Arawak) 395 Baños Baños, José 277 Barcélo, Gérard 118 Barcis (Friuli) 206 Baré 409, 411 Barese, Bari 120, 171–172, 177, 200, 214, 219, 222, 237 Barros, Maria de 396, 411 Basilicatese (southern Italy) 42, 108 Battaglia, Salvatore 82 Baudouin de Courtenay, Jan 309 Bauer, Brigitte 12, 57 Baxter, Alan 394 Beard, Robert 4 Beavers, John 338, 339 Béarnais (Gascony) 125 Bélanger, Monique 211 Belém (Brazil) 393, 396 Bengali 364 Benincà, Paola 22, 78, 146, 154, 165, 202, 203, 341, 350, 371 Bentein, Klaas 74, 81, 85 Bentley, Delia 7, 108, 183, 200, 211, 215, 231, 335, 336, 337, 338, 341, 359, 361 Benucci, Franco 164

469

Berceo, Gonzalo de 98 Berchem, Theodor 125 Berea-Găgeanu, Elena 258 Berizzi, Mariachiara 55 Bernini, Giuliano 340, 341 Bertinetto, Pier Marco 28, 45, 46, 50, 63, 74, 147, 257, 339, 404 Bertocci, Davide 147 Bettendorff 396 Bianchi, Valentina 182, 337, 338, 361 Bierwisch, Manfred 4 bilingualism 411, 412 Biondelli, Bernardino 163 Birchall, Joshua 398 Bitettese, Bitetto (Bari) 199, 204–205, 208, 209, 219–220, 221, 222, 223, 224, 226, 227, 228, 230, 231, 235, 236, 237 Blake, Barry 68 Blanche-Benveniste, Claire 131 Blasco Ferrer, Eduardo 140, 148 bleaching (see also grammaticalization) 146, 175, 182, 284, 409 Blevins, James 18, 25, 197, 206 Blondeau 354 Bloomfield, Leonard 305, 366, 375 Boa Vista (Brazil) 393 Bobaljik, Jonathan 17, 32 Bollée, Annegret 131 Bollig 61 Bolognese, Bologna 340, 342, 343 eastern 340 Bonami, Olivier 198, 216, 233, 235, 236 bondedness 94 Bonet, Eulàlia 368, 372, 373, 376 Bonorvese, Bonorva (Sardinia) 225–226, 235, 236 Bonvini, Emilio 395 Booij, Geert 62–63, 215 Borgomanero (Piedmont) 340 Bo¨rjars, Kersti 13, 43, 169 borrowing (see also contact) 70 Bosque, Ignacio 45 Bossong, Georg 28 Bottasso, Enzo 161 Bourciez, Édouard 12 Boxer, Charles 396 Breimaier, Federica 213 Bres, Jacques 48, 49, 80, 139, 140, 142, 144 Breton 307 Brandi, Luciana 337 Brewer, Mary 392 Brindisi 120, 177 Britain, David 382, 383 Broselow, Ellen 364

470

index

Brown, Dunstan 15, 21, 25, 26, 29, 31, 33, 34, 37, 39, 73, 74, 75, 77, 78, 79, 85, 89, 169, 171, 195, 196, 198, 199 Brown, Josh 382, 394 Brucale, Luisa 183 Bulgarian 44, 69, 80, 244, 268 Burgos 382 Buridant, Claude 164, 241, 277 Burzio, Luigi 163, 377 Butt, Miriam 164 Buxton, Tiziana 146 Bybee, Joan 44, 46, 141, 142, 155, 244, 392, 409, 410 Cabral, Ana 396, 398 Cabré, Teresa 70 Caha, Pavel 5 Cairese, Cairo Montenotte (Liguria) 95, 148, 149–150 Calabrese, Andrea 62, 174, 175, 176, 177, 179, 180, 185, 188 Calabrian 151, 154, 172, 179, 202 Caldarola, Laura 213 Calderara di Reno (Bologna) 343–344, 350–351 Campania 318 Campbell, Lyle 87, 409, 410 Campi Salentina (Salento) 174 Camproux, Charles 64 Canale, Michael 211 Canavese (Piedmont) 149 canonical typology 37 Cappellaro, Chiara 39, 61, 310 Caragiu Marioțeanu, Matilda 243 Carbonell, Pere Miquel 112 Cardeira, Esperança 384 Carden, Guy 165 cardinal numerals (see also quantifier) 14, 16, 287–288, 289 Cardinaletti, Anna 22, 83, 158, 172, 173, 177, 180, 181, 335, 338, 361, 368, 369, 373 Carratala´, Ernesto 81 Carruthers, Janice 78, 88, 89, 242, 249 Carstairs, Andrew 310, 311–312, 313, 314, 316 Carstairs-McCarthy, Andrew 204 cartography 59 Casasnovas, Afonso 395 case (see also accusative; dative; genitive; obliques; vocative) 22, 62, 63, 68–69, 70, 195, 319, 332 adnominal 273, 279, 280, 282, 284, 286, 290–292, 293, 294, 297–298, 299 adnominal vs adverbal 272–273 adverbal 273, 290, 294, inflexional marking 13–14, 272–302

inherent/lexical 21 marker 277, 281, 284, 296 scattered vs syncretic realization 15–17 structural 21 Castellani, Arrigo 157 Castellazzo Bormida (Piedmont) 203, 204 Castelletto Merli (Piedmont) 203, 204 Castilian (see also Spanish) 57, 382 old 103 Catalan (see also Alguerès; Valencian) 40–41, 42, 82, 85, 87, 97, 99, 101, 102, 105–107, 108, 109, 110–118, 139, 142, 148, 249, 372, 373, 374, 375, 376, 383 Balearic 78, 114 Formentera 362, 376 Menorca 107, 116 central 370 early 106, 112 eastern 115 Gironı´ 82 northern 78 old 101, 105, 125, 133 southern 107 western 103 Cava dei Tirreni (Campania) 318, 319 Cazzano di Tramigna (Veneto) 203 Cennamo, Michela 7, 123, 146, 158, 164, 165, 200, 227, 335, 340, 341, 354 Cerano 227 Cerullo, Maria 340 Champion, James 129, 130 Chapman, Carol 13, 167 Chomsky, Noam 4, 59 Chumakina, Marina 37, 43 Ciconte, Francesco 335, 341 Cinque, Guglielmo 82, 86, 148, 164, 165, 242 Ciorănescu, Alexandru 277 Cisternino (Apulia) 177–178, 179 Clackson, James 126 Claudi, Ulrike 155 clipping 70 clitic (see also mesoclisis) 34, 71–72, 94, 100, 373, 385, 390 climbing 148, 149, 154, 155, 158, 162, 165, 166, 167, 173, 268 enclisis 73, 149, 154, 158, 371, 374, 376 interrogative conjugation 371 object 370 doubling 20–21, 56, 210, 287 placement 36 proclisis 73, 158, 368, 371 subject 72, 336–337, 340, 345, 346, 347, 349, 350, 351, 352, 353, 354, 355, 367, 368 cliticization 121, 166–167, 176, 179, 306

index Cocchi, Gloria 215 Coene, Martine 244, 256, 258, 262, 266 Colantoni, Laura 363, 364 Colella, Gianluca 50 Collins, Chris 5 Colombia 395 Colón, Germà 125, 133 colon 73 Columbus 385 Company Company, Concepción 81 comparative 16 complementizer 406, 412 deletion 24–25 prepositional 42, 118 compositional(ity) 27–30, 33, 34, 37, 51–54, 83, 85, 243, 250, 262 compounding 63 Comrie, Bernard 63, 75, 76, 243, 262 Conceição de Paiva, Maria da 388 concord 176 conditional (see also auxiliary) 52–53, 121, 244–245, 266 future-in-the-past 244–245, 258, 266 Confais, Jean-Paul 132 contact (see also borrowing) 3, 7, 11, 43, 56–59, 306, 309, 381, 390, 395, 411, 412 anti-contact 59 Balkan 59 between Brazilian Portuguese and Tupi-Guarani 394–411 -induced change 384, 394 morphological copy 409 contiguity 81–83, 85 Conversano (Bari) 38, 39, 175, 177–178 coordination (see also pseudo-coordination) 20–21, 24, 35, 57, 120, 154, 172, 278, 284–285 Corbea 292 Corbett, Greville 28, 37, 43, 63, 171, 195, 196, 198, 210, 211, 213, 215, 228, 231, 312, 313, 315, 335 Cordin, Patrizia 22, 337 Cornilescu, Alexandru 20, 268 Cornu, Maurice 78 Coromines, Joan 148 Corr, Alice 361 Correggio (Emilia-Romagna) 348–349, 354–355, 356 Corsican 55 Cosentino, Cosenza (Calabria) 38, 109, 151, 163, 165 Coseriu, Eugenio 12, 77, 241 Costa, Lairson Barbosa da 393 Craig, Colette 45, 46

471

Cravens, Thomas 76 creolization 130, 384 Croft, William 385 Cruschina, Silvio 4, 6, 83, 84, 120, 146, 169, 170, 171, 172, 173, 175, 176, 177, 178, 179, 180, 181, 182, 187, 188, 335, 337, 338, 359, 361 Cruz, Aline 395, 396–397, 398, 404, 405–406, 407, 409 Culbertson, Jennifer 72 Cusic, David 147 Cuzzolin, Pierluigi 306, 319 D’Agostino, Rosa 146 D’Alessandro, Roberta 31, 32, 78, 200, 381 D’Hulst, Yves 244, 256, 258, 262, 266 D’Ingeo, Aurelia 213 Daco-Romance 6, 108, 241 dialects 258–262 Dahl, Östen 77 Dahl’s paradox 52 Dale, Rick 412 Dalmatian 3 Daniel, João 396 Danish 14 Dardel, Robert de 329 Darrigrand, Robert 32 dative (see also case: adnominal) 13–14, 15, 16, 18–19, 20, 21, 22, 55, 68, 196, 277, 279, 280, 284, 287, 294, 299, 306, 316, 317, 324, 328, 329, 330, 405, 412 clitic 19–21, 299 dative-genitive case forms (see also case: adnominal) 18, 22, 70, 272, 285, 299 Davies, William 234 Davis, Stuart 371 de Lacy, Paul 364 De Mauro, Tullio 148 de Melo, Wolfgang 277 De Wit, Astrid 257 decategorialization (see also grammaticalization) 146, 162–166, 176 declensional class 22, 25–26, 39, 195, 316, 320 default marking 30, 194, 198, 203, 204–205, 250, 341 definiteness 14, 16, 17, 18, 19, 22, 26, 40, 226, 235, 272–275, 279, 280, 281, 282, 291, 294, 297, 298, 299, 301, 337, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352–353, 354, 355, 356, 360, hierarchy 342 DeGraff, Michel 131 degrammaticalization 43–44, 76 Del Gobbo, Francesca 371, 373

472

index

Del Prete, Fabio 172, 173 Delia (Sicily) 184 demonstrative 273, 397–398 Demuth, Katherine 364 Denis, Derek 131 Densusianu, Ovid 33, 244, 247, 249, 261, 281 desemanticization (see also grammaticalization) 74, 160–163, 172, 176, 181 Deshaies, Denise 132 desyntacticization 35, 36, 37 Detges, Ulrich 63, 110, 125, 139 Di Caro, Vincenzo 171, 172, 173, 177, 178, 179, 183, 184, 186 Di Fabio, Elvira (see also DiFabio, Elvira) 377 Di Vita, Rosario 369 diachrony and synchrony 59–60 dialect mixing (see also koinéization) 382, 384 Diémoz, Federica 306 Dietrich, Wolf 395 Diez, Friedrich 125, 137, 216 DiFabio, Elvira (see also Di Fabio, Elvira) 377 differential object marking (DOM) 55–56, 226 diglossia 71 diminutive 32, 63–65, 66, 90, 366, 375, 378 past participle 64–65 Dimitrescu, Florica 100, 103 directionality of change 6, 43–44, 51, 309 distributed exponence 31–32, 33, 74, 80–81, 84, 85, 169 Distributed Morphology (DM) 2, 5, 6, 15–17, 18, 20, 28, 30, 31, 32, 33, 37, 62, 120 Dobrovie-Sorin, Carmen 262, 264, 268, 269 Dols, Nicolau 40 Double Access Reading 244 Doubly Inflected Construction (DIC) (see also pseudo-coordination) 83, 171–182 Dowty, David 338, 339 Dragomirescu, Adina 6, 33, 48, 49, 58, 247, 250, 261, 265 Dressler, Wolfgang 4, 371, 378 Du Cange, C. Du Fresne 156 Duarte, Fabio 127 Durham, Mercedes 131 Dyirbal 214 Eckardt, Regine 94 Eckert, Penelope 381 Egerland, Verner 158 Einhorn, E. 127, 128 Elcock, W. D. 54 Elson, Mark 244, 267 Embick, David 15, 16, 18, 30 Emilian-Romagnol, Emilia Romagna 337, 338, 340, 341, 355, 356, 360, 361

Engadinian (Romansh, Switzerland) 235 Engel, Dulcie 86 English 1, 2, 3, 16, 39, 44, 45, 47, 65, 67, 69, 83, 131, 148, 160, 161, 165, 214, 229, 310, 314, 317, 338, 363, 364, 365, 366, 372, 389 epenthesis 226, 362–380 in Brazilian Portuguese 365–366, 372, 375, 376 in Catalan 370–371, 372, 373, 374, 375, 376 copy vowel 364 default consonant 365, 366 default vowel 365, 367, 368, 369, 370, 371, 374 in Italian 369–370, 371, 372, 373, 374, 375 morphologically-conditioned 373–376, 378 in Padua 368, 369, 370, 371, 372, 373, 374, 375 phonologically-conditioned 364–365, 378 in San Marino 364–365, 366–368, 369–370, 371, 372, 374–375 types 379 Eremita, Salvatore 213 Ernout, Alfred 306 erosion 38, 150, 166–167, 176, 182, 244 Erstarrung 309 Erteschik-Shir, Nomi 361 Esher, Louise 53 exaptation 382 excrescent vowel 363, 364 Experiencer 299, 301 extended exponence 32, 33 extension 87–88, 146, 160–162 Eythórsson, Thórhallur 200, 211, 215, 231 Fabricius-Hansen, Cathrine 147 Faggin, Giorgio 151, 155 Fagundes, Sidney 395, 396, 398 Faraco, Carlos 392 Faria, Francisco 396 Farkas, Donka 342 feature-spreading 31 Felizzano 203 Fernandes, Maria 384 Ferorelli, Dina 213 Ferrarese, Ferrara 340, 341, 347–348, 352–354, 356 Ferrari, Franca 373, 374 Ferreira, Epimonides 396 Ferreiro, Manuel 97 Ferrero, Carlo 161, 163 Figueira 396 Filipponio, Lorenzo 340 Finbow, Tom 7, 59 Finco, Franco 146, 155

index finiteness 25, 29 Fink, Robert 69 von Fintel, Kai 53 Fiorentino, Fernando 38 Fischer, Olga 94 fission 5, 32, Fleischman, Suzanne 129, 131, 134 flexeme 219 Florentine old 50, 157–158, 159, 160, 212, 217, 235 Floricic, Franck 69, 70, 71 focus 22, 351, 397 mirative 83, 84, 181–182 sentence-focus 351 thetic/broad focus 337, 351, 361 Fogarasi, Miklós 82 Foggia 120 folk etymology 3, 51, 52 form–function relations 311–312, 320, 321, 325 Formentin, Vittorio 79, 100, 216 Foulet, Lucien 127 Fradin, Bernard 219 Fraenkel, Eduard 73 Francoprovençal 78, 101, 306, 308 Frâncu, Constantin 243, 247, 248, 254 Freire, José 395, 396 French 2, 3, 6, 15, 34, 44, 48–49, 51, 52, 53, 54, 56, 57, 64, 65, 66, 67, 69, 70, 72, 75, 77–78, 79, 80, 82, 83, 85, 86, 89, 90, 94, 95, 96, 97, 100–101, 108, 109, 118, 119, 123, 126–132, 133, 137–138, 140, 142, 144, 156, 170, 194, 197, 201, 211, 214, 217, 229, 233–234, 235, 236, 241–242, 249, 266, 301, 363, 382 Acadian 89, 101 Prince Edward Island 132 colloquial 72 European 211 Laurentian 89, 132, 211, 234 middle 84, 129–131 old 57, 71, 96, 97, 102, 108, 114, 125, 127, 128, 133, 164, 277 standard 72 Friedman, Victor 268 Friulian 78, 151, 153, 155, 162, 202, 249 Fuèc (see also future; conditional) 53 functional categories 123, 131, 172, 176, 243, 268 fusion 5, 15, 16, 17, 21 future (see also auxiliary: future) 15–16, 44, 52–53, 71, 72, 89, 93, 96, 108, 121, 126, 129, 243–244, 268, 404, 406–411 synthetic vs analytic 131–132, 170

473

Gabelentz, Georg von der 94 Gaeng, Paul 324 Galante, Maria 318 Galician 97, 98, 101, 102, 109, 383, 385, 387 Gallo-Italian 55, 203 Gallo-Romance 3, 97, 123, 127 oïl varieties (see also Wallon) 72, 211, 234 Garcia, Guilherme 365, 372 Garcı´a Jurado, Maria 364 Gardani, Francesco 59 Garrapa 371 Gasca Queirazza, Giuliano 160, 164 Gascon 32, 83, 125, 130, 148 Gatti, Tiziana 217 Geana, Ionuț 241 gender 22, 25, 30, 33, 62, 63, 64, 65, 66–67, 68, 74, 75, 90, 210, 250, 274, 293, 299, 319, 341, 372, 383 allomorphy 372 correlations with fruits and trees 66–67 metaphorical vs literal meanings 66 sex 66, 67 shape 66, 67 size 66, 67 lexical 67 marking 1, 307 genitive (see also case: adnominal) 18–19, 22, 68, 196, 277, 278, 279, 294, 301, 306, 316, 319, 320, 322, 324, 328, 329 genitive-dative case forms (see also case: adnominal) 18, 22, 70, 272, 285, 299 Genoese 161, 163 Gentile, Salvatore 164 German 317 gerund 57–58, 243 Gévaudan 64–65 Gheorghe, Mihaela 269 Gheție, Ion 281 Giacalone Ramat, Anna 46, 48, 145, 146, 148, 165, 166, 270 Giannini, Gianni 213 Gildea, Spike 409, 410 Gili y Gaya, Samuel 81 Giorgi, Alessandra 16, 17, 24 Gisborne, Nikolas 199, 204 Giudici, Alberto 59 Giusti, Giuliana 83, 123, 172, 173, 177, 180, 181 Givón, Talmy 71 Goal 124, 126, 133, 137, 139, 144, 160, 301, 339, 360 Godard, Thomas 61 Gougenheim, Georges 125, 128, 129, 130, 131, 137, 138, 139

474

index

Goosse, André 131, 132 gradience 29, 34–37, 169, 354 grammaticalization 29, 34, 38, 42–51, 52, 56, 57, 81, 84, 87–88, 105, 111–112, 120, 121, 123, 124, 125, 129, 133, 139, 142, 145–146, 155–167, 169, 171, 176, 181, 182, 184, 187, 242, 243, 244, 270, 277, 279, 284, 300, 302, 360, 404–411 cline 34, 35, 83, 84, 85, 94, 146, 160, 284 cognitive schema 124, 139, 144 divergence 93–94, 96, 103, 119, 121, 146 event schema 145, 156 layering 93–94, 121, 122, 160 path(way) 131, 139, 144, 148, 244 subjectification 146 Grandgent, C. H. 97 Grandi, Nicola 63, 66, 340, 354 Gravinese, Gravina in Puglia 38, 39, 219, 220–221, 222, 223, 224, 226, 227, 230, 235, 237 Greek 226, 243, 244, 253, 307–308 Ancient 74, 363 Classical 362 Greenberg, Joseph 382 Gregores, Emma 404 Grenand, Françoise 396 Grevisse, Maurice 131, 132 Griffiths, Bruce 166 Gro¨ber, Gustav 371 Groothuis, Kim 16 Grosio 336 Guardiano, Cristina 340 Guardiol, Guardia Piemontese (see also Occitan) 59, 117, 125 Guasch, Antonio 404 Guirao, Miguelina 364 Guy, Gregory 384 Haase, Martin 28 Haberland, Hartmut 147 Hacken, Pius ten 62 Hajek, John 337, 340, 341 Hall, Nancy 363 Hall, R. A. 54 Halle, Morris 15, 30, 62 Harley, Heidi 62 Harris, Alice 87, 409, 410 Harris, Martin 12, 83, 123, 361 Hartt, Charles 395, 396, 410 Haspelmath, Martin 18, 22, 26, 29, 38, 39, 73, 74, 76, 79, 84, 88, 169, 170, 241 Hay, Jennifer 339 Hebrew 385 Heidinger, Steffen 234

Heine, Bernd 94, 124, 139, 145, 150, 155, 156, 160, 162, 166, 242, 257, 409, 410 Hemming, John 396 Henrichsen, Arne-Johan 137 Herman, József 329 heteroclisis 3, 39, 193–194, 198, 201, 202, 203, 204–210, 212, 214, 219, 230, 231, 232, 306, 371 by overabundance 203, 206–208 heterosemy 44, 146 Hill, Virginia (see also Motapanyane, Virginia) 255 Hippisley, Andrew 199, 204 Hjelmslev, Louis 305 Hoberman, Robert 378 Hock, Hans 392 Hockett, Charles 74 Hofmann, J. B. 306 homonymy 307, 308, 313 Hopper, Paul 34, 60, 93, 94, 119, 123, 145, 146, 153 Hornstein, Norbert 75, 76 Houaïlou 45 Hu¨nnemeyer, Friederike 155 Humboldt, Wilhelm von 12 Humboldt’s Universal 311 Hummel, Martin 57 Hungarian 244 hybridization 381 hypotaxis (see also infinitival construction) 38–39 Iatridou, Sabine 53 Ibero-Romance 3, 73, 109, 247, 394, 409, 412 early 55 old 98, 99 Iliescu, Maria 129, 279 impoverishment 5, 179, 188 indefiniteness 18, 19, 22, 26, 214, 235, 273, 282, 287, 288, 297, 298, 301, 337, 341, 342, 343, 344, 345, 347, 348, 349, 350, 351, 352, 353, 354, 355, 373 indirect object (see also Recipient) 277, 270, 281, 283, 287, 295, 297, 301, 328, 329, 386, 386–387, 390, 391 Indo-European 315 inflexion 12, 14 attrition 35, 36, 38, 176, 182, 184 boundaries of 62–73 classes 193–212, 233, 235, 320, 371 segregated 197 contextual 62–63 defectivity/defectiveness 22, 37, 53, 171, 173, 315

index inherent 62–63 of late Latin nominal systems 319–331 atrophization 322 hypercharacterization of grammatical relation 330 overextension 329 of inflexion 323–324 of stem 323 phonetic factors 322 syntactic factors 329–330 loss 120–121 noun 194 vs periphrasis 12–21, 61, 411 simplification 262, 264, 265 suffix 371 uninflectability 36, 37, 38, 145, 173 infinitival construction 171–172, 175, 176, 177, 179, 181, 243–246 interpolation 35, 81–83, 85, 88, 226–227 intersectivity 13, 14, 18, 21–27, 31, 32, 33, 60, 74, 75–79, 84, 88–89, 90, 119, 169–170, intransitive (clause/verb)/intransitivity 133, 145, 156, 233, 301, 329, 398, 400, 401, 403, 404 split intransitivity 336 intrusive vowels 363–364, 379 Ionescu-Ruxându, Liliana 99 Iorga Mihail, Ana-Maria 14 Iovino, Rossella 70, 123 Irsina 42 Ischitano, Ischia (southern Italy) 1 Italian 1, 3, 22–26, 34, 45, 46, 48, 49–50, 51, 52, 53–54, 56, 57, 58, 59, 63, 64, 66, 68, 82–83, 84, 85, 86, 95, 97, 100, 102, 108, 109–110, 119, 146, 148, 153, 154, 157, 163, 165–166, 185–186, 194, 197, 199, 201, 202, 204, 210, 211, 215, 216, 218, 226, 231–232, 234, 235, 236, 307, 310, 313–315, 338, 361, 362, 369–370, 371, 372, 373, 374, 375, 376, 377–378, 382 American varieties spoken in USA 364 central varieties 24 literary 313 northern varieties 24, 25, 26, 86 old 46–47, 50, 53, 58–59, 97, 185, 231, 232, 233 Italian dialects 148, 164, 167, 335 central 41, 157, 213 northern 50–51, 119, 202, 335 north-western 6 southern 41, 107, 109, 120, 154, 155, 213 Italo-Romance 34, 38, 70, 98, 102, 108, 145, 171, 178, 185, 194, 197, 199–204, 211, 214, 219, 227 central-southern 70, 363

475

early 55, 99, 100, 319 north-eastern 78, 151, 374 northern 72, 337, 374 southern 83, 84, 87, 100 Item-and-Arrangement 2 Item-and-Process 2 Ivănescu, Gheorghe 244, 245 Jacobs, Bart 59, 117, 118, 125 Janda, Richard 35, 213 Jeanroy, Alfred 143 Jendraschek, Gerd 411 Jensen, Frede 124, 127, 129, 137, 142, 164 Jiménez, Jesús 370 Johnson, Daniel 359 Jolivet, Rémi 78, 89 Jones, Dafydd Glyn 166 Jones, Michael Allan 32, 151, 166 Joseph, Brian 35, 43, 56, 57 Juge, Matthew 125 Kabardian 312 Kamp, Hans 147 Karlsson, Keith 56, 57 K(ase) 15–16, 17, 21 Kathman, David 213 Kaufman, Terrence 70 Kaye, Steven 194, 198 Kayne, Richard 5, 163, 215, 216, 217, 218, 224, 230 Keenan, Elinor 384 Keniston, Hayward 81 Kennedy, Christopher 339 Kerswill, Paul 381 Kiefer, Ferenc 4 King, Ruth 132 Kiparsky, Paul 15, 38, 59 Kitto, Catherine 364 Klein, Wolfgang 75, 361 Klein-Andreu, Flora 76 koiné 382, 390 koinéization 381, 383, 412 in Portuguese 384–394 in Romance 381–384 Komatsu, Elsuke 305 Korotkova, Natalia 62 Ko¨rtvélyessy 66 Kra¨mer, Martin 371 Kristol, Andres 306, 308 Kroch, Anthony 71 Kruszewski, Mikołaj 309 Kunert, Hans 59, 117, 118, 125 Kuryłowicz, Jerzy 93, 317 Kuteva, Tania 124, 139, 242, 244, 409, 410

476

index

La Fauci, Nunzio 216, 234, 235 Labeau, Emmanuelle 48, 49, 80, 139, 140, 142, 144 Lacross, Elisabeth 131 Ladin 151, 165 Laforge, Éve 132 Lancelot, Claude 130 Lander, Yuri 62 Langacker, Ronald 51, 87, 146 Langfur, Hal 394 LaPolla, Randy 151, 161, 335, 338, 339, 361 Larson, Richard 328 Late Insertion 5 Latin 1, 3, 11, 12–13, 16, 19, 27, 30, 34, 44, 51, 52, 53, 54, 55, 56, 57, 68, 69, 70, 73, 76, 77, 121, 123, 126, 130, 137, 145, 154, 167, 169–170, 172, 180, 195, 197, 227, 228, 241, 244, 247, 273, 292, 306, 307, 315–316, 317, 331, 377, 378, 383, 385 late 71, 73, 83, 87, 95, 96, 98, 100, 102, 106, 108, 127–128, 156, 277, 281, 319 late-medieval 318–331 medieval 7 sub-literary 127–128 Lass, Roger 60, 309, 382 Lăzărescu, Paul 260 Leccese, Lecce 36, 38, 120, 174, 175, 180, 217, 235 Ledgeway, Adam 6, 12, 14, 22, 29, 30, 31, 32, 33, 34, 35, 37, 38, 42, 55, 61, 78, 79, 93, 99, 102, 109, 120, 123, 129, 154, 155, 156, 165, 166, 172, 173, 174, 175, 179, 182, 200, 203, 210, 211, 212, 214, 215, 219, 228, 241, 242, 249, 255, 260, 266, 274, 277 Lee, M. Kittiya 398 Lefebvre, Claire 131 Legendre, Géraldine 200 Lehmann, Christian 94, 146, 155, 167, 188, 360 Lenci, Alessandro 147 Lengadocian 64–65, 81, 85, 132 Leone, Alfonso 154, 172 Leone, Angela 213 Leone, Francesco 213 Leonese 382 Lepschy, Giulio 146, 156 Leumann, Manu 306, 324, 329 levelling 382, 386, 389, 392 Levin, Beth 338, 339, 340, Levin, Juliette 363, 364 lexeme ancillary 216, 219 vs flexeme 219 split 231 Lexical Integrity Hypothesis 5

Lexical Phonology 366 lexical verb reinforcement 118–119 lexicalization 64 lexico-aspectual structure 338–340 accomplishment 338, 339, 349, 352, 358–359 achievement 338, 339, 349, 352, 358–359 activity 338, 339, 345, 357, 358–359 state 338, 339, 349, 357, 358 state-based vs non-state-based 338, 340 Lichtenberk, Frantisek 44, 46, 47, 48, 50, 51, 146, 161 Lieber, Rochelle 15 Ligurian 101, 146, 148, 149, 160, 164, 166, 336, 354 old 160–161, 162 lingua franca 394, 396, 411 lı´ngua geral (amazônica) (see also Yẽgatú) 395, 396 Lisbon 384, 385 listemes 15 Lizio-Bruno, Letterio 154 Lizzano in Belvedere 342–343, 349–350 Lloret, Maria-Rosa 368, 370, 372, 372 Lloyd, Paul 97 Lodge, Anthony 382 Lo¨fstedt, Einar 329 Logudorese, Logudoro 217, 225, 234, 235 Lois, Ximena 216 Lombard (dialect) 42, 78, 336 early 157 Lombard, Alf 102 Lombardi, Alessandra 56, 146 Lombardy 318 López Sanz, Rafael 409 Loporcaro, Michele 1, 6, 35, 42, 59, 85, 108, 182, 184, 199, 200, 201, 202, 205, 206, 207, 208, 212, 214, 215, 216, 217, 218, 219, 227, 228, 231, 233, 234, 235, 236, 341, 370, 371, 372, 372 Loregian-Penkal, Loremi 393 Lorusso, Paolo 173, 175, 177, 178, 179 Lucanian see Basilicatese Lucchesi, Dante 384, 394 Luı´s, Ana 72 Lunn, Patricia 76 Lupyan, Gary 412 Lyons, John 69 Macapa´ 393 Macedonian 69, 80–81, 268 Macerata 208 Macron, Emmanuel 65 Madrid 384 Magalhães, José 395, 396, 409, 410

index Magistrale, Francesco 318 Maiden, Martin 1–7, 14, 18, 28, 36, 39, 52, 53, 59, 61, 62, 70, 89, 98, 99, 100, 102, 106, 109, 122, 146, 171, 177, 178, 179, 185, 189, 193, 196, 213, 230, 232, 241, 247, 255, 258, 262, 272, 277, 297, 305, 321, 335, 363, 377 Malmberg, Bertil 363 Malouf, Rob 18 Maltese 378 Manambu, Papua New Guinea 67 Manasantivongs, Peter 23 Manaus (Brazil) 393, 397 Manea, Dana 82 Manning, Christopher 164 Manoliu, Maria 243 Mantuan 59 Manzari, Giovanni 213, 220 Manzini, Rita 35, 55, 173, 174, 175, 177, 179, 194, 199, 200, 201, 203, 204, 205, 206, 207, 209, 213, 215, 216, 219, 220, 221, 222, 223, 224, 226, 227, 228, 230, 335, 341, 345, 349 Maragua´ people 395 Marantz. Alec 15, 62 Marcato, Gianna 152 Marchigiano, Marche 202, 208 Marchello-Nizia, Christiane 129 Mardale, Alexandru 55, 56, 301 Mărgărit, Iulia 259, 260, 261, 262 Marin, Maria 259, 260, 261 Marina di Ragusa 173, 174 Markopoulos, Theodore 244 Marotta, Giovanna 371 Marsalese, Marsala 172, 177 Martin, Fabienne 337, 338 Martina Franca 36, 120, 172, 175, 178 Martinet, André 380 Martı´nez Cortés 117 Mascaró, Joan 371, 372, 373 Matalon, Zuan 155 Matino (Apulia) 36, 38 Matos, Gabriela 127 Matthews, P. H. 4, 138, 305 Matras, Yaron 70 Matushansky, Ora 62 Maupas, Charles 130 Maurer, Philippe 131 Mayerthaler, Willi 311 Mazzoleni, Marco 70 Mazzucco, Giancarlo 146 Mazzucco, Iolanda 146 McCrary, Kristie 371 McCrary Kambourakis, Kristie 371 Meillet, Antoine 12, 44, 94 Melchior, Luca 249

477

Melo, Juliana 409 Menéndez Pidal, R. 97 Menon, Odete 393 Menoni, Viviana 57 Mensching, Guido 129 merge(r) 15, 16, 172, 173 Merlini Barbaresi, Lavinia 64, 378 Mesagnese, Mesagne (Apulia) 121, 180 mesoclisis 72–73, 127 metaphony 1–2, 62 metaphor 63, 66, 67, 83, 124, 146, 147, 155–156, 156 metonym 47, 83, 124, 146, 301 Meyer-Lu¨bke, Wilhelm 125, 142 Miatto, Veronica 363 Michalias, Regis 87 Michelotti, Alexander 364, 365, 366, 371, 374 Milanese 70, 335–336 old 43, 158 Miremont, Pierre 34 Mirror Principle 15 Mithun, Marianne 398, 409 Mocciaro, Egle 46, 47, 183 Modena 340, 346–347, 352, 356 Mok, Quirinus 151 Moll, Francesc 78 Monachesi, Paola 164 Monaci, Ernesto 99 Monserrat, Ruth 396 Monteiro, John 394 Monville-Burston, Monique 86 mood 22, 94, 270 marking on auxiliaries 264, 266–267, 268, 270 Mooney, Damien 44, 118, 132 Moore, Denny 395, 396, 397, 398 Moradi, Sedigheh 364 Morais Silva, António de 148 Moravcsik, Edith 70, 335 Morelli, Frida 371 morphologization 29, 35, 37, 169, 170, 187–188 morphology 4–5, a-morphous 4 autonomy 2, 4, 28, 62, 233, 314, 321, 323 derivational 61, 63, 64, 65, 70, 84 evaluative (see also augmentative; diminutive) 63–64, 65, 66, 70, 366, 375, 378 inflexional vs derivational 62, 64 limits 233–236 natural 311 periphrastic 41 separationist 4 -syntax interface 15, 32–33 theories

478

index

morphology(Continued) incremental 15, 18 inferential 5, 18–21 lexical 5, 15–17 phonological 62 realizational 5, 15, 18, 194 morphome 2, 3, 4, 5, 6, 24, 39, 52, 53, 89, 90, 106, 122, 141, 170, 171, 177, 178, 180, 181, 185, 186, 187, 188, 189, 193, 208, 212, 233, 314 attraction 171, 183, 200 E-pattern 185–186 G-morphome 106 L-pattern 99, 102, 106 N-pattern 2, 39, 62, 89, 98, 100, 102, 103, 108, 109, 110, 112, 121, 122, 177–182, 187, 188, 232, 233 periphrastic 170, 171, 179, 187, 188 sociolinguistic variant 89 U-pattern 36, 178 W-pattern 186, 187 morphonology 1 Mortelmans, Tanja 257 Motapanyane, Virginia (see also Hill, Virginia) 269 Mougeon, Raymond 211 movement (see also verb: of motion/movement) 5, 87 head 15, 16, 32, 176, 268 Mufwene, Salikoko 381, 394 multilingualism 394, 412 Munaro, Nicola 371 Mussomelese, Mussomeli (Sicily) 172, 177, 180, 181–182, 183, 184, 185 Nadasdi, Terry 132 Nagy C., Katalin 125, 139, 142 nanosyntax 5, 59 Naples (see also Neapolitan) 322 Napoleon 54 Napoli, Donna Jo 378 Nardò 180 Naro, Anthony 384 Narrog, Heiko 94 Navarrese 382 Navarro, Donatello 213 Navarro, Eduardo de Almeida 404, 408 Neagoe, Victorela 259, 260, 261, 262 Neapolitan 99, 101, 103, 156, 165, 209–210 old 55, 79, 159, 164, 165, 249, 277 Negrão, Esmeralda 394 Nemi (Lazio) 377 Nettle, Daniel 412 Nevaci, Manuela 33

Newmeyer, Frederick 87 Nheengatu see Yẽgatú Nicolae, Alexandru 6, 48, 49, 58, 242, 247, 256, 264, 268, 269 Nicula 243 Niculescu, Dana 247, 252, 257 No Blur Principle 204 Nobre, Wagner 395, 398 Noceto (Parma) 342, 349 Noll, Volker 395 nominative 320, 329, 331–332 nominative-accusative case forms (see also case: adverbal) 18 non-compositional(ity) 27–30, 32, 33, 51, 74, 83, 85, 169 Noonan, Michael 155 Norberg, Dag 329 Norde, Muriel 382 Noyer, Rolf 16, 62 number 1, 15, 18, 22, 25, 26, 30, 31, 62, 63, 67–68, 74, 75, 88, 90, 94, 99, 107, 114, 176, 195, 199, 200, 210, 250, 258, 274, 293, 299, 307, 317, 319, 321, 323, 332, 336, 337, 341, 361, 366, 373, 383, 400, 403 lexical distinctions 68, 90, 307 marking 1, 30, 67–68, 331, 373, 376 pluriform 401, 402, 412 polyfunctionalism 321, 327, 330 Nyrop, K. 97 object 7, 324, 326, 330, 385, 386–387, 390, 391, 401, 402, 403 of preposition 326–327, 329, 330 obliques 329 analytic marking 279–286 competition between prepositions 283–284 double marking 277–278, 284–285 inflexional vs prepositional marking 272–302 mixed marking 278, 284–285, 286–287, 293–294 Occitan (see also Lengadocian; Périgord) 6, 34, 35, 59, 78, 86, 87, 97, 101, 102, 108, 124–125, 130, 137, 139, 142, 144, 148, 151, 249 medieval 127 northern 72 old 57, 102, 108, 114, 132–137, 139, 140–141, 142, 143 Oceanic 44, 146, 161 Ojeda, Almerindo 68 Old Church Slavonic 243, 246 Oliveira, Fa´tima 77 Oliveira, Gabriel Barros Viana de 398

index Oltenia 28 O’Neill, Paul 3, 7, 59, 62, 179 Oostendorp, Marc van 70 opacity 27 Origlia, Luciano 146 overabundance 7, 39, 40, 77, 89, 90, 179, 203, 206, 208, 214, 230–232, 305, 308–311, 312, 313, 314, 389 overdifferentiation 26 overexhaustivity 40 Paciaroni, Tania 194, 202, 208, 216 Paduan 152–153, 166, 203, 354, 368, 369, 370, 371, 372, 373, 374, 375 Paesano, Nicolò 183, 184 Pagliuca, William 155, 244 Pagotto, Emilio 392 Pallarès, Pallars (Catalan) 370–371, 374 Palmer, Frank 243 Pamfil, Viorica 249 Pană Dindelegan, Gabriela 6, 33, 82, 292 Paoli, Sandra 6, 118, 119 paradigm 2–3, 18–21, 22–23, 33, 34, 37, 61, 62, 63, 69, 84, 90, 169, 178, 182, 188, 194–195, 215, 242, 307, 308, 310, 312, 314, 317, 320–321, 324, 328, 331, 392 content 194–197, 201, 210 defective 186 extended 25, 26, 27 form 194–197, 210 hypodifferentiation 325–328 local 25, 26 micro-paradigm 318, 332 realized 194–197 split 96–97 sub-paradigm 25, 26, 27, 197 symmetry 22 paradigmaticity 18, 21, 37–42, 73, 309, 318, 392 Paradigm Function Morphology (PFM) 4, 5, 18, 28, 32, 72–73 paragogy 309 parameter 212 parataxis (see also: pseudo-coordination) 38–39, 83, 154–155 Parma 340 Parodi, Ernesto 161 Parry, Mair 6, 38, 50, 95, 149, 202, 203, 335, 336, 340, 341, 354, 355, 361 Partee, Barbara 4 participle Latin in -urus 30, 108 perfective 29, 41

479

agreement 35, 64–65, 81, 84–85, 86, 88, 163, 164, 210, 250–251, 261, 265, 337, 341, 354 diminutive 64–65 passive voice 2, 29–30, 32, 33, 36, 38, 41, 46–47, 48, 58–59, 109, 164, 169–170 Pau, Jeroni 112 Pellegrini, Giovan Battista 203 Pennsylvania German 44 Penny, Ralph 382, 383 Peperkamp, Sharon 368 Pérez Saldanya 78 perfect (see also auxiliary) 85–86 existential 78 Périgord (Dordogne) 34 periphrasis (see also auxiliary) 14, 17 vs analysis 12–21 boundaries 73–89 sociolinguistic variables 84–87 categorial 76, 84, 169, 170, 173, 241–270 compound (see also surcomposé forms) 31, 32–34, 248–250, 253, 260, 261, 264, 269, 270 hypercompound/double compound 79 defectiveness/defectivity 37–38, 173, 179, 180, 181, 188 fuzziness 83–84 with gerund/present participle 252–255, 261–262 grammatical 14 with infinitive 243–246, 258–259 morphomic 170, 171, 179, 187, 188 with past participle 246–251, 259–260 suppletive 169, 170, 171, 188, 201, 219 uninflectability 37–38 periphrastic perfect 25, 26, 27, 34, 75–79, 246–251 iterative interpretation 77 Perkins, Revere 244 Perlmutter, David 216, 234 Pernicone, Vincenzo 82 person (see also: auxiliary: perfective: selection) 22, 132, 135–137, 139, 140, 141, 142, 177, 179, 182, 184, 215 Pescarini, Diego 202, 203, 218, 219, 231, 340, 341, 350 Pesetsky, David 165 Pessoa, Marlos 384 Petracco Sicardi, Giulia 328 Petter, Maria 395 Philippide, Alexandru 244 phonology-free syntax principle 224 Piacenza 340, 345 Pianesi, Fabio 16, 17, 24

480

index

Picernese, Picerno (Basilicata, southern Italy) 217 Piedmontese 42, 78, 101, 146–147, 148, 149–150, 160, 162, 162, 166, 202, 203, 336, 340, 354, 355 Alpine varieties 306 old 160, 161, 164, 166 Pinkster, Harm 123, 126, 277 Pinto, Manuela 361 Pires, Nadia 395, 396, 398 Pisano, Simone 78 Plank, Frans 179, 308, 310, 315 pluperfect 27, 31, 85 analytic 77 compound periphrasis 33, 41, 249, 260, 270 synthetic 76, 77 Poggi, Giuseppe 163 Poggio Imperiale 208–209, 210, 211 polarity 89, 132 Poletto, Cecilia 24, 78, 213, 250, 341, 371 polyfunctionalism 305–308, 313, 318–321, 326, 327, 331–332 case/number 321 in diachrony 317–318 number 330–331 paradigmatic and syntagmatic relations 316–317 syntactic function 330–331 polymorphism 305–310, 318–321, 325, 331–332 in diachrony 317–318 endogenous 309 exogenous 309 free variation 312–314, 329–330, 332 paradigmatic and syntagmatic relations 316–317 polysemy 146, 156 polysynthesis 62 polythematism 307 Pomino, Natascha 18 Pompeii 322, 331 Ponte dell’Olio (Piacenza) 345–346, 351–352, 355, 356 Ponte di Legno (Brescia) 118–119 Popescu-Marin, Magdalena 70 Poplack, Sharon 89, 131, 132 Popoli (Abruzzo) 206, 207 Popova, Gergana 14, 21, 26, 29, 31, 36, 37, 40, 43, 63, 65, 80, 81, 83, 84, 85, 88 Porto Alegre (Brazil) 390, 391, 392 Porto Velho (Brazil) 393 Portuguese 3, 7, 32, 52–53, 57–58, 73, 77, 79, 83, 97, 98, 101, 102, 103, 109, 148, 381, 383, 384–394, 406–407, 411–412

Brazilian 365–366, 372, 375, 376, 378, 381, 384, 389–390, 392, 394–411, 412 northern 393 second-person singular non-deferential address 390–394 south-eastern 398 southern 392–393 European 72, 127, 384, 390, 391, 392 northern 386 second-person plural pronominal and verbal forms 384-390 southern 389 Posner, Rebecca 12, 131 possessor 277, 299, 300, 301 possessive 403 constructions 274–275, 277, 280–281, 283, 295–296, 297, 299, 412 pronouns 382–384, 386, 387, 388, 390, 391, 402 Pountain, Christopher 127 Prati 378 preterite (see also auxiliary: preterite: go) 27, 28, 86, 102, 109, 113–114, 126, 171, 185–186, 246 periphrastic vs synthetic 40, 139 Price, Glanville 127 pro-drop 22–23 progressive (see also auxiliary: progressive) 32, 33, 35–36, 38, 42, 171, 175, 178, 182, 254 Proietti, Domenico 318 pronoun (see also clitic) 398 proper name 69, 274, 375 toponym 281 proto-Romance 54, 73, 96–97, 103, 107, 108, 219, 329 pseudo-coordination (see also Doubly Inflected Construction) 35–36, 38, 120–121, 154–155 Pugliese 35–36, 38, 42, 120, 171, 174, 175, 177, 178, 179, 182, 209 Apulo-Barese 178, 204, 209 Dauno-Appennine 209 Pullum, Geoffrey K. 224, 229 Putignano 35–36, 120, 174, 175 Putzu, Ignazio 216 PYTA (perfecto y tiempos afines) 3, 106 quantifier 16, 287–294 homogenous vs heterogeneous constructions 290–292 Quarto di Altino (Treviso) 306 Quer, Josep 266 raddoppiamento/rafforzamento fonosintattico 184, 200

index Ræto-Romance 72, 108, 249 Raiol, Domingos 396 Rama (Nicaragua) 45 Ramat, Paolo 146 Raoult, Didier 54 Rappaport Hovav, Malka 338, 340 Rea, Béatrice 211 reallocation 382, 383, 384, 386, 393 reanalysis 87, 88, 146 Rebotier, Aude 132 Recasens i Vives, Daniel 97 Recipient (see also indirect object) 13–14, 277, 285, 299, 300, 301 reconstruction 43, 54–56, 96, 122 refunctionalization 332, 382 Reggio Emilia 340, 348, 354 Régnier-Desmarais, François-Séraphin 139 Reich, Uli 398 Reichenbach, Hans 75 register 16, 28, 40, 58, 72, 79, 86, 114, 127, 132, 137, 141, 249, 275, 295, 296, 297, 300, 301, 302, 308, 319, 323, 324 regrammaticalization 119, 382 relabelling 87–88 Relational Grammar 217 Remacle, Louis 78 Remberger, Eva-Maria 18, 130 Renzi, Lorenzo 99, 371 Repetti, Lori 7, 335, 364, 368, 369, 373 restructuring 148, 149, 158, 165–166, 175, 176 Reynolds, Bill 379 Ribeiro, Ilsa 394 Rinollo, Melanie 187 Rio Branco (Brazil) 393 Rio de Janeiro 398 Rio Negro (Brazil) 395, 397, 398, 406, 409, 410, 411, 412 Rizzi, Luigi 163, 337, 338, 361 Roberts, Ian 15, 31, 72, 123, 200 Rodrigues, Aryon Dall’Igna 395, 396, 397 Rodrigues, João Barbosa 396, 410 Rohlfs, Gerhard 55, 78, 87, 97, 99, 119, 148, 164, 172, 182, 377 Roller, Heather 394 Romagnol (north-eastern Italy) 101, 340 Romagnoli, Serena 335 Romance 11–13, 34, 47, 54, 57, 70, 71, 73, 85, 93, 100, 101, 108, 109, 121, 123, 148, 156, 158, 185, 211, 214, 216–217, 234, 241, 251, 277, 301, 308, 318, 323, 329, 330, 335, 363, 377, 378 western 108, 258 Romanesco old 234

481

Romanian 3, 6, 13–17, 18–20, 22, 26–27, 28, 29, 30, 33, 35, 36, 39, 40, 41, 44, 48, 49, 55, 58, 70, 76, 77, 82, 85, 86, 99, 101, 102, 103, 108, 230, 241, 242, 245, 250, 254, 255–258, 272–275, 294–300, 301–302 colloquial see substandard high register 16 old 55, 56, 58, 99–100, 241, 242, 243–255, 259, 273, 275, 276–294, 301–302 dialectal/regional varieties 241, 243, 258–262, 265, 297, 299, 301 Apuseni Mountains 261 Banat 33, 259, 261 southern 286–287 Valea Almăjului 260 Bucovina 259 Crișana 33, 259 western 261 Dobrodja 261 Maramureș 33, 230, 259 Fundătura 230 Tăureni 230 Muntenia 33 northern 260, 281 North Moldavia 33 Oltenia 28, 261 north-western 260–261 Sălaj 259 southern 260, 261, 281 sub-Danubian 241 Transylvania 33, 261 Wallachia 261 western 260 substandard 16, 40, 260, 295–296, 298–299, 301 Romansh see Engadinian; Surmiran; Sursilvan; Vallader Ronjat, Jules 148 Rosa, Carlota 395 Rose, Yvan 364 Rosemeyer, Malte 145, 157, 158, 159, 161 Rosen, Carol 216, 234 Rosén, Hannah 319 Rosetti, Alexandru 100, 244, 267, 291 Rossdeutscher, Antje 147 Rossi, Silvia 70, 147 Roussou, Anna 15, 123 Russian 70 Rutigliano 38, 39 Ruvese, Ruvo di Puglia 219–220, 221, 222, 223, 224, 226, 228–229, 230, 235, 237 Sabatini, Francesco 318 Saccon, Graziella 335

482

index

Sadler, Louisa 169 Salentino 35–36, 38, 120, 166, 171, 174, 175, 177, 180, 182 Salvador 393 Salvi, Giampaolo 99, 277 Samek-Lodovici, Vieri 335 Sampaio 394 Sampson, Rodney 146 San Donșa del Piave (Venice) 306 San Giovanni in Persiceto (Bologna) 344–345, 351 San Marco in Lamis (Apulia) 178 San Marino 364–365, 366–368, 369–370, 371, 372, 374–375 San Vito di Cadore (Ladin) 151–152 Sandfeld, Kristian 246, 257 Sankoff, Gillian 89, 132, 211, 354 Sanskrit 363 Sansò, Andrea 48 Santoro, Damiana 213 Santoro, Salvatore 213 São Gabriel da Cachoeira (Brazil) 395 São Paulo 390, 392–393, 394, 396, 398 Sapir, Edward 366, 375 Sardinian 32, 44, 70, 78, 87, 101, 103, 108, 109, 151, 166, 234, 235–236 old 55, 56 Sateré-Mawé people 395 Saussure, Ferdinand de 51, 59, 305 Saussure, Louis de 249 Savoia, Leonardo 35, 55, 173, 174, 175, 177, 179, 194, 199, 200, 201, 203, 204, 205, 206, 207, 209, 214, 215, 216, 219, 220, 221, 222, 223, 224, 226, 227, 228, 230, 335, 341, 345, 349 Scaglione, Massimo 162 Scaglione, Tommaso 146 Scalise, Sergio 63 Scandinavian 40 Schaden, Gerhard 28 Scherre, Maria 384, 393, 394 Schifano, Norma 86, 242, 268 Schilling-Estes, Natalie 392 Schirru, Giancarlo 107 Schlegel, August 241 Schmidt-Riese 398 Schogt, H. G. 86 Schwade, Michéli 395 Schwegler, Armin 12, 123, 241 Scorrano 36 Scorretti, Mauro 24 Seifert, Eva 87 Seki, Lucy 398 Selkirk, Lisa 368 serial construction 154–155, 409

Serianni, Luca 310 Setterberg-Jørgensen, Birgit 156, 157 Seville 382, 384, 385 Sgroi, Salvatore 70 Sheehan, Michelle 361 Sibille, Jean 117 Sicilian 99, 109, 154, 171, 172, 173, 177, 179, 180, 181, 183–187 central 185 eastern 178–179 Gallo-Italic varieties 55 old 99, 159, 183 Siddiqi, Daniel 17, 32, 62 Siegel, Jeff 381 Silvestri, Giuseppina 16, 201 Simintendi, Arrigo 57 Slavonic 69, 70–71, 279, 281, 288 Slovenian 69 Sluckin, Benjamin 337, 338, 361 Smith, Dafydd Glyn 166 Smith, Jennifer 131 Smith, John 362 Smith, John Charles 6, 35, 83, 85, 87, 89, 93, 101, 103, 119, 131, 211, 354, 382 Solà i Pujols, Jaume 78 So¨ll, Ludwig 132 Sorace, Antonella 153, 165, 233, 335, 340, 354 Sornicola, Rosanna 7, 55, 154, 172, 305, 308, 309, 318, 321, 322, 327, 328 Sotho 354 sound change 97–98 Sound Pattern of English (SPE) 366 spanning 176 Spanish 3, 34, 35, 45–46, 49–50, 53–54, 55, 56–57, 64, 66, 68, 75, 76, 79, 81–82, 83, 85, 86, 87, 88, 95, 101, 102, 103–105, 108, 109, 118, 148, 156, 170, 197, 201, 216, 217, 362, 364, 375, 378, 381, 382, 383, 384, 385 Argentinian 363 central-northern varieties 384 early 148 Latin American 76, 383 morisco 99 northern 249 old 72, 76, 81, 97, 98–99, 103 Spencer, Andrew 1, 14, 21, 26, 28, 29, 31, 36, 37, 38, 40, 43, 63, 65, 72, 79, 80, 81, 83, 84, 85, 88, 169, 201, 213 Squartini, Mario 28, 50, 241, 242, 249, 268, 339 Stan, Camelia 14, 277, 281, 297 standardization 394 Stati, Sorin 262 Steele, Jeffrey 363, 364 Steele, Susan 18

index Stéfanini, Jean 78 Stewart, Thomas 17, 18 Sthioul, Bertrand 249 Štichauer, Pavel 6, 200, 203, 206, 207, 213, 215, 216, 219, 231, 233 Stotz, Peter 329 Stradelli, Ernano 395, 396, 410 Stragapede, Pietro 213 Strasbourg Oaths 71, 96, 108, 127 Strudsholm, Erling 183 stress 107, 121, 185, 291, 377 Stump, Gregory 4, 5, 15, 17, 18, 21, 27, 28, 29, 31, 32, 39, 63, 69, 74, 80, 85, 169, 194, 195, 196, 197, 198, 206, 210, 219, 231, 371 Sturcken, H.T. 81 Sturtevant’s paradox 317 Sua´rez 404 subject (see also verb–subject inversion) 7, 72, 156, 324, 326, 329, 330, 386, 387, 390, 391, 400, 401, 403 position 24–25 of predication 361 subjunctive 79 Catalan 40–41 periphrastic perfective paradigm 40–41, 115–117 future 79 Italian 22–25 Romanian 30, 36, 245, 250, 256, 261, 265 subordination 405–406 substrate 395, 409, 412 superlative 16 superstrate 412 suppletion 2–3, 18, 38–39, 51, 75, 109, 117, 121, 164, 169, 171, 174, 179, 180, 181, 182, 188, 201, 219, 230, 273, 307–308, 313, 314–316 surcomposé forms 32, 52, 78–80, 88–89, 242, 249 Surmiran (Romansh, Switzerland) 5, 62 Sursilvan (Romansh, Switzerland) 101, 108 svarabhakti 363 Svenonius, Peter 176 Sweet, James 385 switch-reference 408–409 syllable 363, 368, 369, 376, 379 Sympson, Pedro 396 syncretism 35, 39, 102, 125, 232–233, 264, 285, 299, 307, 313–316, 321, 336, 341 synchrony and diachrony 59–60 syncope 366 synæresis 382 synonymy 307, 308 synthetic(ity), synthesis 6, 11, 12–14, 71, 169–170, 176, 270

483

vs analytic 61, 77, 126, 241, 242, 302 Szantyr, Anton 306 Tagliamonte, Sali 131 TAM 115, 156, 243, 262, 269, 270 chain 145, 160 Tarantino 120 Tastevin, Constantin 396 Taylor, Catherine 13, 25, 28, 29, 39 Taylor, Gerald 395, 396, 404 Tekavčic´, Pavao 97, 375 Telmon, Tullio 148 tense 22, 31, 62, 63, 75–76, 94, on auxiliaries 264, 265, 266, 269, 270 marking 62–63, 125 nominal 404 sequence of 62, 244 switching 134 Terracini, Benvenuto 306, 308–309 thematic roles 338–340 theme (see also topic) 7 Thibault, Pierrette 211 Thomason, Sarah 70 Thornton, Anna 39, 77, 89, 179, 206, 213, 219, 231, 232, 310, 312, 313, 314 Timberlake, Alan 87 Timotin, Emanuela 243 Todaro, Giuseppina 172, 173 Todi 33 Toledo 382–383, 384 topic 7, 56, 361 situational topic 337, 361 spatio-temporal topic 337, 361 topic-shift 22 topic time 75 topicality 400 Torres-Tamarit, Francesc 376 Tortora, Christina 335, 340, 341 Toso, Fiorenzo 146, 147 Tranel, Bernard 371, 373 transcategorization 146 transderivationality 18 transitional segment 363 transitional vowel 363 transitive (clause/verb)/transitivity 103, 106, 133, 145, 149, 153, 156, 164, 167, 194, 199, 200, 201, 202, 203, 205, 206, 207, 208, 209, 211, 212, 215, 216, 218, 220, 251, 400, 401, 402, 403 Traugott, Elizabeth 34, 60, 123, 145, 146, 153, 382 Traves 203 Trecate 227 Trento 217

484

index

Trevisan 119 Trousdale, Graeme 123 Trubetzkoy, Nikolai 312 Trudgill, Peter 381, 382, 383, 384, 394 Tukano (East Tukanoan) 395 Tupi–Guarani 381, 394–411, 412 old Tupi 395–411 Turinese 163 Turpin, Daniella 89, 131, 132 Tuscan 100, 313 early 157 northern 72 Tuten, Donald 382, 383 Umbrian dialects 55, 363 unaccusative (clause/verb)/unaccusativity 79, 149, 151, 152, 156, 162, 165, 166, 194, 199, 200, 201, 204, 205, 206, 207, 208, 209, 212, 214, 215, 216, 217, 218, 220, 221, 222, 231, 233–234, 235, 236, 246, 251, 329, 335, 336 Unaccusative Hypothesis 216, 217 underexhaustivity 38, 79 unergative (clause/verb)/unergativity 149, 194, 199, 200, 202, 204, 205, 206, 207, 209, 211, 212, 214, 215, 216, 217, 218, 220, 221, 222, 231, 251, 335, 336, 337 uniformitarianism 71 univerbation 35, 51–54, 94, 121 Universal Grammar (UG) 5, 59 universal spine 59 Urițescu, Dorin 259, 260, 261 Ursini, Flavia 152 Uță Bărbulescu, Oana 6, 291 Va¨a¨na¨nen, Veikko 277, 322 Vairel, Hélène 69 Vaissière, Françoise 61 Valencian 106, 107 Vallader (Romansh, Switzerland) 101 Van Gelderen, Elly 123 Van Valin, Robert 151, 161, 335, 338, 339, 340, 361 Vanelli, Laura 371, 373 Vanrell, Maria del Mar 70 Varvaro, Alberto 54, 71, 127 Vasiliu, Emanuel 99 Vecchio, Paola 216 Velo Veronese 203 Vendler, Zeno 161, 338 Venetan, Veneto 154, 202, 306, 368 Venetian, Venice 152, 153 old 79, 157, 159 Venezuela 395 Venneman, Theo 311, 314, 317

verb agreement (see also verb–subject inversion) 7, 176, 319, 335 number 15, 336, 337 functional 175, 181 light 146, 164 of motion/movement (see also: auxiliary: go; come; return) 49, 79, 83, 84, 95, 109, 110, 120, 124, 128, 133, 135, 136, 139, 144, 145, 146, 147, 148, 149, 151, 152, 154, 155–156, 158, 159, 160, 163, 164–166, 167, 171–177, 181, 182, 409 predominance of synthetic forms 241–270 reflexive subclasses 202–203 semi-functional 164, 172 Verb Second 158 verb–subject inversion 335–337, 340 agreement 335, 336, 337 definiteness 337, 341, 342, 343, 344, 347, 348, 349, 350, 351, 352–353, 354, 355, 356–357 free variation 349, 354 generational gradient 354 lexico-aspectual structure 335–337, 355, 357, 358, 359–360, 361[– state] verbs 342–349, 355[+state] verbs 349–355 lexico-semantic structure 337, 360, 361 pronominal subject 337, 341, 342, 343, 344, 345, 347, 348, 350, 351, 355–357, 359–360 Vernet, Florian 132 Verratti, Vittore 31 Vet, Co 77 Viaplana, Joaquim 373 Vigna, Amedeo 162 Vincent, Nigel 6, 12, 13, 14, 17, 18, 21, 25, 28, 29, 30, 31, 32, 34, 35, 37, 38, 39, 41, 42, 43, 51, 52, 54, 56, 59, 61, 78, 84, 108, 123, 169, 179, 243, 251, 315, 319, 382 Viotti, Evani de Carvalho 394 Virdis, Maurizio 56 Vlad, Daciana 267 Vliet, Edward van 86 Vocabulary Insertion 15, 17 vocative 24, 68–71, 74, 90, 225, 229, 316, 393 neo-vocative 70 Vogel, Irene 377 voice (see also: active voice; passive voice) 22, 29–30, 41, 169–170 Vulpe, Magdalena 260, 261 vyanjanabhakti 363

index Wackernagel Law 73 Wagner, Max Leopold 226 Wagner, Michael 365, 366, 372, 375, 376 Wagner, Suzanne Evans 89, 132 Wakker, Gerry 74 Wa¨lchi, Bernard 147, 159 Walkden, George 54 Wallon La Gleize 79 Wanner, Dieter 24 Warekena 409 Waugh, Linda 86 Weigand, Gustav 244 Weiss, Michael 307, 317 Welsh 166 Westerbergh, Ulla 328 Wheeler, Max 6, 40, 61, 78, 84, 106, 110, 113, 125, 151, 373 Willis, David 43, 167 Wilmet, Marc 63, 128, 129, 130 Wolf, George 305 Wolfe, Sam 6, 61, 72, 158 Wolfram, Walt 392

485

Woodrow Wilson, Thomas 1, 213 Woods, Rebecca 158 Word-and-Paradigm (WP) 18, 32, Wurzel, Wolfgang 4, 232, 311, 317 Xavier, Carla 365 Yanomani 395 Yates, Alan 40 Yẽgatú 381, 395, 396, 397, 398–404, 406, 407, 409, 410, 411 Yildiz, Anil 213 Yllera, Alicia 148, 156 Yvon, Henri 86 Zafiu, Rodica 6, 33, 82, 243, 244, 246, 254, 255, 257, 266, 267 Zamboni, Alberto 329, 369, 371, 374, 377 Zamfir, Dana-Mihaela 244, 247, 248, 249, 252, 253 Zilles, Ana Maria 392 Zulu 156 Zwicky, Arnold 1, 69, 213, 224–225, 229

OX F OR D ST U DI E S I N DIAC H RON IC A N D H I STOR IC A L L I NG U I ST IC S

General editors Adam Ledgeway and Ian Roberts, University of Cambridge Advisory editors Cynthia L. Allen, Australian National University; Ricardo Bermúdez-Otero, The University of Manchester; Theresa Biberauer, University of Cambridge; Charlotte Galves, University of Campinas; Geoff Horrocks, University of Cambridge; Paul Kiparsky, Stanford University; David Lightfoot, Georgetown University; Giuseppe Longobardi, University of York; George Walkden, University of Konstanz; David Willis, University of Oxford published 1 From Latin to Romance Morphosyntactic Typology and Change Adam Ledgeway 2 Parameter Theory and Linguistic Change Edited by Charlotte Galves, Sonia Cyrino, Ruth Lopes, Filomena Sandalo, and Juanito Avelar 3 Case in Semitic Roles, Relations, and Reconstruction Rebecca Hasselbach 4 The Boundaries of Pure Morphology Diachronic and Synchronic Perspectives Edited by Silvio Cruschina, Martin Maiden, and John Charles Smith 5 The History of Negation in the Languages of Europe and the Mediterranean Volume I: Case Studies Edited by David Willis, Christopher Lucas, and Anne Breitbarth 6 Constructionalization and Constructional Changes Elizabeth Closs Traugott and Graeme Trousdale 7 Word Order in Old Italian Cecilia Poletto 8 Diachrony and Dialects Grammatical Change in the Dialects of Italy Edited by Paola Benincà, Adam Ledgeway, and Nigel Vincent

9 Discourse and Pragmatic Markers from Latin to the Romance Languages Edited by Chiara Ghezzi and Piera Molinelli 10 Vowel Length from Latin to Romance Michele Loporcaro 11 The Evolution of Functional Left Peripheries in Hungarian Syntax Edited by Katalin É. Kiss 12 Syntactic Reconstruction and Proto-Germanic George Walkden 13 The History of Low German Negation Anne Breitbarth 14 Arabic Indefinites, Interrogatives, and Negators A Linguistic History of Western Dialects David Wilmsen 15 Syntax over Time Lexical, Morphological, and Information-Structural Interactions Edited by Theresa Biberauer and George Walkden 16 Syllable and Segment in Latin Ranjan Sen 17 Participles in Rigvedic Sanskrit The Syntax and Semantics of Adjectival Verb Forms John J. Lowe 18 Verb Movement and Clause Structure in Old Romanian Virginia Hill and Gabriela Alboiu 19 The Syntax of Old Romanian Edited by Gabriela Pană Dindelegan 20 Grammaticalization and the Rise of Configurationality in Indo-Aryan Uta Reino¨hl 21 The Rise and Fall of Ergativity in Aramaic Cycles of Alignment Change Eleanor Coghill

22 Portuguese Relative Clauses in Synchrony and Diachrony Adriana Cardoso 23 Micro-change and Macro-change in Diachronic Syntax Edited by Eric Mathieu and Robert Truswell 24 The Development of Latin Clause Structure A Study of the Extended Verb Phrase Lieven Danckaert 25 Transitive Nouns and Adjectives Evidence from Early Indo-Aryan John. J. Lowe 26 Quantitative Historical Linguistics A Corpus Framework Gard B. Jenset and Barbara McGillivray 27 Gender from Latin to Romance History, Geography, Typology Michele Loporcaro 28 Clause Structure and Word Order in the History of German Edited by Agnes Ja¨ger, Gisella Ferraresi, and Helmut Weiß 29 Word Order Change Edited by Ana Maria Martins and Adriana Cardoso 30 Arabic Historical Dialectology Linguistic and Sociolinguistic Approaches Edited by Clive Holes 31 Grammaticalization from a Typological Perspective Edited by Heiko Narrog and Bernd Heine 32 Negation and Nonveridicality in the History of Greek Katerina Chatzopoulou 33 Indefinites between Latin and Romance Chiara Gianollo

34 Verb Second in Medieval Romance Sam Wolfe 35 Referential Null Subjects in Early English Kristian A. Rusten 36 Word Order and Parameter Change in Romanian A Comparative Romance Perspective Alexandru Nicolae 37 Cycles in Language Change Edited by Miriam Bouzouita, Anne Breitbarth, Lieven Danckaert, and Elisabeth Witzenhausen 38 Palatal Sound Change in the Romance Languages Diachronic and Synchronic Perspectives André Zampaulo 39 Dative External Possessors in Early English Cynthia L. Allen 40 The History of Negation in the Languages of Europe and the Mediterranean Volume II: Patterns and Processes Anne Breitbarth, Christopher Lucas, and David Willis 41 Variation and Change in Gallo-Romance Grammar Edited by Sam Wolfe and Martin Maiden 42 Phonetic Causes of Sound Change The Palatalization and Assibilation of Obstruents Daniel Recasens 43 Syntactic Features and the Limits of Syntactic Change Edited by Jóhannes Gı´sli Jónsson and Thórhallur Eythórsson 44 Romance Object Clitics Microvariation and Linguistic Change Diego Pescarini 45 The Diachrony of Differential Object Marking in Romanian Virginia Hill and Alexandru Mardale

46 Noun-based Constructions in the History of Portuguese and Spanish Patrı´cia Amaral and Manuel Delicado Cantero 47 Syntactic Change in French Sam Wolfe 48 Periphrasis and Inflexion in Diachrony A View from Romance Edited by Adam Ledgeway, John Charles Smith, and Nigel Vincent in preparation Redevelopment of Case Systems in Indo-Aryan Miriam Butt Functional Heads Across Time Edited by Barbara Egedi and Veronika Hegedűs Classical Portuguese Grammar and History Charlotte Galves, Aroldo de Andrade, Christiane Namiuti, and Maria Clara Paixão de Sousa Morphological Borrowing Francesco Gardani Nominal Expressions and Language Change From Early Latin to Modern Romance Giuliana Giusti A Study in Grammatical Change The Modern Greek Weak Subject Pronoun τoς and its Implications for Language Change and Structure Brian D. Joseph Reconstructing Pre-Islamic Arabic Dialects Alexander Magidow Iranian Syntax in Classical Armenian Robin Meyer